From 1ec5083ab4d94c758227c22be052f69e2ae2cf49 Mon Sep 17 00:00:00 2001 From: mateo-berri <277851410+mateo-berri@users.noreply.github.com> Date: Tue, 8 Sep 2026 12:41:07 +0000 Subject: [PATCH] refactor(types): replace Any with real types across 54 more backend files Third batch of the fifth basedpyright Any reduction round. Every change is typing-only and leaves runtime behavior identical. Provider transformation configs, video and rerank base classes, OTel metadata and the guardrail and realtime type modules move their payload, header and optional-parameter annotations from Any to object, Mapping[str, object] or the concrete model the call site already produces. Repositories and endpoints that reached Prisma through an untyped handle now name the actions they call with the repo's own TableActions protocol. The pydantic field retypes were checked against pydantic to confirm object and Any validate, serialize and generate JSON schema identically. --- .../providers/pydantic_ai_agents/config.py | 4 +-- litellm/integrations/argilla.py | 7 +++-- litellm/integrations/dynamodb.py | 19 +++++++++++-- litellm/integrations/focus/database.py | 6 ++-- .../focus/destinations/vantage_destination.py | 5 ++-- .../langfuse/langfuse_otel_attributes.py | 3 +- litellm/integrations/langsmith.py | 8 +++--- litellm/integrations/otel/model/metadata.py | 14 +++++----- litellm/integrations/otel/mount.py | 19 +++++++++++-- .../litellm_core_utils/coroutine_checker.py | 6 ++-- .../exception_mapping_utils.py | 6 ++-- .../huggingface_template_handler.py | 26 +++++++++++++---- .../anthropic/count_tokens/handler.py | 2 +- .../llms/base_llm/videos/transformation.py | 10 +++---- .../count_tokens/bedrock_token_counter.py | 11 ++++---- litellm/llms/bedrock/files/transformation.py | 6 ++-- .../guardrail_translation/handler.py | 5 +++- litellm/llms/chatgpt/common_utils.py | 4 +-- .../responses/transformation.py | 7 +++-- .../llms/hosted_vllm/chat/transformation.py | 21 ++++++-------- litellm/llms/openai/common_utils.py | 6 ++-- litellm/llms/snowflake/chat/transformation.py | 4 +-- .../stability/image_edit/transformations.py | 4 +-- .../llms/triton/completion/transformation.py | 8 +++--- .../llms/vertex_ai/rerank/transformation.py | 4 +-- .../volcengine/embedding/transformation.py | 9 +++--- litellm/llms/watsonx/rerank/transformation.py | 8 +++--- litellm/llms/xai/realtime/transformation.py | 18 ++++++------ .../analytics_endpoints/cache_activity.py | 20 +++++++++---- .../proxy/client/cli/commands/model_groups.py | 18 ++++++++---- litellm/proxy/client/cli/commands/up.py | 3 +- .../common_utils/cache_pydantic_utils.py | 2 +- .../common_utils/proxy_rate_limit_error.py | 4 +-- .../proxy/container_endpoints/endpoints.py | 10 ++++--- litellm/proxy/db/exception_handler.py | 16 +++++++++-- litellm/proxy/db/spend_log_tool_index.py | 2 +- .../crowdstrike_aidr/crowdstrike_aidr.py | 16 +++++------ .../guardrail_hooks/enkryptai/enkryptai.py | 4 +-- .../guardrail_hooks/qualifire/qualifire.py | 2 +- .../semantic_guard/route_loader.py | 3 +- .../proxy/hooks/parallel_request_limiter.py | 6 ++-- .../common_daily_activity.py | 7 +++-- litellm/proxy/ocr_endpoints/endpoints.py | 14 +++++----- litellm/repositories/budget_repository.py | 28 +++++++++++++++---- .../repositories/organization_repository.py | 28 +++++++++++++++---- litellm/repositories/project_repository.py | 11 ++++---- .../adaptive_router/adaptive_router.py | 9 +++--- .../adaptive_router/signals.py | 8 +++--- .../adaptive_router/update_queue.py | 11 ++++---- litellm/types/containers/main.py | 2 +- litellm/types/guardrails.py | 16 +++++------ litellm/types/integrations/prometheus.py | 8 +++--- litellm/types/realtime.py | 16 +++++------ litellm/vector_store_files/utils.py | 11 ++++---- 54 files changed, 323 insertions(+), 202 deletions(-) diff --git a/litellm/a2a_protocol/providers/pydantic_ai_agents/config.py b/litellm/a2a_protocol/providers/pydantic_ai_agents/config.py index b7546e1a2a1..20404e3702b 100644 --- a/litellm/a2a_protocol/providers/pydantic_ai_agents/config.py +++ b/litellm/a2a_protocol/providers/pydantic_ai_agents/config.py @@ -23,7 +23,7 @@ class PydanticAIProviderConfig(BaseA2AProviderConfig): params: dict[str, Any], api_base: str | None = None, **kwargs: Any, - ) -> dict[str, Any]: + ) -> dict[str, object]: """Handle non-streaming request to Pydantic AI agent.""" if api_base is None: raise ValueError("api_base is required for PydanticAIProviderConfig") @@ -41,7 +41,7 @@ class PydanticAIProviderConfig(BaseA2AProviderConfig): params: dict[str, Any], api_base: str | None = None, **kwargs, - ) -> AsyncIterator[dict[str, Any]]: + ) -> AsyncIterator[dict[str, object]]: """Handle streaming request with fake streaming.""" if not api_base: raise ValueError("api_base is required for Pydantic AI agents") diff --git a/litellm/integrations/argilla.py b/litellm/integrations/argilla.py index 9a87a94cf0b..664ef8efda1 100644 --- a/litellm/integrations/argilla.py +++ b/litellm/integrations/argilla.py @@ -7,7 +7,8 @@ import json import os import random import types -from typing import Any, Final +from collections.abc import Mapping +from typing import Final import httpx from pydantic import BaseModel @@ -69,7 +70,7 @@ class ArgillaLogger(CustomBatchLogger): self.flush_lock = asyncio.Lock() super().__init__(**kwargs, flush_lock=self.flush_lock) - def validate_argilla_transformation_object(self, argilla_transformation_object: dict[str, Any]): + def validate_argilla_transformation_object(self, argilla_transformation_object: Mapping[str, object]): if not isinstance(argilla_transformation_object, dict): raise Exception("'argilla_transformation_object' must be a dictionary, to log your payload to Argilla.") @@ -115,7 +116,7 @@ class ArgillaLogger(CustomBatchLogger): ARGILLA_DATASET_NAME=_credentials_dataset_name, ) - def get_chat_messages(self, payload: StandardLoggingPayload) -> list[dict[str, Any]]: + def get_chat_messages(self, payload: StandardLoggingPayload) -> list[dict[str, object]]: payload_messages: Final = payload.get("messages", None) if payload_messages is None: diff --git a/litellm/integrations/dynamodb.py b/litellm/integrations/dynamodb.py index 38f5924a233..3401ced4efb 100644 --- a/litellm/integrations/dynamodb.py +++ b/litellm/integrations/dynamodb.py @@ -3,12 +3,25 @@ import os import traceback -from typing import Any, Final +from collections.abc import Mapping +from typing import Final, Protocol import litellm from litellm._uuid import uuid +class _DynamoTable(Protocol): + """The one boto3 DynamoDB table call this logger makes.""" + + def put_item(self, *, Item: Mapping[str, object]) -> object: ... + + +class _DynamoResource(Protocol): + """The one boto3 DynamoDB resource call this logger makes.""" + + def Table(self, name: str) -> _DynamoTable: ... + + class DyanmoDBLogger: # Class variables or attributes @@ -16,7 +29,7 @@ class DyanmoDBLogger: # Instance variables import boto3 - self.dynamodb: Any = boto3.resource("dynamodb", region_name=os.environ["AWS_REGION_NAME"]) + self.dynamodb: Final[_DynamoResource] = boto3.resource("dynamodb", region_name=os.environ["AWS_REGION_NAME"]) if litellm.dynamodb_table_name is None: raise ValueError( "LiteLLM Error, trying to use DynamoDB but not table name passed. Create a table and set `litellm.dynamodb_table_name=`" @@ -41,7 +54,7 @@ class DyanmoDBLogger: id: Final = response_obj.get("id", str(uuid.uuid4())) # Build the initial payload - payload: Final = { + payload: Final[dict[str, object]] = { "id": id, "call_type": call_type, "startTime": start_time, diff --git a/litellm/integrations/focus/database.py b/litellm/integrations/focus/database.py index 657c7e0d264..891318f1c54 100644 --- a/litellm/integrations/focus/database.py +++ b/litellm/integrations/focus/database.py @@ -3,7 +3,7 @@ from __future__ import annotations from datetime import datetime -from typing import Any, Final +from typing import Final import polars as pl @@ -32,7 +32,7 @@ class FocusLiteLLMDatabase: client: Final = self._ensure_prisma_client() where_clauses: Final[list[str]] = [] - query_params: Final[list[Any]] = [] + query_params: Final[list[datetime | int]] = [] placeholder_index = 1 if start_time_utc: where_clauses.append(f"dus.updated_at >= ${placeholder_index}::timestamptz") @@ -112,7 +112,7 @@ class FocusLiteLLMDatabase: except Exception as exc: raise RuntimeError(f"Error retrieving usage data: {exc}") from exc - async def get_table_info(self) -> dict[str, Any]: + async def get_table_info(self) -> dict[str, object]: """Return metadata about the spend table for diagnostics.""" client: Final = self._ensure_prisma_client() diff --git a/litellm/integrations/focus/destinations/vantage_destination.py b/litellm/integrations/focus/destinations/vantage_destination.py index 132f27779c2..68b0d399975 100644 --- a/litellm/integrations/focus/destinations/vantage_destination.py +++ b/litellm/integrations/focus/destinations/vantage_destination.py @@ -4,7 +4,8 @@ from __future__ import annotations import csv import io -from typing import Any, Final +from collections.abc import Mapping +from typing import Final import httpx # noqa: F401 - used at runtime (AsyncClient, HTTPStatusError) @@ -94,7 +95,7 @@ class FocusVantageDestination(FocusDestination): self, *, prefix: str, - config: dict[str, Any] | None = None, + config: Mapping[str, object] | None = None, ) -> None: config = config or {} api_key: Final = config.get("api_key") diff --git a/litellm/integrations/langfuse/langfuse_otel_attributes.py b/litellm/integrations/langfuse/langfuse_otel_attributes.py index 70fea1abb3b..1eb3ce1a9c2 100644 --- a/litellm/integrations/langfuse/langfuse_otel_attributes.py +++ b/litellm/integrations/langfuse/langfuse_otel_attributes.py @@ -5,6 +5,7 @@ Relevant Issue: https://github.com/BerriAI/litellm/issues/13764 """ import json +from collections.abc import Mapping from typing import TYPE_CHECKING, Any, Final from pydantic import BaseModel @@ -40,7 +41,7 @@ def get_output_content_by_type( | HttpxBinaryResponseContent | ResponsesAPIResponse | list, - kwargs: dict[str, Any] | None = None, + kwargs: Mapping[str, object] | None = None, ) -> str: """ Extract output content from response objects based on their type. diff --git a/litellm/integrations/langsmith.py b/litellm/integrations/langsmith.py index 9607eccef52..2c3837b12ec 100644 --- a/litellm/integrations/langsmith.py +++ b/litellm/integrations/langsmith.py @@ -75,9 +75,9 @@ class LangsmithLogger(CustomBatchLogger): if _batch_size: self.batch_size = int(_batch_size) self.log_queue: list[LangsmithQueueObject] = [] - self._flush_task: asyncio.Task[Any] | None = self._start_periodic_flush_task() + self._flush_task: asyncio.Task[None] | None = self._start_periodic_flush_task() - def _start_periodic_flush_task(self) -> asyncio.Task[Any] | None: + def _start_periodic_flush_task(self) -> asyncio.Task[None] | None: """Start the periodic flush task only when an event loop is already running.""" try: loop: Final = asyncio.get_running_loop() @@ -152,9 +152,9 @@ class LangsmithLogger(CustomBatchLogger): return self._redact_metadata(extra_metadata) - def _build_outputs_with_usage(self, payload: StandardLoggingPayload) -> dict[str, Any]: + def _build_outputs_with_usage(self, payload: StandardLoggingPayload) -> dict[str, object]: response: Final = payload["response"] - outputs: dict[str, Any] + outputs: dict[str, object] if isinstance(response, dict): outputs = {**response} else: diff --git a/litellm/integrations/otel/model/metadata.py b/litellm/integrations/otel/model/metadata.py index ee116aca46b..9eab3021d58 100644 --- a/litellm/integrations/otel/model/metadata.py +++ b/litellm/integrations/otel/model/metadata.py @@ -36,7 +36,7 @@ model. They coincide on the SDK path, which is correct. from __future__ import annotations -from collections.abc import Iterator, Mapping +from collections.abc import Callable, Iterator, Mapping from dataclasses import dataclass, field from types import MappingProxyType from typing import TYPE_CHECKING, Any, Final, cast @@ -57,7 +57,7 @@ class RequestIdentity: # The team's free-form metadata, carried raw (empty/missing -> None) and # filtered to an operator allowlist only at Baggage-promotion time, so an # unconfigured deployment never promotes any of it. - team_metadata: Mapping[str, Any] | None = None + team_metadata: Mapping[str, object] | None = None key_hash: str | None = None end_user: str | None = None # The model litellm dispatched to the provider. Only known once the call @@ -103,7 +103,7 @@ class RequestIdentity: ``user_api_key_*`` names that ``baggage.DEFAULT_BAGGAGE_METADATA_KEYS`` promotes. """ - get: Final = lambda name: getattr(auth, name, None) # noqa: E731 + get: Final[Callable[[str], object]] = lambda name: getattr(auth, name, None) # noqa: E731 metadata: Final = { meta_key: str(value) for meta_key, attr in ( @@ -217,7 +217,7 @@ class LLMCallEvent: time_to_first_chunk_seconds: float | None @classmethod - def from_dict(cls, kwargs: Mapping[str, Any]) -> LLMCallEvent: + def from_dict(cls, kwargs: Mapping[str, object]) -> LLMCallEvent: raw_payload: Final = kwargs.get("standard_logging_object") payload: Final = cast("StandardLoggingPayload", raw_payload) if raw_payload else None operation: Final = resolve_operation(as_str(kwargs.get("call_type"))) @@ -239,7 +239,7 @@ def time_to_first_chunk_seconds(kwargs: Mapping[str, Any]) -> float | None: to the first streamed chunk (``completion_start_time``); ``None`` for non-streaming calls, where ``completion_start_time`` is backfilled with the end time and would not measure first-chunk latency.""" - optional_params: Final = cast(Mapping[str, Any], kwargs.get("optional_params") or {}) + optional_params: Final = cast(Mapping[str, object], kwargs.get("optional_params") or {}) if not optional_params.get("stream"): return None api_call_start: Final = to_seconds(kwargs.get("api_call_start_time")) @@ -307,7 +307,7 @@ def _metadata_dicts( ) -def _call_id(payload: StandardLoggingPayload | None, kwargs: Mapping[str, Any]) -> str | None: +def _call_id(payload: StandardLoggingPayload | None, kwargs: Mapping[str, object]) -> str | None: """The call id from the payload (when closed) or the bare kwargs (at pre_call).""" if payload is not None: call_id: Final = as_str(payload.get("litellm_call_id")) or as_str(payload.get("id")) @@ -351,7 +351,7 @@ def _model_info_id(model_info: object) -> str | None: return None -def _team_metadata_dict(value: object) -> Mapping[str, Any] | None: +def _team_metadata_dict(value: object) -> Mapping[str, object] | None: """The team's free-form metadata as a raw mapping, or ``None`` when missing or empty. diff --git a/litellm/integrations/otel/mount.py b/litellm/integrations/otel/mount.py index ac647c2c4f6..776d8722d14 100644 --- a/litellm/integrations/otel/mount.py +++ b/litellm/integrations/otel/mount.py @@ -12,11 +12,14 @@ when the feature gate is off. """ import os -from typing import Any, Final +from typing import TYPE_CHECKING, Final, Protocol from litellm._logging import verbose_logger from litellm.integrations.otel.model.config import is_otel_v2_enabled +if TYPE_CHECKING: + from fastapi import FastAPI + # Routes excluded from server-span tracing by default: high-frequency pollers and # static UI/docs assets, none of which are LLM traffic. Entries are substring-matched # against the request path (unanchored, so they survive a ``server_root_path`` prefix @@ -65,7 +68,17 @@ PASSTHROUGH_PREFIXES: Final = frozenset( ) -def _passthrough_span_name_hook(span: Any, scope: dict) -> None: +class _RenameableSpan(Protocol): + """The span surface the passthrough naming hook drives.""" + + def is_recording(self) -> bool: ... + + def update_name(self, name: str) -> None: ... + + def set_attribute(self, key: str, value: str) -> None: ... + + +def _passthrough_span_name_hook(span: "_RenameableSpan | None", scope: dict) -> None: """FastAPI ``server_request_hook``: give passthrough server spans a useful name. The instrumentation matches the route at span creation, so both the span name @@ -88,7 +101,7 @@ def _passthrough_span_name_hook(span: Any, scope: dict) -> None: pass -def instrument_fastapi_app(app: Any) -> None: +def instrument_fastapi_app(app: "FastAPI") -> None: """Attach OTel server-span instrumentation to the proxy FastAPI app. Safe no-op when the V2 gate is off or ``opentelemetry-instrumentation-fastapi`` diff --git a/litellm/litellm_core_utils/coroutine_checker.py b/litellm/litellm_core_utils/coroutine_checker.py index 52fc44ba8dc..a5d4f8b968f 100644 --- a/litellm/litellm_core_utils/coroutine_checker.py +++ b/litellm/litellm_core_utils/coroutine_checker.py @@ -16,7 +16,7 @@ class CoroutineChecker: """ def __init__(self): - self._cache = WeakKeyDictionary() + self._cache: WeakKeyDictionary[object, bool] = WeakKeyDictionary() self._max_size = COROUTINE_CHECKER_MAX_SIZE_IN_MEMORY def is_async_callable(self, callback: Any) -> bool: @@ -33,10 +33,10 @@ class CoroutineChecker: pass # Determine target - optimized path for common cases - target = callback + target: object = callback if not inspect.isfunction(target) and not inspect.ismethod(target): try: - call_attr: Final = getattr(target, "__call__", None) + call_attr: Final[object] = getattr(target, "__call__", None) if call_attr is not None: target = call_attr except Exception: diff --git a/litellm/litellm_core_utils/exception_mapping_utils.py b/litellm/litellm_core_utils/exception_mapping_utils.py index 82708d412c9..92f07057da5 100644 --- a/litellm/litellm_core_utils/exception_mapping_utils.py +++ b/litellm/litellm_core_utils/exception_mapping_utils.py @@ -1,7 +1,7 @@ import json import re import traceback -from typing import Any, Final, Protocol, cast +from typing import Final, Protocol, cast import httpx @@ -191,7 +191,7 @@ def _get_response_headers(original_exception: Exception) -> httpx.Headers | None _response_headers: httpx.Headers | None = None try: _response_headers = getattr(original_exception, "headers", None) - error_response: Final = getattr(original_exception, "response", None) + error_response: Final[object] = getattr(original_exception, "response", None) if not _response_headers and error_response: _response_headers = getattr(error_response, "headers", None) if not _response_headers: @@ -203,7 +203,7 @@ def _get_response_headers(original_exception: Exception) -> httpx.Headers | None def extract_and_raise_litellm_exception( - response: Any | None, + response: object | None, error_str: str, model: str, custom_llm_provider: str, diff --git a/litellm/litellm_core_utils/prompt_templates/huggingface_template_handler.py b/litellm/litellm_core_utils/prompt_templates/huggingface_template_handler.py index 3878b36cd91..8f8228d6dfd 100644 --- a/litellm/litellm_core_utils/prompt_templates/huggingface_template_handler.py +++ b/litellm/litellm_core_utils/prompt_templates/huggingface_template_handler.py @@ -1,6 +1,8 @@ import json from datetime import datetime -from typing import Any, Final +from typing import Any, Final, Literal + +from typing_extensions import NotRequired, ReadOnly, TypedDict from litellm.llms.custom_httpx.http_handler import ( _get_httpx_client, @@ -9,6 +11,20 @@ from litellm.llms.custom_httpx.http_handler import ( from litellm.types.llms.custom_http import httpxSpecialProvider +class _TokenizerConfigResult(TypedDict): + """Outcome of a tokenizer_config.json fetch, carrying the parsed document when the fetch succeeded.""" + + status: ReadOnly[Literal["success", "failure"]] + tokenizer: NotRequired[ReadOnly[object]] + + +class _ChatTemplateFileResult(TypedDict): + """Outcome of a chat template file fetch, carrying the template body when the fetch succeeded.""" + + status: ReadOnly[Literal["success", "failure"]] + chat_template: NotRequired[ReadOnly[str]] + + def strftime_now(fmt: str) -> str: """ Custom function for templates that need current date/time formatting (e.g., gpt-oss) @@ -22,7 +38,7 @@ def strftime_now(fmt: str) -> str: return datetime.now().strftime(fmt) -def _get_tokenizer_config(hf_model_name: str) -> dict[str, Any]: +def _get_tokenizer_config(hf_model_name: str) -> _TokenizerConfigResult: """ Fetch tokenizer_config.json from HuggingFace (sync) @@ -45,7 +61,7 @@ def _get_tokenizer_config(hf_model_name: str) -> dict[str, Any]: return {"status": "failure"} -async def _aget_tokenizer_config(hf_model_name: str) -> dict[str, Any]: +async def _aget_tokenizer_config(hf_model_name: str) -> _TokenizerConfigResult: """ Fetch tokenizer_config.json from HuggingFace (async) @@ -70,7 +86,7 @@ async def _aget_tokenizer_config(hf_model_name: str) -> dict[str, Any]: return {"status": "failure"} -def _get_chat_template_file(hf_model_name: str) -> dict[str, Any]: +def _get_chat_template_file(hf_model_name: str) -> _ChatTemplateFileResult: """ Fetch chat template from separate .jinja file (sync) @@ -98,7 +114,7 @@ def _get_chat_template_file(hf_model_name: str) -> dict[str, Any]: return {"status": "failure"} -async def _aget_chat_template_file(hf_model_name: str) -> dict[str, Any]: +async def _aget_chat_template_file(hf_model_name: str) -> _ChatTemplateFileResult: """ Fetch chat template from separate .jinja file (async) diff --git a/litellm/llms/azure_ai/anthropic/count_tokens/handler.py b/litellm/llms/azure_ai/anthropic/count_tokens/handler.py index 3cc90823af9..36d5a56db0d 100644 --- a/litellm/llms/azure_ai/anthropic/count_tokens/handler.py +++ b/litellm/llms/azure_ai/anthropic/count_tokens/handler.py @@ -33,7 +33,7 @@ class AzureAIAnthropicCountTokensHandler(AzureAIAnthropicCountTokensConfig): litellm_params: dict[str, Any] | None = None, timeout: float | httpx.Timeout | None = None, tools: list[dict[str, Any]] | None = None, - system: Any | None = None, + system: object = None, ) -> dict[str, Any]: """ Handle a CountTokens request using httpx with Azure authentication. diff --git a/litellm/llms/base_llm/videos/transformation.py b/litellm/llms/base_llm/videos/transformation.py index f725b295d0f..88cdcd61dd6 100644 --- a/litellm/llms/base_llm/videos/transformation.py +++ b/litellm/llms/base_llm/videos/transformation.py @@ -180,7 +180,7 @@ class BaseVideoConfig(ABC): api_base: str, litellm_params: GenericLiteLLMParams, headers: dict, - extra_body: dict[str, Any] | None = None, + extra_body: dict[str, object] | None = None, ) -> tuple[str, dict]: """ Transform the video remix request into a URL and data @@ -207,7 +207,7 @@ class BaseVideoConfig(ABC): after: str | None = None, limit: int | None = None, order: str | None = None, - extra_query: dict[str, Any] | None = None, + extra_query: dict[str, object] | None = None, ) -> tuple[str, dict]: """ Transform the video list request into a URL and params @@ -342,8 +342,8 @@ class BaseVideoConfig(ABC): litellm_params: GenericLiteLLMParams, headers: dict, video_file: FileContent | None = None, - extra_body: dict[str, Any] | None = None, - prefetched_source_data: dict[str, Any] | None = None, + extra_body: dict[str, object] | None = None, + prefetched_source_data: dict[str, object] | None = None, ) -> tuple[str, Mapping[str, object], RequestFiles | None]: """ Transform the video edit request into a URL plus either JSON data or @@ -373,7 +373,7 @@ class BaseVideoConfig(ABC): api_base: str, litellm_params: GenericLiteLLMParams, headers: dict, - extra_body: dict[str, Any] | None = None, + extra_body: dict[str, object] | None = None, ) -> tuple[str, dict]: """ Transform the video extension request into a URL and JSON data. diff --git a/litellm/llms/bedrock/count_tokens/bedrock_token_counter.py b/litellm/llms/bedrock/count_tokens/bedrock_token_counter.py index 9c5211ed072..2d02b152c61 100644 --- a/litellm/llms/bedrock/count_tokens/bedrock_token_counter.py +++ b/litellm/llms/bedrock/count_tokens/bedrock_token_counter.py @@ -2,6 +2,7 @@ Bedrock Token Counter implementation using the CountTokens API. """ +from collections.abc import Mapping, Sequence from typing import Any, Final from litellm._logging import verbose_logger @@ -26,12 +27,12 @@ class BedrockTokenCounter(BaseTokenCounter): async def count_tokens( self, model_to_use: str, - messages: list[dict[str, Any]] | None, - contents: list[dict[str, Any]] | None, + messages: Sequence[Mapping[str, object]] | None, + contents: Sequence[Mapping[str, object]] | None, deployment: dict[str, Any] | None = None, request_model: str = "", - tools: list[dict[str, Any]] | None = None, - system: Any | None = None, + tools: Sequence[Mapping[str, object]] | None = None, + system: object | None = None, ) -> TokenCountResponse | None: """ Count tokens using AWS Bedrock's CountTokens API. @@ -56,7 +57,7 @@ class BedrockTokenCounter(BaseTokenCounter): litellm_params: Final = deployment.get("litellm_params", {}) # Build request data in the format expected by BedrockCountTokensHandler - request_data: Final[dict[str, Any]] = { + request_data: Final[dict[str, object]] = { "model": model_to_use, "messages": messages, } diff --git a/litellm/llms/bedrock/files/transformation.py b/litellm/llms/bedrock/files/transformation.py index 33b27943ad8..f50d63c9f6f 100644 --- a/litellm/llms/bedrock/files/transformation.py +++ b/litellm/llms/bedrock/files/transformation.py @@ -247,7 +247,7 @@ def _validate_file_id_against_configured_buckets( return validate_against(configured_bucket_names[-1]) -def _uploaded_object_size(litellm_params: Mapping[str, object], raw_response: Response) -> int: +def _uploaded_object_size(litellm_params: Mapping[str, object], response_headers: Mapping[str, str]) -> int: """ S3 answers PutObject with an empty body, so the stored object size comes from the signed request recorded by `transform_create_file_request`, not the response headers. @@ -255,7 +255,7 @@ def _uploaded_object_size(litellm_params: Mapping[str, object], raw_response: Re uploaded_size: Final = litellm_params.get(UPLOAD_CONTENT_LENGTH_PARAM) if isinstance(uploaded_size, int): return uploaded_size - response_content_length: Final = raw_response.headers.get("Content-Length", "0") + response_content_length: Final = response_headers.get("Content-Length", "0") return int(response_content_length) if response_content_length.isdigit() else 0 @@ -1161,7 +1161,7 @@ class BedrockFilesConfig(BaseAWSLLM, BaseFilesConfig): filename=filename, created_at=int(time.time()), # Current timestamp status="uploaded", - bytes=_uploaded_object_size(litellm_params=litellm_params, raw_response=raw_response), + bytes=_uploaded_object_size(litellm_params=litellm_params, response_headers=raw_response.headers), object="file", ) diff --git a/litellm/llms/bedrock/passthrough/guardrail_translation/handler.py b/litellm/llms/bedrock/passthrough/guardrail_translation/handler.py index b87f6196e51..eac8afd767c 100644 --- a/litellm/llms/bedrock/passthrough/guardrail_translation/handler.py +++ b/litellm/llms/bedrock/passthrough/guardrail_translation/handler.py @@ -14,6 +14,9 @@ from litellm.types.utils import GenericGuardrailAPIInputs if TYPE_CHECKING: from litellm.integrations.custom_guardrail import CustomGuardrail from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj + from litellm.llms.pass_through.guardrail_translation.handler import ( + PassThroughEndpointHandler, + ) from litellm.proxy._types import UserAPIKeyAuth from litellm.proxy.utils import ProxyLogging @@ -27,7 +30,7 @@ def _is_converse_endpoint(endpoint: str) -> bool: return bool(parts) and parts[-1] in _CONVERSE_ACTIONS -def _generic_passthrough_handler() -> BaseTranslation: +def _generic_passthrough_handler() -> "PassThroughEndpointHandler": """ Fallback for non-Converse Bedrock routes (e.g. invoke). The generic handler scans the full request/response payload so blocking guardrails diff --git a/litellm/llms/chatgpt/common_utils.py b/litellm/llms/chatgpt/common_utils.py index 35e32e4172f..fe33219f110 100644 --- a/litellm/llms/chatgpt/common_utils.py +++ b/litellm/llms/chatgpt/common_utils.py @@ -268,7 +268,7 @@ def _normalize_litellm_params(litellm_params: Any | None) -> dict: return {} -def get_chatgpt_session_id(litellm_params: Any | None) -> str | None: +def get_chatgpt_session_id(litellm_params: object) -> str | None: params: Final = _normalize_litellm_params(litellm_params) for key in ("litellm_session_id", "session_id"): value = params.get(key) @@ -286,5 +286,5 @@ def get_chatgpt_session_id(litellm_params: Any | None) -> str | None: return None -def ensure_chatgpt_session_id(litellm_params: Any | None) -> str: +def ensure_chatgpt_session_id(litellm_params: object) -> str: return get_chatgpt_session_id(litellm_params) or str(uuid4()) diff --git a/litellm/llms/github_copilot/responses/transformation.py b/litellm/llms/github_copilot/responses/transformation.py index 5a4bb798851..8b85b668cba 100644 --- a/litellm/llms/github_copilot/responses/transformation.py +++ b/litellm/llms/github_copilot/responses/transformation.py @@ -19,6 +19,7 @@ from litellm.llms.openai.responses.transformation import OpenAIResponsesAPIConfi from litellm.types.llms.openai import ( ResponseInputParam, ResponsesAPIOptionalRequestParams, + ResponsesAPIStreamingResponse, ) from litellm.types.router import GenericLiteLLMParams from litellm.types.utils import LlmProviders @@ -129,7 +130,7 @@ class GithubCopilotResponsesAPIConfig(OpenAIResponsesAPIConfig): model: str, parsed_chunk: dict, logging_obj: LiteLLMLoggingObj, - ) -> Any: + ) -> ResponsesAPIStreamingResponse: parsed_chunk = self._normalize_stream_item_id(parsed_chunk) return super().transform_streaming_response( model=model, @@ -262,7 +263,7 @@ class GithubCopilotResponsesAPIConfig(OpenAIResponsesAPIConfig): # Return the responses endpoint return f"{effective_api_base}/responses" - def _handle_reasoning_item(self, item: dict[str, Any]) -> dict[str, Any]: + def _handle_reasoning_item(self, item: dict[str, object]) -> dict[str, object]: """ Handle reasoning items for GitHub Copilot, preserving encrypted_content. @@ -280,7 +281,7 @@ class GithubCopilotResponsesAPIConfig(OpenAIResponsesAPIConfig): # Filter out None values for known problematic fields, # but preserve encrypted_content even if it exists - filtered_item: Final[dict[str, Any]] = {} + filtered_item: Final[dict[str, object]] = {} for k, v in item.items(): # Always include encrypted_content if present (even if None) if k == "encrypted_content": diff --git a/litellm/llms/hosted_vllm/chat/transformation.py b/litellm/llms/hosted_vllm/chat/transformation.py index 29dc485732f..92bc857e385 100644 --- a/litellm/llms/hosted_vllm/chat/transformation.py +++ b/litellm/llms/hosted_vllm/chat/transformation.py @@ -28,12 +28,12 @@ from ...openai.chat.gpt_transformation import OpenAIGPTConfig class HostedVLLMChatConfig(OpenAIGPTConfig): - def _convert_custom_tools_to_function_tools(self, tools: list[dict[str, Any]]) -> list[dict[str, Any]]: + def _convert_custom_tools_to_function_tools(self, tools: list[dict[str, Any]]) -> list[dict[str, object]]: """ vLLM chat completions currently accepts only OpenAI function tools. Convert custom tools into function tools so request validation does not fail. """ - converted_tools: Final[list[dict[str, Any]]] = [] + converted_tools: Final[list[dict[str, object]]] = [] for idx, tool in enumerate(tools): if not isinstance(tool, dict): converted_tools.append(tool) @@ -63,17 +63,14 @@ class HostedVLLMChatConfig(OpenAIGPTConfig): "required": ["input"], } - function_tool: dict[str, Any] = { - "type": "function", - "function": { - "name": str(tool_name), - "parameters": tool_parameters, - }, + function_definition: dict[str, object] = { + "name": str(tool_name), + "parameters": tool_parameters, } if isinstance(tool_description, str): - function_tool["function"]["description"] = tool_description + function_definition["description"] = tool_description - converted_tools.append(function_tool) + converted_tools.append({"type": "function", "function": function_definition}) return converted_tools @@ -148,7 +145,7 @@ class HostedVLLMChatConfig(OpenAIGPTConfig): @overload def _transform_messages( self, messages: list[AllMessageValues], model: str, is_async: Literal[True] - ) -> Coroutine[Any, Any, list[AllMessageValues]]: ... + ) -> Coroutine[object, object, list[AllMessageValues]]: ... @overload def _transform_messages( @@ -160,7 +157,7 @@ class HostedVLLMChatConfig(OpenAIGPTConfig): def _transform_messages( self, messages: list[AllMessageValues], model: str, is_async: bool = False - ) -> list[AllMessageValues] | Coroutine[Any, Any, list[AllMessageValues]]: + ) -> list[AllMessageValues] | Coroutine[object, object, list[AllMessageValues]]: """ Support translating: - video files from file_id or file_data to video_url diff --git a/litellm/llms/openai/common_utils.py b/litellm/llms/openai/common_utils.py index 2db6d78a218..e60b6145b44 100644 --- a/litellm/llms/openai/common_utils.py +++ b/litellm/llms/openai/common_utils.py @@ -10,7 +10,7 @@ import ssl import time import uuid from collections.abc import AsyncIterator, Iterator, Mapping -from typing import TYPE_CHECKING, Any, Final, Literal, NamedTuple, Optional +from typing import TYPE_CHECKING, Final, Literal, NamedTuple, Optional from urllib.parse import urlsplit import httpx @@ -87,8 +87,8 @@ class OpenAIError(BaseLLMException): ################################################################### def drop_params_from_unprocessable_entity_error( e: openai.UnprocessableEntityError | httpx.HTTPStatusError, - data: dict[str, Any], -) -> dict[str, Any]: + data: Mapping[str, object], +) -> dict[str, object]: """ Helper function to read OpenAI UnprocessableEntityError and drop the params that raised an error from the error message. diff --git a/litellm/llms/snowflake/chat/transformation.py b/litellm/llms/snowflake/chat/transformation.py index f65b0876202..aeff902f655 100644 --- a/litellm/llms/snowflake/chat/transformation.py +++ b/litellm/llms/snowflake/chat/transformation.py @@ -315,7 +315,7 @@ class SnowflakeConfig(SnowflakeBaseConfig, OpenAIGPTConfig): for msg in messages: if isinstance(msg, dict): role = msg.get("role", "") - content: Any = msg.get("content", "") + content: object = msg.get("content", "") msg_cache_control: object = msg.get("cache_control") else: role = getattr(msg, "role", "") @@ -463,7 +463,7 @@ class SnowflakeConfig(SnowflakeBaseConfig, OpenAIGPTConfig): return body - def _transform_tool_choice_to_anthropic(self, tool_choice: Any) -> dict[str, Any]: + def _transform_tool_choice_to_anthropic(self, tool_choice: object) -> Mapping[str, object]: """ Convert tool_choice from OpenAI format to Anthropic format. diff --git a/litellm/llms/stability/image_edit/transformations.py b/litellm/llms/stability/image_edit/transformations.py index 0b6052ad593..94711d21b50 100644 --- a/litellm/llms/stability/image_edit/transformations.py +++ b/litellm/llms/stability/image_edit/transformations.py @@ -74,7 +74,7 @@ class StabilityImageEditConfig(BaseImageEditConfig): } # Create a copy to not mutate original - convert TypedDict to regular dict - mapped_params: Final[dict[str, Any]] = dict(image_edit_optional_params) + mapped_params: Final[dict[str, object]] = dict(image_edit_optional_params) for k, v in image_edit_optional_params.items(): if k in param_mapping: @@ -182,7 +182,7 @@ class StabilityImageEditConfig(BaseImageEditConfig): # Build Stability request # Populate multipart form-data as separate text fields (data) and files. # Stability expects prompt/output_format/etc. as normal form fields, not file parts. - data: Final[dict[str, Any]] = { + data: Final[dict[str, object]] = { "output_format": "png", # Default to PNG } diff --git a/litellm/llms/triton/completion/transformation.py b/litellm/llms/triton/completion/transformation.py index 98a68ba2c36..3c868b3a96f 100644 --- a/litellm/llms/triton/completion/transformation.py +++ b/litellm/llms/triton/completion/transformation.py @@ -4,7 +4,7 @@ Translates from OpenAI's `/v1/chat/completions` endpoint to Triton's `/generate` import json from collections.abc import AsyncIterator, Iterator -from typing import TYPE_CHECKING, Any, Final, Literal +from typing import TYPE_CHECKING, Final, Literal from httpx import Headers, Response @@ -172,7 +172,7 @@ class TritonConfig(BaseConfig): streaming_response: Iterator[str] | AsyncIterator[str] | ModelResponse, sync_stream: bool, json_mode: bool | None = False, - ) -> Any: + ) -> "TritonResponseIterator": return TritonResponseIterator( streaming_response=streaming_response, sync_stream=sync_stream, @@ -195,14 +195,14 @@ class TritonGenerateConfig(TritonConfig): ) -> dict: inference_params: Final = optional_params.copy() stream: Final = inference_params.pop("stream", False) - data_for_triton: Final[dict[str, Any]] = { + data_for_triton: Final[dict[str, object]] = { "text_input": prompt_factory(model=model, messages=messages), "parameters": { "max_tokens": int(optional_params.get("max_tokens", DEFAULT_MAX_TOKENS_FOR_TRITON)), + **inference_params, }, "stream": bool(stream), } - data_for_triton["parameters"].update(inference_params) return data_for_triton def transform_response( diff --git a/litellm/llms/vertex_ai/rerank/transformation.py b/litellm/llms/vertex_ai/rerank/transformation.py index 2ec4f2da79b..2b1ed3a29a8 100644 --- a/litellm/llms/vertex_ai/rerank/transformation.py +++ b/litellm/llms/vertex_ai/rerank/transformation.py @@ -5,7 +5,7 @@ Why separate file? Make it easy to see how transformation works """ from collections.abc import Mapping -from typing import Any, Final +from typing import Final import httpx @@ -227,7 +227,7 @@ class VertexAIRerankConfig(BaseRerankConfig, VertexBase): model: str, drop_params: bool, query: str, - documents: list[str | dict[str, Any]], + documents: list[str | dict[str, object]], custom_llm_provider: str | None = None, top_n: int | None = None, rank_fields: list[str] | None = None, diff --git a/litellm/llms/volcengine/embedding/transformation.py b/litellm/llms/volcengine/embedding/transformation.py index 091b9dfd334..7c626c66e9d 100644 --- a/litellm/llms/volcengine/embedding/transformation.py +++ b/litellm/llms/volcengine/embedding/transformation.py @@ -3,7 +3,8 @@ Volcengine Embedding Transformation Transforms OpenAI embedding requests to Volcengine format """ -from typing import Any, Final +from collections.abc import Mapping +from typing import Final import httpx @@ -83,11 +84,11 @@ class VolcEngineEmbeddingConfig(BaseEmbeddingConfig): def map_openai_params( self, - non_default_params: dict[str, Any], - optional_params: dict[str, Any], + non_default_params: Mapping[str, object], + optional_params: dict[str, object], model: str, drop_params: bool, - ) -> dict[str, Any]: + ) -> dict[str, object]: """ Map OpenAI embedding parameters to Volcengine format. diff --git a/litellm/llms/watsonx/rerank/transformation.py b/litellm/llms/watsonx/rerank/transformation.py index 293880b188d..0c81d50a5fe 100644 --- a/litellm/llms/watsonx/rerank/transformation.py +++ b/litellm/llms/watsonx/rerank/transformation.py @@ -5,8 +5,8 @@ Docs - https://cloud.ibm.com/apidocs/watsonx-ai#text-rerank """ import uuid -from collections.abc import Mapping -from typing import Any, Final, cast +from collections.abc import Mapping, Sequence +from typing import Final, cast import httpx @@ -96,7 +96,7 @@ class IBMWatsonXRerankConfig(IBMWatsonXMixin, BaseRerankConfig): model: str, drop_params: bool, query: str, - documents: list[str | dict[str, Any]], + documents: Sequence[str | Mapping[str, object]], custom_llm_provider: str | None = None, top_n: int | None = None, rank_fields: list[str] | None = None, @@ -178,7 +178,7 @@ class IBMWatsonXRerankConfig(IBMWatsonXMixin, BaseRerankConfig): transformed_results: Final = [] for result in _results: - transformed_result: dict[str, Any] = { + transformed_result: dict[str, object] = { "index": result["index"], "relevance_score": result["score"], } diff --git a/litellm/llms/xai/realtime/transformation.py b/litellm/llms/xai/realtime/transformation.py index e9d16daad7c..5efe125ee60 100644 --- a/litellm/llms/xai/realtime/transformation.py +++ b/litellm/llms/xai/realtime/transformation.py @@ -16,7 +16,7 @@ construction time (see ``handler.py``) so all normalization is isolated here and ``RealTimeStreaming`` stays provider-agnostic. """ -from typing import Any, Final +from typing import Final class XAIRealtimeNormalizer: @@ -58,7 +58,7 @@ class XAIRealtimeNormalizer: # Cache content-part objects keyed by (response_id, item_id, content_index) # so that ``response.content_part.done`` events missing ``part`` can be # back-filled from earlier ``content_part.added`` / delta-done events. - self._content_part_by_key: dict[tuple, dict[str, Any]] = {} + self._content_part_by_key: dict[tuple, dict[str, object]] = {} # --------------------------------------------------------------------------- # Public interface consumed by RealTimeStreaming @@ -140,7 +140,7 @@ class XAIRealtimeNormalizer: } self._content_part_by_key[key] = updated - def _resolve_content_part(self, event: dict) -> dict[str, Any]: + def _resolve_content_part(self, event: dict) -> dict[str, object]: part: Final = event.get("part") if isinstance(part, dict): return part @@ -214,7 +214,7 @@ class XAIRealtimeNormalizer: needs_content: Final = event_type in self._EVENTS_NEEDING_CONTENT_INDEX if not needs_output and not needs_content: return event - patch: Final[dict[str, Any]] = {} + patch: Final[dict[str, object]] = {} if needs_output and "output_index" not in event: patch["output_index"] = 0 if needs_content and "content_index" not in event: @@ -228,8 +228,8 @@ class XAIRealtimeNormalizer: # --------------------------------------------------------------------------- @staticmethod - def _default_ga_usage() -> dict[str, Any]: - default_details: Final[dict[str, Any]] = { + def _default_ga_usage() -> dict[str, object]: + default_details: Final[dict[str, int]] = { "cached_tokens": 0, "text_tokens": 0, "audio_tokens": 0, @@ -243,7 +243,7 @@ class XAIRealtimeNormalizer: } @staticmethod - def _normalize_usage(usage: object, *, empty_as_null: bool) -> dict[str, Any] | None: + def _normalize_usage(usage: object, *, empty_as_null: bool) -> dict[str, object] | None: """Coerce a usage object into the full OpenAI GA shape. ``empty_as_null=True`` for ``response.created`` (usage optional). @@ -253,12 +253,12 @@ class XAIRealtimeNormalizer: return None if not usage: return None if empty_as_null else XAIRealtimeNormalizer._default_ga_usage() - default_details: Final[dict[str, Any]] = { + default_details: Final[dict[str, int]] = { "cached_tokens": 0, "text_tokens": 0, "audio_tokens": 0, } - normalized: Final[dict[str, Any]] = { + normalized: Final[dict[str, object]] = { "total_tokens": usage.get("total_tokens", 0), "input_tokens": usage.get("input_tokens", 0), "output_tokens": usage.get("output_tokens", 0), diff --git a/litellm/proxy/analytics_endpoints/cache_activity.py b/litellm/proxy/analytics_endpoints/cache_activity.py index b87b8eac3ef..952ef5bdcfc 100644 --- a/litellm/proxy/analytics_endpoints/cache_activity.py +++ b/litellm/proxy/analytics_endpoints/cache_activity.py @@ -2,16 +2,26 @@ import asyncio import json from collections.abc import Sequence from datetime import datetime -from typing import TYPE_CHECKING, Final +from typing import Final, Protocol from pydantic import BaseModel, TypeAdapter -if TYPE_CHECKING: - from litellm.proxy.utils import PrismaClient - UNKNOWN_CALL_TYPE: Final = "Unknown" +class _SupportsQueryRaw(Protocol): + """The single database operation the cache-activity queries issue.""" + + async def query_raw(self, query: str, *args: object) -> Sequence[object]: ... + + +class _SupportsRawQueryDb(Protocol): + """A prisma client handle, narrowed to the raw-query surface used here.""" + + @property + def db(self) -> _SupportsQueryRaw: ... + + class CacheActivityGroup(BaseModel): call_type: str api_requests: int @@ -143,7 +153,7 @@ def compute_totals(groups: Sequence[CacheActivityGroup]) -> CacheActivityTotals: async def get_cache_activity( - prisma_client: "PrismaClient", + prisma_client: _SupportsRawQueryDb, start_date: datetime, end_date: datetime, key_aliases: Sequence[str], diff --git a/litellm/proxy/client/cli/commands/model_groups.py b/litellm/proxy/client/cli/commands/model_groups.py index c904e5bed49..367c2063b6b 100644 --- a/litellm/proxy/client/cli/commands/model_groups.py +++ b/litellm/proxy/client/cli/commands/model_groups.py @@ -1,3 +1,4 @@ +from collections.abc import Mapping from typing import Final, Literal import click @@ -5,10 +6,17 @@ import rich import rich.table from ... import Client +from ._cli_context import cli_context_values def create_client(ctx: click.Context) -> Client: - return Client(base_url=ctx.obj["base_url"], api_key=ctx.obj["api_key"]) + context: Final = cli_context_values(ctx) + return Client(base_url=context["base_url"], api_key=context["api_key"]) + + +def _rendered_field(group: Mapping[str, object], key: str, default: str) -> str: + """The rendered value of one model group field, or ``default`` when the group omits it.""" + return str(group.get(key, default)) @click.group(name="model-groups") @@ -46,10 +54,10 @@ def list_model_groups(ctx: click.Context, output_format: Literal["table", "json" for group in groups: table.add_row( - str(group.get("model_group", "")), - str(group.get("mode", "chat")), - str(group.get("input_cost_per_token", "")), - str(group.get("output_cost_per_token", "")), + _rendered_field(group, "model_group", ""), + _rendered_field(group, "mode", "chat"), + _rendered_field(group, "input_cost_per_token", ""), + _rendered_field(group, "output_cost_per_token", ""), ) rich.print(table) diff --git a/litellm/proxy/client/cli/commands/up.py b/litellm/proxy/client/cli/commands/up.py index b7c02866d6f..45d21b0c0b5 100644 --- a/litellm/proxy/client/cli/commands/up.py +++ b/litellm/proxy/client/cli/commands/up.py @@ -166,7 +166,8 @@ def up(ctx: click.Context) -> None: is already running (this does not start one for you). Cursor is not supported: it has no equivalent file-based config to patch. """ - base_url: Final = ctx.obj["base_url"] + ctx_obj: Final[CliContextObj] = ctx.obj + base_url: Final = ctx_obj["base_url"] try: _ensure_fresh_login(ctx) diff --git a/litellm/proxy/common_utils/cache_pydantic_utils.py b/litellm/proxy/common_utils/cache_pydantic_utils.py index 725c2b61145..3703cf7c916 100644 --- a/litellm/proxy/common_utils/cache_pydantic_utils.py +++ b/litellm/proxy/common_utils/cache_pydantic_utils.py @@ -37,7 +37,7 @@ class CacheCodec: """ @staticmethod - def serialize(value: Any, model_type: type[T] | None = None) -> Any: + def serialize(value: object, model_type: type[T] | None = None) -> object: """ Encode a value for DualCache / Redis (``json.dumps``-safe). diff --git a/litellm/proxy/common_utils/proxy_rate_limit_error.py b/litellm/proxy/common_utils/proxy_rate_limit_error.py index c109da6f571..8d3a587a4cd 100644 --- a/litellm/proxy/common_utils/proxy_rate_limit_error.py +++ b/litellm/proxy/common_utils/proxy_rate_limit_error.py @@ -66,7 +66,7 @@ def map_v3_rate_limit_type( return None -def _coerce_message(detail: Any) -> str: +def _coerce_message(detail: object) -> str: """Best-effort, JSON-friendly stringification of an HTTPException-style detail.""" if detail is None: return "" @@ -144,7 +144,7 @@ class ProxyRateLimitError(HTTPException, RateLimitError): def __init__( self, detail: Any, - headers: Mapping[str, Any] | None = None, + headers: Mapping[str, object] | None = None, category: str | RateLimitErrorCategory = RateLimitErrorCategory.LITELLM_RATE_LIMIT, rate_limit_type: str | RateLimitType | None = None, model: str | None = None, diff --git a/litellm/proxy/container_endpoints/endpoints.py b/litellm/proxy/container_endpoints/endpoints.py index e852eb5d6f9..c1407979f29 100644 --- a/litellm/proxy/container_endpoints/endpoints.py +++ b/litellm/proxy/container_endpoints/endpoints.py @@ -106,7 +106,7 @@ async def create_container( # Process request using ProxyBaseLLMRequestProcessing processor: Final = ProxyBaseLLMRequestProcessing(data=data) try: - response: Final = await processor.base_process_llm_request( + response: Final[object] = await processor.base_process_llm_request( request=request, fastapi_response=fastapi_response, user_api_key_dict=user_api_key_dict, @@ -216,7 +216,7 @@ async def list_containers( or get_custom_llm_provider_from_request_query(request=request) or "openai" ) - data: Final[dict[str, Any]] = { + data: Final[dict[str, object]] = { "query_params": query_params, "model": query_params.get("model"), "order": order, @@ -341,7 +341,7 @@ async def retrieve_container( # Process request using ProxyBaseLLMRequestProcessing processor: Final = ProxyBaseLLMRequestProcessing(data=data) try: - return await processor.base_process_llm_request( + container: Final[object] = await processor.base_process_llm_request( request=request, fastapi_response=fastapi_response, user_api_key_dict=user_api_key_dict, @@ -366,6 +366,7 @@ async def retrieve_container( proxy_logging_obj=proxy_logging_obj, version=version, ) + return container @router.delete( @@ -446,7 +447,7 @@ async def delete_container( # Process request using ProxyBaseLLMRequestProcessing processor: Final = ProxyBaseLLMRequestProcessing(data=data) try: - return await processor.base_process_llm_request( + deleted_container: Final[object] = await processor.base_process_llm_request( request=request, fastapi_response=fastapi_response, user_api_key_dict=user_api_key_dict, @@ -471,6 +472,7 @@ async def delete_container( proxy_logging_obj=proxy_logging_obj, version=version, ) + return deleted_container # Register JSON-configured container file endpoints diff --git a/litellm/proxy/db/exception_handler.py b/litellm/proxy/db/exception_handler.py index f469587ab8e..de20416ad9c 100644 --- a/litellm/proxy/db/exception_handler.py +++ b/litellm/proxy/db/exception_handler.py @@ -1,5 +1,5 @@ from collections.abc import Awaitable, Callable, Iterator -from typing import Any, Final, TypeVar +from typing import Final, Protocol, TypeVar from litellm._logging import verbose_proxy_logger from litellm.proxy._types import ( @@ -407,8 +407,20 @@ def _coerce_timeout(value: object, fallback: float) -> float: _ReadResultT: Final = TypeVar("_ReadResultT") +class _DBReconnectClient(Protocol): + """The one method `call_with_db_reconnect_retry` needs from a Prisma client.""" + + async def attempt_db_reconnect( + self, + *, + reason: str, + timeout_seconds: float | None = None, + lock_timeout_seconds: float | None = None, + ) -> bool: ... + + async def call_with_db_reconnect_retry( - prisma_client: Any, + prisma_client: _DBReconnectClient, coro_factory: Callable[[], Awaitable[_ReadResultT]], *, reason: str, diff --git a/litellm/proxy/db/spend_log_tool_index.py b/litellm/proxy/db/spend_log_tool_index.py index 93bbc567430..7a7982481fb 100644 --- a/litellm/proxy/db/spend_log_tool_index.py +++ b/litellm/proxy/db/spend_log_tool_index.py @@ -148,4 +148,4 @@ async def flush_tool_usage_transactions( except DB_RETRY_SAFE_ERROR_TYPES: if attempt >= n_retry_times: raise - await asyncio.sleep(2**attempt + random.uniform(0, 1)) + await asyncio.sleep(2.0**attempt + random.uniform(0, 1)) diff --git a/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py b/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py index f7f500b1adc..db6fc238b9b 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py +++ b/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py @@ -6,7 +6,7 @@ from typing import TYPE_CHECKING, Annotated, Final, Literal, NamedTuple, Optiona from fastapi import HTTPException from pydantic import BaseModel, ConfigDict, Field, ValidationError -from typing_extensions import Any, override +from typing_extensions import override from litellm._logging import verbose_proxy_logger from litellm.integrations.custom_guardrail import ( @@ -78,7 +78,7 @@ class _GuardChatCompletionsResult(BaseModel): """Whether or not the prompt triggered a block detection.""" transformed: bool | None = None """Whether or not the original input was transformed.""" - detectors: dict[str, Any] | None = None + detectors: dict[str, object] | None = None """Result of the policy analyzing and input prompt.""" @@ -146,8 +146,8 @@ def _extract_text_from_message(message: _Message) -> str: return "\n".join(part.text for part in content if isinstance(part, _TextContentPart)) -def _merge_metadata_bags(request_data: Mapping[str, Any]) -> Mapping[str, Any] | None: - merged: Final[dict[str, Any]] = {} +def _merge_metadata_bags(request_data: Mapping[str, object]) -> Mapping[str, object] | None: + merged: Final[dict[str, object]] = {} present = False for bag in (request_data.get("metadata"), request_data.get("litellm_metadata")): if isinstance(bag, Mapping): @@ -313,7 +313,7 @@ class CrowdStrikeAIDRHandler(CustomGuardrail): self._set_streaming_params(streaming_params_from_litellm_params(litellm_params)) async def _call_crowdstrike_aidr_guard( - self, payload: dict[str, Any], hook_name: str + self, payload: dict[str, object], hook_name: str ) -> _GuardChatCompletionsResult: """ Makes the API call to the CrowdStrike AIDR AI Guard endpoint. @@ -423,7 +423,7 @@ class CrowdStrikeAIDRHandler(CustomGuardrail): return [_extract_text_from_message(msg) for msg in tail] async def _call_or_fail_open( - self, payload: dict[str, Any], hook_name: str, request_data: dict[str, object] + self, payload: dict[str, object], hook_name: str, request_data: dict[str, object] ) -> _GuardChatCompletionsResult: start_time: Final = time.time() try: @@ -506,7 +506,7 @@ class CrowdStrikeAIDRHandler(CustomGuardrail): event_type = "output" hook_name = "apply_guardrail (response)" - ai_guard_payload: Final[dict[str, Any]] = { + ai_guard_payload: Final[dict[str, object]] = { "guard_input": guard_input.model_dump(mode="json"), "event_type": event_type, } @@ -521,7 +521,7 @@ class CrowdStrikeAIDRHandler(CustomGuardrail): if user_id: ai_guard_payload["user_id"] = user_id - extra_info: Final[dict[str, str]] = {} + extra_info: Final[dict[str, object]] = {} user_email: Final = metadata.get("user_api_key_user_email") if user_email: extra_info["user_name"] = user_email diff --git a/litellm/proxy/guardrails/guardrail_hooks/enkryptai/enkryptai.py b/litellm/proxy/guardrails/guardrail_hooks/enkryptai/enkryptai.py index 89afecafb0f..efe959bd186 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/enkryptai/enkryptai.py +++ b/litellm/proxy/guardrails/guardrail_hooks/enkryptai/enkryptai.py @@ -6,7 +6,7 @@ # +-------------------------------------------------------------+ import os -from collections.abc import AsyncGenerator +from collections.abc import AsyncGenerator, AsyncIterable from datetime import datetime from typing import TYPE_CHECKING, Any, Final, Literal, Optional @@ -465,7 +465,7 @@ class EnkryptAIGuardrails(CustomGuardrail): async def async_post_call_streaming_iterator_hook( self, user_api_key_dict: UserAPIKeyAuth, - response: Any, + response: AsyncIterable[ModelResponseStream], request_data: dict, ) -> AsyncGenerator[ModelResponseStream, None]: """ diff --git a/litellm/proxy/guardrails/guardrail_hooks/qualifire/qualifire.py b/litellm/proxy/guardrails/guardrail_hooks/qualifire/qualifire.py index d82944c44ed..eceb54681f6 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/qualifire/qualifire.py +++ b/litellm/proxy/guardrails/guardrail_hooks/qualifire/qualifire.py @@ -383,7 +383,7 @@ class QualifireGuardrail(CustomGuardrail): result: Final = response.json() # Extract response info for logging - qualifire_response: Final = { + qualifire_response: Final[dict[str, object]] = { "score": result.get("score"), "status": result.get("status"), } diff --git a/litellm/proxy/guardrails/guardrail_hooks/semantic_guard/route_loader.py b/litellm/proxy/guardrails/guardrail_hooks/semantic_guard/route_loader.py index 8d5923d1302..b2139779925 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/semantic_guard/route_loader.py +++ b/litellm/proxy/guardrails/guardrail_hooks/semantic_guard/route_loader.py @@ -6,6 +6,7 @@ then builds a SemanticRouter for prompt matching. """ import os +from collections.abc import Mapping, Sequence from typing import TYPE_CHECKING, Any, Final import yaml @@ -66,7 +67,7 @@ class SemanticGuardRouteLoader: cls, route_templates: list[str] | None, custom_routes_file: str | None, - custom_routes: list[dict[str, Any]] | None, + custom_routes: Sequence[Mapping[str, object]] | None, global_threshold: float = DEFAULT_SEMANTIC_GUARD_SIMILARITY_THRESHOLD, ) -> list["Route"]: """Build semantic-router Route objects from templates + custom config.""" diff --git a/litellm/proxy/hooks/parallel_request_limiter.py b/litellm/proxy/hooks/parallel_request_limiter.py index b313cb64c3f..d41acadc4dd 100644 --- a/litellm/proxy/hooks/parallel_request_limiter.py +++ b/litellm/proxy/hooks/parallel_request_limiter.py @@ -27,7 +27,7 @@ if TYPE_CHECKING: from litellm.proxy.utils import InternalUsageCache as _InternalUsageCache - Span = _Span | Any + Span = _Span InternalUsageCache = _InternalUsageCache else: Span = Any @@ -75,7 +75,7 @@ class _PROXY_MaxParallelRequestsHandler(CustomLogger): current: dict | None, request_count_api_key: str, rate_limit_type: Literal["key", "model_per_key", "user", "customer", "team"], - values_to_update_in_cache: list[tuple[Any, Any]], + values_to_update_in_cache: list[tuple[str, object]], ) -> dict: verbose_proxy_logger.info("Current Usage of %s in this minute: %s", rate_limit_type, current) if current is None: @@ -266,7 +266,7 @@ class _PROXY_MaxParallelRequestsHandler(CustomLogger): rpm_limit = sys.maxsize values_to_update_in_cache: list[ - tuple[Any, Any] + tuple[str, object] ] = [] # values that need to get updated in cache, will run a batch_set_cache after this function # ------------ diff --git a/litellm/proxy/management_endpoints/common_daily_activity.py b/litellm/proxy/management_endpoints/common_daily_activity.py index ce6a97708ab..6841cadf972 100644 --- a/litellm/proxy/management_endpoints/common_daily_activity.py +++ b/litellm/proxy/management_endpoints/common_daily_activity.py @@ -17,6 +17,7 @@ from litellm.proxy.spend_tracking.key_metadata_recovery import ( ) from litellm.proxy.spend_tracking.ptu_feature_flag import is_ptu_cost_attribution_enabled from litellm.proxy.utils import PrismaClient +from litellm.repositories.prisma_protocols import TableActions from litellm.repositories.table_repositories import DeletedVerificationTokenRepository from litellm.repositories.verification_token_repository import ( VerificationTokenRepository, @@ -1155,8 +1156,10 @@ async def get_daily_activity( include_current_utc_day=include_current_utc_day, ) + spend_table: Final[TableActions[DailySpendRecord]] = getattr(prisma_client.db, table_name) + # Get total count for pagination - total_count: Final[int] = await getattr(prisma_client.db, table_name).count(where=where_conditions) + total_count: Final[int] = await spend_table.count(where=where_conditions) # Fetch paginated results. # ``date`` alone is not a unique sort key -- a busy tenant has many @@ -1168,7 +1171,7 @@ async def get_daily_activity( # total. Adding ``id`` (the row's UUID primary key, present on both # LiteLLM_DailyUserSpend and LiteLLM_DailyTeamSpend) as a tiebreaker # gives every page a stable cursor (#30164). - daily_spend_data: Final[Sequence[DailySpendRecord]] = await getattr(prisma_client.db, table_name).find_many( + daily_spend_data: Final[Sequence[DailySpendRecord]] = await spend_table.find_many( where=where_conditions, order=[ {"date": "desc"}, diff --git a/litellm/proxy/ocr_endpoints/endpoints.py b/litellm/proxy/ocr_endpoints/endpoints.py index ebf4d988fdd..1d0108cc2ba 100644 --- a/litellm/proxy/ocr_endpoints/endpoints.py +++ b/litellm/proxy/ocr_endpoints/endpoints.py @@ -2,7 +2,7 @@ import json from collections.abc import Mapping -from typing import Any, Final, cast +from typing import Final, cast import orjson from fastapi import APIRouter, Depends, HTTPException, Request, Response, UploadFile @@ -48,7 +48,7 @@ def _build_document_from_upload( ) -def _with_request_format(data: Mapping[str, Any], request: Request) -> Mapping[str, Any]: +def _with_request_format(data: Mapping[str, object], request: Request) -> Mapping[str, object]: """ Resolve the requested response format from the body or the `x-req-format` header. @@ -90,7 +90,7 @@ def _native_response(response: object, fastapi_response: Response) -> Response | ) -async def _parse_multipart_form(request: Request) -> dict[str, Any]: +async def _parse_multipart_form(request: Request) -> dict[str, object]: """ Extract OCR data from a multipart form request. @@ -130,7 +130,7 @@ async def _parse_multipart_form(request: Request) -> dict[str, Any]: content_type=uploaded_file.content_type, ) - data: Final[dict[str, Any]] = {"document": document} + data: Final[dict[str, object]] = {"document": document} for field_name, field_value in form.items(): if field_name in ("file", "document"): @@ -154,12 +154,12 @@ async def _parse_multipart_form(request: Request) -> dict[str, Any]: return data -async def _parse_ocr_request(request: Request) -> Mapping[str, Any]: +async def _parse_ocr_request(request: Request) -> Mapping[str, object]: """Parse an OCR request and apply the `x-req-format` header, if any.""" return _with_request_format(await _parse_ocr_request_body(request), request) -async def _parse_ocr_request_body(request: Request) -> dict[str, Any]: +async def _parse_ocr_request_body(request: Request) -> dict[str, object]: """ Parse an OCR request, supporting both JSON and multipart form data. @@ -320,7 +320,7 @@ async def ocr( # Process request using ProxyBaseLLMRequestProcessing processor = ProxyBaseLLMRequestProcessing(data=data) - response: Final = await processor.base_process_llm_request( + response: Final[object] = await processor.base_process_llm_request( request=request, fastapi_response=fastapi_response, user_api_key_dict=user_api_key_dict, diff --git a/litellm/repositories/budget_repository.py b/litellm/repositories/budget_repository.py index 62632ffb5f6..205646c8393 100644 --- a/litellm/repositories/budget_repository.py +++ b/litellm/repositories/budget_repository.py @@ -2,7 +2,8 @@ Budget repository for database operations on LiteLLM_BudgetTable. """ -from typing import TYPE_CHECKING, Any, Final +from collections.abc import Mapping +from typing import TYPE_CHECKING, Final, Protocol from litellm.models.budget import LiteLLM_BudgetTable from litellm.repositories.base_repository import BaseRepository @@ -12,12 +13,27 @@ if TYPE_CHECKING: from prisma import models as prisma_models +class _BudgetDb(Protocol): + """The single Prisma table this repository reaches for on ``prisma_client.db``.""" + + @property + def litellm_budgettable(self) -> TableActions["prisma_models.LiteLLM_BudgetTable"]: ... + + +class _PrismaClientView(Protocol): + """The one attribute this repository reads off the untyped Prisma client wrapper.""" + + @property + def db(self) -> _BudgetDb: ... + + class BudgetRepository(BaseRepository[LiteLLM_BudgetTable]): """Repository for budget database operations.""" @property def table(self) -> TableActions["prisma_models.LiteLLM_BudgetTable"]: - return self.prisma_client.db.litellm_budgettable + client: Final[_PrismaClientView] = self.prisma_client + return client.db.litellm_budgettable @property def model_class(self) -> type[LiteLLM_BudgetTable]: @@ -34,12 +50,12 @@ class BudgetRepository(BaseRepository[LiteLLM_BudgetTable]): max_parallel_requests: int | None = None, tpm_limit: int | None = None, rpm_limit: int | None = None, - model_max_budget: dict[str, Any] | None = None, + model_max_budget: Mapping[str, object] | None = None, budget_duration: str | None = None, allowed_models: list[str] | None = None, ) -> LiteLLM_BudgetTable: """Create a new budget record.""" - data: Final[dict[str, Any]] = { + data: Final[dict[str, object]] = { "created_by": created_by, "updated_by": created_by, } @@ -71,12 +87,12 @@ class BudgetRepository(BaseRepository[LiteLLM_BudgetTable]): max_parallel_requests: int | None = None, tpm_limit: int | None = None, rpm_limit: int | None = None, - model_max_budget: dict[str, Any] | None = None, + model_max_budget: Mapping[str, object] | None = None, budget_duration: str | None = None, allowed_models: list[str] | None = None, ) -> LiteLLM_BudgetTable | None: """Update an existing budget record.""" - data: Final[dict[str, Any]] = {"updated_by": updated_by} + data: Final[dict[str, object]] = {"updated_by": updated_by} if max_budget is not None: data["max_budget"] = max_budget if soft_budget is not None: diff --git a/litellm/repositories/organization_repository.py b/litellm/repositories/organization_repository.py index 5a9bd3724e0..47eb8f4a609 100644 --- a/litellm/repositories/organization_repository.py +++ b/litellm/repositories/organization_repository.py @@ -2,7 +2,8 @@ Organization repository for database operations on LiteLLM_OrganizationTable. """ -from typing import TYPE_CHECKING, Any, Final +from collections.abc import Mapping +from typing import TYPE_CHECKING, Final, Protocol from litellm.models.organization import LiteLLM_OrganizationTable from litellm.repositories.base_repository import BaseRepository @@ -12,12 +13,27 @@ if TYPE_CHECKING: from prisma import models as prisma_models +class _OrganizationDb(Protocol): + """The single Prisma table this repository reaches for on ``prisma_client.db``.""" + + @property + def litellm_organizationtable(self) -> TableActions["prisma_models.LiteLLM_OrganizationTable"]: ... + + +class _PrismaClientView(Protocol): + """The one attribute this repository reads off the untyped Prisma client wrapper.""" + + @property + def db(self) -> _OrganizationDb: ... + + class OrganizationRepository(BaseRepository[LiteLLM_OrganizationTable]): """Repository for organization database operations.""" @property def table(self) -> TableActions["prisma_models.LiteLLM_OrganizationTable"]: - return self.prisma_client.db.litellm_organizationtable + client: Final[_PrismaClientView] = self.prisma_client + return client.db.litellm_organizationtable @property def model_class(self) -> type[LiteLLM_OrganizationTable]: @@ -39,12 +55,12 @@ class OrganizationRepository(BaseRepository[LiteLLM_OrganizationTable]): budget_id: str, created_by: str, organization_id: str | None = None, - metadata: dict[str, Any] | None = None, + metadata: Mapping[str, object] | None = None, models: list[str] | None = None, object_permission_id: str | None = None, ) -> LiteLLM_OrganizationTable: """Create a new organization.""" - data: Final[dict[str, Any]] = { + data: Final[dict[str, object]] = { "organization_alias": organization_alias, "budget_id": budget_id, "created_by": created_by, @@ -67,12 +83,12 @@ class OrganizationRepository(BaseRepository[LiteLLM_OrganizationTable]): updated_by: str, organization_alias: str | None = None, budget_id: str | None = None, - metadata: dict[str, Any] | None = None, + metadata: Mapping[str, object] | None = None, models: list[str] | None = None, object_permission_id: str | None = None, ) -> LiteLLM_OrganizationTable | None: """Update an organization.""" - data: Final[dict[str, Any]] = {"updated_by": updated_by} + data: Final[dict[str, object]] = {"updated_by": updated_by} if organization_alias is not None: data["organization_alias"] = organization_alias if budget_id is not None: diff --git a/litellm/repositories/project_repository.py b/litellm/repositories/project_repository.py index 48e55efd258..905e813f35e 100644 --- a/litellm/repositories/project_repository.py +++ b/litellm/repositories/project_repository.py @@ -2,7 +2,8 @@ Project repository for database operations on LiteLLM_ProjectTable. """ -from typing import TYPE_CHECKING, Any, Final +from collections.abc import Mapping +from typing import TYPE_CHECKING, Final from litellm.models.project import LiteLLM_ProjectTable from litellm.repositories.base_repository import BaseRepository @@ -43,14 +44,14 @@ class ProjectRepository(BaseRepository[LiteLLM_ProjectTable]): description: str | None = None, team_id: str | None = None, budget_id: str | None = None, - metadata: dict[str, Any] | None = None, + metadata: Mapping[str, object] | None = None, models: list[str] | None = None, model_rpm_limit: dict[str, int] | None = None, model_tpm_limit: dict[str, int] | None = None, object_permission_id: str | None = None, ) -> LiteLLM_ProjectTable: """Create a new project.""" - data: Final[dict[str, Any]] = { + data: Final[dict[str, object]] = { "created_by": created_by, "updated_by": created_by, } @@ -85,7 +86,7 @@ class ProjectRepository(BaseRepository[LiteLLM_ProjectTable]): description: str | None = None, team_id: str | None = None, budget_id: str | None = None, - metadata: dict[str, Any] | None = None, + metadata: Mapping[str, object] | None = None, models: list[str] | None = None, model_rpm_limit: dict[str, int] | None = None, model_tpm_limit: dict[str, int] | None = None, @@ -93,7 +94,7 @@ class ProjectRepository(BaseRepository[LiteLLM_ProjectTable]): object_permission_id: str | None = None, ) -> LiteLLM_ProjectTable | None: """Update a project.""" - data: Final[dict[str, Any]] = {"updated_by": updated_by} + data: Final[dict[str, object]] = {"updated_by": updated_by} if project_alias is not None: data["project_alias"] = project_alias if description is not None: diff --git a/litellm/router_strategy/adaptive_router/adaptive_router.py b/litellm/router_strategy/adaptive_router/adaptive_router.py index 12ccacbbc1d..8727c3a69f0 100644 --- a/litellm/router_strategy/adaptive_router/adaptive_router.py +++ b/litellm/router_strategy/adaptive_router/adaptive_router.py @@ -16,6 +16,7 @@ from __future__ import annotations import asyncio import time from collections import OrderedDict +from collections.abc import Mapping from dataclasses import asdict, dataclass from typing import Any, Final, cast @@ -122,7 +123,7 @@ class AdaptiveRouter: prefs = self.model_to_prefs.get(model) or _default_prefs() self._cells[(rt, model)] = initial_cell(prefs, rt) - async def load_state_from_db(self, prisma_client: Any) -> None: + async def load_state_from_db(self, prisma_client: object) -> None: """Add each row's persisted delta to a freshly computed cold-start prior. A row holds an accumulated delta, not a full posterior, and can be one-sided @@ -237,7 +238,7 @@ class AdaptiveRouter: cost_weight=self.config.weights.cost, ) - async def get_state_snapshot(self) -> dict[str, Any]: + async def get_state_snapshot(self) -> dict[str, object]: """In-memory snapshot for the introspection endpoint. Cheap; no DB hit.""" cells: Final = [] for (rt, model), cell in sorted(self._cells.items(), key=lambda kv: (kv[0][0].value, kv[0][1])): @@ -278,7 +279,7 @@ class AdaptiveRouter: @staticmethod def _extract_min_quality_tier( - request_kwargs: dict[str, Any], + request_kwargs: Mapping[str, object], ) -> int | None: """Pull `min_quality_tier` from request headers or metadata. @@ -484,7 +485,7 @@ class AdaptiveRouter: return combined_delta @staticmethod - def _persistable_session_snapshot(state: SessionState) -> dict[str, Any]: + def _persistable_session_snapshot(state: SessionState) -> dict[str, object]: snapshot: Final = asdict(state) for sensitive in ( "last_user_content", diff --git a/litellm/router_strategy/adaptive_router/signals.py b/litellm/router_strategy/adaptive_router/signals.py index c28613b54eb..72e8d27d2bf 100644 --- a/litellm/router_strategy/adaptive_router/signals.py +++ b/litellm/router_strategy/adaptive_router/signals.py @@ -92,7 +92,7 @@ class Turn: user_content: str | None = None assistant_content: str | None = None - tool_calls: list[dict[str, Any]] = field(default_factory=list) + tool_calls: Sequence[Mapping[str, object]] = field(default_factory=list[Mapping[str, object]]) tool_results: Sequence[Mapping[str, object]] = field(default_factory=list) response_status: int | None = None @@ -174,7 +174,7 @@ def _detect_failure(tool_results: Sequence[Mapping[str, object]]) -> bool: return False -def _signature(call: dict[str, Any]) -> str: +def _signature(call: Mapping[str, Any]) -> str: """Stable signature for loop detection: name + sorted JSON-ish args.""" name: Final = call.get("name") or call.get("function", {}).get("name", "") call_args = call.get("arguments") @@ -185,7 +185,7 @@ def _signature(call: dict[str, Any]) -> str: return f"{name}({call_args})" -def _detect_loop(history: list[str], new_calls: list[dict[str, Any]]) -> bool: +def _detect_loop(history: list[str], new_calls: Sequence[Mapping[str, object]]) -> bool: """Fires if any new call's signature appears >= LOOP_REPEAT_THRESHOLD-1 times in recent history (so this call would be the Nth).""" if not new_calls: @@ -238,7 +238,7 @@ def detect_response_signals( previous_assistant_content: str | None, current_assistant_content: str | None, tool_call_history: list[str], - tool_calls: list[dict[str, Any]], + tool_calls: Sequence[Mapping[str, object]], tool_results: Sequence[Mapping[str, object]], response_status: int | None, ) -> SignalDelta: diff --git a/litellm/router_strategy/adaptive_router/update_queue.py b/litellm/router_strategy/adaptive_router/update_queue.py index 1b9fce284ac..e28f2379f9c 100644 --- a/litellm/router_strategy/adaptive_router/update_queue.py +++ b/litellm/router_strategy/adaptive_router/update_queue.py @@ -19,7 +19,8 @@ to the in-memory aggregator). Flush is async and batched. from __future__ import annotations import asyncio -from typing import Any, Final +from collections.abc import Mapping +from typing import Final from litellm._logging import verbose_router_logger from litellm.repositories.table_repositories import ( @@ -39,7 +40,7 @@ class AdaptiveRouterUpdateQueue: def __init__(self) -> None: self._state_agg: dict[StateKey, dict[str, float]] = {} - self._session_agg: dict[SessionKey, dict[str, Any]] = {} + self._session_agg: dict[SessionKey, Mapping[str, object]] = {} self._lock = asyncio.Lock() self._max_state_size_seen = 0 self._max_session_size_seen = 0 @@ -77,7 +78,7 @@ class AdaptiveRouterUpdateQueue: session_id: str, router_name: str, model_name: str, - state_dict: dict[str, Any], + state_dict: Mapping[str, object], ) -> None: """ Last-write-wins per session row. The state_dict is a snapshot of the @@ -91,7 +92,7 @@ class AdaptiveRouterUpdateQueue: # ---- Flushers (called by background task) ---------------------------- - async def flush_state_to_db(self, prisma_client: Any) -> int: + async def flush_state_to_db(self, prisma_client: object) -> int: """ Drain state aggregator and apply to LiteLLM_AdaptiveRouterState. Returns number of cells flushed. @@ -147,7 +148,7 @@ class AdaptiveRouterUpdateQueue: return len(batch) - async def flush_session_to_db(self, prisma_client: Any) -> int: + async def flush_session_to_db(self, prisma_client: object) -> int: """ Drain session aggregator and upsert into LiteLLM_AdaptiveRouterSession. Returns number of session rows flushed. diff --git a/litellm/types/containers/main.py b/litellm/types/containers/main.py index 6a339fd2eac..62ef524a435 100644 --- a/litellm/types/containers/main.py +++ b/litellm/types/containers/main.py @@ -140,7 +140,7 @@ class ContainerFileObject(BaseModel): created_at: int path: str source: str - _hidden_params: dict[str, Any] = {} + _hidden_params: dict[str, builtins.object] = {} def __contains__(self, key: str) -> bool: return hasattr(self, key) diff --git a/litellm/types/guardrails.py b/litellm/types/guardrails.py index 02dee40f2a3..175226f8d4b 100644 --- a/litellm/types/guardrails.py +++ b/litellm/types/guardrails.py @@ -2,10 +2,10 @@ from collections.abc import Mapping from datetime import datetime from enum import Enum from types import MappingProxyType -from typing import Any, Final, Literal +from typing import Final, Literal from pydantic import BaseModel, ConfigDict, Field, field_validator, model_validator -from typing_extensions import Required, TypedDict +from typing_extensions import ReadOnly, Required, TypedDict from litellm.constants import BEDROCK_APPLY_GUARDRAIL_CHUNK_BUDGET_CHARS from litellm.types.proxy.guardrails.guardrail_hooks.akto import ( @@ -935,7 +935,7 @@ class BaseLitellmParams(ContentFilterConfigModel): # works for new and patch up ), ) - additional_provider_specific_params: dict[str, Any] | None = Field( + additional_provider_specific_params: dict[str, object] | None = Field( default=None, description="Additional provider-specific parameters for generic guardrail APIs", ) @@ -1157,7 +1157,7 @@ class GuardrailEventHooks(str, Enum): class DynamicGuardrailParams(TypedDict): - extra_body: dict[str, Any] + extra_body: ReadOnly[dict[str, object]] class GUARDRAIL_DEFINITION_LOCATION(str, Enum): @@ -1188,7 +1188,7 @@ class GuardrailUIAddGuardrailSettings(BaseModel): supported_modes: list[str] supported_modes_by_provider: dict[str, list[str]] pii_entity_categories: list[PiiEntityCategoryMap] - content_filter_settings: dict[str, Any] | None = None + content_filter_settings: dict[str, object] | None = None class PresidioPerRequestConfig(BaseModel): @@ -1206,8 +1206,8 @@ class ApplyGuardrailRequest(BaseModel): language: str | None = None entities: list[PiiEntityType] | None = None input_type: str = "request" - messages: list[dict[str, Any]] | None = None - metadata: dict[str, Any] | None = None + messages: list[dict[str, object]] | None = None + metadata: dict[str, object] | None = None class ApplyGuardrailResponse(BaseModel): @@ -1217,4 +1217,4 @@ class ApplyGuardrailResponse(BaseModel): class PatchGuardrailRequest(BaseModel): guardrail_name: str | None = None litellm_params: BaseLitellmParams | None = None - guardrail_info: dict[str, Any] | None = None + guardrail_info: dict[str, object] | None = None diff --git a/litellm/types/integrations/prometheus.py b/litellm/types/integrations/prometheus.py index 8498b6f6d00..323f43f4b01 100644 --- a/litellm/types/integrations/prometheus.py +++ b/litellm/types/integrations/prometheus.py @@ -44,7 +44,7 @@ def _sanitize_prometheus_label_name(label: str) -> str: _PROMETHEUS_LABEL_VALUE_TRANSLATE_V1: Final = str.maketrans("\n", " ", "\r\u2028\u2029") -def _sanitize_prometheus_label_value(value: Any | None) -> str | None: +def _sanitize_prometheus_label_value(value: object | None) -> str | None: """ Same semantics as :func:`_sanitize_prometheus_label_value`, implemented with ``str.translate`` plus a single escape pass instead of chained ``replace``. @@ -1023,7 +1023,7 @@ class UserAPIKeyLabelValues: ``hashed_api_key``. This supports ``**standard_logging_payload`` in tests. """ field_names: Final = {f.name for f in fields(self)} - merged: Final[dict[str, Any]] = {} + merged: Final[dict[str, object]] = {} for f in fields(self): if f.default_factory is not MISSING: merged[f.name] = f.default_factory() @@ -1060,9 +1060,9 @@ class UserAPIKeyLabelValues: # stays cheap. (Dataclass default `str()` delegates to `__repr__`.) return "" - def model_dump(self) -> dict[str, Any]: + def model_dump(self) -> dict[str, object]: """Same shape as the former Pydantic ``model_dump()`` (plain dict, list tags).""" - d: Final[dict[str, Any]] = {f.name: getattr(self, f.name) for f in fields(self)} + d: Final[dict[str, object]] = {f.name: getattr(self, f.name) for f in fields(self)} d["tags"] = list(self.tags) d["custom_metadata_labels"] = dict(self.custom_metadata_labels) return d diff --git a/litellm/types/realtime.py b/litellm/types/realtime.py index 17dc70126f3..bda3865c46a 100644 --- a/litellm/types/realtime.py +++ b/litellm/types/realtime.py @@ -78,15 +78,15 @@ class RealtimeSessionConfig(BaseModel): type: str | None = None model: str | None = None instructions: str | None = None - audio: dict[str, Any] | None = None + audio: dict[str, object] | None = None include: list[str] | None = None max_output_tokens: int | str | None = None output_modalities: list[str] | None = None - tool_choice: Any | None = None - tools: list[dict[str, Any]] | None = None - tracing: Any | None = None - truncation: Any | None = None - prompt: dict[str, Any] | None = None + tool_choice: object | None = None + tools: list[dict[str, object]] | None = None + tracing: object | None = None + truncation: object | None = None + prompt: dict[str, object] | None = None class RealtimeClientSecretRequest(BaseModel): @@ -114,7 +114,7 @@ class RealtimeClientSecretResponse(BaseModel): expires_at: int | None = None value: str - session: dict[str, Any] | None = None + session: dict[str, object] | None = None class RealtimeTranscriptionSessionRequest(BaseModel): @@ -151,7 +151,7 @@ class RealtimeTranscriptionSessionResponse(BaseModel): model_config = {"extra": "allow"} - client_secret: dict[str, Any] | None = None + client_secret: dict[str, object] | None = None class RealtimeErrorDetail(TypedDict): diff --git a/litellm/vector_store_files/utils.py b/litellm/vector_store_files/utils.py index 94ad5c0ecdf..8b4bff921f8 100644 --- a/litellm/vector_store_files/utils.py +++ b/litellm/vector_store_files/utils.py @@ -1,4 +1,5 @@ -from typing import Any, Final, cast, get_type_hints +from collections.abc import Mapping +from typing import Final, cast, get_type_hints from litellm.types.vector_store_files import ( VectorStoreFileCreateRequest, @@ -11,25 +12,25 @@ class VectorStoreFileRequestUtils: """Helper utilities for constructing vector store file requests.""" @staticmethod - def _filter_params(params: dict[str, Any], model: Any) -> dict[str, Any]: + def _filter_params(params: Mapping[str, object], model: type[object]) -> dict[str, object]: valid_keys: Final = get_type_hints(model).keys() return {key: value for key, value in params.items() if key in valid_keys and value is not None} @staticmethod def get_create_request_params( - params: dict[str, Any], + params: Mapping[str, object], ) -> VectorStoreFileCreateRequest: filtered: Final = VectorStoreFileRequestUtils._filter_params(params=params, model=VectorStoreFileCreateRequest) return cast(VectorStoreFileCreateRequest, filtered) @staticmethod - def get_list_query_params(params: dict[str, Any]) -> VectorStoreFileListQueryParams: + def get_list_query_params(params: Mapping[str, object]) -> VectorStoreFileListQueryParams: filtered = VectorStoreFileRequestUtils._filter_params(params=params, model=VectorStoreFileListQueryParams) return cast(VectorStoreFileListQueryParams, filtered) @staticmethod def get_update_request_params( - params: dict[str, Any], + params: Mapping[str, object], ) -> VectorStoreFileUpdateRequest: filtered: Final = VectorStoreFileRequestUtils._filter_params(params=params, model=VectorStoreFileUpdateRequest) return cast(VectorStoreFileUpdateRequest, filtered)