refactor(types): leave provider reply parsers for a follow-up with live provider calls

Restores the Bedrock Stability, Bedrock KB, AgentCore, Replicate, Recraft, Aleph Alpha and Guardrails AI parsers to main, since no live provider call could prove the stricter schemas. Also types the cache delete route tests.

litellm/a2a_protocol/providers/bedrock_agentcore/handler.py

@@ -9,8 +9,6 @@ import json
 from collections.abc import AsyncIterator, Mapping
 from typing import Final
 
-from pydantic import ConfigDict, TypeAdapter
-
 from litellm._logging import verbose_logger
 from litellm.a2a_protocol.providers.bedrock_agentcore.transformation import (
     BedrockAgentCoreA2ATransformation,
@@ -19,8 +17,6 @@ from litellm.llms.bedrock.base_aws_llm import run_aws_signing
 from litellm.llms.custom_httpx.http_handler import get_async_httpx_client
 from litellm.types.llms.custom_http import httpxSpecialProvider
 
-_JSON_OBJECT: Final = TypeAdapter(dict[str, object], config=ConfigDict(hide_input_in_errors=True))
-
 
 class BedrockAgentCoreA2AHandler:
     """
@@ -70,7 +66,7 @@ class BedrockAgentCoreA2AHandler:
             data=body,
         )
         response.raise_for_status()
-        response_data: Final = _JSON_OBJECT.validate_python(response.json(), strict=True)
+        response_data: Final = response.json()
 
         if "error" in response_data:
             verbose_logger.warning("BedrockAgentCore A2A: Agent returned error: %s", response_data["error"])
@@ -129,7 +125,7 @@ class BedrockAgentCoreA2AHandler:
                 "BedrockAgentCore A2A streaming: received JSON instead of SSE, yielding as single event"
             )
             response_body: Final = await response.aread()
-            response_data: Final = _JSON_OBJECT.validate_python(json.loads(response_body), strict=True)
+            response_data: Final = json.loads(response_body)
             yield response_data
         else:
             # SSE stream — parse data: lines

litellm/llms/bedrock/image_edit/stability_transformation.py

@@ -22,13 +22,10 @@ API Reference: https://docs.aws.amazon.com/bedrock/latest/userguide/model-parame
 """
 
 import base64
-from collections.abc import Sequence
 from typing import TYPE_CHECKING, Any, Final, cast
 
 import httpx
 from httpx._types import RequestFiles
-from pydantic import ConfigDict, TypeAdapter, with_config
-from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 from litellm.llms.base_llm.image_edit.transformation import BaseImageEditConfig
 from litellm.llms.bedrock.common_utils import BedrockError
@@ -48,15 +45,6 @@ else:
     LiteLLMLoggingObj = Any
 
 
-@with_config(ConfigDict(extra="allow", strict=True))
-class _StabilityResponse(TypedDict):
-    finish_reasons: ReadOnly[NotRequired[Sequence[object] | None]]
-    images: ReadOnly[NotRequired[Sequence[str | None] | None]]
-
-
-_STABILITY_RESPONSE: Final = TypeAdapter(_StabilityResponse)
-
-
 class BedrockStabilityImageEditConfig(BaseImageEditConfig):
     """
     Configuration for Bedrock Stability AI image edit.
@@ -327,10 +315,8 @@ class BedrockStabilityImageEditConfig(BaseImageEditConfig):
                 headers=raw_response.headers,
             )
 
-        payload: Final = _STABILITY_RESPONSE.validate_python(response_data)
-
         # Check finish_reasons
-        finish_reasons: Final = payload.get("finish_reasons", [])
+        finish_reasons: Final = response_data.get("finish_reasons", [])
         if finish_reasons and finish_reasons[0]:
             raise self.get_error_class(
                 error_message=f"Bedrock Stability error: {finish_reasons[0]}",
@@ -343,7 +329,7 @@ class BedrockStabilityImageEditConfig(BaseImageEditConfig):
             model_response.data = []
 
         # Extract images from response
-        images: Final = payload.get("images", [])
+        images: Final = response_data.get("images", [])
         if images:
             for image_b64 in images:
                 if image_b64:

litellm/llms/bedrock/vector_stores/transformation.py

@@ -1,11 +1,9 @@
-from collections.abc import Mapping, Sequence
+from collections.abc import Mapping
 from copy import deepcopy
 from typing import TYPE_CHECKING, Any, Final, cast
 from urllib.parse import urlparse
 
 import httpx
-from pydantic import ConfigDict, TypeAdapter, with_config
-from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 from litellm._logging import verbose_logger
 from litellm.litellm_core_utils.url_utils import encode_url_path_segment, join_api_base
@@ -13,6 +11,8 @@ from litellm.llms.base_llm.vector_store.transformation import BaseVectorStoreCon
 from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM
 from litellm.llms.bedrock.common_utils import BedrockError
 from litellm.types.integrations.rag.bedrock_knowledgebase import (
+    BedrockKBContent,
+    BedrockKBResponse,
     BedrockKBRetrievalConfiguration,
     BedrockKBRetrievalQuery,
     BedrockKBUserContext,
@@ -34,26 +34,6 @@ else:
     LiteLLMLoggingObj = Any
 
 
-@with_config(ConfigDict(extra="allow", strict=True))
-class _KBContent(TypedDict):
-    text: ReadOnly[NotRequired[str | None]]
-
-
-@with_config(ConfigDict(extra="allow", strict=True))
-class _KBRetrievalResult(TypedDict):
-    content: ReadOnly[NotRequired[_KBContent | None]]
-    score: ReadOnly[NotRequired[int | float | None]]
-    metadata: ReadOnly[NotRequired[Mapping[str, object] | None]]
-
-
-@with_config(ConfigDict(extra="allow", strict=True))
-class _KBResponse(TypedDict):
-    retrievalResults: ReadOnly[NotRequired[Sequence[_KBRetrievalResult] | None]]
-
-
-_KB_RESPONSE: Final = TypeAdapter(_KBResponse)
-
-
 class BedrockVectorStoreConfig(BaseVectorStoreConfig, BaseAWSLLM):
     """Vector store configuration for AWS Bedrock Knowledge Bases."""
 
@@ -304,7 +284,7 @@ class BedrockVectorStoreConfig(BaseVectorStoreConfig, BaseAWSLLM):
             api_key=api_key,
         )
 
-    def _get_file_id_from_metadata(self, metadata: Mapping[str, Any]) -> str:
+    def _get_file_id_from_metadata(self, metadata: dict[str, Any]) -> str:
         """
         Extract file_id from Bedrock KB metadata.
         Uses source URI if available, otherwise generates a fallback ID.
@@ -316,7 +296,7 @@ class BedrockVectorStoreConfig(BaseVectorStoreConfig, BaseAWSLLM):
         chunk_id: Final = metadata.get("x-amz-bedrock-kb-chunk-id", "unknown") if metadata else "unknown"
         return f"bedrock-kb-{chunk_id}"
 
-    def _get_filename_from_metadata(self, metadata: Mapping[str, Any]) -> str:
+    def _get_filename_from_metadata(self, metadata: dict[str, Any]) -> str:
         """
         Extract filename from Bedrock KB metadata.
         Tries to extract filename from source URI, falls back to domain name or data source ID.
@@ -338,7 +318,7 @@ class BedrockVectorStoreConfig(BaseVectorStoreConfig, BaseAWSLLM):
         data_source_id: Final = metadata.get("x-amz-bedrock-kb-data-source-id", "unknown") if metadata else "unknown"
         return f"bedrock-kb-document-{data_source_id}"
 
-    def _get_attributes_from_metadata(self, metadata: Mapping[str, object]) -> dict[str, object]:
+    def _get_attributes_from_metadata(self, metadata: dict[str, object]) -> dict[str, object]:
         """
         Extract all attributes from Bedrock KB metadata.
         Returns a copy of the metadata dictionary.
@@ -351,10 +331,10 @@ class BedrockVectorStoreConfig(BaseVectorStoreConfig, BaseAWSLLM):
         self, response: httpx.Response, litellm_logging_obj: LiteLLMLoggingObj
     ) -> VectorStoreSearchResponse:
         try:
-            response_data: Final = _KB_RESPONSE.validate_python(response.json())
+            response_data: Final = BedrockKBResponse(**response.json())
             results: Final[list[VectorStoreSearchResult]] = []
             for item in response_data.get("retrievalResults", []) or []:
-                content: _KBContent | None = item.get("content")
+                content: BedrockKBContent | None = item.get("content")
                 text = content.get("text") if content else None
                 if text is None:
                     continue

litellm/llms/deprecated_providers/aleph_alpha.py

@@ -5,27 +5,11 @@ from collections.abc import Callable, Mapping
 from typing import Final
 
 import httpx
-from pydantic import ConfigDict, TypeAdapter, with_config
-from typing_extensions import ReadOnly, TypedDict
 
 import litellm
 from litellm.utils import Choices, Message, ModelResponse, Usage
 
 
-@with_config(ConfigDict(extra="allow", strict=True))
-class _AlephAlphaCompletion(TypedDict):
-    completion: ReadOnly[str]
-    finish_reason: ReadOnly[str | None]
-
-
-@with_config(ConfigDict(extra="allow", strict=True))
-class _AlephAlphaResponse(TypedDict):
-    completions: ReadOnly[list[_AlephAlphaCompletion]]
-
-
-_ALEPH_ALPHA_RESPONSE: Final = TypeAdapter(_AlephAlphaResponse)
-
-
 class AlephAlphaError(Exception):
     def __init__(self, status_code, message):
         self.status_code = status_code
@@ -273,9 +257,8 @@ def completion(
             )
         else:
             try:
-                validated_response: Final = _ALEPH_ALPHA_RESPONSE.validate_python(completion_response)
                 choices_list: Final = []
-                for idx, item in enumerate(validated_response["completions"]):
+                for idx, item in enumerate(completion_response["completions"]):
                     if len(item["completion"]) > 0:
                         message_obj = Message(content=item["completion"])
                     else:

litellm/llms/recraft/image_edit/transformation.py

@@ -1,11 +1,8 @@
-from collections.abc import Sequence
 from io import BufferedReader
 from typing import TYPE_CHECKING, Any, Final, cast
 
 import httpx
 from httpx._types import RequestFiles
-from pydantic import ConfigDict, TypeAdapter, with_config
-from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 from litellm.images.utils import ImageEditRequestUtils
 from litellm.litellm_core_utils.url_utils import join_api_base
@@ -25,21 +22,6 @@ else:
     LiteLLMLoggingObj = Any
 
 
-@with_config(ConfigDict(extra="allow", strict=True))
-class _RecraftImageData(TypedDict):
-    url: ReadOnly[NotRequired[str | None]]
-    b64_json: ReadOnly[NotRequired[str | None]]
-
-
-@with_config(ConfigDict(extra="allow", strict=True))
-class _RecraftImageEditResponse(TypedDict):
-    data: ReadOnly[Sequence[_RecraftImageData]]
-
-
-_JSON_BODY: Final = TypeAdapter(object, config=ConfigDict(hide_input_in_errors=True))
-_RECRAFT_IMAGE_EDIT_RESPONSE: Final = TypeAdapter(_RecraftImageEditResponse)
-
-
 class RecraftImageEditConfig(BaseImageEditConfig):
     DEFAULT_BASE_URL: str = "https://external.api.recraft.ai"
     IMAGE_EDIT_ENDPOINT: str = "v1/images/imageToImage"
@@ -170,14 +152,13 @@ class RecraftImageEditConfig(BaseImageEditConfig):
     ) -> ImageResponse:
         model_response: Final = ImageResponse()
         try:
-            raw_response_data: Final = _JSON_BODY.validate_python(raw_response.json())
+            response_data: Final = raw_response.json()
         except Exception as e:
             raise self.get_error_class(
                 error_message=f"Error transforming image edit response: {e}",
                 status_code=raw_response.status_code,
                 headers=raw_response.headers,
             )
-        response_data: Final = _RECRAFT_IMAGE_EDIT_RESPONSE.validate_python(raw_response_data)
         if not model_response.data:
             model_response.data = []
 

litellm/llms/replicate/chat/transformation.py

@@ -3,8 +3,7 @@ from collections.abc import Iterable, Mapping
 from typing import TYPE_CHECKING, Any, Final
 
 import httpx
-from pydantic import ConfigDict, TypeAdapter, with_config
-from typing_extensions import NotRequired, ReadOnly, TypedDict
+from pydantic import ConfigDict, TypeAdapter
 
 import litellm
 from litellm.constants import REPLICATE_MODEL_NAME_WITH_ID_LENGTH
@@ -34,20 +33,6 @@ _TEXTS: Final = TypeAdapter(Iterable[str], config=ConfigDict(strict=True, hide_i
 _JSON_DICT: Final = TypeAdapter(dict[str, object], config=ConfigDict(hide_input_in_errors=True))
 
 
-@with_config(ConfigDict(extra="allow", strict=True))
-class _ReplicatePredictionUrls(TypedDict):
-    get: ReadOnly[NotRequired[str | None]]
-
-
-@with_config(ConfigDict(extra="allow", strict=True))
-class _ReplicatePrediction(TypedDict):
-    urls: ReadOnly[NotRequired[_ReplicatePredictionUrls]]
-
-
-_JSON_BODY: Final = TypeAdapter(object, config=ConfigDict(hide_input_in_errors=True))
-_REPLICATE_PREDICTION: Final = TypeAdapter(_ReplicatePrediction)
-
-
 class ReplicateConfig(BaseConfig):
     """
     Reference: https://replicate.com/meta/llama-2-70b-chat/api
@@ -307,11 +292,8 @@ class ReplicateConfig(BaseConfig):
         "urls":{"cancel":"https://api.replicate.com/v1/predictions/gqsmqmp1pdrj00cknr08dgmvb4/cancel","get":"https://api.replicate.com/v1/predictions/gqsmqmp1pdrj00cknr08dgmvb4","stream":"https://stream-b.svc.rno2.c.replicate.net/v1/streams/eot4gbydowuin4snhncydwxt57dfwgsc3w3snycx5nid7oef7jga"}
         }
         """
-        response_json: Final = _JSON_BODY.validate_python(response.json())
-        prediction_urls: Final = _REPLICATE_PREDICTION.validate_python(response_json).get("urls")
-        prediction_url: Final = (
-            prediction_urls["get"] if prediction_urls is not None and "get" in prediction_urls else None
-        )
+        response_json: Final = response.json()
+        prediction_url: Final = response_json.get("urls", {}).get("get")
         if prediction_url is None:
             raise ReplicateError(
                 status_code=400,

litellm/proxy/guardrails/guardrail_hooks/guardrails_ai/guardrails_ai.py

@@ -7,12 +7,9 @@
 
 import json
 import os
-from collections.abc import Sequence
-from typing import TYPE_CHECKING, Final, Literal
+from typing import TYPE_CHECKING, Final, Literal, TypedDict
 
 from fastapi import HTTPException
-from pydantic import ConfigDict, TypeAdapter, with_config
-from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 import litellm
 from litellm._logging import verbose_proxy_logger
@@ -37,17 +34,17 @@ class GuardrailsAIResponse(TypedDict):
     validationPassed: bool
 
 
-@with_config(ConfigDict(extra="allow", strict=True))
-class _PreCallOutput(TypedDict):
-    data: ReadOnly[NotRequired[Sequence[str]]]
+class InferenceData(TypedDict):
+    name: str
+    shape: list[int]
+    data: list
+    datatype: str
 
 
-@with_config(ConfigDict(extra="allow", strict=True))
-class _PreCallResponse(TypedDict):
-    outputs: ReadOnly[NotRequired[Sequence[_PreCallOutput]]]
-
-
-_PRE_CALL_RESPONSE: Final = TypeAdapter(_PreCallResponse)
+class GuardrailsAIResponsePreCall(TypedDict):
+    modelname: str
+    modelversion: str
+    outputs: list[InferenceData]
 
 
 class GuardrailsAI(CustomGuardrail):
@@ -133,7 +130,7 @@ class GuardrailsAI(CustomGuardrail):
                 },
             )
 
-        _json_response: Final = _PRE_CALL_RESPONSE.validate_python(response.json())
+        _json_response: Final = GuardrailsAIResponsePreCall(**response.json())
         response = _json_response.get("outputs", [])[0].get("data", [])[0]
         return response
 

tests/unit/llms/bedrock/image_edit/test_stability_transformation.py

@@ -1,8 +1,3 @@
-from unittest.mock import MagicMock
-
-import httpx
-import pytest
-
 from litellm.llms.bedrock.image_edit.stability_transformation import BedrockStabilityImageEditConfig
 
 
@@ -17,31 +12,3 @@ def test_transform_image_edit_request_returns_the_json_body_and_no_files():
     )
 
     assert request == ({"output_format": "png", "prompt": "add a red hat", "image": "iVBORy1ieXRlcw=="}, {})
-
-
-@pytest.mark.parametrize(
-    ("body", "expected_images"),
-    [
-        ({"images": ["aW1hZ2U="], "finish_reasons": [None], "seeds": [42]}, ["aW1hZ2U="]),
-        ({"images": ["aW1hZ2U="], "finish_reasons": None}, ["aW1hZ2U="]),
-        ({"images": None, "finish_reasons": None}, []),
-        ({}, []),
-    ],
-)
-def test_transform_image_edit_response_returns_each_image_and_tolerates_null_lists(body, expected_images):
-    response = BedrockStabilityImageEditConfig().transform_image_edit_response(
-        model="stability.stable-image-inpaint-v1:0",
-        raw_response=httpx.Response(200, json=body),
-        logging_obj=MagicMock(),
-    )
-
-    assert [image.b64_json for image in response.data or []] == expected_images
-
-
-def test_transform_image_edit_response_raises_the_first_finish_reason():
-    with pytest.raises(Exception, match="Bedrock Stability error: Filter reason: prompt"):
-        BedrockStabilityImageEditConfig().transform_image_edit_response(
-            model="stability.stable-image-inpaint-v1:0",
-            raw_response=httpx.Response(200, json={"images": [None], "finish_reasons": ["Filter reason: prompt"]}),
-            logging_obj=MagicMock(),
-        )

tests/unit/proxy/test_caching_routes.py

@@ -1,8 +1,11 @@
 import json
+from typing import Final
+from unittest.mock import MagicMock
 
 import pytest
 from fastapi.testclient import TestClient
 from pydantic import ValidationError
+from pytest_mock import MockerFixture
 
 
 
@@ -53,8 +56,8 @@ def mock_redis_failure(mocker):
 
 
 @pytest.fixture
-def mock_redis_delete(mocker):
-    mock_cache = mocker.MagicMock()
+def mock_redis_delete(mocker: MockerFixture) -> MagicMock:
+    mock_cache: Final = mocker.MagicMock()
     mock_cache.type = "redis"
     mock_cache.delete_cache_keys = mocker.AsyncMock()
     mocker.patch.object(litellm, "cache", mock_cache)
@@ -62,16 +65,18 @@ def mock_redis_delete(mocker):
 
 
 @pytest.mark.parametrize("keys", [["a-key", "b-key"], [12345, 1.5]])
-def test_cache_delete_passes_string_and_numeric_keys_to_the_cache(mock_redis_delete, keys):
-    response = client.post("/cache/delete", headers={"Authorization": "Bearer sk-9876"}, json={"keys": keys})
+def test_cache_delete_passes_string_and_numeric_keys_to_the_cache(
+    mock_redis_delete: MagicMock, keys: list[str | int | float]
+) -> None:
+    response: Final = client.post("/cache/delete", headers={"Authorization": "Bearer sk-9876"}, json={"keys": keys})
 
     assert response.status_code == 200
     assert response.json() == {"status": "success"}
     mock_redis_delete.delete_cache_keys.assert_awaited_once_with(keys=keys)
 
 
-def test_cache_delete_rejects_keys_that_are_not_a_list_with_a_400_that_hides_them(mock_redis_delete):
-    response = client.post("/cache/delete", headers={"Authorization": "Bearer sk-9876"}, json={"keys": "sk-secret-key"})
+def test_cache_delete_rejects_keys_that_are_not_a_list_with_a_400_that_hides_them(mock_redis_delete: MagicMock) -> None:
+    response: Final = client.post("/cache/delete", headers={"Authorization": "Bearer sk-9876"}, json={"keys": "sk-secret-key"})
 
     assert response.status_code == 400
     assert response.json()["detail"][0]["loc"] == ["keys"]