diff --git a/litellm/proxy/_lazy_features.py b/litellm/proxy/_lazy_features.py index faf95397fa5..d6b81b7908a 100644 --- a/litellm/proxy/_lazy_features.py +++ b/litellm/proxy/_lazy_features.py @@ -208,6 +208,7 @@ LAZY_FEATURES: Final[tuple[LazyFeature, ...]] = ( "/nvidia_nim/", "/openai/", "/openai_passthrough/", + "/transcribe", "/vertex-ai/", "/vertex_ai/", "/vllm/", diff --git a/litellm/proxy/_lazy_openapi_snapshot.json b/litellm/proxy/_lazy_openapi_snapshot.json index 74f38b3ca6d..085093ddb30 100644 --- a/litellm/proxy/_lazy_openapi_snapshot.json +++ b/litellm/proxy/_lazy_openapi_snapshot.json @@ -20373,6 +20373,77 @@ ] } }, + "/transcribe": { + "post": { + "description": "AWS-SDK-shaped pass-through for Amazon Transcribe: point the SDK's `endpoint_url`\nat `/transcribe` and the operation is read from the `X-Amz-Target` header, per the\nAWS JSON 1.1 protocol.\n\n[Docs](https://docs.litellm.ai/docs/pass_through/transcribe)", + "operationId": "transcribe_sdk_proxy_route_transcribe_post", + "responses": { + "200": { + "content": { + "application/json": { + "schema": {} + } + }, + "description": "Successful Response" + } + }, + "security": [ + { + "APIKeyHeader": [] + } + ], + "summary": "Transcribe Sdk Proxy Route", + "tags": [ + "llm_passthrough" + ] + } + }, + "/transcribe/{operation}": { + "post": { + "description": "Pass-through for the Amazon Transcribe API, e.g. `POST /transcribe/StartTranscriptionJob`.\n\nThe request body is forwarded as-is to the AWS JSON 1.1 API and signed with SigV4\nusing the proxy's AWS credentials. Streaming transcription (`transcribestreaming`)\nuses a separate HTTP/2 event-stream protocol and is not served by this route.\n\n[Docs](https://docs.litellm.ai/docs/pass_through/transcribe)", + "operationId": "transcribe_proxy_route_transcribe__operation__post", + "parameters": [ + { + "in": "path", + "name": "operation", + "required": true, + "schema": { + "title": "Operation", + "type": "string" + } + } + ], + "responses": { + "200": { + "content": { + "application/json": { + "schema": {} + } + }, + "description": "Successful Response" + }, + "422": { + "content": { + "application/json": { + "schema": { + "$ref": "#/components/schemas/HTTPValidationError" + } + } + }, + "description": "Validation Error" + } + }, + "security": [ + { + "APIKeyHeader": [] + } + ], + "summary": "Transcribe Proxy Route", + "tags": [ + "llm_passthrough" + ] + } + }, "/vertex_ai/discovery/{endpoint}": { "delete": { "description": "Call any vertex discovery endpoint using the proxy.\n\nJust use `{PROXY_BASE_URL}/vertex_ai/discovery/{endpoint:path}`\n\nTarget url: `https://discoveryengine.googleapis.com`", diff --git a/litellm/proxy/_types.py b/litellm/proxy/_types.py index 14e3635f079..d6c6d45260c 100644 --- a/litellm/proxy/_types.py +++ b/litellm/proxy/_types.py @@ -468,6 +468,7 @@ class LiteLLMRoutes(enum.Enum): mapped_pass_through_routes = [ "/bedrock", "/comprehendmedical", + "/transcribe", "/vertex-ai", "/vertex_ai", "/cohere", diff --git a/litellm/proxy/middleware/billable_request_metrics_middleware.py b/litellm/proxy/middleware/billable_request_metrics_middleware.py index ac119e81d9c..96c3276efac 100644 --- a/litellm/proxy/middleware/billable_request_metrics_middleware.py +++ b/litellm/proxy/middleware/billable_request_metrics_middleware.py @@ -93,6 +93,7 @@ _LLM_ROUTE_EXACT: Final[tuple[str, ...]] = ( "/interactions", # Google Interactions create; /{id} reads and /cancel do not match "/v1beta/interactions", "/comprehendmedical", # AWS-SDK-shaped passthrough: the operation rides in the X-Amz-Target header + "/transcribe", ) # Provider passthrough prefixes (e.g. /bedrock/..., /vertex-ai/...) carry real diff --git a/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py b/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py index 0fe9d1cc626..4d0932794a1 100644 --- a/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py +++ b/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py @@ -1200,7 +1200,7 @@ async def bedrock_proxy_route( COMPREHEND_MEDICAL_TARGET_PREFIX: Final = "ComprehendMedical_20181030" -def _resolve_comprehend_medical_region() -> str | None: +def _resolve_aws_passthrough_region() -> str | None: region_candidates: Final = ( get_secret_str(secret_name="AWS_REGION_NAME"), get_secret_str(secret_name="AWS_REGION"), @@ -1240,7 +1240,7 @@ async def comprehend_medical_proxy_route( ), ) - aws_region_name: Final = _resolve_comprehend_medical_region() + aws_region_name: Final = _resolve_aws_passthrough_region() if aws_region_name is None: raise HTTPException( status_code=400, @@ -1317,6 +1317,121 @@ async def comprehend_medical_sdk_proxy_route( ) +@router.post( + "/transcribe/{operation}", + tags=["Amazon Transcribe Pass-through", "pass-through"], # mutable-ok: fastapi route tags must be a list +) +async def transcribe_proxy_route( + operation: str, + request: Request, + fastapi_response: Response, + user_api_key_dict: Annotated[UserAPIKeyAuth, Depends(user_api_key_auth)], +): + """ + Pass-through for the Amazon Transcribe API, e.g. `POST /transcribe/StartTranscriptionJob`. + + The request body is forwarded as-is to the AWS JSON 1.1 API and signed with SigV4 + using the proxy's AWS credentials. Streaming transcription (`transcribestreaming`) + uses a separate HTTP/2 event-stream protocol and is not served by this route. + + [Docs](https://docs.litellm.ai/docs/pass_through/transcribe) + """ + from .llm_provider_handlers.transcribe_passthrough_logging_handler import ( + TRANSCRIBE_CUSTOM_LLM_PROVIDER, + TRANSCRIBE_TARGET_PREFIX, + transcribe_supported_operations, + ) + + if operation not in transcribe_supported_operations(): + raise HTTPException( + status_code=400, + detail=( + f"Unsupported Amazon Transcribe operation: {operation}. " + f"Supported operations: {', '.join(sorted(transcribe_supported_operations()))}" + ), + ) + + aws_region_name: Final = _resolve_aws_passthrough_region() + if aws_region_name is None: + raise HTTPException( + status_code=400, + detail="AWS region not found. Set AWS_REGION_NAME in the proxy environment.", + ) + + try: + data: Final = await _json_request_body(request) + except ValueError as e: + raise HTTPException(status_code=400, detail=f"Request body must be valid JSON: {e}") + + if not isinstance(data, dict): + raise HTTPException(status_code=400, detail="Request body must be a JSON object") + if "stream" in data: + raise HTTPException(status_code=400, detail="'stream' is not an Amazon Transcribe request member") + + from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM, run_aws_signing, sign_aws_json_post + + target_url: Final = f"https://transcribe.{aws_region_name}.{get_aws_dns_suffix(aws_region_name)}/" + prepped: Final = await run_aws_signing( + sign_aws_json_post, + get_credentials=partial(BaseAWSLLM().get_credentials, aws_region_name=aws_region_name), + service_name="transcribe", + aws_region_name=aws_region_name, + url=target_url, + body=json.dumps(data), + headers=MappingProxyType( + { + "Content-Type": "application/x-amz-json-1.1", + "X-Amz-Target": f"{TRANSCRIBE_TARGET_PREFIX}.{operation}", + } + ), + ) + + endpoint_func: Final = create_pass_through_route( + endpoint=operation, + target=str(prepped.url), + custom_headers=prepped.headers, + custom_llm_provider=TRANSCRIBE_CUSTOM_LLM_PROVIDER, + ) + setattr(request.state, LITELLM_PASS_THROUGH_CUSTOM_BODY_STATE_KEY, data) + setattr(request.state, LITELLM_PASS_THROUGH_RAW_BODY_STATE_KEY, prepped.body) + return await endpoint_func(request, fastapi_response, user_api_key_dict) + + +@router.post( + "/transcribe", + tags=["Amazon Transcribe Pass-through", "pass-through"], # mutable-ok: fastapi route tags must be a list +) +async def transcribe_sdk_proxy_route( + request: Request, + fastapi_response: Response, + user_api_key_dict: Annotated[UserAPIKeyAuth, Depends(user_api_key_auth)], +): + """ + AWS-SDK-shaped pass-through for Amazon Transcribe: point the SDK's `endpoint_url` + at `/transcribe` and the operation is read from the `X-Amz-Target` header, per the + AWS JSON 1.1 protocol. + + [Docs](https://docs.litellm.ai/docs/pass_through/transcribe) + """ + from .llm_provider_handlers.transcribe_passthrough_logging_handler import ( + TRANSCRIBE_TARGET_PREFIX, + ) + + target_header: Final = request.headers.get("x-amz-target", "") + target_prefix, _, operation = target_header.partition(".") + if target_prefix != TRANSCRIBE_TARGET_PREFIX or not operation: + raise HTTPException( + status_code=400, + detail=f"Expected an X-Amz-Target header of the form {TRANSCRIBE_TARGET_PREFIX}.", + ) + return await transcribe_proxy_route( + operation=operation, + request=request, + fastapi_response=fastapi_response, + user_api_key_dict=user_api_key_dict, + ) + + def _resolve_vertex_model_from_router( model_id: str, llm_router: litellm.Router | None, diff --git a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/transcribe_passthrough_logging_handler.py b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/transcribe_passthrough_logging_handler.py new file mode 100644 index 00000000000..0cf593d28df --- /dev/null +++ b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/transcribe_passthrough_logging_handler.py @@ -0,0 +1,90 @@ +from collections.abc import Mapping +from datetime import datetime +from functools import lru_cache +from typing import Final + +import httpx + +from litellm._logging import verbose_proxy_logger +from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +from litellm.litellm_core_utils.litellm_logging import ( + get_standard_logging_object_payload, +) +from litellm.proxy._types import PassThroughEndpointLoggingTypedDict +from litellm.types.utils import StandardPassThroughResponseObject + +TRANSCRIBE_TARGET_PREFIX: Final = "Transcribe" +TRANSCRIBE_CUSTOM_LLM_PROVIDER: Final = "transcribe" + + +@lru_cache(maxsize=1) +def transcribe_supported_operations() -> frozenset[str]: + """ + Operation names of the Amazon Transcribe JSON 1.1 API, read from the botocore + service model so the allowlist tracks the installed SDK instead of a hand-typed copy. + """ + from botocore.session import get_session + + return frozenset(get_session().get_service_model("transcribe").operation_names) + + +class TranscribePassthroughLoggingHandler: + @staticmethod + def _operation_from_response(httpx_response: httpx.Response) -> str: + target: Final = httpx_response.request.headers.get("x-amz-target", "") + return target.split(".")[-1] + + @staticmethod + def transcribe_passthrough_handler( + httpx_response: httpx.Response, + logging_obj: LiteLLMLoggingObj, + url_route: str, + result: str, + start_time: datetime, + end_time: datetime, + cache_hit: bool, + request_body: Mapping[str, object], + **kwargs: object, # kwargs-ok: the passthrough logging dispatch forwards shared logging kwargs to every handler + ) -> PassThroughEndpointLoggingTypedDict: + """ + Records model and provider for an Amazon Transcribe control-plane call. Transcribe + bills per second of audio once a job finishes, which no request or response on this + path carries, so response_cost is recorded as 0.0 rather than estimated. + """ + try: + operation: Final = TranscribePassthroughLoggingHandler._operation_from_response(httpx_response) + model_name: Final = f"{TRANSCRIBE_CUSTOM_LLM_PROVIDER}/{operation}" + + updated_kwargs: Final = { # mutable-ok: the logging pipeline requires a plain kwargs dict + **kwargs, + "model": model_name, + "custom_llm_provider": TRANSCRIBE_CUSTOM_LLM_PROVIDER, + "response_cost": 0.0, + } + logging_obj.model_call_details.update( + model=model_name, + custom_llm_provider=TRANSCRIBE_CUSTOM_LLM_PROVIDER, + response_cost=0.0, + ) + + standard_logging_object: Final = get_standard_logging_object_payload( + kwargs=updated_kwargs, + init_response_obj=StandardPassThroughResponseObject(response=result), + start_time=start_time, + end_time=end_time, + logging_obj=logging_obj, + status="success", + ) + + handler_payload: Final[PassThroughEndpointLoggingTypedDict] = { + "result": StandardPassThroughResponseObject(response=result), + "kwargs": {**updated_kwargs, "standard_logging_object": standard_logging_object}, + } + except Exception as e: # noqa: BLE001 # logging must never fail the forwarded request + verbose_proxy_logger.exception("Error in Amazon Transcribe passthrough logging handler: %s", e) + fallback_payload: Final[PassThroughEndpointLoggingTypedDict] = { + "result": StandardPassThroughResponseObject(response=result), + "kwargs": kwargs, + } + return fallback_payload + return handler_payload diff --git a/litellm/proxy/pass_through_endpoints/success_handler.py b/litellm/proxy/pass_through_endpoints/success_handler.py index 76a471302f4..7dfada592b8 100644 --- a/litellm/proxy/pass_through_endpoints/success_handler.py +++ b/litellm/proxy/pass_through_endpoints/success_handler.py @@ -27,6 +27,10 @@ from .llm_provider_handlers.cursor_passthrough_logging_handler import ( from .llm_provider_handlers.gemini_passthrough_logging_handler import ( GeminiPassthroughLoggingHandler, ) +from .llm_provider_handlers.transcribe_passthrough_logging_handler import ( + TRANSCRIBE_CUSTOM_LLM_PROVIDER, + TranscribePassthroughLoggingHandler, +) from .llm_provider_handlers.vertex_passthrough_logging_handler import ( VertexPassthroughLoggingHandler, ) @@ -256,6 +260,20 @@ class PassThroughEndpointLogging: ) standard_logging_response_object = comprehend_medical_handler_result["result"] # rebind-ok: elif-chain kwargs = comprehend_medical_handler_result["kwargs"] # rebind-ok: elif-chain contract + elif self.is_transcribe_route(custom_llm_provider): + transcribe_handler_result: Final = TranscribePassthroughLoggingHandler.transcribe_passthrough_handler( + httpx_response=httpx_response, + logging_obj=logging_obj, + url_route=url_route, + result=result, + start_time=start_time, + end_time=end_time, + cache_hit=cache_hit, + request_body=request_body, + **kwargs, + ) + standard_logging_response_object = transcribe_handler_result["result"] # rebind-ok: elif-chain + kwargs = transcribe_handler_result["kwargs"] # rebind-ok: elif-chain contract elif self.is_vertex_ai_live_route(url_route): from .llm_provider_handlers.vertex_ai_live_passthrough_logging_handler import ( VertexAILivePassthroughLoggingHandler, @@ -389,6 +407,9 @@ class PassThroughEndpointLogging: def is_comprehend_medical_route(self, custom_llm_provider: str | None) -> bool: return custom_llm_provider == "comprehendmedical" + def is_transcribe_route(self, custom_llm_provider: str | None) -> bool: + return custom_llm_provider == TRANSCRIBE_CUSTOM_LLM_PROVIDER + def is_langfuse_route(self, url_route: str): parsed_url: Final = urlparse(url_route) for route in self.TRACKED_LANGFUSE_ROUTES: diff --git a/tests/pass_through_unit_tests/test_pass_through_unit_tests.py b/tests/pass_through_unit_tests/test_pass_through_unit_tests.py index ed04b63000f..dd8a6486e4f 100644 --- a/tests/pass_through_unit_tests/test_pass_through_unit_tests.py +++ b/tests/pass_through_unit_tests/test_pass_through_unit_tests.py @@ -411,6 +411,8 @@ async def test_pass_through_request_logging_failure_with_stream( PROTOCOL_CONSTRAINED_PASS_THROUGH_ROUTES = { "/comprehendmedical": {"POST"}, "/comprehendmedical/{operation}": {"POST"}, + "/transcribe": {"POST"}, + "/transcribe/{operation}": {"POST"}, } @@ -419,8 +421,8 @@ def test_pass_through_routes_support_all_methods(): A pass-through route fronts a whole provider API, so narrowing its method set turns a request the upstream would have accepted into a 405. The exceptions are providers whose wire protocol admits only one method: Amazon - Comprehend Medical speaks AWS JSON 1.1, which is POST-only, so there is no - other method to forward. + Comprehend Medical and Amazon Transcribe speak AWS JSON 1.1, which is + POST-only, so there is no other method to forward. """ from litellm.proxy.pass_through_endpoints.llm_passthrough_endpoints import ( router as llm_router, diff --git a/tests/test_litellm/proxy/middleware/test_billable_request_metrics_middleware.py b/tests/test_litellm/proxy/middleware/test_billable_request_metrics_middleware.py index 9c61412bd6e..c6af900d263 100644 --- a/tests/test_litellm/proxy/middleware/test_billable_request_metrics_middleware.py +++ b/tests/test_litellm/proxy/middleware/test_billable_request_metrics_middleware.py @@ -116,6 +116,8 @@ def test_is_pure_asgi_not_base_http_middleware(): # Bare AWS-SDK-shaped route carries the operation in X-Amz-Target and writes SpendLogs ("/comprehendmedical", (BillableCategory.LLM, "/comprehendmedical")), ("/comprehendmedical/DetectEntitiesV2", (BillableCategory.LLM, "/comprehendmedical")), + ("/transcribe", (BillableCategory.LLM, "/transcribe")), + ("/transcribe/StartTranscriptionJob", (BillableCategory.LLM, "/transcribe")), ("/mcp", (BillableCategory.MCP, "/mcp")), ("/mcp/", (BillableCategory.MCP, "/mcp")), ("/mcp/tools/list", (BillableCategory.MCP, "/mcp")), diff --git a/tests/test_litellm/proxy/pass_through_endpoints/llm_provider_handlers/test_transcribe_passthrough_logging_handler.py b/tests/test_litellm/proxy/pass_through_endpoints/llm_provider_handlers/test_transcribe_passthrough_logging_handler.py new file mode 100644 index 00000000000..6dd9794344e --- /dev/null +++ b/tests/test_litellm/proxy/pass_through_endpoints/llm_provider_handlers/test_transcribe_passthrough_logging_handler.py @@ -0,0 +1,110 @@ +from datetime import datetime +from unittest.mock import MagicMock + +import httpx + +from litellm.proxy.pass_through_endpoints.llm_provider_handlers.transcribe_passthrough_logging_handler import ( + TranscribePassthroughLoggingHandler, + transcribe_supported_operations, +) +from litellm.proxy.pass_through_endpoints.success_handler import ( + PassThroughEndpointLogging, +) + + +def _make_response(operation: str) -> httpx.Response: + request = httpx.Request( + "POST", + "https://transcribe.us-west-2.amazonaws.com/", + headers={"X-Amz-Target": f"Transcribe.{operation}"}, + ) + return httpx.Response(200, request=request, text='{"TranscriptionJob": {}}') + + +def _make_logging_obj() -> MagicMock: + logging_obj = MagicMock() + logging_obj.litellm_call_id = "test-call-id" + logging_obj.model_call_details = {} + return logging_obj + + +class TestTranscribeSupportedOperations: + def test_matches_the_installed_botocore_service_model(self): + from botocore.session import get_session + + assert transcribe_supported_operations() == frozenset( + get_session().get_service_model("transcribe").operation_names + ) + assert "StartTranscriptionJob" in transcribe_supported_operations() + + +class TestTranscribePassthroughHandler: + def test_records_model_provider_and_zero_cost(self): + logging_obj = _make_logging_obj() + request_body = {"TranscriptionJobName": "litellm-job-1"} + + handler_result = TranscribePassthroughLoggingHandler.transcribe_passthrough_handler( + httpx_response=_make_response("StartTranscriptionJob"), + logging_obj=logging_obj, + url_route="https://transcribe.us-west-2.amazonaws.com/", + result='{"TranscriptionJob": {}}', + start_time=datetime.now(), + end_time=datetime.now(), + cache_hit=False, + request_body=request_body, + ) + + assert handler_result["result"] == {"response": '{"TranscriptionJob": {}}'} + assert handler_result["kwargs"]["model"] == "transcribe/StartTranscriptionJob" + assert handler_result["kwargs"]["custom_llm_provider"] == "transcribe" + assert handler_result["kwargs"]["response_cost"] == 0.0 + assert "standard_logging_object" in handler_result["kwargs"] + assert logging_obj.model_call_details["model"] == "transcribe/StartTranscriptionJob" + assert logging_obj.model_call_details["custom_llm_provider"] == "transcribe" + assert logging_obj.model_call_details["response_cost"] == 0.0 + assert request_body == {"TranscriptionJobName": "litellm-job-1"} + + +class TestIsTranscribeRoute: + def test_matches_by_provider_tag(self): + assert PassThroughEndpointLogging().is_transcribe_route("transcribe") + + def test_does_not_match_other_providers(self): + assert not PassThroughEndpointLogging().is_transcribe_route("comprehendmedical") + + def test_dispatch_reaches_transcribe_handler(self): + logging_obj = _make_logging_obj() + + normalized = PassThroughEndpointLogging().normalize_llm_passthrough_logging_payload( + httpx_response=_make_response("GetTranscriptionJob"), + response_body={"TranscriptionJob": {}}, + request_body={"TranscriptionJobName": "litellm-job-1"}, + logging_obj=logging_obj, + url_route="https://transcribe.us-west-2.amazonaws.com/", + result='{"TranscriptionJob": {}}', + start_time=datetime.now(), + end_time=datetime.now(), + cache_hit=False, + custom_llm_provider="transcribe", + ) + + assert normalized["kwargs"]["model"] == "transcribe/GetTranscriptionJob" + assert normalized["kwargs"]["response_cost"] == 0.0 + + def test_config_driven_passthrough_to_transcribe_host_is_not_claimed(self): + logging_obj = _make_logging_obj() + + normalized = PassThroughEndpointLogging().normalize_llm_passthrough_logging_payload( + httpx_response=_make_response("GetTranscriptionJob"), + response_body={"TranscriptionJob": {}}, + request_body={"TranscriptionJobName": "litellm-job-1"}, + logging_obj=logging_obj, + url_route="https://transcribe.us-west-2.amazonaws.com/", + result='{"TranscriptionJob": {}}', + start_time=datetime.now(), + end_time=datetime.now(), + cache_hit=False, + custom_llm_provider=None, + ) + + assert normalized["kwargs"].get("model") != "transcribe/GetTranscriptionJob" diff --git a/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py b/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py index e0785b002b2..8b860acf189 100644 --- a/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py +++ b/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py @@ -5135,6 +5135,137 @@ class TestComprehendMedicalProxyRoute: assert exc_info.value.status_code == 400 +TRANSCRIBE_UPSTREAM = "https://transcribe.us-west-2.amazonaws.com/" + + +@pytest.fixture +def transcribe_client(monkeypatch: pytest.MonkeyPatch) -> Iterator[TestClient]: + from litellm.proxy.proxy_server import app + + monkeypatch.setenv("AWS_REGION_NAME", "us-west-2") + monkeypatch.setenv("AWS_ACCESS_KEY_ID", "test-access-key") + monkeypatch.setenv("AWS_SECRET_ACCESS_KEY", "test-secret-key") + monkeypatch.delenv("AWS_SESSION_TOKEN", raising=False) + monkeypatch.delenv("SERVER_ROOT_PATH", raising=False) + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + litellm.in_memory_llm_clients_cache.flush_cache() + monkeypatch.setitem(app.dependency_overrides, user_api_key_auth, lambda: UserAPIKeyAuth(api_key="sk-virtual")) + yield TestClient(app) + + +class TestTranscribeProxyRoute: + START_JOB_BODY: Final = MappingProxyType( + { + "TranscriptionJobName": "litellm-job-1", + "LanguageCode": "en-US", + "Media": {"MediaFileUri": "s3://bucket/audio.wav"}, + } + ) + + def test_signs_and_forwards_start_transcription_job(self, transcribe_client: TestClient) -> None: + upstream_body = {"TranscriptionJob": {"TranscriptionJobName": "litellm-job-1", "TranscriptionJobStatus": "IN_PROGRESS"}} + with respx.mock(assert_all_called=True) as upstream: + route = upstream.post(TRANSCRIBE_UPSTREAM).mock(return_value=httpx.Response(200, json=upstream_body)) + response = transcribe_client.post( + "/transcribe/StartTranscriptionJob", + json=dict(self.START_JOB_BODY), + headers={"Authorization": "Bearer sk-virtual"}, + ) + + assert (response.status_code, response.json()) == (200, upstream_body) + sent = route.calls.last.request + assert json.loads(sent.content) == dict(self.START_JOB_BODY) + assert sent.headers["x-amz-target"] == "Transcribe.StartTranscriptionJob" + assert sent.headers["content-type"] == "application/x-amz-json-1.1" + assert sent.headers["authorization"].startswith("AWS4-HMAC-SHA256 Credential=test-access-key/") + assert "/us-west-2/transcribe/aws4_request" in sent.headers["authorization"] + assert "x-amz-date" in sent.headers + + def test_sdk_route_reads_operation_from_x_amz_target_and_resigns(self, transcribe_client: TestClient) -> None: + with respx.mock(assert_all_called=True) as upstream: + route = upstream.post(TRANSCRIBE_UPSTREAM).mock( + return_value=httpx.Response(200, json={"TranscriptionJob": {"TranscriptionJobStatus": "COMPLETED"}}) + ) + response = transcribe_client.post( + "/transcribe", + json={"TranscriptionJobName": "litellm-job-1"}, + headers={ + "Authorization": "AWS4-HMAC-SHA256 Credential=sk-virtual/20260101/us-west-2/transcribe/aws4_request", + "X-Amz-Target": "Transcribe.GetTranscriptionJob", + "Content-Type": "application/x-amz-json-1.1", + }, + ) + + assert (response.status_code, response.json()) == (200, {"TranscriptionJob": {"TranscriptionJobStatus": "COMPLETED"}}) + sent = route.calls.last.request + assert sent.headers["x-amz-target"] == "Transcribe.GetTranscriptionJob" + assert "Credential=test-access-key/" in sent.headers["authorization"] + assert "sk-virtual" not in sent.headers["authorization"] + + def test_upstream_error_status_and_body_are_returned(self, transcribe_client: TestClient) -> None: + aws_error = {"__type": "BadRequestException", "Message": "The requested job couldn't be found."} + with respx.mock(assert_all_called=True) as upstream: + upstream.post(TRANSCRIBE_UPSTREAM).mock(return_value=httpx.Response(400, json=aws_error)) + response = transcribe_client.post("/transcribe/GetTranscriptionJob", json={"TranscriptionJobName": "missing"}) + + assert (response.status_code, response.json()) == (400, aws_error) + + @pytest.mark.parametrize( + "operation", + ["Start-Transcription-Job", "Transcribe.StartTranscriptionJob", "a" * 200, "starttranscriptionjob", "DetectEntitiesV2"], + ) + def test_rejects_unsupported_operations_without_calling_aws(self, transcribe_client: TestClient, operation: str) -> None: + with respx.mock(assert_all_called=False) as upstream: + route = upstream.post(TRANSCRIBE_UPSTREAM) + response = transcribe_client.post(f"/transcribe/{operation}", json={}) + + assert response.status_code == 400 + assert "Unsupported Amazon Transcribe operation" in response.json()["detail"] + assert not route.called + + @pytest.mark.parametrize( + "raw_body", + ['{"MaxResults": 5, "stream": true}', '{"MaxResults": 5, "stream": false}', '["x"]', "not json"], + ) + def test_rejects_bad_bodies_without_calling_aws(self, transcribe_client: TestClient, raw_body: str) -> None: + with respx.mock(assert_all_called=False) as upstream: + route = upstream.post(TRANSCRIBE_UPSTREAM) + response = transcribe_client.post( + "/transcribe/ListTranscriptionJobs", content=raw_body, headers={"Content-Type": "application/json"} + ) + + assert response.status_code == 400 + assert not route.called + + def test_missing_region_returns_400_without_calling_aws( + self, transcribe_client: TestClient, monkeypatch: pytest.MonkeyPatch + ) -> None: + for name in ("AWS_REGION_NAME", "AWS_REGION", "AWS_DEFAULT_REGION"): + monkeypatch.delenv(name, raising=False) + with respx.mock(assert_all_called=False) as upstream: + route = upstream.post(TRANSCRIBE_UPSTREAM) + response = transcribe_client.post("/transcribe/ListTranscriptionJobs", json={}) + + assert response.status_code == 400 + assert "AWS region" in response.json()["detail"] + assert not route.called + + @pytest.mark.parametrize("target_header", ["", "Transcribe", "ComprehendMedical_20181030.DetectPHI", "Transcribe."]) + def test_sdk_route_rejects_bad_x_amz_target(self, transcribe_client: TestClient, target_header: str) -> None: + with respx.mock(assert_all_called=False) as upstream: + route = upstream.post(TRANSCRIBE_UPSTREAM) + response = transcribe_client.post("/transcribe", json={}, headers={"X-Amz-Target": target_header}) + + assert response.status_code == 400 + assert "X-Amz-Target" in response.json()["detail"] + assert not route.called + + def test_transcribe_is_a_mapped_pass_through_route(self) -> None: + from litellm.proxy._types import LiteLLMRoutes + + assert "/transcribe" in LiteLLMRoutes.mapped_pass_through_routes.value + + LIVE_RESOURCE_PATH = "projects/proj-db/locations/global/publishers/google/models/gemini-live-2.5-flash" diff --git a/ui/litellm-dashboard/src/lib/http/schema.d.ts b/ui/litellm-dashboard/src/lib/http/schema.d.ts index 872875cc535..df9a5b6a9ff 100644 --- a/ui/litellm-dashboard/src/lib/http/schema.d.ts +++ b/ui/litellm-dashboard/src/lib/http/schema.d.ts @@ -16437,6 +16437,56 @@ export interface paths { patch: operations["toolset_mcp_route_toolset__toolset_name__mcp_patch"]; trace?: never; }; + "/transcribe": { + parameters: { + query?: never; + header?: never; + path?: never; + cookie?: never; + }; + get?: never; + put?: never; + /** + * Transcribe Sdk Proxy Route + * @description AWS-SDK-shaped pass-through for Amazon Transcribe: point the SDK's `endpoint_url` + * at `/transcribe` and the operation is read from the `X-Amz-Target` header, per the + * AWS JSON 1.1 protocol. + * + * [Docs](https://docs.litellm.ai/docs/pass_through/transcribe) + */ + post: operations["transcribe_sdk_proxy_route_transcribe_post"]; + delete?: never; + options?: never; + head?: never; + patch?: never; + trace?: never; + }; + "/transcribe/{operation}": { + parameters: { + query?: never; + header?: never; + path?: never; + cookie?: never; + }; + get?: never; + put?: never; + /** + * Transcribe Proxy Route + * @description Pass-through for the Amazon Transcribe API, e.g. `POST /transcribe/StartTranscriptionJob`. + * + * The request body is forwarded as-is to the AWS JSON 1.1 API and signed with SigV4 + * using the proxy's AWS credentials. Streaming transcription (`transcribestreaming`) + * uses a separate HTTP/2 event-stream protocol and is not served by this route. + * + * [Docs](https://docs.litellm.ai/docs/pass_through/transcribe) + */ + post: operations["transcribe_proxy_route_transcribe__operation__post"]; + delete?: never; + options?: never; + head?: never; + patch?: never; + trace?: never; + }; "/update/default_team_settings": { parameters: { query?: never; @@ -61441,6 +61491,57 @@ export interface operations { }; }; }; + transcribe_sdk_proxy_route_transcribe_post: { + parameters: { + query?: never; + header?: never; + path?: never; + cookie?: never; + }; + requestBody?: never; + responses: { + /** @description Successful Response */ + 200: { + headers: { + [name: string]: unknown; + }; + content: { + "application/json": unknown; + }; + }; + }; + }; + transcribe_proxy_route_transcribe__operation__post: { + parameters: { + query?: never; + header?: never; + path: { + operation: string; + }; + cookie?: never; + }; + requestBody?: never; + responses: { + /** @description Successful Response */ + 200: { + headers: { + [name: string]: unknown; + }; + content: { + "application/json": unknown; + }; + }; + /** @description Validation Error */ + 422: { + headers: { + [name: string]: unknown; + }; + content: { + "application/json": components["schemas"]["HTTPValidationError"]; + }; + }; + }; + }; update_default_team_settings_update_default_team_settings_patch: { parameters: { query?: never;