104 lines
3.3 KiB
Python
104 lines
3.3 KiB
Python
from __future__ import annotations
|
|
|
|
from contextlib import asynccontextmanager
|
|
|
|
import httpx
|
|
from fastapi import Depends, FastAPI, HTTPException, Request
|
|
|
|
from app.exceptions import ProviderAPICallError, ProviderConfigurationError
|
|
from app.models import (
|
|
DataImportAnalysisRequest,
|
|
DataImportAnalysisResponse,
|
|
LLMRequest,
|
|
LLMResponse,
|
|
)
|
|
from app.services import LLMGateway
|
|
from app.services.import_analysis import build_import_messages, resolve_provider_from_model
|
|
|
|
|
|
@asynccontextmanager
|
|
async def lifespan(app: FastAPI):
|
|
client = httpx.AsyncClient(timeout=httpx.Timeout(30.0))
|
|
gateway = LLMGateway()
|
|
try:
|
|
app.state.http_client = client # type: ignore[attr-defined]
|
|
app.state.gateway = gateway # type: ignore[attr-defined]
|
|
yield
|
|
finally:
|
|
await client.aclose()
|
|
|
|
|
|
def create_app() -> FastAPI:
|
|
application = FastAPI(
|
|
title="Unified LLM Gateway",
|
|
version="0.1.0",
|
|
lifespan=lifespan,
|
|
)
|
|
|
|
@application.post(
|
|
"/v1/chat/completions",
|
|
response_model=LLMResponse,
|
|
summary="Dispatch chat completion to upstream provider",
|
|
)
|
|
async def create_chat_completion(
|
|
payload: LLMRequest,
|
|
gateway: LLMGateway = Depends(get_gateway),
|
|
client: httpx.AsyncClient = Depends(get_http_client),
|
|
) -> LLMResponse:
|
|
try:
|
|
return await gateway.chat(payload, client)
|
|
except ProviderConfigurationError as exc:
|
|
raise HTTPException(status_code=422, detail=str(exc)) from exc
|
|
except ProviderAPICallError as exc:
|
|
raise HTTPException(status_code=502, detail=str(exc)) from exc
|
|
|
|
@application.post(
|
|
"/v1/import/analyze",
|
|
response_model=DataImportAnalysisResponse,
|
|
summary="Analyze import sample data via configured LLM",
|
|
)
|
|
async def analyze_import_data(
|
|
payload: DataImportAnalysisRequest,
|
|
gateway: LLMGateway = Depends(get_gateway),
|
|
client: httpx.AsyncClient = Depends(get_http_client),
|
|
) -> DataImportAnalysisResponse:
|
|
try:
|
|
provider, model_name = resolve_provider_from_model(payload.llm_model)
|
|
except ValueError as exc:
|
|
raise HTTPException(status_code=422, detail=str(exc)) from exc
|
|
|
|
messages = build_import_messages(payload)
|
|
|
|
llm_request = LLMRequest(
|
|
provider=provider,
|
|
model=model_name,
|
|
messages=messages,
|
|
temperature=payload.temperature if payload.temperature is not None else 0.2,
|
|
max_tokens=payload.max_tokens,
|
|
)
|
|
|
|
try:
|
|
llm_response = await gateway.chat(llm_request, client)
|
|
except ProviderConfigurationError as exc:
|
|
raise HTTPException(status_code=422, detail=str(exc)) from exc
|
|
except ProviderAPICallError as exc:
|
|
raise HTTPException(status_code=502, detail=str(exc)) from exc
|
|
|
|
return DataImportAnalysisResponse(
|
|
import_record_id=payload.import_record_id,
|
|
llm_response=llm_response,
|
|
)
|
|
|
|
return application
|
|
|
|
|
|
async def get_gateway(request: Request) -> LLMGateway:
|
|
return request.app.state.gateway # type: ignore[return-value, attr-defined]
|
|
|
|
|
|
async def get_http_client(request: Request) -> httpx.AsyncClient:
|
|
return request.app.state.http_client # type: ignore[return-value, attr-defined]
|
|
|
|
|
|
app = create_app()
|