AR 15 M4
a7f78d71b2
FastAPI + ChromaDB + E5-large + DeepSeek по паттерну work-pcs-dr-cdss, адаптированному под пациентский контекст: - services: embeddings (E5-large с префиксами), vectorstore (коллекция operators_wiki), document_processor (PDF/DOCX/TXT/MD + чанкер с FAQ- паттерном под wiki), llm_client (системный промпт ассистента клиники), rag_pipeline (одиночный вопрос → retrieval → ответ). - routers: /health, /documents (upload, list, chunks, delete), /query. - static/index.html: шапка со статусом, блок базы знаний с раскрытием чанков по клику, блок тест-вопроса с 3-колоночным ответом (чанки со score / собранный промпт / ответ LLM). - Порт 8003 (8001 занят CDSS, 8002 — voicenote). E2E проверен: загрузка wiki_test.md → 2 чанка, вопрос «как записать ребёнка к лору?» → top score 84.8%, корректный ответ DeepSeek. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
56 lines
1.7 KiB
Python
56 lines
1.7 KiB
Python
import logging
|
|
from contextlib import asynccontextmanager
|
|
|
|
from fastapi import FastAPI
|
|
from fastapi.middleware.cors import CORSMiddleware
|
|
from fastapi.staticfiles import StaticFiles
|
|
|
|
from config import settings
|
|
from services.embeddings import EmbeddingService
|
|
from services.vectorstore import VectorStoreService
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
embedding_service: EmbeddingService | None = None
|
|
vectorstore_service: VectorStoreService | None = None
|
|
|
|
|
|
@asynccontextmanager
|
|
async def lifespan(app: FastAPI):
|
|
global embedding_service, vectorstore_service
|
|
logging.basicConfig(level=getattr(logging, settings.log_level.upper(), logging.INFO))
|
|
logger.info("Loading embedding model: %s", settings.embedding_model)
|
|
embedding_service = EmbeddingService(settings.embedding_model)
|
|
logger.info("Embedding model loaded")
|
|
vectorstore_service = VectorStoreService(
|
|
persist_dir=settings.chroma_persist_dir,
|
|
embedding_service=embedding_service,
|
|
)
|
|
logger.info("ChromaDB initialized at %s", settings.chroma_persist_dir)
|
|
yield
|
|
logger.info("Shutting down")
|
|
|
|
|
|
app = FastAPI(
|
|
title="Chat Agent for Patients — Tuning Tool",
|
|
description="RAG-ядро и инструмент настройки пациентского чат-агента",
|
|
version="0.1.0",
|
|
lifespan=lifespan,
|
|
)
|
|
|
|
app.add_middleware(
|
|
CORSMiddleware,
|
|
allow_origins=["*"],
|
|
allow_credentials=True,
|
|
allow_methods=["*"],
|
|
allow_headers=["*"],
|
|
)
|
|
|
|
from routers import documents, health, query # noqa: E402
|
|
|
|
app.include_router(health.router)
|
|
app.include_router(documents.router)
|
|
app.include_router(query.router)
|
|
|
|
app.mount("/", StaticFiles(directory="static", html=True), name="static")
|