from contextlib import asynccontextmanager

from fastapi import FastAPI

from app.agent.graph import run_agent
from app.config import get_settings
from app.db.postgres import close_pool, open_pool
from app.db.neo4j import verify_connectivity, close_driver
from app.schemas.request import ChatRequest
from app.schemas.response import ChatResponse
from app.memory.redis_checkpointer import get_checkpointer, close_checkpointer
from app.agent.graph import build_agent

settings = get_settings()


@asynccontextmanager
async def lifespan(app: FastAPI):
    await open_pool()
    await verify_connectivity()
    checkpointer = await get_checkpointer()
    build_agent(checkpointer)
    yield
    await close_pool()
    await close_driver()
    await close_checkpointer()

app = FastAPI(title="chat-service", version="0.1.0", lifespan=lifespan)


@app.get("/health")
def health() -> dict:
    return {"status": "ok"}


@app.post("/chat", response_model=ChatResponse)
async def chat(payload: ChatRequest) -> ChatResponse:
    """
    Milestone 2: routes through the LangGraph ReAct agent, which can call
    sql_query_tool to answer questions grounded in the analyzed-articles DB.
    """
    result = await run_agent(
        message=payload.message,
        session_id=payload.session_id,
        user_id=payload.user_id,
    )
    return ChatResponse(
        session_id=payload.session_id,
        answer=result["answer"],
        sources=result["sources"],
    )