from contextlib import AsyncExitStack, asynccontextmanager

from fastapi.staticfiles import StaticFiles
import requests, warnings
from dotenv import load_dotenv
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from fastapi.responses import FileResponse, StreamingResponse
from bs4 import BeautifulSoup

import requests as req_lib
from huggingface_hub.utils import configure_http_backend

from schemas import *
from indexers import (
    get_folder_name,
    get_spec_3gpp_indexer,
    get_spec_etsi_indexer,
    get_tdoc_indexer,
    indexing_locks,
    meetings_mapping,
)
from mcp_indexing import indexing_mcp

def hf_client_factory() -> req_lib.Session:
    session = req_lib.Session()
    session.verify = False
    return session

configure_http_backend(backend_factory=hf_client_factory)

warnings.filterwarnings("ignore")
load_dotenv()


@asynccontextmanager
async def lifespan(app: FastAPI):
    # The MCP server's session manager owns its own background task group; it must be
    # entered as part of this app's lifespan since mounting its Starlette sub-app does not
    # run its lifespan on its own.
    async with AsyncExitStack() as stack:
        await stack.enter_async_context(indexing_mcp.session_manager.run())
        yield


app = FastAPI(lifespan=lifespan)
app.add_middleware(CORSMiddleware, allow_credentials=True, allow_headers=["*"], allow_origins=["*"])
app.mount("/static", StaticFiles(directory="static"), name="static")
# Claude Code (or any MCP client) connects over Streamable HTTP at /index/mcp
app.mount("/index", indexing_mcp.streamable_http_app())

@app.get('/')
def main():
    return FileResponse("index.html")

@app.get("/get_meetings/{working_group}")
def get_meetings(working_group: str):
    category, wg_number = get_folder_name(working_group)
    folder = meetings_mapping[category][wg_number]
    url = f"https://www.3gpp.org/ftp/{meetings_mapping[category][0]}/{folder}"
    response = requests.get(url, verify=False, timeout=(10, 30))
    responseHTML = response.text
    soup = BeautifulSoup(responseHTML, "html.parser")
    return {"url": url, "meetings": [item.get_text() for item in soup.select("tr td a") if item.get_text().startswith("TSG") or item.get_text().startswith("CT")]}

@app.post("/index_tdocs/working_group")
def index_tdocs_wg_progress(req: IndexTDoc):
    if not req.wg:
        raise HTTPException(status_code=400, detail="Working Group not defined !")
    if not indexing_locks["tdoc"].acquire(blocking=False):
        raise HTTPException(status_code=409, detail="TDoc indexing already in progress")
    category, wg_number = get_folder_name(req.wg)
    folder = meetings_mapping[category][wg_number]
    url = f"https://www.3gpp.org/ftp/{meetings_mapping[category][0]}"
    indexer = get_tdoc_indexer()
    def generate_events():
        try:
            yield f"event: info\ndata: {req.wg}\n\n"
            for content in indexer.process_workgroup(folder, url):
                yield content
            indexer.save_indexer()
            yield "event: end\ndata: Indexation ended successfully !\n\n"
        finally:
            indexing_locks["tdoc"].release()
    return StreamingResponse(generate_events(), media_type="text/event-stream")

@app.post("/index_tdocs/meeting")
def index_tdocs_meeting_progress(req: IndexTDoc):
    if not req.wg:
        raise HTTPException(status_code=400, detail="Working Group not defined !")
    if not req.meetings:
        raise HTTPException(status_code=400, detail="Meetings not defined !")
    if not indexing_locks["tdoc"].acquire(blocking=False):
        raise HTTPException(status_code=409, detail="TDoc indexing already in progress")

    category, wg_number = get_folder_name(req.wg)
    folder = meetings_mapping[category][wg_number]
    url = f"https://www.3gpp.org/ftp/{meetings_mapping[category][0]}/{folder}"
    indexer = get_tdoc_indexer()
    def generate_events():
        try:
            yield f"event: get-maximum\ndata: {len(req.meetings)}\n\n"
            for i, meet in enumerate(req.meetings):
                yield f"event: info\ndata: {req.wg}-{meet}\n\n"
                indexer.process_meeting(meet, url)
                yield f"event: progress\ndata: {i+1}\n\n"
            indexer.save_indexer()
            yield "event: end\ndata: Indexation ended successfully !\n\n"
        finally:
            indexing_locks["tdoc"].release()
    return StreamingResponse(generate_events(), media_type="text/event-stream")


@app.post("/index_tdocs/all")
def index_all_tdocs_progress():
    if not indexing_locks["tdoc"].acquire(blocking=False):
        raise HTTPException(status_code=409, detail="TDoc indexing already in progress")
    indexer = get_tdoc_indexer()
    def generate_events():
        try:
            for content in indexer.index_all_tdocs():
                yield content
            indexer.save_indexer()
            yield "event: end\ndata: Indexation ended successfully !\n\n"
        finally:
            indexing_locks["tdoc"].release()
    return StreamingResponse(generate_events(), media_type="text/event-stream")


@app.post("/index_specs/3gpp")
def index_3gpp_specs_progress():
    if not indexing_locks["3gpp"].acquire(blocking=False):
        raise HTTPException(status_code=409, detail="3GPP spec indexing already in progress")
    indexer = get_spec_3gpp_indexer()
    def generate_events():
        try:
            for content in indexer.run():
                yield content
            yield "event: info\ndata: Saving index ...\n\n"
            yield "event: get-maximum\ndata: 1\n\n"
            yield "event: progress\ndata: 1\n\n"
            indexer.save()
            yield "event: info\ndata: Creating BM25 models ...\n\n"
            yield "event: get-maximum\ndata: 1\n\n"
            yield "event: progress\ndata: 1\n\n"
            indexer.create_bm25_index()
            yield "event: end\ndata: Indexation ended successfully !\n\n"
        finally:
            indexing_locks["3gpp"].release()
    return StreamingResponse(generate_events(), media_type="text/event-stream")

@app.post("/index_specs/etsi")
def index_etsi_specs_progress():
    if not indexing_locks["etsi"].acquire(blocking=False):
        raise HTTPException(status_code=409, detail="ETSI spec indexing already in progress")
    indexer = get_spec_etsi_indexer()
    def generate_events():
        try:
            for content in indexer.run():
                yield content
            yield "event: info\ndata: Saving index ...\n\n"
            yield "event: get-maximum\ndata: 1\n\n"
            yield "event: progress\ndata: 1\n\n"
            indexer.save()
            yield "event: info\ndata: Creating BM25 models ...\n\n"
            yield "event: get-maximum\ndata: 1\n\n"
            yield "event: progress\ndata: 1\n\n"
            indexer.create_bm25_index()
            yield "event: end\ndata: Indexation ended successfully !\n\n"
        finally:
            indexing_locks["etsi"].release()
    return StreamingResponse(generate_events(), media_type="text/event-stream")