app.py

"""OKF Screener — Gradio Server entrypoint.

Serves a custom HTML frontend and exposes a single streaming API endpoint
that runs three compliance-checking lanes (RAG, CSV, OKF) concurrently.
"""

import json
import os
import sys

from dotenv import load_dotenv

load_dotenv()
os.environ.setdefault("GRADIO_ANALYTICS_ENABLED", "False")

_ROOT = os.path.dirname(os.path.abspath(__file__))
if _ROOT not in sys.path:
    sys.path.insert(0, _ROOT)

import gradio as gr
from fastapi import Request
from fastapi.responses import FileResponse, JSONResponse, StreamingResponse

from backend.race import race_events

app = gr.Server()
_DIST = os.path.join(_ROOT, "frontend", "dist")


@app.post("/race/stream")
async def race_stream(request: Request):
    """Stream NDJSON events from the three-lane compliance race."""
    try:
        body = await request.json()
    except Exception:
        body = {}

    query = body.get("query", "").strip()
    model_id = body.get("model_id", "meta-llama/Llama-3.3-70B-Instruct").strip()

    def generate():
        try:
            for event in race_events(query, model_id):
                yield json.dumps(event, ensure_ascii=True) + "\n"
        except Exception as exc:
            yield json.dumps({"type": "error", "message": str(exc)}) + "\n"

    return StreamingResponse(
        generate(),
        media_type="application/x-ndjson",
        headers={"Cache-Control": "no-store"},
    )


@app.get("/")
def index():
    path = os.path.join(_DIST, "index.html")
    if os.path.isfile(path):
        return FileResponse(path)
    return JSONResponse({"error": "Frontend not built."}, status_code=503)


if __name__ == "__main__":
    app.launch(
        server_name="0.0.0.0",
        server_port=int(os.environ.get("PORT", "7860")),
        show_error=True,
    )