from fastapi import FastAPI
import requests, os
from qdrant_client import QdrantClient
from qdrant_client.models import PointStruct
import hashlib
import json

app = FastAPI()

QDRANT_URL = os.getenv("QDRANT_URL")
OLLAMA_API = os.getenv("OLLAMA_API")
COLLECTION_NAME = os.getenv("COLLECTION_NAME", "chat-memory")

client = QdrantClient(url=QDRANT_URL)

@app.get("/health")
def health():
    return {"status": "ok", "qdrant": QDRANT_URL, "ollama": OLLAMA_API}

def embed(text):
    r = requests.post(f"{OLLAMA_API}/api/embeddings", json={"model":"nomic-embed-text","prompt":text})
    return r.json()["embedding"]

@app.post("/store")
def store(item: dict):
    text = item["text"]
    metadata = item.get("metadata", {})
    vec = embed(text)
    pid = hashlib.sha256(text.encode()).hexdigest()
    client.upsert(collection_name=COLLECTION_NAME, points=[PointStruct(id=pid, vector=vec, payload={"text": text, **metadata})])
    return {"stored": True}

@app.post("/search")
def search(query: dict):
    q = query["text"]
    top_k = query.get("top_k", 5)
    vec = embed(q)
    result = client.search(collection_name=COLLECTION_NAME, query_vector=vec, limit=top_k)
    return [{"score": r.score, "text": r.payload["text"]} for r in result]