import os
import uuid
from dotenv import load_dotenv
from sentence_transformers import SentenceTransformer
from qdrant_client import QdrantClient
from qdrant_client.http import models

import uuid
from datetime import datetime, timezone

load_dotenv()
QDRANT_URL = os.getenv("QDRANT_URL")
QDRANT_API_KEY = os.getenv("QDRANT_API_KEY")
COLLECTION = "docs"

model = SentenceTransformer("all-MiniLM-L6-v2") #fast embedder
qdrant = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY)

# Ensure collection exists (safe init .... we dont wipe on reload)
def _ensure_collection():
    try:
        coll_info = qdrant.get_collection(collection_name=COLLECTION)
        if not coll_info:
            raise Exception("Collection not found")
    except Exception:
        print(f"⚠️ Collection '{COLLECTION}' not found. Creating fresh collection...")
        qdrant.create_collection(
            collection_name=COLLECTION,
            vectors_config=models.VectorParams(
                size=384,
                distance=models.Distance.COSINE
            ),
        )

_ensure_collection()

def embed_and_upsert(chunks, source="user", timestamp=None):
    if timestamp is None:
        timestamp = datetime.now(timezone.utc).isoformat()

    embeddings = model.encode(chunks).tolist()
    points = []
    for i, (chunk, emb) in enumerate(zip(chunks, embeddings)):
        points.append(
            models.PointStruct(
                id=str(uuid.uuid4()),
                vector=emb,
                payload={
                    "text": chunk,
                    "source": source,
                    "timestamp": timestamp,
                    "chunk_id": i
                }
            )
        )
    qdrant.upsert(collection_name=COLLECTION, points=points, wait=True)
    print(f"✅ Stored {len(points)} chunks in Qdrant (source={source}, ts={timestamp})")
    return True

def search(query: str, top_k: int = 5):
    q_emb = model.encode([query])[0].tolist()
    results = qdrant.search(
        collection_name=COLLECTION,
        query_vector=q_emb,
        limit=top_k,
        with_payload=True
    )
    return [
        {
            "chunk": r.payload.get("text", ""),
            "cosine": r.score,
            "timestamp": r.payload.get("timestamp"),
            "source": r.payload.get("source")
        }
        for r in results
    ]