import json
from pathlib import Path
from typing import List, Dict
from sentence_transformers import SentenceTransformer
import numpy as np

def load_index(env: Dict):
    import faiss, json
    index_path = Path(env["INDEX_DIR"]) / "faiss.index"
    meta_path = Path(env["INDEX_DIR"]) / "meta.json"
    if not index_path.exists():
        raise RuntimeError("Index not found. Run ingest first.")
    index = faiss.read_index(str(index_path))
    metas = json.load(open(meta_path))
    return index, metas

def embed(texts: List[str]):
    model = SentenceTransformer("all-MiniLM-L6-v2")
    return model.encode(texts, convert_to_numpy=True, normalize_embeddings=True)

def search(q: str, env: Dict, top_k: int = 15, filters: Dict = None) -> List[Dict]:
    index, metas = load_index(env)
    qv = embed([q])
    scores, idxs = index.search(qv, top_k)
    results = []
    for score, idx in zip(scores[0], idxs[0]):
        if idx == -1: continue
        m = metas[idx]
        if filters:
            if "geo" in filters and filters["geo"] and m.get("geo") not in filters["geo"]:
                continue
            if "categories" in filters and filters["categories"]:
                if not set(filters["categories"]).intersection(set(m.get("categories",[]))):
                    continue
        m["score"] = float(score)
        results.append(m)
    return results