import os
import json
import faiss
import numpy as np
from sentence_transformers import SentenceTransformer

# Fix caching permissions for Hugging Face
os.environ["HF_HOME"] = "./cache"
os.environ["TRANSFORMERS_CACHE"] = "./cache"
os.environ["SENTENCE_TRANSFORMERS_HOME"] = "./cache"

# Load products
with open("products.json", "r", encoding="utf-8") as f:
    products = json.load(f)

print(f"📦 Loaded {len(products)} products from products.json")

# Load CLIP model
print("🧠 Loading CLIP model...")
model = SentenceTransformer("sentence-transformers/clip-ViT-B-32", cache_folder="./cache")

# Encode product names (normalize for cosine sim)
print("🔎 Encoding product features...")
product_names = [p["name"] for p in products]
embeddings = model.encode(
    product_names,
    convert_to_numpy=True,
    show_progress_bar=True,
    normalize_embeddings=True,   # important for cosine similarity
)

# Build FAISS index with cosine similarity (inner product on normalized vectors)
dimension = embeddings.shape[1]
index = faiss.IndexFlatIP(dimension)
index.add(embeddings)

# Save index
faiss.write_index(index, "products.index")
print("✅ Saved FAISS index to products.index")

# Save mapping
with open("id_mapping.json", "w", encoding="utf-8") as f:
    json.dump(products, f, ensure_ascii=False, indent=2)

print("✅ Saved product ID mapping to id_mapping.json")