services/embedding.py

from sentence_transformers import SentenceTransformer
from PIL import Image
from fastapi import UploadFile
from typing import List, Optional
import torch
import os

from huggingface_hub import HfFileSystem
import transformers.utils.hub

# Set cache directory globally for huggingface_hub
transformers.utils.hub.HUGGINGFACE_HUB_CACHE = "/app/cache/hub"

# Load model
model = SentenceTransformer("clip-ViT-B-32")
# Load model with custom cache directory
# model = SentenceTransformer("clip-ViT-B-32")

def get_text_embedding(text: str) -> Optional[List[float]]:
    try:
        embedding = model.encode(text, convert_to_tensor=True).cpu().numpy().tolist()
        return embedding
    except Exception as e:
        print(f"Error generating text embedding: {e}")
        return None

def get_image_embedding(image_file: UploadFile) -> Optional[List[float]]:
    try:
        image = Image.open(image_file.file).convert("RGB").resize((224, 224))
        embedding = model.encode(image, convert_to_tensor=True).cpu().numpy().tolist()
        return embedding
    except Exception as e:
        print(f"Error generating image embedding: {e}")
        return None