handler.py

from typing import Dict, List, Any
from io import BytesIO
import base64
import logging
from PIL import Image
import numpy as np
from transformers import AutoModel


class EndpointHandler():
    def __init__(self, path=""):
        self.model = AutoModel.from_pretrained('Blueway/Inference-endpoint-for-jina-clip-v1', trust_remote_code=True)

    def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
        """
        data args:
            image (:obj:`string`)
            candidates (:obj:`list`)
        Return:
            A :obj:`list`: une liste permettant de passer les embedding 
        """
        inputs_request = data.pop("inputs", data)

        # decode base64 image to PIL
        image = Image.open(BytesIO(base64.b64decode(inputs_request['image'])))
        text = inputs_request['text']
        if text is not None:
            text_embedding = self.model.encode_text(text)
        if image is not None:
            image_embedding = self.model.encode_image(image)
        # Convert embeddings to lists of floats
        serializable_results = {
            'text_embedding': (text_embedding.tolist() if isinstance(text_embedding, np.ndarray) else text_embedding) if text_embedding is not None else [],
            'image_embedding': (image_embedding.tolist() if isinstance(image_embedding, np.ndarray) else image_embedding) if image_embeddingis not None else []
        }

        return serializable_results