src/embedding_function.py

"""
This module provides functionality to embed texts using the Hugging Face API.
It includes an EmbeddingFunction class for asynchronous embedding and a sync_embed function for synchronous embedding.
"""
from huggingface_hub import InferenceClient

import os
from typing import List, Optional, Union
import os
from huggingface_hub import InferenceClient
from dotenv import load_dotenv

load_dotenv()


TextType = Union[str, List[str]]


class EmbeddingFunction:
    """
    A class to handle embedding functions using the Hugging Face API.
    """

    def __init__(
        self,
        model: str,
        api_key: Optional[str] = None,
        batch_size: int = 50,
        api_url: Optional[str] = None,
    ):
        """
        Initialize the EmbeddingFunction.

        Args:
            model (str): The model to use for embedding.
            api_key (Optional[str]): The API key for the Hugging Face API. If not provided, 
                it will be fetched from the environment variable `HF_API_KEY`.
            batch_size (int): The number of texts to process in a single batch. Default is 50.
            api_url (Optional[str]): Custom API URL for Hugging Face inference endpoint.
        """
  

def sync_embed(texts: str, model: str, api_key: str) -> list:
    """
    Extrait les embeddings d'un texte via l'API Inference de Hugging Face.

    Args:
        texts (str): Le texte à encoder.
        model (str): Le modèle Hugging Face à utiliser.
        api_key (str): La clé API Hugging Face.

    Returns:
        list: Les embeddings du texte.
    """
    client = InferenceClient(provider="hf-inference", api_key=api_key)
    result = client.feature_extraction(texts, model=model)
    return result[0]  # Retourne le premier embedding