from tools.interfaces.interface_audio_provider import AudioTranscriptionProvider
import whisper
import requests
import tempfile
import os

class WhisperAudioTranscription(AudioTranscriptionProvider):
    def __init__(self, huggingface_token: str, base_url: str,is_local=False):
        self.huggingface_token = huggingface_token
        self.base_url = base_url
        self.supported_formats = ['.mp3']
        self.model = whisper.load_model("base")  
        self.is_local=is_local

    def transcribe(self, audio_path_or_url: str) -> str:
        try:
            # Vérifie le format
            if not any(audio_path_or_url.lower().endswith(ext) for ext in self.supported_formats):
                return f"Format non supporté. Formats acceptés : {', '.join(self.supported_formats)}"

            # Si fichier local
            if self.is_local:
                if os.path.exists(audio_path_or_url):
                    tmp_path = audio_path_or_url
            else:
                # Sinon, on télécharge depuis Hugging Face
                headers = {"Authorization": f"Bearer {self.huggingface_token}"}
                response = requests.get(self.base_url + audio_path_or_url, headers=headers)
                response.raise_for_status()

                with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as tmp:
                    tmp.write(response.content)
                    tmp_path = tmp.name

            # Transcription
            result = self.model.transcribe(tmp_path)

            # Supprime le fichier temporaire si on l’a téléchargé
            if not os.path.exists(audio_path_or_url):
                os.remove(tmp_path)

            return result["text"]

        except Exception as e:
            return f"Erreur de transcription : {str(e)}"