llm_ticket3/utils/translate_utils.py

# utils/translate_utils.py

from deep_translator import GoogleTranslator
import json
import os
from datetime import datetime
import logging

logger = logging.getLogger("Translate")

def fr_to_en(text: str) -> str:
    try:
        if not text.strip():
            return ""
        return GoogleTranslator(source="fr", target="en").translate(text)
    except Exception as e:
        logger.error(f"Traduction FR->EN échouée: {e}")
        return ""

def en_to_fr(text: str) -> str:
    try:
        if not text.strip():
            return ""
        return GoogleTranslator(source="en", target="fr").translate(text)
    except Exception as e:
        logger.error(f"Traduction EN->FR échouée: {e}")
        return ""

def sauvegarder_ocr_traduction(
    image_path: str,
    ticket_id: str,
    ocr_fr: str,
    ocr_en: str,
    ocr_en_back_fr: str = "",  # <- Ajout facultatif
    base_dir: str = "reports"
) -> None:
    """
    Sauvegarde les résultats OCR + TRAD en JSON + ajoute une ligne dans le fichier texte global.
    Inclut éventuellement une traduction EN → FR.
    """
    try:
        image_name = os.path.basename(image_path)
        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
        rapport_dir = os.path.join(base_dir, ticket_id, "pipeline", "ocr_traduction")
        os.makedirs(rapport_dir, exist_ok=True)

        result = {
            "image_name": image_name,
            "ocr_fr": ocr_fr,
            "translation_en": ocr_en,
            "translation_en_back_fr": ocr_en_back_fr,
            "metadata": {
                "ticket_id": ticket_id,
                "timestamp": timestamp,
                "source_module": "ocr_utils + translate_utils",
                "lang_detected": "fr"
            }
        }

        # Fichier JSON par image
        with open(os.path.join(rapport_dir, f"{image_name}.json"), "w", encoding="utf-8") as f:
            json.dump(result, f, ensure_ascii=False, indent=2)

        # Append texte global
        ligne = (
            f"{image_name}\n"
            f"[FR] {ocr_fr or '_'}\n"
            f"[EN] {ocr_en or '_'}\n"
            f"[EN→FR] {ocr_en_back_fr or '_'}\n\n"
        )
        with open(os.path.join(rapport_dir, "ocr_traduction.txt"), "a", encoding="utf-8") as f:
            f.write(ligne)

    except Exception as e:
        logger.error(f"Erreur sauvegarde OCR+TRAD pour {image_path}: {e}")