utils/translation.py

"""
Text Translation Module using NLLB-3.3B model
Handles text segmentation and batch translation
"""

import logging
logger = logging.getLogger(__name__)

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

def translate_text(text):
    """
    Translate English text to Simplified Chinese
    Args:
        text: Input English text
    Returns:
        Translated Chinese text
    """
    logger.info(f"Starting translation for text length: {len(text)}")
    
    try:
        # Model initialization
        logger.info("Loading NLLB model")
        tokenizer = AutoTokenizer.from_pretrained("facebook/nllb-200-3.3B")
        model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-3.3B")
        logger.debug("Translation model loaded")

        # Text processing
        max_chunk_length = 1000
        text_chunks = [text[i:i+max_chunk_length] for i in range(0, len(text), max_chunk_length)]
        logger.info(f"Split text into {len(text_chunks)} chunks")

        translated_chunks = []
        for i, chunk in enumerate(text_chunks):
            logger.debug(f"Processing chunk {i+1}/{len(text_chunks)}")
            inputs = tokenizer(
                chunk, 
                return_tensors="pt", 
                max_length=1024, 
                truncation=True
            )
            
            outputs = model.generate(
                **inputs,
                forced_bos_token_id=tokenizer.lang_code_to_id["zho_Hans"],
                max_new_tokens=1024
            )
            translated = tokenizer.decode(outputs[0], skip_special_tokens=True)
            translated_chunks.append(translated)
            logger.debug(f"Chunk {i+1} translated successfully")

        result = "".join(translated_chunks)
        logger.info(f"Translation completed. Total length: {len(result)}")
        return result

    except Exception as e:
        logger.error(f"Translation failed: {str(e)}", exc_info=True)
        raise