core/base_processor.py

from abc import ABC, abstractmethod
from typing import List, Tuple, Optional, Generator
from pathlib import Path
from core.base_translator import BaseTranslator
from core.exceptions import ProcessorError
import os

class DocumentProcessor(ABC):
    """Abstract base class for document processors"""
    
    def __init__(self, translator: BaseTranslator):
        self.translator = translator
    
    @abstractmethod
    def extract_text_elements(self, file_path: Path) -> Generator[Tuple[str, dict], None, None]:
        """
        Extract text elements from document
        
        Args:
            file_path: Path to the document
            
        Yields:
            Tuple of (text_content, metadata) for each translatable element
        """
        pass
    
    @abstractmethod
    def apply_translations(self, file_path: Path, translations: List[Tuple[str, dict]]) -> Path:
        """
        Apply translations back to the document
        
        Args:
            file_path: Path to original document
            translations: List of (translated_text, metadata) tuples
            
        Returns:
            Path to the translated document
        """
        pass
    
    def process_document(
        self, 
        file_path: Path, 
        source_lang: str, 
        target_lang: str,
        progress_callback: Optional[callable] = None
    ) -> Tuple[Path, str]:
        """
        Process entire document translation
        
        Args:
            file_path: Path to document
            source_lang: Source language
            target_lang: Target language  
            progress_callback: Optional progress callback
            
        Returns:
            Tuple of (output_file_path, summary_text)
        """
        try:
            # Extract text elements
            text_elements = list(self.extract_text_elements(file_path))
            total_elements = len(text_elements)
            
            if total_elements == 0:
                raise ProcessorError("No translatable text found in document")
            
            # Translate each element
            translations = []
            all_translated_text = ""
            
            for i, (text, metadata) in enumerate(text_elements):
                if text.strip():  # Only translate non-empty text
                    translated = self.translator.translate_text(text, source_lang, target_lang)
                    translations.append((translated, metadata))
                    all_translated_text += translated + "\n"
                else:
                    translations.append((text, metadata))  # Keep empty text as-is
                
                # Update progress
                if progress_callback:
                    progress_callback((i + 1) / total_elements, f"Translating element {i + 1}/{total_elements}")
            
            # Apply translations to document
            output_path = self.apply_translations(file_path, translations)
            
            return output_path, all_translated_text
            
        except Exception as e:
            raise ProcessorError(f"Document processing failed: {str(e)}")
    
    def generate_output_path(self, original_path: Path, suffix: str = "translated") -> Path:
        """
        Generate output file path
        
        Args:
            original_path: Original file path
            suffix: Suffix to add to filename
            
        Returns:
            New file path with suffix
        """
        stem = original_path.stem
        extension = original_path.suffix
        directory = original_path.parent
        
        return directory / f"{stem}_{suffix}{extension}"
    
    @property
    @abstractmethod
    def supported_extensions(self) -> List[str]:
        """Return list of supported file extensions"""
        pass