Function perform_ocr_task

api/main.py:561–629

Perform OCR task on uploaded file

(file: UploadFile, task_type: str)

    await asyncio.get_event_loop().run_in_executor(None, create_zip_sync)

async def perform_ocr_task(file: UploadFile, task_type: str) -> TaskResponse:
    """Perform OCR task on uploaded file"""
    try:
        monkey_ocr_model = model_manager.get_model()
        supports_async = model_manager.get_async_support()
        model_lock = model_manager.get_model_lock()
        
        if not monkey_ocr_model:
            raise HTTPException(status_code=500, detail="Model not initialized")
        
        # Validate file type
        allowed_extensions = {'.pdf', '.jpg', '.jpeg', '.png'}
        file_ext = Path(file.filename).suffix.lower()
        if file_ext not in allowed_extensions:
            raise HTTPException(
                status_code=400, 
                detail=f"Unsupported file type: {file_ext}. Allowed: {', '.join(allowed_extensions)}"
            )
        
        # Save uploaded file temporarily with unique name
        import uuid
        unique_suffix = str(uuid.uuid4())[:8]
        
        with tempfile.NamedTemporaryFile(delete=False, suffix=file_ext, prefix=f"ocr_{unique_suffix}_") as temp_file:
            content = await file.read()
            temp_file.write(content)
            temp_file_path = temp_file.name
        
        try:
            # Create output directory with unique name
            output_dir = tempfile.mkdtemp(prefix=f"monkeyocr_{task_type}_{unique_suffix}_")
            
            # Use optimized async single task recognition
            result_dir = await async_single_task_recognition(temp_file_path, output_dir, task_type)
            
            # Read result file
            def read_result_sync():
                result_files = [f for f in os.listdir(result_dir) if f.endswith(f'_{task_type}_result.md')]
                if not result_files:
                    raise Exception("No result file generated")
                
                result_file_path = os.path.join(result_dir, result_files[0])
                with open(result_file_path, 'r', encoding='utf-8') as f:
                    return f.read()
            
            content = await asyncio.get_event_loop().run_in_executor(None, read_result_sync)
            
            return TaskResponse(
                success=True,
                task_type=task_type,
                content=content,
                message=f"{task_type.capitalize()} extraction completed successfully"
            )
            
        finally:
            # Clean up temporary file
            try:
                os.unlink(temp_file_path)