Function prepare_task_info

RealChart2Code_eval/evaluate_task2.py:788–868

(task_file: Path, max_data_rows: int)

def prepare_task_info(task_file: Path, max_data_rows: int) -> Optional[Dict]:
    
    try:
        
        category, instruction, data_filenames = parse_task_file(task_file)
        
        if not instruction:
            print(f"⚠️  Skip task {task_file}: missing instruction content")
            return None
        
        if not data_filenames:
            print(f"⚠️  Skip task {task_file}: missing data file list")
            return None
        
        data_directory = task_file.parent
        data_files = find_data_files(data_directory, data_filenames)
        
        if not data_files:
            print(f"⚠️  Skip task {task_file}: data files not found {data_filenames} in directory {data_directory}")
            return None
        
        
        missing_files = set(data_filenames) - set(data_files.keys())
        if missing_files:
            print(f"⚠️  Skip task {task_file}: missing data files {missing_files}")
            return None
        
        
        data_info = {}
        for filename, filepath in data_files.items():
            try:
                data_info[filename] = read_data_file(filepath, max_data_rows)
            except Exception as e:
                print(f"⚠️  Skip task {task_file}: read data file {filename} Failed: {e}")
                return None
        
        difficulty = task_file.stem.split('_')[-1] if not task_file.stem.endswith('_mul') else task_file.stem.split('_')[-2]
        is_multi = task_file.stem.endswith('_mul')
        
        image_path=None
        refine_folders = [f for f in data_directory.glob("refine*") if f.is_dir()]
        if refine_folders:
            for refine_folder in refine_folders:
                if refine_folder.is_dir():
                    if not is_multi:
                        png_files = list(refine_folder.glob(f"*{difficulty}.png"))
                        if png_files:
                            image_path = png_files[0]
                    else:
                        png_files = list(refine_folder.glob(f"*{difficulty}_mul.png"))
                        if png_files:
                            image_path = png_files[0]
        if not image_path:
            if not is_multi:
                png_files = list(data_directory.glob(f"*{difficulty}.png"))
                if png_files:
                    image_path = png_files[0]
            else: