Function diagnose_task_file_issues

RealChart2Code_eval/evaluate_task2.py:137–194

(task_file: Path)

def diagnose_task_file_issues(task_file: Path) -> Dict[str, any]:
    
    diagnosis = {
        'task_file': str(task_file),
        'exists': task_file.exists(),
        'readable': False,
        'has_category': False,
        'has_instruction': False,
        'has_files': False,
        'data_filenames': [],
        'data_directory': str(task_file.parent),
        'found_data_files': {},
        'missing_data_files': [],
        'errors': []
    }
    
    try:
        if not task_file.exists():
            diagnosis['errors'].append("task file does not exist")
            return diagnosis
        
        
        with open(task_file, 'r', encoding='utf-8') as f:
            content = f.read()
        diagnosis['readable'] = True
        
        
        category_match = re.search(r'## Category\s*\n(.+)', content)
        if category_match:
            diagnosis['has_category'] = True
        
        instruction_match = re.search(r'## Instruction\s*\n(.+?)(?=\n##|\n---|$)', content, re.DOTALL)
        if instruction_match and instruction_match.group(1).strip():
            diagnosis['has_instruction'] = True
        
        files_match = re.search(r'## Files\s*\n(.+?)(?=\n##|\n---|$)', content, re.DOTALL)
        if files_match:
            files_text = files_match.group(1).strip()
            if files_text:
                diagnosis['has_files'] = True
                data_files = []
                for f in files_text.split('\n'):
                    f = f.strip()
                    if f and (f.endswith('.csv') or f.endswith('.xlsx')):
                        data_files.append(f)
                diagnosis['data_filenames'] = data_files
        
        
        if diagnosis['data_filenames']:
            data_directory = task_file.parent
            found_files = find_data_files(data_directory, diagnosis['data_filenames'])
            diagnosis['found_data_files'] = {k: str(v) for k, v in found_files.items()}
            diagnosis['missing_data_files'] = list(set(diagnosis['data_filenames']) - set(found_files.keys()))
        
    except Exception as e:
        diagnosis['errors'].append(f"error reading file: {str(e)}")
    
    return diagnosis