validate_data.py

def validate_inputs(snippets):
    """数据验证工具"""
    errors = []
    valid_data = []
    
    for idx, s in enumerate(snippets):
        # 类型检查
        if not isinstance(s, str):
            errors.append(f"索引 {idx}: 类型错误 ({type(s)}) → 已转换")
            s = str(s)
            
        # 清理数据
        cleaned = s.replace("...", "").strip()
        
        # 有效性检查
        if len(cleaned) < 10:  # 最小代码长度
            errors.append(f"索引 {idx}: 代码过短 ({len(cleaned)} 字符)")
            continue
            
        valid_data.append(cleaned)
    
    # 输出报告
    print(f"验证结果：\n成功 {len(valid_data)} 条\n错误 {len(errors)} 条")
    for err in errors[:3]:  # 显示前3个错误
        print(f"  - {err}")
    if len(errors) > 3:
        print(f"  ...（共 {len(errors)} 个错误）")
    
    return valid_data

if __name__ == "__main__":
    # 测试数据
    test_data = [
        "def example(): pass",
        12345,  # 错误类型
        "print(...)",  # 包含...
        "   ",  # 空字符串
        """def valid_func():
            return '正确代码'"""
    ]
    
    validated = validate_inputs(test_data)
    print("\n有效数据示例：")
    for i, data in enumerate(validated[:2]):
        print(f"[{i}] {data[:50]}...")