validate_data.py

#!/usr/bin/env python3
"""
验证Python错误数据JSON文件的完整性和格式
"""

import json
from typing import Dict, List, Set

def load_json_file(filename: str) -> Dict:
    """加载JSON文件"""
    try:
        with open(filename, 'r', encoding='utf-8') as f:
            return json.load(f)
    except FileNotFoundError:
        print(f"错误：找不到文件 {filename}")
        return {}
    except json.JSONDecodeError as e:
        print(f"错误：JSON格式无效 - {e}")
        return {}

def validate_json_format(data: Dict) -> bool:
    """验证JSON基本格式"""
    required_keys = ['version', 'categories', 'exceptions']

    for key in required_keys:
        if key not in data:
            print(f"✗ 缺少必需字段: {key}")
            return False

    print("✓ JSON格式验证通过")
    return True

def validate_data_integrity(data: Dict) -> bool:
    """验证数据完整性"""
    issues = []
    
    # 检查分类ID的唯一性
    category_ids = [cat['id'] for cat in data.get('categories', [])]
    if len(category_ids) != len(set(category_ids)):
        issues.append("分类ID存在重复")
    
    # 检查异常ID的唯一性
    exception_ids = [exc['id'] for exc in data.get('exceptions', [])]
    if len(exception_ids) != len(set(exception_ids)):
        issues.append("异常ID存在重复")
    
    # 检查异常的category_id是否都存在
    valid_category_ids = set(category_ids)
    for exc in data.get('exceptions', []):
        if exc.get('category_id') not in valid_category_ids:
            issues.append(f"异常 {exc.get('id')} 的category_id '{exc.get('category_id')}' 不存在")
    
    # 检查每个异常是否有基本信息
    for exc in data.get('exceptions', []):
        exc_id = exc.get('id', 'unknown')
        if not exc.get('exception_name'):
            issues.append(f"异常 {exc_id} 缺少exception_name")
        if not exc.get('description_en'):
            issues.append(f"异常 {exc_id} 缺少英文描述")
        if not exc.get('description_zh'):
            issues.append(f"异常 {exc_id} 缺少中文描述")
        if not exc.get('common_messages'):
            issues.append(f"异常 {exc_id} 缺少常见错误消息")
    
    if issues:
        print("✗ 数据完整性检查发现问题:")
        for issue in issues:
            print(f"  - {issue}")
        return False
    else:
        print("✓ 数据完整性检查通过")
        return True

def generate_statistics(data: Dict) -> None:
    """生成统计信息"""
    print("\n=== 数据统计 ===")
    print(f"分类数量: {len(data.get('categories', []))}")
    print(f"异常数量: {len(data.get('exceptions', []))}")
    
    # 按分类统计异常数量
    category_counts = {}
    for exc in data.get('exceptions', []):
        cat_id = exc.get('category_id')
        category_counts[cat_id] = category_counts.get(cat_id, 0) + 1
    
    print("\n按分类统计:")
    categories = {cat['id']: cat['name_zh'] for cat in data.get('categories', [])}
    for cat_id, count in category_counts.items():
        cat_name = categories.get(cat_id, cat_id)
        print(f"  {cat_name}: {count} 个异常")
    
    # 统计错误消息数量
    total_messages = sum(len(exc.get('common_messages', [])) for exc in data.get('exceptions', []))
    print(f"\n总错误消息数量: {total_messages}")

def main():
    """主函数"""
    print("Python错误数据验证工具")
    print("=" * 30)
    
    # 加载数据文件
    data = load_json_file('python_errors_data.json')
    if not data:
        return

    # 验证JSON格式
    validate_json_format(data)
    
    # 验证数据完整性
    validate_data_integrity(data)
    
    # 生成统计信息
    generate_statistics(data)

if __name__ == "__main__":
    main()