|
| 1 | +#!/usr/bin/env python3 |
| 2 | +# -*- coding: utf-8 -*- |
| 3 | +""" |
| 4 | +JavaScript源文件对比工具 |
| 5 | +功能:对比js和js_dr2目录中的文件,找出js_dr2_old目录中的重复文件并提供删除功能 |
| 6 | +作者:AI Assistant |
| 7 | +创建时间:2025年1月 |
| 8 | +""" |
| 9 | + |
| 10 | +import argparse |
| 11 | +import re |
| 12 | +import sys |
| 13 | +from datetime import datetime |
| 14 | +from pathlib import Path |
| 15 | +from typing import Set, List, Dict |
| 16 | + |
| 17 | + |
| 18 | +class SourceFileComparator: |
| 19 | + """JavaScript源文件对比器""" |
| 20 | + |
| 21 | + def __init__(self, base_dir: str = None): |
| 22 | + """ |
| 23 | + 初始化对比器 |
| 24 | + |
| 25 | + Args: |
| 26 | + base_dir: 项目根目录,默认为当前脚本所在项目根目录 |
| 27 | + """ |
| 28 | + if base_dir is None: |
| 29 | + # 获取项目根目录(脚本在scripts/python目录下) |
| 30 | + script_dir = Path(__file__).parent |
| 31 | + self.base_dir = script_dir.parent.parent |
| 32 | + else: |
| 33 | + self.base_dir = Path(base_dir) |
| 34 | + |
| 35 | + # 定义三个目录路径 |
| 36 | + self.js_dir = self.base_dir / "spider" / "js" |
| 37 | + self.js_dr2_dir = self.base_dir / "spider" / "js_dr2" |
| 38 | + self.js_dr2_old_dir = self.base_dir / "spider" / "js_dr2_old" |
| 39 | + |
| 40 | + # 存储文件信息 |
| 41 | + self.js_files: Set[str] = set() |
| 42 | + self.js_dr2_files: Set[str] = set() |
| 43 | + self.js_dr2_old_files: Set[str] = set() |
| 44 | + self.duplicate_files: List[str] = [] |
| 45 | + |
| 46 | + # 存储标准化文件名映射 |
| 47 | + self.js_normalized_map: Dict[str, str] = {} |
| 48 | + self.js_dr2_normalized_map: Dict[str, str] = {} |
| 49 | + self.js_dr2_old_normalized_map: Dict[str, str] = {} |
| 50 | + |
| 51 | + def _normalize_filename(self, filename: str) -> str: |
| 52 | + """ |
| 53 | + 标准化文件名,去除中括号及其内容 |
| 54 | + |
| 55 | + Args: |
| 56 | + filename: 原始文件名 |
| 57 | + |
| 58 | + Returns: |
| 59 | + str: 标准化后的文件名 |
| 60 | + """ |
| 61 | + # 去除中括号及其内容,例如:可可影视[优].js -> 可可影视.js |
| 62 | + normalized = re.sub(r'\[.*?\]', '', filename) |
| 63 | + return normalized |
| 64 | + |
| 65 | + def _is_valid_js_file(self, filename: str) -> bool: |
| 66 | + """ |
| 67 | + 检查是否为有效的JavaScript文件(非_开头的.js文件) |
| 68 | + |
| 69 | + Args: |
| 70 | + filename: 文件名 |
| 71 | + |
| 72 | + Returns: |
| 73 | + bool: 是否为有效的JavaScript文件 |
| 74 | + """ |
| 75 | + return (filename.endswith('.js') and |
| 76 | + not filename.startswith('_') and |
| 77 | + not filename.startswith('.')) |
| 78 | + |
| 79 | + def _scan_directory(self, directory: Path, normalized_map: Dict[str, str] = None) -> Set[str]: |
| 80 | + """ |
| 81 | + 扫描目录中的有效JavaScript文件 |
| 82 | + |
| 83 | + Args: |
| 84 | + directory: 要扫描的目录路径 |
| 85 | + normalized_map: 用于存储标准化文件名映射的字典 |
| 86 | + |
| 87 | + Returns: |
| 88 | + Set[str]: 文件名集合 |
| 89 | + """ |
| 90 | + files = set() |
| 91 | + if not directory.exists(): |
| 92 | + print(f"警告:目录不存在 - {directory}") |
| 93 | + return files |
| 94 | + |
| 95 | + try: |
| 96 | + for file_path in directory.iterdir(): |
| 97 | + if file_path.is_file() and self._is_valid_js_file(file_path.name): |
| 98 | + files.add(file_path.name) |
| 99 | + # 如果提供了映射字典,则建立标准化文件名映射 |
| 100 | + if normalized_map is not None: |
| 101 | + normalized_name = self._normalize_filename(file_path.name) |
| 102 | + normalized_map[normalized_name] = file_path.name |
| 103 | + except PermissionError: |
| 104 | + print(f"错误:没有权限访问目录 - {directory}") |
| 105 | + except Exception as e: |
| 106 | + print(f"错误:扫描目录时发生异常 - {directory}: {e}") |
| 107 | + |
| 108 | + return files |
| 109 | + |
| 110 | + def scan_all_directories(self): |
| 111 | + """扫描所有目录中的JavaScript文件""" |
| 112 | + print("正在扫描目录...") |
| 113 | + print(f"项目根目录: {self.base_dir}") |
| 114 | + |
| 115 | + # 扫描js目录 |
| 116 | + print(f"扫描目录: {self.js_dir}") |
| 117 | + self.js_files = self._scan_directory(self.js_dir, self.js_normalized_map) |
| 118 | + print(f"找到 {len(self.js_files)} 个有效JS文件") |
| 119 | + |
| 120 | + # 扫描js_dr2目录 |
| 121 | + print(f"扫描目录: {self.js_dr2_dir}") |
| 122 | + self.js_dr2_files = self._scan_directory(self.js_dr2_dir, self.js_dr2_normalized_map) |
| 123 | + print(f"找到 {len(self.js_dr2_files)} 个有效JS文件") |
| 124 | + |
| 125 | + # 扫描js_dr2_old目录 |
| 126 | + print(f"扫描目录: {self.js_dr2_old_dir}") |
| 127 | + self.js_dr2_old_files = self._scan_directory(self.js_dr2_old_dir, self.js_dr2_old_normalized_map) |
| 128 | + print(f"找到 {len(self.js_dr2_old_files)} 个有效JS文件") |
| 129 | + |
| 130 | + def find_duplicates(self): |
| 131 | + """查找重复文件(基于标准化文件名)""" |
| 132 | + print("\n正在分析重复文件...") |
| 133 | + |
| 134 | + # 合并js和js_dr2目录的标准化文件名 |
| 135 | + source_normalized_names = set(self.js_normalized_map.keys()).union(set(self.js_dr2_normalized_map.keys())) |
| 136 | + |
| 137 | + # 查找js_dr2_old中与源目录重复的文件(基于标准化文件名) |
| 138 | + self.duplicate_files = [] |
| 139 | + for normalized_name, original_name in self.js_dr2_old_normalized_map.items(): |
| 140 | + if normalized_name in source_normalized_names: |
| 141 | + self.duplicate_files.append(original_name) |
| 142 | + |
| 143 | + print(f"发现 {len(self.duplicate_files)} 个重复文件(基于标准化文件名匹配)") |
| 144 | + |
| 145 | + def generate_report(self) -> Dict: |
| 146 | + """ |
| 147 | + 生成详细报告 |
| 148 | + |
| 149 | + Returns: |
| 150 | + Dict: 包含统计信息的报告 |
| 151 | + """ |
| 152 | + report = { |
| 153 | + 'scan_time': datetime.now().strftime('%Y-%m-%d %H:%M:%S'), |
| 154 | + 'directories': { |
| 155 | + 'js': str(self.js_dir), |
| 156 | + 'js_dr2': str(self.js_dr2_dir), |
| 157 | + 'js_dr2_old': str(self.js_dr2_old_dir) |
| 158 | + }, |
| 159 | + 'file_counts': { |
| 160 | + 'js': len(self.js_files), |
| 161 | + 'js_dr2': len(self.js_dr2_files), |
| 162 | + 'js_dr2_old': len(self.js_dr2_old_files), |
| 163 | + 'duplicates': len(self.duplicate_files) |
| 164 | + }, |
| 165 | + 'duplicate_files': sorted(self.duplicate_files), |
| 166 | + 'unique_files_in_old': sorted(list(self.js_dr2_old_files - set(self.duplicate_files))) |
| 167 | + } |
| 168 | + return report |
| 169 | + |
| 170 | + def print_report(self): |
| 171 | + """打印详细报告""" |
| 172 | + report = self.generate_report() |
| 173 | + |
| 174 | + print("\n" + "=" * 60) |
| 175 | + print("JavaScript文件对比报告") |
| 176 | + print("=" * 60) |
| 177 | + print(f"扫描时间: {report['scan_time']}") |
| 178 | + print(f"项目根目录: {self.base_dir}") |
| 179 | + |
| 180 | + print(f"\n目录统计:") |
| 181 | + print(f" js目录: {report['file_counts']['js']} 个文件") |
| 182 | + print(f" js_dr2目录: {report['file_counts']['js_dr2']} 个文件") |
| 183 | + print(f" js_dr2_old目录: {report['file_counts']['js_dr2_old']} 个文件") |
| 184 | + |
| 185 | + print(f"\n重复文件分析:") |
| 186 | + print(f" 重复文件数量: {report['file_counts']['duplicates']} 个") |
| 187 | + print(f" js_dr2_old中唯一文件: {len(report['unique_files_in_old'])} 个") |
| 188 | + |
| 189 | + if self.duplicate_files: |
| 190 | + print(f"\n重复文件清单:") |
| 191 | + for i, filename in enumerate(report['duplicate_files'], 1): |
| 192 | + print(f" {i:2d}. {filename}") |
| 193 | + |
| 194 | + if report['unique_files_in_old']: |
| 195 | + print(f"\njs_dr2_old中的唯一文件(不会被删除):") |
| 196 | + for i, filename in enumerate(report['unique_files_in_old'], 1): |
| 197 | + print(f" {i:2d}. {filename}") |
| 198 | + |
| 199 | + def delete_duplicates(self, confirm: bool = False) -> bool: |
| 200 | + """ |
| 201 | + 删除重复文件 |
| 202 | + |
| 203 | + Args: |
| 204 | + confirm: 是否已确认删除 |
| 205 | + |
| 206 | + Returns: |
| 207 | + bool: 删除是否成功 |
| 208 | + """ |
| 209 | + if not self.duplicate_files: |
| 210 | + print("没有发现重复文件,无需删除。") |
| 211 | + return True |
| 212 | + |
| 213 | + if not confirm: |
| 214 | + print(f"\n即将删除 {len(self.duplicate_files)} 个重复文件:") |
| 215 | + for filename in self.duplicate_files: |
| 216 | + print(f" - {filename}") |
| 217 | + |
| 218 | + response = input(f"\n确认删除这些文件吗?(y/N): ").strip().lower() |
| 219 | + if response not in ['y', 'yes', '是']: |
| 220 | + print("取消删除操作。") |
| 221 | + return False |
| 222 | + |
| 223 | + # 执行删除 |
| 224 | + deleted_count = 0 |
| 225 | + failed_files = [] |
| 226 | + |
| 227 | + for filename in self.duplicate_files: |
| 228 | + file_path = self.js_dr2_old_dir / filename |
| 229 | + try: |
| 230 | + if file_path.exists(): |
| 231 | + file_path.unlink() |
| 232 | + deleted_count += 1 |
| 233 | + print(f"已删除: {filename}") |
| 234 | + else: |
| 235 | + print(f"文件不存在: {filename}") |
| 236 | + except Exception as e: |
| 237 | + failed_files.append((filename, str(e))) |
| 238 | + print(f"删除失败: {filename} - {e}") |
| 239 | + |
| 240 | + print(f"\n删除完成:") |
| 241 | + print(f" 成功删除: {deleted_count} 个文件") |
| 242 | + if failed_files: |
| 243 | + print(f" 删除失败: {len(failed_files)} 个文件") |
| 244 | + for filename, error in failed_files: |
| 245 | + print(f" - {filename}: {error}") |
| 246 | + |
| 247 | + return len(failed_files) == 0 |
| 248 | + |
| 249 | + |
| 250 | +def main(): |
| 251 | + """主函数""" |
| 252 | + parser = argparse.ArgumentParser( |
| 253 | + description="JavaScript源文件对比工具", |
| 254 | + formatter_class=argparse.RawDescriptionHelpFormatter, |
| 255 | + epilog=""" |
| 256 | +使用示例: |
| 257 | + python compare_source.py # 扫描并显示报告 |
| 258 | + python compare_source.py --delete # 扫描并交互式删除重复文件 |
| 259 | + python compare_source.py --delete --force # 扫描并强制删除重复文件(无确认) |
| 260 | + python compare_source.py --base-dir /path # 指定项目根目录 |
| 261 | + """ |
| 262 | + ) |
| 263 | + |
| 264 | + parser.add_argument( |
| 265 | + '--base-dir', |
| 266 | + type=str, |
| 267 | + help='项目根目录路径(默认为脚本所在项目根目录)' |
| 268 | + ) |
| 269 | + parser.add_argument( |
| 270 | + '--delete', |
| 271 | + action='store_true', |
| 272 | + help='删除重复文件' |
| 273 | + ) |
| 274 | + parser.add_argument( |
| 275 | + '--force', |
| 276 | + action='store_true', |
| 277 | + help='强制删除,不需要确认(需要与--delete一起使用)' |
| 278 | + ) |
| 279 | + parser.add_argument( |
| 280 | + '--quiet', |
| 281 | + action='store_true', |
| 282 | + help='静默模式,只显示关键信息' |
| 283 | + ) |
| 284 | + |
| 285 | + args = parser.parse_args() |
| 286 | + |
| 287 | + try: |
| 288 | + # 创建对比器实例 |
| 289 | + comparator = SourceFileComparator(args.base_dir) |
| 290 | + |
| 291 | + # 扫描目录 |
| 292 | + if not args.quiet: |
| 293 | + comparator.scan_all_directories() |
| 294 | + else: |
| 295 | + comparator.js_files = comparator._scan_directory(comparator.js_dir) |
| 296 | + comparator.js_dr2_files = comparator._scan_directory(comparator.js_dr2_dir) |
| 297 | + comparator.js_dr2_old_files = comparator._scan_directory(comparator.js_dr2_old_dir) |
| 298 | + |
| 299 | + # 查找重复文件 |
| 300 | + comparator.find_duplicates() |
| 301 | + |
| 302 | + # 显示报告 |
| 303 | + if not args.quiet: |
| 304 | + comparator.print_report() |
| 305 | + else: |
| 306 | + print(f"发现 {len(comparator.duplicate_files)} 个重复文件") |
| 307 | + |
| 308 | + # 删除重复文件 |
| 309 | + if args.delete: |
| 310 | + success = comparator.delete_duplicates(confirm=args.force) |
| 311 | + if success: |
| 312 | + print("\n重复文件清理完成!") |
| 313 | + return 0 |
| 314 | + else: |
| 315 | + print("\n重复文件清理失败!") |
| 316 | + return 1 |
| 317 | + else: |
| 318 | + if comparator.duplicate_files: |
| 319 | + print(f"\n提示:使用 --delete 参数可以删除这 {len(comparator.duplicate_files)} 个重复文件") |
| 320 | + |
| 321 | + return 0 |
| 322 | + |
| 323 | + except KeyboardInterrupt: |
| 324 | + print("\n\n操作被用户中断。") |
| 325 | + return 1 |
| 326 | + except Exception as e: |
| 327 | + print(f"\n错误:{e}") |
| 328 | + return 1 |
| 329 | + |
| 330 | + |
| 331 | +if __name__ == "__main__": |
| 332 | + sys.exit(main()) |
0 commit comments