crawlo 1.4.7__py3-none-any.whl → 1.4.8__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.


This version of crawlo might be problematic. Click here for more details.

Files changed (348) hide show
  1. crawlo/__init__.py +90 -90
  2. crawlo/__version__.py +1 -1
  3. crawlo/cli.py +75 -75
  4. crawlo/commands/__init__.py +14 -14
  5. crawlo/commands/check.py +594 -594
  6. crawlo/commands/genspider.py +186 -186
  7. crawlo/commands/help.py +140 -140
  8. crawlo/commands/list.py +155 -155
  9. crawlo/commands/run.py +379 -379
  10. crawlo/commands/startproject.py +460 -460
  11. crawlo/commands/stats.py +187 -187
  12. crawlo/commands/utils.py +196 -196
  13. crawlo/config.py +320 -320
  14. crawlo/config_validator.py +277 -277
  15. crawlo/core/__init__.py +52 -52
  16. crawlo/core/engine.py +451 -451
  17. crawlo/core/processor.py +47 -47
  18. crawlo/core/scheduler.py +290 -290
  19. crawlo/crawler.py +698 -698
  20. crawlo/data/__init__.py +5 -5
  21. crawlo/data/user_agents.py +194 -194
  22. crawlo/downloader/__init__.py +280 -280
  23. crawlo/downloader/aiohttp_downloader.py +233 -233
  24. crawlo/downloader/cffi_downloader.py +250 -250
  25. crawlo/downloader/httpx_downloader.py +265 -265
  26. crawlo/downloader/hybrid_downloader.py +212 -212
  27. crawlo/downloader/playwright_downloader.py +425 -425
  28. crawlo/downloader/selenium_downloader.py +486 -486
  29. crawlo/event.py +45 -45
  30. crawlo/exceptions.py +214 -214
  31. crawlo/extension/__init__.py +64 -64
  32. crawlo/extension/health_check.py +141 -141
  33. crawlo/extension/log_interval.py +94 -94
  34. crawlo/extension/log_stats.py +70 -70
  35. crawlo/extension/logging_extension.py +53 -53
  36. crawlo/extension/memory_monitor.py +104 -104
  37. crawlo/extension/performance_profiler.py +133 -133
  38. crawlo/extension/request_recorder.py +107 -107
  39. crawlo/factories/__init__.py +27 -27
  40. crawlo/factories/base.py +68 -68
  41. crawlo/factories/crawler.py +104 -104
  42. crawlo/factories/registry.py +84 -84
  43. crawlo/factories/utils.py +134 -134
  44. crawlo/filters/__init__.py +170 -170
  45. crawlo/filters/aioredis_filter.py +347 -347
  46. crawlo/filters/memory_filter.py +261 -261
  47. crawlo/framework.py +306 -306
  48. crawlo/initialization/__init__.py +44 -44
  49. crawlo/initialization/built_in.py +391 -391
  50. crawlo/initialization/context.py +141 -141
  51. crawlo/initialization/core.py +240 -240
  52. crawlo/initialization/phases.py +229 -229
  53. crawlo/initialization/registry.py +143 -143
  54. crawlo/initialization/utils.py +48 -48
  55. crawlo/interfaces.py +23 -23
  56. crawlo/items/__init__.py +23 -23
  57. crawlo/items/base.py +23 -23
  58. crawlo/items/fields.py +52 -52
  59. crawlo/items/items.py +104 -104
  60. crawlo/logging/__init__.py +42 -42
  61. crawlo/logging/config.py +280 -276
  62. crawlo/logging/factory.py +175 -175
  63. crawlo/logging/manager.py +104 -104
  64. crawlo/middleware/__init__.py +87 -87
  65. crawlo/middleware/default_header.py +132 -132
  66. crawlo/middleware/download_delay.py +104 -104
  67. crawlo/middleware/middleware_manager.py +142 -142
  68. crawlo/middleware/offsite.py +123 -123
  69. crawlo/middleware/proxy.py +209 -209
  70. crawlo/middleware/request_ignore.py +86 -86
  71. crawlo/middleware/response_code.py +150 -150
  72. crawlo/middleware/response_filter.py +136 -136
  73. crawlo/middleware/retry.py +124 -124
  74. crawlo/mode_manager.py +287 -287
  75. crawlo/network/__init__.py +21 -21
  76. crawlo/network/request.py +408 -376
  77. crawlo/network/response.py +598 -569
  78. crawlo/pipelines/__init__.py +52 -52
  79. crawlo/pipelines/base_pipeline.py +452 -452
  80. crawlo/pipelines/bloom_dedup_pipeline.py +145 -146
  81. crawlo/pipelines/console_pipeline.py +39 -39
  82. crawlo/pipelines/csv_pipeline.py +316 -316
  83. crawlo/pipelines/database_dedup_pipeline.py +196 -197
  84. crawlo/pipelines/json_pipeline.py +218 -218
  85. crawlo/pipelines/memory_dedup_pipeline.py +104 -105
  86. crawlo/pipelines/mongo_pipeline.py +140 -139
  87. crawlo/pipelines/mysql_pipeline.py +468 -469
  88. crawlo/pipelines/pipeline_manager.py +100 -100
  89. crawlo/pipelines/redis_dedup_pipeline.py +155 -155
  90. crawlo/project.py +347 -347
  91. crawlo/queue/__init__.py +9 -9
  92. crawlo/queue/pqueue.py +38 -38
  93. crawlo/queue/queue_manager.py +591 -591
  94. crawlo/queue/redis_priority_queue.py +518 -518
  95. crawlo/settings/__init__.py +7 -7
  96. crawlo/settings/default_settings.py +287 -284
  97. crawlo/settings/setting_manager.py +219 -219
  98. crawlo/spider/__init__.py +658 -657
  99. crawlo/stats_collector.py +81 -81
  100. crawlo/subscriber.py +129 -129
  101. crawlo/task_manager.py +138 -138
  102. crawlo/templates/crawlo.cfg.tmpl +10 -10
  103. crawlo/templates/project/__init__.py.tmpl +1 -1
  104. crawlo/templates/project/items.py.tmpl +13 -13
  105. crawlo/templates/project/middlewares.py.tmpl +38 -38
  106. crawlo/templates/project/pipelines.py.tmpl +35 -35
  107. crawlo/templates/project/settings.py.tmpl +113 -109
  108. crawlo/templates/project/settings_distributed.py.tmpl +160 -156
  109. crawlo/templates/project/settings_gentle.py.tmpl +174 -170
  110. crawlo/templates/project/settings_high_performance.py.tmpl +175 -171
  111. crawlo/templates/project/settings_minimal.py.tmpl +102 -98
  112. crawlo/templates/project/settings_simple.py.tmpl +172 -168
  113. crawlo/templates/project/spiders/__init__.py.tmpl +9 -9
  114. crawlo/templates/run.py.tmpl +23 -23
  115. crawlo/templates/spider/spider.py.tmpl +32 -32
  116. crawlo/templates/spiders_init.py.tmpl +4 -4
  117. crawlo/tools/__init__.py +86 -86
  118. crawlo/tools/date_tools.py +289 -289
  119. crawlo/tools/distributed_coordinator.py +384 -384
  120. crawlo/tools/scenario_adapter.py +262 -262
  121. crawlo/tools/text_cleaner.py +232 -232
  122. crawlo/utils/__init__.py +74 -50
  123. crawlo/utils/batch_processor.py +276 -276
  124. crawlo/utils/config_manager.py +442 -442
  125. crawlo/utils/controlled_spider_mixin.py +439 -439
  126. crawlo/utils/db_helper.py +250 -250
  127. crawlo/utils/encoding_helper.py +190 -0
  128. crawlo/utils/error_handler.py +410 -410
  129. crawlo/utils/fingerprint.py +121 -121
  130. crawlo/utils/func_tools.py +82 -82
  131. crawlo/utils/large_scale_helper.py +344 -344
  132. crawlo/utils/leak_detector.py +335 -335
  133. crawlo/utils/misc.py +81 -81
  134. crawlo/utils/mongo_connection_pool.py +157 -157
  135. crawlo/utils/mysql_connection_pool.py +197 -197
  136. crawlo/utils/performance_monitor.py +285 -285
  137. crawlo/utils/queue_helper.py +175 -175
  138. crawlo/utils/redis_checker.py +90 -90
  139. crawlo/utils/redis_connection_pool.py +578 -578
  140. crawlo/utils/redis_key_validator.py +198 -198
  141. crawlo/utils/request.py +278 -278
  142. crawlo/utils/request_serializer.py +225 -225
  143. crawlo/utils/resource_manager.py +337 -337
  144. crawlo/utils/response_helper.py +113 -0
  145. crawlo/utils/selector_helper.py +138 -137
  146. crawlo/utils/singleton.py +69 -69
  147. crawlo/utils/spider_loader.py +201 -201
  148. crawlo/utils/text_helper.py +94 -94
  149. {crawlo-1.4.7.dist-info → crawlo-1.4.8.dist-info}/METADATA +831 -689
  150. crawlo-1.4.8.dist-info/RECORD +347 -0
  151. examples/__init__.py +7 -7
  152. tests/__init__.py +7 -7
  153. tests/advanced_tools_example.py +217 -217
  154. tests/authenticated_proxy_example.py +110 -110
  155. tests/baidu_performance_test.py +108 -108
  156. tests/baidu_test.py +59 -59
  157. tests/bug_check_test.py +250 -250
  158. tests/cleaners_example.py +160 -160
  159. tests/comprehensive_framework_test.py +212 -212
  160. tests/comprehensive_test.py +81 -81
  161. tests/comprehensive_testing_summary.md +186 -186
  162. tests/config_validation_demo.py +142 -142
  163. tests/controlled_spider_example.py +205 -205
  164. tests/date_tools_example.py +180 -180
  165. tests/debug_configure.py +69 -69
  166. tests/debug_framework_logger.py +84 -84
  167. tests/debug_log_config.py +126 -126
  168. tests/debug_log_levels.py +63 -63
  169. tests/debug_pipelines.py +66 -66
  170. tests/detailed_log_test.py +233 -233
  171. tests/direct_selector_helper_test.py +96 -96
  172. tests/distributed_dedup_test.py +467 -467
  173. tests/distributed_test.py +66 -66
  174. tests/distributed_test_debug.py +76 -76
  175. tests/dynamic_loading_example.py +523 -523
  176. tests/dynamic_loading_test.py +104 -104
  177. tests/error_handling_example.py +171 -171
  178. tests/explain_mysql_update_behavior.py +76 -76
  179. tests/final_comprehensive_test.py +151 -151
  180. tests/final_log_test.py +260 -260
  181. tests/final_validation_test.py +182 -182
  182. tests/fix_log_test.py +142 -142
  183. tests/framework_performance_test.py +202 -202
  184. tests/log_buffering_test.py +111 -111
  185. tests/log_generation_timing_test.py +153 -153
  186. tests/monitor_redis_dedup.sh +72 -72
  187. tests/ofweek_scrapy/ofweek_scrapy/items.py +12 -12
  188. tests/ofweek_scrapy/ofweek_scrapy/middlewares.py +100 -100
  189. tests/ofweek_scrapy/ofweek_scrapy/pipelines.py +13 -13
  190. tests/ofweek_scrapy/ofweek_scrapy/settings.py +84 -84
  191. tests/ofweek_scrapy/ofweek_scrapy/spiders/__init__.py +4 -4
  192. tests/ofweek_scrapy/scrapy.cfg +11 -11
  193. tests/optimized_performance_test.py +211 -211
  194. tests/performance_comparison.py +244 -244
  195. tests/queue_blocking_test.py +113 -113
  196. tests/queue_test.py +89 -89
  197. tests/redis_key_validation_demo.py +130 -130
  198. tests/request_params_example.py +150 -150
  199. tests/response_improvements_example.py +144 -144
  200. tests/scrapy_comparison/ofweek_scrapy.py +138 -138
  201. tests/scrapy_comparison/scrapy_test.py +133 -133
  202. tests/simple_cli_test.py +54 -54
  203. tests/simple_command_test.py +119 -119
  204. tests/simple_crawlo_test.py +126 -126
  205. tests/simple_follow_test.py +38 -38
  206. tests/simple_log_test2.py +137 -137
  207. tests/simple_optimization_test.py +128 -128
  208. tests/simple_queue_type_test.py +41 -41
  209. tests/simple_response_selector_test.py +94 -94
  210. tests/simple_selector_helper_test.py +154 -154
  211. tests/simple_selector_test.py +207 -207
  212. tests/simple_spider_test.py +49 -49
  213. tests/simple_url_test.py +73 -73
  214. tests/simulate_mysql_update_test.py +139 -139
  215. tests/spider_log_timing_test.py +177 -177
  216. tests/test_advanced_tools.py +148 -148
  217. tests/test_all_commands.py +230 -230
  218. tests/test_all_pipeline_fingerprints.py +133 -133
  219. tests/test_all_redis_key_configs.py +145 -145
  220. tests/test_asyncmy_usage.py +56 -56
  221. tests/test_batch_processor.py +178 -178
  222. tests/test_cleaners.py +54 -54
  223. tests/test_cli_arguments.py +118 -118
  224. tests/test_component_factory.py +174 -174
  225. tests/test_config_consistency.py +80 -80
  226. tests/test_config_merge.py +152 -152
  227. tests/test_config_validator.py +182 -182
  228. tests/test_controlled_spider_mixin.py +79 -79
  229. tests/test_crawler_process_import.py +38 -38
  230. tests/test_crawler_process_spider_modules.py +47 -47
  231. tests/test_crawlo_proxy_integration.py +114 -114
  232. tests/test_date_tools.py +123 -123
  233. tests/test_dedup_fix.py +220 -220
  234. tests/test_dedup_pipeline_consistency.py +124 -124
  235. tests/test_default_header_middleware.py +313 -313
  236. tests/test_distributed.py +65 -65
  237. tests/test_double_crawlo_fix.py +204 -204
  238. tests/test_double_crawlo_fix_simple.py +124 -124
  239. tests/test_download_delay_middleware.py +221 -221
  240. tests/test_downloader_proxy_compatibility.py +272 -272
  241. tests/test_edge_cases.py +305 -305
  242. tests/test_encoding_core.py +56 -56
  243. tests/test_encoding_detection.py +126 -126
  244. tests/test_enhanced_error_handler.py +270 -270
  245. tests/test_enhanced_error_handler_comprehensive.py +245 -245
  246. tests/test_error_handler_compatibility.py +112 -112
  247. tests/test_factories.py +252 -252
  248. tests/test_factory_compatibility.py +196 -196
  249. tests/test_final_validation.py +153 -153
  250. tests/test_fingerprint_consistency.py +135 -135
  251. tests/test_fingerprint_simple.py +51 -51
  252. tests/test_get_component_logger.py +83 -83
  253. tests/test_hash_performance.py +99 -99
  254. tests/test_integration.py +169 -169
  255. tests/test_item_dedup_redis_key.py +122 -122
  256. tests/test_large_scale_helper.py +235 -235
  257. tests/test_logging_enhancements.py +374 -374
  258. tests/test_logging_final.py +184 -184
  259. tests/test_logging_integration.py +312 -312
  260. tests/test_logging_system.py +282 -282
  261. tests/test_middleware_debug.py +141 -141
  262. tests/test_mode_consistency.py +51 -51
  263. tests/test_multi_directory.py +67 -67
  264. tests/test_multiple_spider_modules.py +80 -80
  265. tests/test_mysql_pipeline_config.py +164 -164
  266. tests/test_mysql_pipeline_error.py +98 -98
  267. tests/test_mysql_pipeline_init_log.py +82 -82
  268. tests/test_mysql_pipeline_integration.py +132 -132
  269. tests/test_mysql_pipeline_refactor.py +143 -143
  270. tests/test_mysql_pipeline_refactor_simple.py +85 -85
  271. tests/test_mysql_pipeline_robustness.py +195 -195
  272. tests/test_mysql_pipeline_types.py +88 -88
  273. tests/test_mysql_update_columns.py +93 -93
  274. tests/test_offsite_middleware.py +244 -244
  275. tests/test_offsite_middleware_simple.py +203 -203
  276. tests/test_optimized_selector_naming.py +100 -100
  277. tests/test_parsel.py +29 -29
  278. tests/test_performance.py +327 -327
  279. tests/test_performance_monitor.py +115 -115
  280. tests/test_pipeline_fingerprint_consistency.py +86 -86
  281. tests/test_priority_behavior.py +211 -211
  282. tests/test_priority_consistency.py +151 -151
  283. tests/test_priority_consistency_fixed.py +249 -249
  284. tests/test_proxy_health_check.py +32 -32
  285. tests/test_proxy_middleware.py +217 -217
  286. tests/test_proxy_middleware_enhanced.py +212 -212
  287. tests/test_proxy_middleware_integration.py +142 -142
  288. tests/test_proxy_middleware_refactored.py +207 -207
  289. tests/test_proxy_only.py +83 -83
  290. tests/test_proxy_providers.py +56 -56
  291. tests/test_proxy_stats.py +19 -19
  292. tests/test_proxy_strategies.py +59 -59
  293. tests/test_proxy_with_downloader.py +152 -152
  294. tests/test_queue_empty_check.py +41 -41
  295. tests/test_queue_manager_double_crawlo.py +173 -173
  296. tests/test_queue_manager_redis_key.py +179 -179
  297. tests/test_queue_naming.py +154 -154
  298. tests/test_queue_type.py +106 -106
  299. tests/test_queue_type_redis_config_consistency.py +130 -130
  300. tests/test_random_headers_default.py +322 -322
  301. tests/test_random_headers_necessity.py +308 -308
  302. tests/test_random_user_agent.py +72 -72
  303. tests/test_redis_config.py +28 -28
  304. tests/test_redis_connection_pool.py +294 -294
  305. tests/test_redis_key_naming.py +181 -181
  306. tests/test_redis_key_validator.py +123 -123
  307. tests/test_redis_queue.py +224 -224
  308. tests/test_redis_queue_name_fix.py +175 -175
  309. tests/test_redis_queue_type_fallback.py +129 -129
  310. tests/test_request_ignore_middleware.py +182 -182
  311. tests/test_request_params.py +111 -111
  312. tests/test_request_serialization.py +70 -70
  313. tests/test_response_code_middleware.py +349 -349
  314. tests/test_response_filter_middleware.py +427 -427
  315. tests/test_response_follow.py +104 -104
  316. tests/test_response_improvements.py +152 -152
  317. tests/test_response_selector_methods.py +92 -92
  318. tests/test_response_url_methods.py +70 -70
  319. tests/test_response_urljoin.py +86 -86
  320. tests/test_retry_middleware.py +333 -333
  321. tests/test_retry_middleware_realistic.py +273 -273
  322. tests/test_scheduler.py +252 -252
  323. tests/test_scheduler_config_update.py +133 -133
  324. tests/test_scrapy_style_encoding.py +112 -112
  325. tests/test_selector_helper.py +100 -100
  326. tests/test_selector_optimizations.py +146 -146
  327. tests/test_simple_response.py +61 -61
  328. tests/test_spider_loader.py +49 -49
  329. tests/test_spider_loader_comprehensive.py +69 -69
  330. tests/test_spider_modules.py +84 -84
  331. tests/test_spiders/test_spider.py +9 -9
  332. tests/test_telecom_spider_redis_key.py +205 -205
  333. tests/test_template_content.py +87 -87
  334. tests/test_template_redis_key.py +134 -134
  335. tests/test_tools.py +159 -159
  336. tests/test_user_agent_randomness.py +176 -176
  337. tests/test_user_agents.py +96 -96
  338. tests/untested_features_report.md +138 -138
  339. tests/verify_debug.py +51 -51
  340. tests/verify_distributed.py +117 -117
  341. tests/verify_log_fix.py +111 -111
  342. tests/verify_mysql_warnings.py +109 -109
  343. crawlo/utils/log.py +0 -80
  344. crawlo/utils/url_utils.py +0 -40
  345. crawlo-1.4.7.dist-info/RECORD +0 -347
  346. {crawlo-1.4.7.dist-info → crawlo-1.4.8.dist-info}/WHEEL +0 -0
  347. {crawlo-1.4.7.dist-info → crawlo-1.4.8.dist-info}/entry_points.txt +0 -0
  348. {crawlo-1.4.7.dist-info → crawlo-1.4.8.dist-info}/top_level.txt +0 -0
@@ -1,411 +1,411 @@
1
- #!/usr/bin/python
2
- # -*- coding:UTF-8 -*-
3
- """
4
- 错误处理工具
5
- 提供详细、一致的错误处理和日志记录机制
6
- """
7
- import traceback
8
- from functools import wraps
9
- from datetime import datetime
10
- from typing import Optional, Callable, Any, Dict, List
11
-
12
- from crawlo.logging import get_logger
13
-
14
-
15
- class ErrorContext:
16
- """错误上下文信息"""
17
-
18
- def __init__(self, context: str = "", module: str = "", function: str = ""):
19
- self.context = context
20
- self.module = module
21
- self.function = function
22
- self.timestamp = datetime.now()
23
-
24
- def __str__(self):
25
- parts = []
26
- if self.module:
27
- parts.append(f"Module: {self.module}")
28
- if self.function:
29
- parts.append(f"Function: {self.function}")
30
- if self.context:
31
- parts.append(f"Context: {self.context}")
32
- parts.append(f"Time: {self.timestamp.strftime('%Y-%m-%d %H:%M:%S')}")
33
- return " | ".join(parts)
34
-
35
-
36
- class DetailedException(Exception):
37
- """带有详细信息的异常基类"""
38
-
39
- def __init__(self, message: str, context: Optional[ErrorContext] = None,
40
- error_code: Optional[str] = None, **kwargs):
41
- super().__init__(message)
42
- self.context = context
43
- self.error_code = error_code
44
- self.details = kwargs
45
- self.timestamp = datetime.now()
46
-
47
- def __str__(self):
48
- base_msg = super().__str__()
49
- if self.context:
50
- return f"{base_msg} ({self.context})"
51
- return base_msg
52
-
53
- def get_full_details(self) -> Dict:
54
- """获取完整的错误详情"""
55
- return {
56
- "message": str(self),
57
- "error_code": self.error_code,
58
- "context": str(self.context) if self.context else None,
59
- "details": self.details,
60
- "timestamp": self.timestamp.isoformat(),
61
- "exception_type": self.__class__.__name__
62
- }
63
-
64
-
65
- class EnhancedErrorHandler:
66
- """增强版错误处理器"""
67
-
68
- def __init__(self, logger_name: str = __name__, log_level: str = 'ERROR'):
69
- self.logger = get_logger(logger_name)
70
- self.error_history: List[Dict] = [] # 错误历史记录
71
- self.max_history_size = 100 # 最大历史记录数
72
-
73
- def handle_error(self, exception: Exception, context: Optional[ErrorContext] = None,
74
- raise_error: bool = True, log_error: bool = True,
75
- extra_info: Optional[Dict] = None) -> Dict:
76
- """
77
- 增强版错误处理
78
-
79
- Args:
80
- exception: 异常对象
81
- context: 错误上下文信息
82
- raise_error: 是否重新抛出异常
83
- log_error: 是否记录错误日志
84
- extra_info: 额外的错误信息
85
-
86
- Returns:
87
- 包含错误详情的字典
88
- """
89
- # 构建错误详情
90
- error_details = {
91
- "exception": exception,
92
- "exception_type": type(exception).__name__,
93
- "message": str(exception),
94
- "context": str(context) if context else None,
95
- "timestamp": datetime.now().isoformat(),
96
- "traceback": traceback.format_exc() if log_error else None,
97
- "extra_info": extra_info or {}
98
- }
99
-
100
- # 记录到历史
101
- self._record_error(error_details)
102
-
103
- # 记录日志
104
- if log_error:
105
- self._log_error(error_details)
106
-
107
- # 重新抛出异常
108
- if raise_error:
109
- raise exception
110
-
111
- return error_details
112
-
113
- def _log_error(self, error_details: Dict):
114
- """记录错误日志"""
115
- # 基本错误信息
116
- context_info = error_details.get("context", "")
117
- message = error_details["message"]
118
- error_msg = f"{message} [{context_info}]" if context_info else message
119
-
120
- # 记录错误
121
- self.logger.error(error_msg)
122
-
123
- # 记录详细信息
124
- if error_details.get("traceback"):
125
- self.logger.debug(f"详细错误信息:\n{error_details['traceback']}")
126
-
127
- # 记录额外信息
128
- if error_details.get("extra_info"):
129
- self.logger.debug(f"额外信息: {error_details['extra_info']}")
130
-
131
- def _record_error(self, error_details: Dict):
132
- """记录错误到历史"""
133
- self.error_history.append(error_details)
134
- # 限制历史记录大小
135
- if len(self.error_history) > self.max_history_size:
136
- self.error_history.pop(0)
137
-
138
- def safe_call(self, func: Callable, *args, default_return=None,
139
- context: Optional[ErrorContext] = None, **kwargs) -> Any:
140
- """
141
- 安全调用函数,捕获并处理异常
142
-
143
- Args:
144
- func: 要调用的函数
145
- *args: 函数参数
146
- default_return: 默认返回值
147
- context: 错误上下文
148
- **kwargs: 函数关键字参数
149
-
150
- Returns:
151
- 函数返回值或默认值
152
- """
153
- try:
154
- return func(*args, **kwargs)
155
- except Exception as e:
156
- self.handle_error(e, context=context, raise_error=False)
157
- return default_return
158
-
159
- def retry_on_failure(self, max_retries: int = 3, delay: float = 1.0,
160
- exceptions: tuple = (Exception,), backoff_factor: float = 1.0,
161
- context: Optional[ErrorContext] = None):
162
- """
163
- 装饰器:失败时重试(增强版)
164
-
165
- Args:
166
- max_retries: 最大重试次数
167
- delay: 初始重试间隔(秒)
168
- exceptions: 需要重试的异常类型
169
- backoff_factor: 退避因子(每次重试间隔乘以此因子)
170
- context: 错误上下文
171
- """
172
- def decorator(func):
173
- @wraps(func)
174
- async def async_wrapper(*args, **kwargs):
175
- last_exception = None
176
- current_delay = delay
177
-
178
- for attempt in range(max_retries + 1):
179
- try:
180
- return await func(*args, **kwargs)
181
- except exceptions as e:
182
- last_exception = e
183
- if attempt < max_retries:
184
- # 记录重试信息
185
- retry_context = ErrorContext(
186
- context=f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1})",
187
- module=context.module if context else "",
188
- function=func.__name__
189
- ) if context else None
190
-
191
- self.logger.warning(
192
- f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1}): {e}"
193
- )
194
-
195
- import asyncio
196
- await asyncio.sleep(current_delay)
197
- current_delay *= backoff_factor # 指数退避
198
- else:
199
- # 最后一次尝试失败
200
- final_context = ErrorContext(
201
- context=f"函数 {func.__name__} 执行失败,已达到最大重试次数",
202
- module=context.module if context else "",
203
- function=func.__name__
204
- ) if context else None
205
-
206
- self.logger.error(
207
- f"函数 {func.__name__} 执行失败,已达到最大重试次数: {e}"
208
- )
209
- raise last_exception
210
-
211
- @wraps(func)
212
- def sync_wrapper(*args, **kwargs):
213
- last_exception = None
214
- current_delay = delay
215
-
216
- for attempt in range(max_retries + 1):
217
- try:
218
- return func(*args, **kwargs)
219
- except exceptions as e:
220
- last_exception = e
221
- if attempt < max_retries:
222
- # 记录重试信息
223
- retry_context = ErrorContext(
224
- context=f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1})",
225
- module=context.module if context else "",
226
- function=func.__name__
227
- ) if context else None
228
-
229
- self.logger.warning(
230
- f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1}): {e}"
231
- )
232
-
233
- import time
234
- time.sleep(current_delay)
235
- current_delay *= backoff_factor # 指数退避
236
- else:
237
- # 最后一次尝试失败
238
- final_context = ErrorContext(
239
- context=f"函数 {func.__name__} 执行失败,已达到最大重试次数",
240
- module=context.module if context else "",
241
- function=func.__name__
242
- ) if context else None
243
-
244
- self.logger.error(
245
- f"函数 {func.__name__} 执行失败,已达到最大重试次数: {e}"
246
- )
247
- raise last_exception
248
-
249
- # 根据函数是否为异步函数返回相应的包装器
250
- import inspect
251
- if inspect.iscoroutinefunction(func):
252
- return async_wrapper
253
- else:
254
- return sync_wrapper
255
-
256
- return decorator
257
-
258
- def get_error_history(self) -> List[Dict]:
259
- """获取错误历史记录"""
260
- return self.error_history.copy()
261
-
262
- def clear_error_history(self):
263
- """清空错误历史记录"""
264
- self.error_history.clear()
265
-
266
-
267
- # 为了向后兼容,提供与旧版error_handler.py相同的接口
268
-
269
- # 别名,保持与旧版接口一致
270
- ErrorHandler = EnhancedErrorHandler
271
-
272
- # 全局增强错误处理器实例
273
- enhanced_error_handler = EnhancedErrorHandler()
274
-
275
- # 为了向后兼容,提供默认错误处理器实例
276
- default_error_handler = enhanced_error_handler
277
-
278
- def handle_exception(context: str = "", module: str = "", function: str = "",
279
- raise_error: bool = True, log_error: bool = True,
280
- error_code: Optional[str] = None):
281
- """
282
- 装饰器:处理函数异常(增强版)
283
-
284
- Args:
285
- context: 错误上下文描述
286
- module: 模块名称
287
- function: 函数名称
288
- raise_error: 是否重新抛出异常
289
- log_error: 是否记录错误日志
290
- error_code: 错误代码
291
- """
292
- def decorator(func):
293
- @wraps(func)
294
- async def async_wrapper(*args, **kwargs):
295
- try:
296
- return await func(*args, **kwargs)
297
- except Exception as e:
298
- error_context = ErrorContext(
299
- context=f"{context} - {func.__name__}",
300
- module=module,
301
- function=func.__name__
302
- )
303
-
304
- # 如果是详细异常,保留原有信息
305
- if isinstance(e, DetailedException):
306
- # 确保上下文信息完整
307
- if not e.context:
308
- e.context = error_context
309
- enhanced_error_handler.handle_error(
310
- e, context=e.context,
311
- raise_error=raise_error, log_error=log_error
312
- )
313
- else:
314
- # 包装为详细异常
315
- detailed_e = DetailedException(
316
- str(e), context=error_context, error_code=error_code
317
- )
318
- enhanced_error_handler.handle_error(
319
- detailed_e, context=error_context,
320
- raise_error=raise_error, log_error=log_error
321
- )
322
- if not raise_error:
323
- return None
324
-
325
- @wraps(func)
326
- def sync_wrapper(*args, **kwargs):
327
- try:
328
- return func(*args, **kwargs)
329
- except Exception as e:
330
- error_context = ErrorContext(
331
- context=f"{context} - {func.__name__}",
332
- module=module,
333
- function=func.__name__
334
- )
335
-
336
- # 如果是详细异常,保留原有信息
337
- if isinstance(e, DetailedException):
338
- # 确保上下文信息完整
339
- if not e.context:
340
- e.context = error_context
341
- enhanced_error_handler.handle_error(
342
- e, context=e.context,
343
- raise_error=raise_error, log_error=log_error
344
- )
345
- else:
346
- # 包装为详细异常
347
- detailed_e = DetailedException(
348
- str(e), context=error_context, error_code=error_code
349
- )
350
- enhanced_error_handler.handle_error(
351
- detailed_e, context=error_context,
352
- raise_error=raise_error, log_error=log_error
353
- )
354
- if not raise_error:
355
- return None
356
-
357
- # 根据函数是否为异步函数返回相应的包装器
358
- import inspect
359
- if inspect.iscoroutinefunction(func):
360
- return async_wrapper
361
- else:
362
- return sync_wrapper
363
-
364
- return decorator
365
-
366
- # 为了完全向后兼容,也提供旧版的handle_exception接口
367
- # 这个版本与error_handler.py中的接口完全一致
368
- def handle_exception_simple(context: str = "", raise_error: bool = True, log_error: bool = True):
369
- """
370
- 装饰器:处理函数异常(简化版,与旧版接口兼容)
371
-
372
- Args:
373
- context: 错误上下文描述
374
- raise_error: 是否重新抛出异常
375
- log_error: 是否记录错误日志
376
- """
377
- def decorator(func):
378
- @wraps(func)
379
- async def async_wrapper(*args, **kwargs):
380
- try:
381
- return await func(*args, **kwargs)
382
- except Exception as e:
383
- error_context = ErrorContext(context=f"{context} - {func.__name__}")
384
- enhanced_error_handler.handle_error(
385
- e, context=error_context,
386
- raise_error=raise_error, log_error=log_error
387
- )
388
- if not raise_error:
389
- return None
390
-
391
- @wraps(func)
392
- def sync_wrapper(*args, **kwargs):
393
- try:
394
- return func(*args, **kwargs)
395
- except Exception as e:
396
- error_context = ErrorContext(context=f"{context} - {func.__name__}")
397
- enhanced_error_handler.handle_error(
398
- e, context=error_context,
399
- raise_error=raise_error, log_error=log_error
400
- )
401
- if not raise_error:
402
- return None
403
-
404
- # 根据函数是否为异步函数返回相应的包装器
405
- import inspect
406
- if inspect.iscoroutinefunction(func):
407
- return async_wrapper
408
- else:
409
- return sync_wrapper
410
-
1
+ #!/usr/bin/python
2
+ # -*- coding:UTF-8 -*-
3
+ """
4
+ 错误处理工具
5
+ 提供详细、一致的错误处理和日志记录机制
6
+ """
7
+ import traceback
8
+ from functools import wraps
9
+ from datetime import datetime
10
+ from typing import Optional, Callable, Any, Dict, List
11
+
12
+ from crawlo.logging import get_logger
13
+
14
+
15
+ class ErrorContext:
16
+ """错误上下文信息"""
17
+
18
+ def __init__(self, context: str = "", module: str = "", function: str = ""):
19
+ self.context = context
20
+ self.module = module
21
+ self.function = function
22
+ self.timestamp = datetime.now()
23
+
24
+ def __str__(self):
25
+ parts = []
26
+ if self.module:
27
+ parts.append(f"Module: {self.module}")
28
+ if self.function:
29
+ parts.append(f"Function: {self.function}")
30
+ if self.context:
31
+ parts.append(f"Context: {self.context}")
32
+ parts.append(f"Time: {self.timestamp.strftime('%Y-%m-%d %H:%M:%S')}")
33
+ return " | ".join(parts)
34
+
35
+
36
+ class DetailedException(Exception):
37
+ """带有详细信息的异常基类"""
38
+
39
+ def __init__(self, message: str, context: Optional[ErrorContext] = None,
40
+ error_code: Optional[str] = None, **kwargs):
41
+ super().__init__(message)
42
+ self.context = context
43
+ self.error_code = error_code
44
+ self.details = kwargs
45
+ self.timestamp = datetime.now()
46
+
47
+ def __str__(self):
48
+ base_msg = super().__str__()
49
+ if self.context:
50
+ return f"{base_msg} ({self.context})"
51
+ return base_msg
52
+
53
+ def get_full_details(self) -> Dict:
54
+ """获取完整的错误详情"""
55
+ return {
56
+ "message": str(self),
57
+ "error_code": self.error_code,
58
+ "context": str(self.context) if self.context else None,
59
+ "details": self.details,
60
+ "timestamp": self.timestamp.isoformat(),
61
+ "exception_type": self.__class__.__name__
62
+ }
63
+
64
+
65
+ class EnhancedErrorHandler:
66
+ """增强版错误处理器"""
67
+
68
+ def __init__(self, logger_name: str = __name__, log_level: str = 'ERROR'):
69
+ self.logger = get_logger(logger_name)
70
+ self.error_history: List[Dict] = [] # 错误历史记录
71
+ self.max_history_size = 100 # 最大历史记录数
72
+
73
+ def handle_error(self, exception: Exception, context: Optional[ErrorContext] = None,
74
+ raise_error: bool = True, log_error: bool = True,
75
+ extra_info: Optional[Dict] = None) -> Dict:
76
+ """
77
+ 增强版错误处理
78
+
79
+ Args:
80
+ exception: 异常对象
81
+ context: 错误上下文信息
82
+ raise_error: 是否重新抛出异常
83
+ log_error: 是否记录错误日志
84
+ extra_info: 额外的错误信息
85
+
86
+ Returns:
87
+ 包含错误详情的字典
88
+ """
89
+ # 构建错误详情
90
+ error_details = {
91
+ "exception": exception,
92
+ "exception_type": type(exception).__name__,
93
+ "message": str(exception),
94
+ "context": str(context) if context else None,
95
+ "timestamp": datetime.now().isoformat(),
96
+ "traceback": traceback.format_exc() if log_error else None,
97
+ "extra_info": extra_info or {}
98
+ }
99
+
100
+ # 记录到历史
101
+ self._record_error(error_details)
102
+
103
+ # 记录日志
104
+ if log_error:
105
+ self._log_error(error_details)
106
+
107
+ # 重新抛出异常
108
+ if raise_error:
109
+ raise exception
110
+
111
+ return error_details
112
+
113
+ def _log_error(self, error_details: Dict):
114
+ """记录错误日志"""
115
+ # 基本错误信息
116
+ context_info = error_details.get("context", "")
117
+ message = error_details["message"]
118
+ error_msg = f"{message} [{context_info}]" if context_info else message
119
+
120
+ # 记录错误
121
+ self.logger.error(error_msg)
122
+
123
+ # 记录详细信息
124
+ if error_details.get("traceback"):
125
+ self.logger.debug(f"详细错误信息:\n{error_details['traceback']}")
126
+
127
+ # 记录额外信息
128
+ if error_details.get("extra_info"):
129
+ self.logger.debug(f"额外信息: {error_details['extra_info']}")
130
+
131
+ def _record_error(self, error_details: Dict):
132
+ """记录错误到历史"""
133
+ self.error_history.append(error_details)
134
+ # 限制历史记录大小
135
+ if len(self.error_history) > self.max_history_size:
136
+ self.error_history.pop(0)
137
+
138
+ def safe_call(self, func: Callable, *args, default_return=None,
139
+ context: Optional[ErrorContext] = None, **kwargs) -> Any:
140
+ """
141
+ 安全调用函数,捕获并处理异常
142
+
143
+ Args:
144
+ func: 要调用的函数
145
+ *args: 函数参数
146
+ default_return: 默认返回值
147
+ context: 错误上下文
148
+ **kwargs: 函数关键字参数
149
+
150
+ Returns:
151
+ 函数返回值或默认值
152
+ """
153
+ try:
154
+ return func(*args, **kwargs)
155
+ except Exception as e:
156
+ self.handle_error(e, context=context, raise_error=False)
157
+ return default_return
158
+
159
+ def retry_on_failure(self, max_retries: int = 3, delay: float = 1.0,
160
+ exceptions: tuple = (Exception,), backoff_factor: float = 1.0,
161
+ context: Optional[ErrorContext] = None):
162
+ """
163
+ 装饰器:失败时重试(增强版)
164
+
165
+ Args:
166
+ max_retries: 最大重试次数
167
+ delay: 初始重试间隔(秒)
168
+ exceptions: 需要重试的异常类型
169
+ backoff_factor: 退避因子(每次重试间隔乘以此因子)
170
+ context: 错误上下文
171
+ """
172
+ def decorator(func):
173
+ @wraps(func)
174
+ async def async_wrapper(*args, **kwargs):
175
+ last_exception = None
176
+ current_delay = delay
177
+
178
+ for attempt in range(max_retries + 1):
179
+ try:
180
+ return await func(*args, **kwargs)
181
+ except exceptions as e:
182
+ last_exception = e
183
+ if attempt < max_retries:
184
+ # 记录重试信息
185
+ retry_context = ErrorContext(
186
+ context=f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1})",
187
+ module=context.module if context else "",
188
+ function=func.__name__
189
+ ) if context else None
190
+
191
+ self.logger.warning(
192
+ f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1}): {e}"
193
+ )
194
+
195
+ import asyncio
196
+ await asyncio.sleep(current_delay)
197
+ current_delay *= backoff_factor # 指数退避
198
+ else:
199
+ # 最后一次尝试失败
200
+ final_context = ErrorContext(
201
+ context=f"函数 {func.__name__} 执行失败,已达到最大重试次数",
202
+ module=context.module if context else "",
203
+ function=func.__name__
204
+ ) if context else None
205
+
206
+ self.logger.error(
207
+ f"函数 {func.__name__} 执行失败,已达到最大重试次数: {e}"
208
+ )
209
+ raise last_exception
210
+
211
+ @wraps(func)
212
+ def sync_wrapper(*args, **kwargs):
213
+ last_exception = None
214
+ current_delay = delay
215
+
216
+ for attempt in range(max_retries + 1):
217
+ try:
218
+ return func(*args, **kwargs)
219
+ except exceptions as e:
220
+ last_exception = e
221
+ if attempt < max_retries:
222
+ # 记录重试信息
223
+ retry_context = ErrorContext(
224
+ context=f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1})",
225
+ module=context.module if context else "",
226
+ function=func.__name__
227
+ ) if context else None
228
+
229
+ self.logger.warning(
230
+ f"函数 {func.__name__} 执行失败 (尝试 {attempt + 1}/{max_retries + 1}): {e}"
231
+ )
232
+
233
+ import time
234
+ time.sleep(current_delay)
235
+ current_delay *= backoff_factor # 指数退避
236
+ else:
237
+ # 最后一次尝试失败
238
+ final_context = ErrorContext(
239
+ context=f"函数 {func.__name__} 执行失败,已达到最大重试次数",
240
+ module=context.module if context else "",
241
+ function=func.__name__
242
+ ) if context else None
243
+
244
+ self.logger.error(
245
+ f"函数 {func.__name__} 执行失败,已达到最大重试次数: {e}"
246
+ )
247
+ raise last_exception
248
+
249
+ # 根据函数是否为异步函数返回相应的包装器
250
+ import inspect
251
+ if inspect.iscoroutinefunction(func):
252
+ return async_wrapper
253
+ else:
254
+ return sync_wrapper
255
+
256
+ return decorator
257
+
258
+ def get_error_history(self) -> List[Dict]:
259
+ """获取错误历史记录"""
260
+ return self.error_history.copy()
261
+
262
+ def clear_error_history(self):
263
+ """清空错误历史记录"""
264
+ self.error_history.clear()
265
+
266
+
267
+ # 为了向后兼容,提供与旧版error_handler.py相同的接口
268
+
269
+ # 别名,保持与旧版接口一致
270
+ ErrorHandler = EnhancedErrorHandler
271
+
272
+ # 全局增强错误处理器实例
273
+ enhanced_error_handler = EnhancedErrorHandler()
274
+
275
+ # 为了向后兼容,提供默认错误处理器实例
276
+ default_error_handler = enhanced_error_handler
277
+
278
+ def handle_exception(context: str = "", module: str = "", function: str = "",
279
+ raise_error: bool = True, log_error: bool = True,
280
+ error_code: Optional[str] = None):
281
+ """
282
+ 装饰器:处理函数异常(增强版)
283
+
284
+ Args:
285
+ context: 错误上下文描述
286
+ module: 模块名称
287
+ function: 函数名称
288
+ raise_error: 是否重新抛出异常
289
+ log_error: 是否记录错误日志
290
+ error_code: 错误代码
291
+ """
292
+ def decorator(func):
293
+ @wraps(func)
294
+ async def async_wrapper(*args, **kwargs):
295
+ try:
296
+ return await func(*args, **kwargs)
297
+ except Exception as e:
298
+ error_context = ErrorContext(
299
+ context=f"{context} - {func.__name__}",
300
+ module=module,
301
+ function=func.__name__
302
+ )
303
+
304
+ # 如果是详细异常,保留原有信息
305
+ if isinstance(e, DetailedException):
306
+ # 确保上下文信息完整
307
+ if not e.context:
308
+ e.context = error_context
309
+ enhanced_error_handler.handle_error(
310
+ e, context=e.context,
311
+ raise_error=raise_error, log_error=log_error
312
+ )
313
+ else:
314
+ # 包装为详细异常
315
+ detailed_e = DetailedException(
316
+ str(e), context=error_context, error_code=error_code
317
+ )
318
+ enhanced_error_handler.handle_error(
319
+ detailed_e, context=error_context,
320
+ raise_error=raise_error, log_error=log_error
321
+ )
322
+ if not raise_error:
323
+ return None
324
+
325
+ @wraps(func)
326
+ def sync_wrapper(*args, **kwargs):
327
+ try:
328
+ return func(*args, **kwargs)
329
+ except Exception as e:
330
+ error_context = ErrorContext(
331
+ context=f"{context} - {func.__name__}",
332
+ module=module,
333
+ function=func.__name__
334
+ )
335
+
336
+ # 如果是详细异常,保留原有信息
337
+ if isinstance(e, DetailedException):
338
+ # 确保上下文信息完整
339
+ if not e.context:
340
+ e.context = error_context
341
+ enhanced_error_handler.handle_error(
342
+ e, context=e.context,
343
+ raise_error=raise_error, log_error=log_error
344
+ )
345
+ else:
346
+ # 包装为详细异常
347
+ detailed_e = DetailedException(
348
+ str(e), context=error_context, error_code=error_code
349
+ )
350
+ enhanced_error_handler.handle_error(
351
+ detailed_e, context=error_context,
352
+ raise_error=raise_error, log_error=log_error
353
+ )
354
+ if not raise_error:
355
+ return None
356
+
357
+ # 根据函数是否为异步函数返回相应的包装器
358
+ import inspect
359
+ if inspect.iscoroutinefunction(func):
360
+ return async_wrapper
361
+ else:
362
+ return sync_wrapper
363
+
364
+ return decorator
365
+
366
+ # 为了完全向后兼容,也提供旧版的handle_exception接口
367
+ # 这个版本与error_handler.py中的接口完全一致
368
+ def handle_exception_simple(context: str = "", raise_error: bool = True, log_error: bool = True):
369
+ """
370
+ 装饰器:处理函数异常(简化版,与旧版接口兼容)
371
+
372
+ Args:
373
+ context: 错误上下文描述
374
+ raise_error: 是否重新抛出异常
375
+ log_error: 是否记录错误日志
376
+ """
377
+ def decorator(func):
378
+ @wraps(func)
379
+ async def async_wrapper(*args, **kwargs):
380
+ try:
381
+ return await func(*args, **kwargs)
382
+ except Exception as e:
383
+ error_context = ErrorContext(context=f"{context} - {func.__name__}")
384
+ enhanced_error_handler.handle_error(
385
+ e, context=error_context,
386
+ raise_error=raise_error, log_error=log_error
387
+ )
388
+ if not raise_error:
389
+ return None
390
+
391
+ @wraps(func)
392
+ def sync_wrapper(*args, **kwargs):
393
+ try:
394
+ return func(*args, **kwargs)
395
+ except Exception as e:
396
+ error_context = ErrorContext(context=f"{context} - {func.__name__}")
397
+ enhanced_error_handler.handle_error(
398
+ e, context=error_context,
399
+ raise_error=raise_error, log_error=log_error
400
+ )
401
+ if not raise_error:
402
+ return None
403
+
404
+ # 根据函数是否为异步函数返回相应的包装器
405
+ import inspect
406
+ if inspect.iscoroutinefunction(func):
407
+ return async_wrapper
408
+ else:
409
+ return sync_wrapper
410
+
411
411
  return decorator