crawlo 1.4.5__py3-none-any.whl → 1.4.7__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 (375) hide show
  1. crawlo/__init__.py +90 -89
  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 -138
  8. crawlo/commands/list.py +155 -155
  9. crawlo/commands/run.py +379 -341
  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 -312
  14. crawlo/config_validator.py +277 -277
  15. crawlo/core/__init__.py +52 -52
  16. crawlo/core/engine.py +451 -438
  17. crawlo/core/processor.py +47 -47
  18. crawlo/core/scheduler.py +290 -291
  19. crawlo/crawler.py +698 -657
  20. crawlo/data/__init__.py +5 -5
  21. crawlo/data/user_agents.py +194 -194
  22. crawlo/downloader/__init__.py +280 -276
  23. crawlo/downloader/aiohttp_downloader.py +233 -233
  24. crawlo/downloader/cffi_downloader.py +250 -245
  25. crawlo/downloader/httpx_downloader.py +265 -259
  26. crawlo/downloader/hybrid_downloader.py +212 -212
  27. crawlo/downloader/playwright_downloader.py +425 -402
  28. crawlo/downloader/selenium_downloader.py +486 -472
  29. crawlo/event.py +45 -11
  30. crawlo/exceptions.py +215 -82
  31. crawlo/extension/__init__.py +65 -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 -61
  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 -103
  42. crawlo/factories/registry.py +84 -84
  43. crawlo/factories/utils.py +135 -0
  44. crawlo/filters/__init__.py +170 -153
  45. crawlo/filters/aioredis_filter.py +348 -264
  46. crawlo/filters/memory_filter.py +261 -276
  47. crawlo/framework.py +306 -292
  48. crawlo/initialization/__init__.py +44 -44
  49. crawlo/initialization/built_in.py +391 -434
  50. crawlo/initialization/context.py +141 -141
  51. crawlo/initialization/core.py +240 -194
  52. crawlo/initialization/phases.py +230 -149
  53. crawlo/initialization/registry.py +143 -145
  54. crawlo/initialization/utils.py +49 -0
  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 -46
  61. crawlo/logging/config.py +277 -197
  62. crawlo/logging/factory.py +175 -171
  63. crawlo/logging/manager.py +104 -112
  64. crawlo/middleware/__init__.py +87 -24
  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 -386
  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 -253
  75. crawlo/network/__init__.py +21 -21
  76. crawlo/network/request.py +375 -379
  77. crawlo/network/response.py +569 -664
  78. crawlo/pipelines/__init__.py +53 -22
  79. crawlo/pipelines/base_pipeline.py +452 -0
  80. crawlo/pipelines/bloom_dedup_pipeline.py +146 -146
  81. crawlo/pipelines/console_pipeline.py +39 -39
  82. crawlo/pipelines/csv_pipeline.py +316 -316
  83. crawlo/pipelines/database_dedup_pipeline.py +197 -197
  84. crawlo/pipelines/json_pipeline.py +218 -218
  85. crawlo/pipelines/memory_dedup_pipeline.py +105 -105
  86. crawlo/pipelines/mongo_pipeline.py +140 -132
  87. crawlo/pipelines/mysql_pipeline.py +470 -326
  88. crawlo/pipelines/pipeline_manager.py +100 -100
  89. crawlo/pipelines/redis_dedup_pipeline.py +155 -156
  90. crawlo/project.py +347 -347
  91. crawlo/queue/__init__.py +10 -0
  92. crawlo/queue/pqueue.py +38 -38
  93. crawlo/queue/queue_manager.py +591 -525
  94. crawlo/queue/redis_priority_queue.py +519 -370
  95. crawlo/settings/__init__.py +7 -7
  96. crawlo/settings/default_settings.py +285 -270
  97. crawlo/settings/setting_manager.py +219 -219
  98. crawlo/spider/__init__.py +657 -657
  99. crawlo/stats_collector.py +82 -73
  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 +2 -4
  104. crawlo/templates/project/items.py.tmpl +13 -17
  105. crawlo/templates/project/middlewares.py.tmpl +38 -38
  106. crawlo/templates/project/pipelines.py.tmpl +35 -36
  107. crawlo/templates/project/settings.py.tmpl +110 -157
  108. crawlo/templates/project/settings_distributed.py.tmpl +156 -161
  109. crawlo/templates/project/settings_gentle.py.tmpl +170 -171
  110. crawlo/templates/project/settings_high_performance.py.tmpl +171 -172
  111. crawlo/templates/project/settings_minimal.py.tmpl +99 -77
  112. crawlo/templates/project/settings_simple.py.tmpl +168 -169
  113. crawlo/templates/project/spiders/__init__.py.tmpl +9 -9
  114. crawlo/templates/run.py.tmpl +23 -30
  115. crawlo/templates/spider/spider.py.tmpl +33 -144
  116. crawlo/templates/spiders_init.py.tmpl +5 -10
  117. crawlo/tools/__init__.py +86 -189
  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 +50 -50
  123. crawlo/utils/batch_processor.py +276 -259
  124. crawlo/utils/config_manager.py +442 -0
  125. crawlo/utils/controlled_spider_mixin.py +439 -439
  126. crawlo/utils/db_helper.py +250 -244
  127. crawlo/utils/error_handler.py +410 -410
  128. crawlo/utils/fingerprint.py +121 -121
  129. crawlo/utils/func_tools.py +82 -82
  130. crawlo/utils/large_scale_helper.py +344 -344
  131. crawlo/utils/leak_detector.py +335 -0
  132. crawlo/utils/log.py +79 -79
  133. crawlo/utils/misc.py +81 -81
  134. crawlo/utils/mongo_connection_pool.py +157 -0
  135. crawlo/utils/mysql_connection_pool.py +197 -0
  136. crawlo/utils/performance_monitor.py +285 -285
  137. crawlo/utils/queue_helper.py +175 -175
  138. crawlo/utils/redis_checker.py +91 -0
  139. crawlo/utils/redis_connection_pool.py +578 -388
  140. crawlo/utils/redis_key_validator.py +198 -198
  141. crawlo/utils/request.py +278 -256
  142. crawlo/utils/request_serializer.py +225 -225
  143. crawlo/utils/resource_manager.py +337 -0
  144. crawlo/utils/selector_helper.py +137 -137
  145. crawlo/utils/singleton.py +70 -0
  146. crawlo/utils/spider_loader.py +201 -201
  147. crawlo/utils/text_helper.py +94 -94
  148. crawlo/utils/{url.py → url_utils.py} +39 -39
  149. crawlo-1.4.7.dist-info/METADATA +689 -0
  150. crawlo-1.4.7.dist-info/RECORD +347 -0
  151. examples/__init__.py +7 -7
  152. tests/__init__.py +7 -7
  153. tests/advanced_tools_example.py +217 -275
  154. tests/authenticated_proxy_example.py +110 -106
  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 -0
  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 +77 -0
  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 -0
  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/scrapy.cfg +11 -11
  192. tests/optimized_performance_test.py +211 -211
  193. tests/performance_comparison.py +244 -244
  194. tests/queue_blocking_test.py +113 -113
  195. tests/queue_test.py +89 -89
  196. tests/redis_key_validation_demo.py +130 -130
  197. tests/request_params_example.py +150 -150
  198. tests/response_improvements_example.py +144 -144
  199. tests/scrapy_comparison/ofweek_scrapy.py +138 -138
  200. tests/scrapy_comparison/scrapy_test.py +133 -133
  201. tests/simple_cli_test.py +55 -0
  202. tests/simple_command_test.py +119 -119
  203. tests/simple_crawlo_test.py +126 -126
  204. tests/simple_follow_test.py +38 -38
  205. tests/simple_log_test2.py +137 -137
  206. tests/simple_optimization_test.py +128 -128
  207. tests/simple_queue_type_test.py +41 -41
  208. tests/simple_response_selector_test.py +94 -94
  209. tests/simple_selector_helper_test.py +154 -154
  210. tests/simple_selector_test.py +207 -207
  211. tests/simple_spider_test.py +49 -49
  212. tests/simple_url_test.py +73 -73
  213. tests/simulate_mysql_update_test.py +140 -0
  214. tests/spider_log_timing_test.py +177 -177
  215. tests/test_advanced_tools.py +148 -148
  216. tests/test_all_commands.py +230 -230
  217. tests/test_all_pipeline_fingerprints.py +133 -133
  218. tests/test_all_redis_key_configs.py +145 -145
  219. tests/test_asyncmy_usage.py +57 -0
  220. tests/test_batch_processor.py +178 -178
  221. tests/test_cleaners.py +54 -54
  222. tests/test_cli_arguments.py +119 -0
  223. tests/test_component_factory.py +174 -174
  224. tests/test_config_consistency.py +80 -80
  225. tests/test_config_merge.py +152 -152
  226. tests/test_config_validator.py +182 -182
  227. tests/test_controlled_spider_mixin.py +79 -79
  228. tests/test_crawler_process_import.py +38 -38
  229. tests/test_crawler_process_spider_modules.py +47 -47
  230. tests/test_crawlo_proxy_integration.py +114 -108
  231. tests/test_date_tools.py +123 -123
  232. tests/test_dedup_fix.py +220 -220
  233. tests/test_dedup_pipeline_consistency.py +124 -124
  234. tests/test_default_header_middleware.py +313 -313
  235. tests/test_distributed.py +65 -65
  236. tests/test_double_crawlo_fix.py +204 -204
  237. tests/test_double_crawlo_fix_simple.py +124 -124
  238. tests/test_download_delay_middleware.py +221 -221
  239. tests/test_downloader_proxy_compatibility.py +272 -268
  240. tests/test_edge_cases.py +305 -305
  241. tests/test_encoding_core.py +56 -56
  242. tests/test_encoding_detection.py +126 -126
  243. tests/test_enhanced_error_handler.py +270 -270
  244. tests/test_enhanced_error_handler_comprehensive.py +245 -245
  245. tests/test_error_handler_compatibility.py +112 -112
  246. tests/test_factories.py +252 -252
  247. tests/test_factory_compatibility.py +196 -196
  248. tests/test_final_validation.py +153 -153
  249. tests/test_fingerprint_consistency.py +135 -135
  250. tests/test_fingerprint_simple.py +51 -51
  251. tests/test_get_component_logger.py +83 -83
  252. tests/test_hash_performance.py +99 -99
  253. tests/test_integration.py +169 -169
  254. tests/test_item_dedup_redis_key.py +122 -122
  255. tests/test_large_scale_helper.py +235 -235
  256. tests/test_logging_enhancements.py +374 -374
  257. tests/test_logging_final.py +184 -184
  258. tests/test_logging_integration.py +312 -312
  259. tests/test_logging_system.py +282 -282
  260. tests/test_middleware_debug.py +141 -141
  261. tests/test_mode_consistency.py +51 -51
  262. tests/test_multi_directory.py +67 -67
  263. tests/test_multiple_spider_modules.py +80 -80
  264. tests/test_mysql_pipeline_config.py +165 -0
  265. tests/test_mysql_pipeline_error.py +99 -0
  266. tests/test_mysql_pipeline_init_log.py +83 -0
  267. tests/test_mysql_pipeline_integration.py +133 -0
  268. tests/test_mysql_pipeline_refactor.py +144 -0
  269. tests/test_mysql_pipeline_refactor_simple.py +86 -0
  270. tests/test_mysql_pipeline_robustness.py +196 -0
  271. tests/test_mysql_pipeline_types.py +89 -0
  272. tests/test_mysql_update_columns.py +94 -0
  273. tests/test_offsite_middleware.py +244 -244
  274. tests/test_offsite_middleware_simple.py +203 -203
  275. tests/test_optimized_selector_naming.py +100 -100
  276. tests/test_parsel.py +29 -29
  277. tests/test_performance.py +327 -327
  278. tests/test_performance_monitor.py +115 -115
  279. tests/test_pipeline_fingerprint_consistency.py +86 -86
  280. tests/test_priority_behavior.py +211 -211
  281. tests/test_priority_consistency.py +151 -151
  282. tests/test_priority_consistency_fixed.py +249 -249
  283. tests/test_proxy_health_check.py +32 -32
  284. tests/test_proxy_middleware.py +217 -121
  285. tests/test_proxy_middleware_enhanced.py +212 -216
  286. tests/test_proxy_middleware_integration.py +142 -137
  287. tests/test_proxy_middleware_refactored.py +207 -184
  288. tests/test_proxy_only.py +84 -0
  289. tests/test_proxy_providers.py +56 -56
  290. tests/test_proxy_stats.py +19 -19
  291. tests/test_proxy_strategies.py +59 -59
  292. tests/test_proxy_with_downloader.py +153 -0
  293. tests/test_queue_empty_check.py +41 -41
  294. tests/test_queue_manager_double_crawlo.py +173 -173
  295. tests/test_queue_manager_redis_key.py +179 -179
  296. tests/test_queue_naming.py +154 -154
  297. tests/test_queue_type.py +106 -106
  298. tests/test_queue_type_redis_config_consistency.py +130 -130
  299. tests/test_random_headers_default.py +322 -322
  300. tests/test_random_headers_necessity.py +308 -308
  301. tests/test_random_user_agent.py +72 -72
  302. tests/test_redis_config.py +28 -28
  303. tests/test_redis_connection_pool.py +294 -294
  304. tests/test_redis_key_naming.py +181 -181
  305. tests/test_redis_key_validator.py +123 -123
  306. tests/test_redis_queue.py +224 -224
  307. tests/test_redis_queue_name_fix.py +175 -175
  308. tests/test_redis_queue_type_fallback.py +129 -129
  309. tests/test_request_ignore_middleware.py +182 -182
  310. tests/test_request_params.py +111 -111
  311. tests/test_request_serialization.py +70 -70
  312. tests/test_response_code_middleware.py +349 -349
  313. tests/test_response_filter_middleware.py +427 -427
  314. tests/test_response_follow.py +104 -104
  315. tests/test_response_improvements.py +152 -152
  316. tests/test_response_selector_methods.py +92 -92
  317. tests/test_response_url_methods.py +70 -70
  318. tests/test_response_urljoin.py +86 -86
  319. tests/test_retry_middleware.py +333 -333
  320. tests/test_retry_middleware_realistic.py +273 -273
  321. tests/test_scheduler.py +252 -252
  322. tests/test_scheduler_config_update.py +133 -133
  323. tests/test_scrapy_style_encoding.py +112 -112
  324. tests/test_selector_helper.py +100 -100
  325. tests/test_selector_optimizations.py +146 -146
  326. tests/test_simple_response.py +61 -61
  327. tests/test_spider_loader.py +49 -49
  328. tests/test_spider_loader_comprehensive.py +69 -69
  329. tests/test_spider_modules.py +84 -84
  330. tests/test_spiders/test_spider.py +9 -9
  331. tests/test_telecom_spider_redis_key.py +205 -205
  332. tests/test_template_content.py +87 -87
  333. tests/test_template_redis_key.py +134 -134
  334. tests/test_tools.py +159 -159
  335. tests/test_user_agent_randomness.py +176 -176
  336. tests/test_user_agents.py +96 -96
  337. tests/untested_features_report.md +138 -138
  338. tests/verify_debug.py +51 -51
  339. tests/verify_distributed.py +117 -117
  340. tests/verify_log_fix.py +111 -111
  341. tests/verify_mysql_warnings.py +110 -0
  342. crawlo/logging/async_handler.py +0 -181
  343. crawlo/logging/monitor.py +0 -153
  344. crawlo/logging/sampler.py +0 -167
  345. crawlo/middleware/simple_proxy.py +0 -65
  346. crawlo/tools/authenticated_proxy.py +0 -241
  347. crawlo/tools/data_formatter.py +0 -226
  348. crawlo/tools/data_validator.py +0 -181
  349. crawlo/tools/encoding_converter.py +0 -127
  350. crawlo/tools/network_diagnostic.py +0 -365
  351. crawlo/tools/request_tools.py +0 -83
  352. crawlo/tools/retry_mechanism.py +0 -224
  353. crawlo/utils/env_config.py +0 -143
  354. crawlo/utils/large_scale_config.py +0 -287
  355. crawlo/utils/system.py +0 -11
  356. crawlo/utils/tools.py +0 -5
  357. crawlo-1.4.5.dist-info/METADATA +0 -329
  358. crawlo-1.4.5.dist-info/RECORD +0 -347
  359. tests/env_config_example.py +0 -134
  360. tests/ofweek_scrapy/ofweek_scrapy/spiders/ofweek_spider.py +0 -162
  361. tests/test_authenticated_proxy.py +0 -142
  362. tests/test_comprehensive.py +0 -147
  363. tests/test_dynamic_downloaders_proxy.py +0 -125
  364. tests/test_dynamic_proxy.py +0 -93
  365. tests/test_dynamic_proxy_config.py +0 -147
  366. tests/test_dynamic_proxy_real.py +0 -110
  367. tests/test_env_config.py +0 -122
  368. tests/test_framework_env_usage.py +0 -104
  369. tests/test_large_scale_config.py +0 -113
  370. tests/test_proxy_api.py +0 -265
  371. tests/test_real_scenario_proxy.py +0 -196
  372. tests/tools_example.py +0 -261
  373. {crawlo-1.4.5.dist-info → crawlo-1.4.7.dist-info}/WHEEL +0 -0
  374. {crawlo-1.4.5.dist-info → crawlo-1.4.7.dist-info}/entry_points.txt +0 -0
  375. {crawlo-1.4.5.dist-info → crawlo-1.4.7.dist-info}/top_level.txt +0 -0
crawlo/logging/factory.py CHANGED
@@ -1,172 +1,176 @@
1
- #!/usr/bin/python
2
- # -*- coding: UTF-8 -*-
3
- """
4
- 日志器工厂 - 创建和缓存Logger实例
5
- """
6
-
7
- import logging
8
- import os
9
- import sys
10
- import threading
11
- from typing import Dict, Optional
12
- from weakref import WeakValueDictionary
13
-
14
- # 尝试导入concurrent-log-handler,如果不可用则回退到标准库
15
- try:
16
- from concurrent_log_handler import ConcurrentRotatingFileHandler
17
- USE_CONCURRENT_HANDLER = True
18
- except ImportError:
19
- from logging.handlers import RotatingFileHandler
20
- USE_CONCURRENT_HANDLER = False
21
-
22
- from .manager import get_config, is_configured, configure
23
- from .config import LogConfig
24
-
25
-
26
- class LoggerFactory:
27
- """
28
- Logger工厂类 - 负责创建和缓存Logger实例
29
-
30
- 特点:
31
- 1. 使用WeakValueDictionary避免内存泄漏
32
- 2. 线程安全的Logger创建
33
- 3. 自动配置管理
34
- 4. 简单的缓存策略
35
- 5. Windows兼容的日志轮转处理
36
- """
37
-
38
- # Logger缓存 - 使用弱引用避免内存泄漏
39
- _logger_cache: WeakValueDictionary = WeakValueDictionary()
40
- _cache_lock = threading.RLock()
41
-
42
- @classmethod
43
- def get_logger(cls, name: str = 'crawlo') -> logging.Logger:
44
- """
45
- 获取Logger实例
46
-
47
- Args:
48
- name: Logger名称
49
-
50
- Returns:
51
- logging.Logger: 配置好的Logger实例
52
- """
53
- # 确保日志系统已配置
54
- if not is_configured():
55
- configure() # 使用默认配置
56
-
57
- # 检查缓存
58
- with cls._cache_lock:
59
- if name in cls._logger_cache:
60
- return cls._logger_cache[name]
61
-
62
- # 创建新的Logger
63
- logger = cls._create_logger(name)
64
- cls._logger_cache[name] = logger
65
- return logger
66
-
67
- @classmethod
68
- def _create_logger(cls, name: str) -> logging.Logger:
69
- """创建新的Logger实例"""
70
- config = get_config()
71
- if not config:
72
- raise RuntimeError("Log system not configured")
73
-
74
- # 创建Logger
75
- logger = logging.getLogger(name)
76
- logger.setLevel(logging.DEBUG) # Logger本身设为最低级别
77
-
78
- # 清除现有handlers(避免重复添加)
79
- logger.handlers.clear()
80
-
81
- # 获取模块级别
82
- module_level = config.get_module_level(name)
83
-
84
- # 创建formatter
85
- formatter = logging.Formatter(config.get_format())
86
-
87
- # 添加控制台Handler
88
- if config.console_enabled:
89
- console_handler = logging.StreamHandler()
90
- console_handler.setFormatter(formatter)
91
- # 使用专门的控制台级别或模块级别
92
- console_level = config.get_console_level()
93
- level = getattr(logging, console_level.upper(), logging.INFO)
94
- console_handler.setLevel(level)
95
- logger.addHandler(console_handler)
96
-
97
- # 添加文件Handler
98
- if config.file_enabled and config.file_path:
99
- try:
100
- # 确保日志目录存在
101
- log_dir = os.path.dirname(config.file_path)
102
- if log_dir and not os.path.exists(log_dir):
103
- os.makedirs(log_dir, exist_ok=True)
104
-
105
- # 根据平台选择合适的Handler
106
- if USE_CONCURRENT_HANDLER:
107
- file_handler = ConcurrentRotatingFileHandler(
108
- filename=config.file_path,
109
- maxBytes=config.max_bytes,
110
- backupCount=config.backup_count,
111
- encoding=config.encoding
112
- )
113
- else:
114
- # 在Windows上给出警告信息
115
- if sys.platform.startswith('win'):
116
- # 检查是否已经有同名的日志文件被其他进程使用
117
- try:
118
- # 尝试以独占模式打开文件来检查是否被占用
119
- with open(config.file_path, 'a'):
120
- pass
121
- except (PermissionError, OSError):
122
- # 如果文件被占用,记录警告信息
123
- console_handler = logging.StreamHandler()
124
- console_handler.setFormatter(formatter)
125
- console_handler.setLevel(logging.WARNING)
126
- logger.addHandler(console_handler)
127
- logger.warning(f"日志文件 {config.file_path} 可能正在被其他进程使用,这可能导致日志轮转失败。建议安装 concurrent-log-handler 库以获得更好的Windows兼容性。")
128
-
129
- file_handler = RotatingFileHandler(
130
- filename=config.file_path,
131
- maxBytes=config.max_bytes,
132
- backupCount=config.backup_count,
133
- encoding=config.encoding
134
- )
135
-
136
- file_handler.setFormatter(formatter)
137
- # 使用专门的文件级别或模块级别
138
- file_level = config.get_file_level()
139
- level = getattr(logging, file_level.upper(), logging.INFO)
140
- file_handler.setLevel(level)
141
- logger.addHandler(file_handler)
142
- except Exception as e:
143
- # 文件Handler创建失败时,至少保证控制台输出
144
- console_handler = logging.StreamHandler()
145
- console_handler.setFormatter(formatter)
146
- console_handler.setLevel(logging.WARNING)
147
- logger.addHandler(console_handler)
148
- logger.warning(f"无法创建文件日志处理器: {e},仅使用控制台输出。")
149
-
150
- # 防止向上传播(避免重复输出)
151
- logger.propagate = False
152
-
153
- return logger
154
-
155
- @classmethod
156
- def clear_cache(cls):
157
- """清空Logger缓存"""
158
- with cls._cache_lock:
159
- cls._logger_cache.clear()
160
-
161
- @classmethod
162
- def refresh_loggers(cls, new_config: LogConfig):
163
- """刷新所有缓存的Logger(配置更新时使用)"""
164
- with cls._cache_lock:
165
- # 清空缓存,强制重新创建
166
- cls._logger_cache.clear()
167
-
168
-
169
- # 便捷函数
170
- def get_logger(name: str = 'crawlo') -> logging.Logger:
171
- """获取Logger实例的便捷函数"""
1
+ #!/usr/bin/python
2
+ # -*- coding: UTF-8 -*-
3
+ """
4
+ 日志器工厂 - 创建和缓存Logger实例
5
+ """
6
+
7
+ import logging
8
+ import os
9
+ import sys
10
+ import threading
11
+ from weakref import WeakValueDictionary
12
+
13
+ # 尝试导入concurrent-log-handler,如果不可用则回退到标准库
14
+ try:
15
+ from concurrent_log_handler import ConcurrentRotatingFileHandler
16
+ USE_CONCURRENT_HANDLER = True
17
+ RotatingFileHandler = ConcurrentRotatingFileHandler # 别名以避免未绑定错误
18
+ except ImportError:
19
+ from logging.handlers import RotatingFileHandler
20
+ USE_CONCURRENT_HANDLER = False
21
+ ConcurrentRotatingFileHandler = RotatingFileHandler # 别名以避免未绑定错误
22
+
23
+ from .manager import get_config, is_configured, configure
24
+ from .config import LogConfig
25
+
26
+
27
+ class LoggerFactory:
28
+ """
29
+ Logger工厂类 - 负责创建和缓存Logger实例
30
+
31
+ 特点:
32
+ 1. 使用WeakValueDictionary避免内存泄漏
33
+ 2. 线程安全的Logger创建
34
+ 3. 自动配置管理
35
+ 4. 简单的缓存策略
36
+ 5. Windows兼容的日志轮转处理
37
+ """
38
+
39
+ # Logger缓存 - 使用弱引用避免内存泄漏
40
+ _logger_cache: WeakValueDictionary = WeakValueDictionary()
41
+ _cache_lock = threading.RLock()
42
+
43
+ @classmethod
44
+ def get_logger(cls, name: str = 'crawlo') -> logging.Logger:
45
+ """
46
+ 获取Logger实例
47
+
48
+ Args:
49
+ name: Logger名称
50
+
51
+ Returns:
52
+ logging.Logger: 配置好的Logger实例
53
+ """
54
+ # 确保日志系统已配置
55
+ if not is_configured():
56
+ configure() # 使用默认配置
57
+
58
+ # 检查缓存
59
+ with cls._cache_lock:
60
+ if name in cls._logger_cache:
61
+ return cls._logger_cache[name]
62
+
63
+ # 创建新的Logger
64
+ logger = cls._create_logger(name)
65
+ cls._logger_cache[name] = logger
66
+ return logger
67
+
68
+ @classmethod
69
+ def _create_logger(cls, name: str) -> logging.Logger:
70
+ """创建新的Logger实例"""
71
+ config = get_config()
72
+ if not config:
73
+ raise RuntimeError("日志系统未配置,请先调用 configure_logging() 进行配置")
74
+
75
+ # 创建Logger
76
+ logger = logging.getLogger(name)
77
+ logger.setLevel(logging.DEBUG) # Logger本身设为最低级别
78
+
79
+ # 清除现有handlers(避免重复添加)
80
+ logger.handlers.clear()
81
+
82
+ # 获取模块级别
83
+ module_level = config.get_module_level(name)
84
+
85
+ # 创建formatter
86
+ formatter = logging.Formatter(config.get_format())
87
+
88
+ # 添加控制台Handler
89
+ if config.console_enabled:
90
+ console_handler = logging.StreamHandler()
91
+ console_handler.setFormatter(formatter)
92
+ # 使用专门的控制台级别或模块级别
93
+ console_level = config.get_console_level()
94
+ level = getattr(logging, console_level.upper(), logging.INFO)
95
+ console_handler.setLevel(level)
96
+ logger.addHandler(console_handler)
97
+
98
+ # 添加文件Handler
99
+ if config.file_enabled and config.file_path:
100
+ try:
101
+ # 确保日志目录存在
102
+ log_dir = os.path.dirname(config.file_path)
103
+ if log_dir and not os.path.exists(log_dir):
104
+ os.makedirs(log_dir, exist_ok=True)
105
+
106
+ # 根据平台选择合适的Handler
107
+ file_handler = None
108
+ if USE_CONCURRENT_HANDLER:
109
+ file_handler = ConcurrentRotatingFileHandler(
110
+ filename=config.file_path,
111
+ maxBytes=config.max_bytes,
112
+ backupCount=config.backup_count,
113
+ encoding=config.encoding
114
+ )
115
+ else:
116
+ # 在Windows上给出警告信息
117
+ if sys.platform.startswith('win'):
118
+ # 检查是否已经有同名的日志文件被其他进程使用
119
+ try:
120
+ # 尝试以独占模式打开文件来检查是否被占用
121
+ with open(config.file_path, 'a'):
122
+ pass
123
+ except (PermissionError, OSError):
124
+ # 如果文件被占用,记录警告信息
125
+ console_handler = logging.StreamHandler()
126
+ console_handler.setFormatter(formatter)
127
+ console_handler.setLevel(logging.WARNING)
128
+ logger.addHandler(console_handler)
129
+ logger.warning(f"日志文件 {config.file_path} 可能正在被其他进程使用,这可能导致日志轮转失败。建议安装 concurrent-log-handler 库以获得更好的Windows兼容性。")
130
+
131
+ file_handler = RotatingFileHandler(
132
+ filename=config.file_path,
133
+ maxBytes=config.max_bytes,
134
+ backupCount=config.backup_count,
135
+ encoding=config.encoding
136
+ )
137
+
138
+ # 添加文件处理器(如果创建成功)
139
+ if file_handler is not None:
140
+ file_handler.setFormatter(formatter)
141
+ # 使用专门的文件级别或模块级别
142
+ file_level = config.get_file_level()
143
+ level = getattr(logging, file_level.upper(), logging.INFO)
144
+ file_handler.setLevel(level)
145
+ logger.addHandler(file_handler)
146
+ except Exception as e:
147
+ # 文件Handler创建失败时,至少保证控制台输出
148
+ console_handler = logging.StreamHandler()
149
+ console_handler.setFormatter(formatter)
150
+ console_handler.setLevel(logging.WARNING)
151
+ logger.addHandler(console_handler)
152
+ logger.warning(f"无法创建文件日志处理器: {e},仅使用控制台输出。")
153
+
154
+ # 防止向上传播(避免重复输出)
155
+ logger.propagate = False
156
+
157
+ return logger
158
+
159
+ @classmethod
160
+ def clear_cache(cls):
161
+ """清空Logger缓存"""
162
+ with cls._cache_lock:
163
+ cls._logger_cache.clear()
164
+
165
+ @classmethod
166
+ def refresh_loggers(cls, new_config: LogConfig):
167
+ """刷新所有缓存的Logger(配置更新时使用)"""
168
+ with cls._cache_lock:
169
+ # 清空缓存,强制重新创建
170
+ cls._logger_cache.clear()
171
+
172
+
173
+ # 便捷函数
174
+ def get_logger(name: str = 'crawlo') -> logging.Logger:
175
+ """获取Logger实例的便捷函数"""
172
176
  return LoggerFactory.get_logger(name)
crawlo/logging/manager.py CHANGED
@@ -1,112 +1,104 @@
1
- #!/usr/bin/python
2
- # -*- coding: UTF-8 -*-
3
- """
4
- 日志管理器 - 核心组件
5
- """
6
-
7
- import threading
8
- from typing import Optional, Any
9
- from .config import LogConfig
10
-
11
-
12
- class LogManager:
13
- """
14
- 日志管理器 - 单例模式
15
-
16
- 职责:
17
- 1. 全局日志配置管理
18
- 2. 配置状态跟踪
19
- 3. 线程安全的配置更新
20
- """
21
-
22
- _instance: Optional['LogManager'] = None
23
- _lock = threading.Lock()
24
-
25
- def __new__(cls) -> 'LogManager':
26
- if cls._instance is None:
27
- with cls._lock:
28
- if cls._instance is None:
29
- cls._instance = super(LogManager, cls).__new__(cls)
30
- cls._instance._initialized = False
31
- return cls._instance
32
-
33
- def __init__(self):
34
- if hasattr(self, '_initialized') and self._initialized:
35
- return
36
-
37
- self._config: Optional[LogConfig] = None
38
- self._configured = False
39
- self._config_lock = threading.RLock()
40
- self._initialized = True
41
-
42
- @property
43
- def config(self) -> Optional[LogConfig]:
44
- """获取当前配置"""
45
- with self._config_lock:
46
- return self._config
47
-
48
- @property
49
- def is_configured(self) -> bool:
50
- """检查是否已配置"""
51
- return self._configured
52
-
53
- def configure(self, settings=None, **kwargs) -> LogConfig:
54
- """
55
- 配置日志系统
56
-
57
- Args:
58
- settings: 配置对象或None
59
- **kwargs: 关键字参数配置
60
-
61
- Returns:
62
- LogConfig: 生效的配置对象
63
- """
64
- with self._config_lock:
65
- # 总是重新配置,即使已经配置过
66
- # 从不同来源创建配置
67
- if settings is not None:
68
- # 检查settings是否已经是LogConfig对象
69
- if isinstance(settings, LogConfig):
70
- config = settings
71
- else:
72
- config = LogConfig.from_settings(settings)
73
- elif kwargs:
74
- config = LogConfig.from_dict(kwargs)
75
- else:
76
- config = LogConfig() # 使用默认配置
77
-
78
- # 验证配置
79
- if not config.validate():
80
- raise ValueError("Invalid log configuration")
81
-
82
- self._config = config
83
- self._configured = True
84
-
85
- return config
86
-
87
- def reset(self):
88
- """重置配置(主要用于测试)"""
89
- with self._config_lock:
90
- self._config = None
91
- self._configured = False
92
-
93
-
94
- # 全局实例
95
- _log_manager = LogManager()
96
-
97
- # 模块级便捷函数
98
- def configure(settings=None, **kwargs) -> LogConfig:
99
- """配置日志系统"""
100
- return _log_manager.configure(settings, **kwargs)
101
-
102
- def is_configured() -> bool:
103
- """检查是否已配置"""
104
- return _log_manager.is_configured
105
-
106
- def get_config() -> Optional[LogConfig]:
107
- """获取当前配置"""
108
- return _log_manager.config
109
-
110
- def reset():
111
- """重置配置"""
112
- _log_manager.reset()
1
+ #!/usr/bin/python
2
+ # -*- coding: UTF-8 -*-
3
+ """
4
+ 日志管理器 - 核心组件
5
+ """
6
+
7
+ import threading
8
+ from typing import Optional
9
+ from .config import LogConfig
10
+ from crawlo.utils.singleton import singleton
11
+
12
+
13
+ @singleton
14
+ class LogManager:
15
+ """
16
+ 日志管理器 - 单例模式
17
+
18
+ 职责:
19
+ 1. 全局日志配置管理
20
+ 2. 配置状态跟踪
21
+ 3. 线程安全的配置更新
22
+ """
23
+
24
+ def __init__(self):
25
+ self._config: Optional[LogConfig] = None
26
+ self._configured = False
27
+ self._config_lock = threading.RLock()
28
+
29
+ @property
30
+ def config(self) -> Optional[LogConfig]:
31
+ """获取当前配置"""
32
+ with self._config_lock:
33
+ return self._config
34
+
35
+ @property
36
+ def is_configured(self) -> bool:
37
+ """检查是否已配置"""
38
+ return self._configured
39
+
40
+ def configure(self, settings=None, **kwargs) -> LogConfig:
41
+ """
42
+ 配置日志系统
43
+
44
+ Args:
45
+ settings: 配置对象或None
46
+ **kwargs: 关键字参数配置
47
+
48
+ Returns:
49
+ LogConfig: 生效的配置对象
50
+ """
51
+ with self._config_lock:
52
+ # 总是重新配置,即使已经配置过
53
+ # 从不同来源创建配置
54
+ if settings is not None:
55
+ # 检查settings是否已经是LogConfig对象
56
+ if isinstance(settings, LogConfig):
57
+ config = settings
58
+ else:
59
+ config = LogConfig.from_settings(settings)
60
+ elif kwargs:
61
+ config = LogConfig.from_dict(kwargs)
62
+ else:
63
+ config = LogConfig() # 使用默认配置
64
+
65
+ # 验证配置
66
+ is_valid, error_msg = config.validate()
67
+ if not is_valid:
68
+ raise ValueError(f"Invalid log configuration: {error_msg}")
69
+
70
+ self._config = config
71
+ self._configured = True
72
+
73
+ return config
74
+
75
+ def reset(self):
76
+ """重置配置(主要用于测试)"""
77
+ with self._config_lock:
78
+ self._config = None
79
+ self._configured = False
80
+
81
+
82
+ # 全局实例
83
+ _log_manager = LogManager()
84
+
85
+
86
+ # 模块级便捷函数
87
+ def configure(settings=None, **kwargs) -> LogConfig:
88
+ """配置日志系统"""
89
+ return _log_manager.configure(settings, **kwargs)
90
+
91
+
92
+ def is_configured() -> bool:
93
+ """检查是否已配置"""
94
+ return _log_manager.is_configured
95
+
96
+
97
+ def get_config() -> Optional[LogConfig]:
98
+ """获取当前配置"""
99
+ return _log_manager.config
100
+
101
+
102
+ def reset():
103
+ """重置配置"""
104
+ _log_manager.reset()