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
@@ -1,375 +1,375 @@
1
- #!/usr/bin/env python
2
- # -*- coding: UTF-8 -*-
3
- """
4
- 日志系统增强功能测试
5
- """
6
-
7
- import os
8
- import sys
9
- import tempfile
10
- import shutil
11
- import threading
12
- import time
13
- from pathlib import Path
14
-
15
- # 添加项目根目录到Python路径
16
- project_root = Path(__file__).parent.parent
17
- sys.path.insert(0, str(project_root))
18
-
19
- from crawlo.logging import (
20
- configure_logging,
21
- get_logger,
22
- LogManager,
23
- get_monitor
24
- )
25
- from crawlo.logging.config import LogConfig
26
- from crawlo.logging.sampler import get_sampler
27
- from crawlo.logging.async_handler import AsyncConcurrentRotatingFileHandler
28
-
29
-
30
- def test_log_configuration():
31
- """测试日志配置功能"""
32
- print("=== 测试日志配置功能 ===")
33
-
34
- # 创建临时目录
35
- temp_dir = tempfile.mkdtemp()
36
- log_file = os.path.join(temp_dir, 'enhanced_test.log')
37
-
38
- try:
39
- # 1. 测试分别设置控制台和文件日志级别
40
- print("1. 测试分别设置控制台和文件日志级别...")
41
- LogManager().reset()
42
-
43
- config = configure_logging(
44
- LOG_LEVEL='INFO',
45
- LOG_CONSOLE_LEVEL='WARNING', # 控制台只显示WARNING及以上
46
- LOG_FILE_LEVEL='DEBUG', # 文件记录DEBUG及以上
47
- LOG_FILE=log_file,
48
- LOG_FILE_ENABLED=True,
49
- LOG_CONSOLE_ENABLED=True,
50
- LOG_INCLUDE_THREAD_ID=True, # 包含线程ID
51
- LOG_INCLUDE_PROCESS_ID=True # 包含进程ID
52
- )
53
-
54
- print(f" 配置详情:")
55
- print(f" 主级别: {config.level}")
56
- print(f" 控制台级别: {config.console_level}")
57
- print(f" 文件级别: {config.file_level}")
58
- print(f" 包含线程ID: {config.include_thread_id}")
59
- print(f" 包含进程ID: {config.include_process_id}")
60
-
61
- # 获取logger并测试输出
62
- logger = get_logger('test.enhanced')
63
-
64
- # 测试不同级别的日志
65
- print("2. 测试不同级别的日志输出...")
66
- logger.debug("DEBUG消息 - 应该在文件中看到,控制台看不到")
67
- logger.info("INFO消息 - 应该在文件中看到,控制台看不到")
68
- logger.warning("WARNING消息 - 应该在文件和控制台都看到")
69
- logger.error("ERROR消息 - 应该在文件和控制台都看到")
70
-
71
- # 检查日志文件内容
72
- print("3. 检查日志文件内容...")
73
- if os.path.exists(log_file):
74
- with open(log_file, 'r', encoding='utf-8') as f:
75
- lines = f.readlines()
76
- print(f" 日志文件行数: {len(lines)}")
77
- for i, line in enumerate(lines):
78
- print(f" {i+1}: {line.strip()}")
79
-
80
- # 验证是否包含线程ID和进程ID
81
- if lines:
82
- first_line = lines[0]
83
- has_thread_id = '[Thread-' in first_line or '[thread:' in first_line.lower()
84
- has_process_id = '[Process-' in first_line or '[process:' in first_line.lower()
85
- print(f" 包含线程ID: {has_thread_id}")
86
- print(f" 包含进程ID: {has_process_id}")
87
- else:
88
- print(" 日志文件不存在!")
89
-
90
- finally:
91
- # 清理
92
- shutil.rmtree(temp_dir, ignore_errors=True)
93
-
94
-
95
- def test_log_performance_monitoring():
96
- """测试日志性能监控功能"""
97
- print("\n=== 测试日志性能监控功能 ===")
98
-
99
- # 创建临时目录
100
- temp_dir = tempfile.mkdtemp()
101
- log_file = os.path.join(temp_dir, 'monitor_test.log')
102
-
103
- try:
104
- # 重置并配置日志系统
105
- LogManager().reset()
106
- configure_logging(
107
- LOG_LEVEL='INFO',
108
- LOG_FILE=log_file,
109
- LOG_FILE_ENABLED=True,
110
- LOG_CONSOLE_ENABLED=False
111
- )
112
-
113
- # 启用性能监控
114
- monitor = get_monitor()
115
- monitor.enable_monitoring()
116
-
117
- # 获取logger并生成一些日志
118
- logger = get_logger('test.monitor')
119
-
120
- print("1. 生成测试日志...")
121
- for i in range(10):
122
- logger.info(f"测试日志消息 {i+1}")
123
- logger.warning(f"测试警告消息 {i+1}")
124
- logger.error(f"测试错误消息 {i+1}")
125
-
126
- # 等待一小段时间
127
- time.sleep(0.1)
128
-
129
- # 获取性能统计
130
- print("2. 获取性能统计...")
131
- stats = monitor.get_statistics()
132
- print(f" 统计信息: {stats}")
133
-
134
- # 获取性能报告
135
- print("3. 获取性能报告...")
136
- report = monitor.get_performance_report()
137
- print(f" 性能报告:\n{report}")
138
-
139
- finally:
140
- # 清理
141
- shutil.rmtree(temp_dir, ignore_errors=True)
142
-
143
-
144
- def test_log_sampling():
145
- """测试日志采样功能"""
146
- print("\n=== 测试日志采样功能 ===")
147
-
148
- # 创建临时目录
149
- temp_dir = tempfile.mkdtemp()
150
- log_file = os.path.join(temp_dir, 'sampler_test.log')
151
-
152
- try:
153
- # 重置并配置日志系统
154
- LogManager().reset()
155
- configure_logging(
156
- LOG_LEVEL='INFO',
157
- LOG_FILE=log_file,
158
- LOG_FILE_ENABLED=True,
159
- LOG_CONSOLE_ENABLED=False
160
- )
161
-
162
- # 配置采样器
163
- sampler = get_sampler()
164
- sampler.set_sample_rate('test.sampler', 0.5) # 50%采样率
165
- sampler.set_rate_limit('test.sampler', 100) # 每秒最多100条日志
166
-
167
- # 获取logger
168
- logger = get_logger('test.sampler')
169
-
170
- print("1. 生成大量日志消息...")
171
- start_time = time.time()
172
- log_count = 0
173
-
174
- for i in range(100):
175
- # 使用采样器判断是否应该记录日志
176
- message = f"采样测试消息 {i+1}"
177
- if sampler.should_log('test.sampler', message):
178
- logger.info(message)
179
- log_count += 1
180
-
181
- end_time = time.time()
182
-
183
- print(f" 生成消息总数: 100")
184
- print(f" 实际记录消息数: {log_count}")
185
- print(f" 记录率: {log_count/100:.2%}")
186
- print(f" 耗时: {end_time - start_time:.3f}秒")
187
-
188
- # 检查日志文件
189
- print("2. 检查日志文件...")
190
- if os.path.exists(log_file):
191
- with open(log_file, 'r', encoding='utf-8') as f:
192
- lines = f.readlines()
193
- print(f" 日志文件实际行数: {len(lines)}")
194
- else:
195
- print(" 日志文件不存在!")
196
-
197
- finally:
198
- # 清理
199
- shutil.rmtree(temp_dir, ignore_errors=True)
200
-
201
-
202
- def test_async_logging():
203
- """测试异步日志处理功能"""
204
- print("\n=== 测试异步日志处理功能 ===")
205
-
206
- # 创建临时目录
207
- temp_dir = tempfile.mkdtemp()
208
- log_file = os.path.join(temp_dir, 'async_test.log')
209
-
210
- try:
211
- # 重置日志管理器
212
- LogManager().reset()
213
-
214
- # 直接测试异步处理器
215
- print("1. 测试异步并发轮转文件处理器...")
216
-
217
- # 创建异步处理器
218
- async_handler = AsyncConcurrentRotatingFileHandler(
219
- filename=log_file,
220
- maxBytes=1024*1024, # 1MB
221
- backupCount=3
222
- )
223
-
224
- # 启动异步处理器
225
- async_handler.start()
226
-
227
- # 创建logger并添加异步处理器
228
- logger = get_logger('test.async')
229
- logger.handlers.clear() # 清除现有处理器
230
- logger.addHandler(async_handler)
231
- logger.setLevel(10) # DEBUG级别
232
-
233
- print("2. 生成异步日志...")
234
- start_time = time.time()
235
-
236
- # 生成大量日志
237
- for i in range(100):
238
- logger.info(f"异步日志消息 {i+1}")
239
-
240
- # 等待日志处理完成
241
- time.sleep(1.0)
242
-
243
- end_time = time.time()
244
-
245
- print(f" 生成100条日志耗时: {end_time - start_time:.3f}秒")
246
-
247
- # 检查日志文件
248
- print("3. 检查日志文件...")
249
- if os.path.exists(log_file):
250
- # 尝试不同的编码方式读取文件
251
- encodings = ['utf-8', 'gbk', 'latin1']
252
- lines = []
253
- for encoding in encodings:
254
- try:
255
- with open(log_file, 'r', encoding=encoding) as f:
256
- lines = f.readlines()
257
- break
258
- except UnicodeDecodeError:
259
- continue
260
-
261
- if lines:
262
- print(f" 日志文件行数: {len(lines)}")
263
- if lines:
264
- print(f" 第一行: {lines[0].strip()}")
265
- print(f" 最后一行: {lines[-1].strip()}")
266
- else:
267
- print(" 无法读取日志文件内容")
268
- else:
269
- print(" 日志文件不存在!")
270
-
271
- # 停止异步处理器
272
- async_handler.stop()
273
-
274
- finally:
275
- # 清理
276
- shutil.rmtree(temp_dir, ignore_errors=True)
277
-
278
-
279
- def test_module_specific_levels():
280
- """测试模块特定日志级别"""
281
- print("\n=== 测试模块特定日志级别 ===")
282
-
283
- # 创建临时目录
284
- temp_dir = tempfile.mkdtemp()
285
- log_file = os.path.join(temp_dir, 'module_levels_test.log')
286
-
287
- try:
288
- # 重置并配置日志系统
289
- LogManager().reset()
290
- configure_logging(
291
- LOG_LEVEL='WARNING',
292
- LOG_FILE=log_file,
293
- LOG_FILE_ENABLED=True,
294
- LOG_CONSOLE_ENABLED=True,
295
- LOG_LEVELS={
296
- 'module.debug': 'DEBUG',
297
- 'module.info': 'INFO',
298
- 'module.error': 'ERROR'
299
- }
300
- )
301
-
302
- # 测试不同模块的日志级别
303
- print("1. 测试DEBUG模块...")
304
- debug_logger = get_logger('module.debug')
305
- debug_logger.debug("DEBUG消息 - 应该显示")
306
- debug_logger.info("INFO消息 - 应该显示")
307
- debug_logger.warning("WARNING消息 - 应该显示")
308
-
309
- print("2. 测试INFO模块...")
310
- info_logger = get_logger('module.info')
311
- info_logger.debug("DEBUG消息 - 不应该显示")
312
- info_logger.info("INFO消息 - 应该显示")
313
- info_logger.warning("WARNING消息 - 应该显示")
314
-
315
- print("3. 测试ERROR模块...")
316
- error_logger = get_logger('module.error')
317
- error_logger.info("INFO消息 - 不应该显示")
318
- error_logger.warning("WARNING消息 - 不应该显示")
319
- error_logger.error("ERROR消息 - 应该显示")
320
-
321
- print("4. 测试默认模块...")
322
- default_logger = get_logger('module.default')
323
- default_logger.info("INFO消息 - 不应该显示")
324
- default_logger.warning("WARNING消息 - 应该显示")
325
- default_logger.error("ERROR消息 - 应该显示")
326
-
327
- # 检查日志文件
328
- print("5. 检查日志文件...")
329
- if os.path.exists(log_file):
330
- with open(log_file, 'r', encoding='utf-8') as f:
331
- lines = f.readlines()
332
- print(f" 日志文件行数: {len(lines)}")
333
- for i, line in enumerate(lines):
334
- print(f" {i+1}: {line.strip()}")
335
- else:
336
- print(" 日志文件不存在!")
337
-
338
- finally:
339
- # 清理
340
- shutil.rmtree(temp_dir, ignore_errors=True)
341
-
342
-
343
- def main():
344
- """主测试函数"""
345
- print("开始测试Crawlo框架日志系统功能...")
346
- print("=" * 60)
347
-
348
- try:
349
- # 运行所有测试
350
- test_log_configuration()
351
- test_log_performance_monitoring()
352
- test_log_sampling()
353
- test_async_logging()
354
- test_module_specific_levels()
355
-
356
- print("\n" + "=" * 60)
357
- print("所有日志系统功能测试完成!")
358
- print("\n测试总结:")
359
- print("✅ 日志配置功能")
360
- print("✅ 日志性能监控")
361
- print("✅ 日志采样")
362
- print("✅ 异步日志处理")
363
- print("✅ 模块特定日志级别")
364
-
365
- except Exception as e:
366
- print(f"\n测试过程中出现错误: {e}")
367
- import traceback
368
- traceback.print_exc()
369
- return 1
370
-
371
- return 0
372
-
373
-
374
- if __name__ == '__main__':
1
+ #!/usr/bin/env python
2
+ # -*- coding: UTF-8 -*-
3
+ """
4
+ 日志系统增强功能测试
5
+ """
6
+
7
+ import os
8
+ import sys
9
+ import tempfile
10
+ import shutil
11
+ import threading
12
+ import time
13
+ from pathlib import Path
14
+
15
+ # 添加项目根目录到Python路径
16
+ project_root = Path(__file__).parent.parent
17
+ sys.path.insert(0, str(project_root))
18
+
19
+ from crawlo.logging import (
20
+ configure_logging,
21
+ get_logger,
22
+ LogManager,
23
+ get_monitor
24
+ )
25
+ from crawlo.logging.config import LogConfig
26
+ from crawlo.logging.sampler import get_sampler
27
+ from crawlo.logging.async_handler import AsyncConcurrentRotatingFileHandler
28
+
29
+
30
+ def test_log_configuration():
31
+ """测试日志配置功能"""
32
+ print("=== 测试日志配置功能 ===")
33
+
34
+ # 创建临时目录
35
+ temp_dir = tempfile.mkdtemp()
36
+ log_file = os.path.join(temp_dir, 'enhanced_test.log')
37
+
38
+ try:
39
+ # 1. 测试分别设置控制台和文件日志级别
40
+ print("1. 测试分别设置控制台和文件日志级别...")
41
+ LogManager().reset()
42
+
43
+ config = configure_logging(
44
+ LOG_LEVEL='INFO',
45
+ LOG_CONSOLE_LEVEL='WARNING', # 控制台只显示WARNING及以上
46
+ LOG_FILE_LEVEL='DEBUG', # 文件记录DEBUG及以上
47
+ LOG_FILE=log_file,
48
+ LOG_FILE_ENABLED=True,
49
+ LOG_CONSOLE_ENABLED=True,
50
+ LOG_INCLUDE_THREAD_ID=True, # 包含线程ID
51
+ LOG_INCLUDE_PROCESS_ID=True # 包含进程ID
52
+ )
53
+
54
+ print(f" 配置详情:")
55
+ print(f" 主级别: {config.level}")
56
+ print(f" 控制台级别: {config.console_level}")
57
+ print(f" 文件级别: {config.file_level}")
58
+ print(f" 包含线程ID: {config.include_thread_id}")
59
+ print(f" 包含进程ID: {config.include_process_id}")
60
+
61
+ # 获取logger并测试输出
62
+ logger = get_logger('test.enhanced')
63
+
64
+ # 测试不同级别的日志
65
+ print("2. 测试不同级别的日志输出...")
66
+ logger.debug("DEBUG消息 - 应该在文件中看到,控制台看不到")
67
+ logger.info("INFO消息 - 应该在文件中看到,控制台看不到")
68
+ logger.warning("WARNING消息 - 应该在文件和控制台都看到")
69
+ logger.error("ERROR消息 - 应该在文件和控制台都看到")
70
+
71
+ # 检查日志文件内容
72
+ print("3. 检查日志文件内容...")
73
+ if os.path.exists(log_file):
74
+ with open(log_file, 'r', encoding='utf-8') as f:
75
+ lines = f.readlines()
76
+ print(f" 日志文件行数: {len(lines)}")
77
+ for i, line in enumerate(lines):
78
+ print(f" {i+1}: {line.strip()}")
79
+
80
+ # 验证是否包含线程ID和进程ID
81
+ if lines:
82
+ first_line = lines[0]
83
+ has_thread_id = '[Thread-' in first_line or '[thread:' in first_line.lower()
84
+ has_process_id = '[Process-' in first_line or '[process:' in first_line.lower()
85
+ print(f" 包含线程ID: {has_thread_id}")
86
+ print(f" 包含进程ID: {has_process_id}")
87
+ else:
88
+ print(" 日志文件不存在!")
89
+
90
+ finally:
91
+ # 清理
92
+ shutil.rmtree(temp_dir, ignore_errors=True)
93
+
94
+
95
+ def test_log_performance_monitoring():
96
+ """测试日志性能监控功能"""
97
+ print("\n=== 测试日志性能监控功能 ===")
98
+
99
+ # 创建临时目录
100
+ temp_dir = tempfile.mkdtemp()
101
+ log_file = os.path.join(temp_dir, 'monitor_test.log')
102
+
103
+ try:
104
+ # 重置并配置日志系统
105
+ LogManager().reset()
106
+ configure_logging(
107
+ LOG_LEVEL='INFO',
108
+ LOG_FILE=log_file,
109
+ LOG_FILE_ENABLED=True,
110
+ LOG_CONSOLE_ENABLED=False
111
+ )
112
+
113
+ # 启用性能监控
114
+ monitor = get_monitor()
115
+ monitor.enable_monitoring()
116
+
117
+ # 获取logger并生成一些日志
118
+ logger = get_logger('test.monitor')
119
+
120
+ print("1. 生成测试日志...")
121
+ for i in range(10):
122
+ logger.info(f"测试日志消息 {i+1}")
123
+ logger.warning(f"测试警告消息 {i+1}")
124
+ logger.error(f"测试错误消息 {i+1}")
125
+
126
+ # 等待一小段时间
127
+ time.sleep(0.1)
128
+
129
+ # 获取性能统计
130
+ print("2. 获取性能统计...")
131
+ stats = monitor.get_statistics()
132
+ print(f" 统计信息: {stats}")
133
+
134
+ # 获取性能报告
135
+ print("3. 获取性能报告...")
136
+ report = monitor.get_performance_report()
137
+ print(f" 性能报告:\n{report}")
138
+
139
+ finally:
140
+ # 清理
141
+ shutil.rmtree(temp_dir, ignore_errors=True)
142
+
143
+
144
+ def test_log_sampling():
145
+ """测试日志采样功能"""
146
+ print("\n=== 测试日志采样功能 ===")
147
+
148
+ # 创建临时目录
149
+ temp_dir = tempfile.mkdtemp()
150
+ log_file = os.path.join(temp_dir, 'sampler_test.log')
151
+
152
+ try:
153
+ # 重置并配置日志系统
154
+ LogManager().reset()
155
+ configure_logging(
156
+ LOG_LEVEL='INFO',
157
+ LOG_FILE=log_file,
158
+ LOG_FILE_ENABLED=True,
159
+ LOG_CONSOLE_ENABLED=False
160
+ )
161
+
162
+ # 配置采样器
163
+ sampler = get_sampler()
164
+ sampler.set_sample_rate('test.sampler', 0.5) # 50%采样率
165
+ sampler.set_rate_limit('test.sampler', 100) # 每秒最多100条日志
166
+
167
+ # 获取logger
168
+ logger = get_logger('test.sampler')
169
+
170
+ print("1. 生成大量日志消息...")
171
+ start_time = time.time()
172
+ log_count = 0
173
+
174
+ for i in range(100):
175
+ # 使用采样器判断是否应该记录日志
176
+ message = f"采样测试消息 {i+1}"
177
+ if sampler.should_log('test.sampler', message):
178
+ logger.info(message)
179
+ log_count += 1
180
+
181
+ end_time = time.time()
182
+
183
+ print(f" 生成消息总数: 100")
184
+ print(f" 实际记录消息数: {log_count}")
185
+ print(f" 记录率: {log_count/100:.2%}")
186
+ print(f" 耗时: {end_time - start_time:.3f}秒")
187
+
188
+ # 检查日志文件
189
+ print("2. 检查日志文件...")
190
+ if os.path.exists(log_file):
191
+ with open(log_file, 'r', encoding='utf-8') as f:
192
+ lines = f.readlines()
193
+ print(f" 日志文件实际行数: {len(lines)}")
194
+ else:
195
+ print(" 日志文件不存在!")
196
+
197
+ finally:
198
+ # 清理
199
+ shutil.rmtree(temp_dir, ignore_errors=True)
200
+
201
+
202
+ def test_async_logging():
203
+ """测试异步日志处理功能"""
204
+ print("\n=== 测试异步日志处理功能 ===")
205
+
206
+ # 创建临时目录
207
+ temp_dir = tempfile.mkdtemp()
208
+ log_file = os.path.join(temp_dir, 'async_test.log')
209
+
210
+ try:
211
+ # 重置日志管理器
212
+ LogManager().reset()
213
+
214
+ # 直接测试异步处理器
215
+ print("1. 测试异步并发轮转文件处理器...")
216
+
217
+ # 创建异步处理器
218
+ async_handler = AsyncConcurrentRotatingFileHandler(
219
+ filename=log_file,
220
+ maxBytes=1024*1024, # 1MB
221
+ backupCount=3
222
+ )
223
+
224
+ # 启动异步处理器
225
+ async_handler.start()
226
+
227
+ # 创建logger并添加异步处理器
228
+ logger = get_logger('test.async')
229
+ logger.handlers.clear() # 清除现有处理器
230
+ logger.addHandler(async_handler)
231
+ logger.setLevel(10) # DEBUG级别
232
+
233
+ print("2. 生成异步日志...")
234
+ start_time = time.time()
235
+
236
+ # 生成大量日志
237
+ for i in range(100):
238
+ logger.info(f"异步日志消息 {i+1}")
239
+
240
+ # 等待日志处理完成
241
+ time.sleep(1.0)
242
+
243
+ end_time = time.time()
244
+
245
+ print(f" 生成100条日志耗时: {end_time - start_time:.3f}秒")
246
+
247
+ # 检查日志文件
248
+ print("3. 检查日志文件...")
249
+ if os.path.exists(log_file):
250
+ # 尝试不同的编码方式读取文件
251
+ encodings = ['utf-8', 'gbk', 'latin1']
252
+ lines = []
253
+ for encoding in encodings:
254
+ try:
255
+ with open(log_file, 'r', encoding=encoding) as f:
256
+ lines = f.readlines()
257
+ break
258
+ except UnicodeDecodeError:
259
+ continue
260
+
261
+ if lines:
262
+ print(f" 日志文件行数: {len(lines)}")
263
+ if lines:
264
+ print(f" 第一行: {lines[0].strip()}")
265
+ print(f" 最后一行: {lines[-1].strip()}")
266
+ else:
267
+ print(" 无法读取日志文件内容")
268
+ else:
269
+ print(" 日志文件不存在!")
270
+
271
+ # 停止异步处理器
272
+ async_handler.stop()
273
+
274
+ finally:
275
+ # 清理
276
+ shutil.rmtree(temp_dir, ignore_errors=True)
277
+
278
+
279
+ def test_module_specific_levels():
280
+ """测试模块特定日志级别"""
281
+ print("\n=== 测试模块特定日志级别 ===")
282
+
283
+ # 创建临时目录
284
+ temp_dir = tempfile.mkdtemp()
285
+ log_file = os.path.join(temp_dir, 'module_levels_test.log')
286
+
287
+ try:
288
+ # 重置并配置日志系统
289
+ LogManager().reset()
290
+ configure_logging(
291
+ LOG_LEVEL='WARNING',
292
+ LOG_FILE=log_file,
293
+ LOG_FILE_ENABLED=True,
294
+ LOG_CONSOLE_ENABLED=True,
295
+ LOG_LEVELS={
296
+ 'module.debug': 'DEBUG',
297
+ 'module.info': 'INFO',
298
+ 'module.error': 'ERROR'
299
+ }
300
+ )
301
+
302
+ # 测试不同模块的日志级别
303
+ print("1. 测试DEBUG模块...")
304
+ debug_logger = get_logger('module.debug')
305
+ debug_logger.debug("DEBUG消息 - 应该显示")
306
+ debug_logger.info("INFO消息 - 应该显示")
307
+ debug_logger.warning("WARNING消息 - 应该显示")
308
+
309
+ print("2. 测试INFO模块...")
310
+ info_logger = get_logger('module.info')
311
+ info_logger.debug("DEBUG消息 - 不应该显示")
312
+ info_logger.info("INFO消息 - 应该显示")
313
+ info_logger.warning("WARNING消息 - 应该显示")
314
+
315
+ print("3. 测试ERROR模块...")
316
+ error_logger = get_logger('module.error')
317
+ error_logger.info("INFO消息 - 不应该显示")
318
+ error_logger.warning("WARNING消息 - 不应该显示")
319
+ error_logger.error("ERROR消息 - 应该显示")
320
+
321
+ print("4. 测试默认模块...")
322
+ default_logger = get_logger('module.default')
323
+ default_logger.info("INFO消息 - 不应该显示")
324
+ default_logger.warning("WARNING消息 - 应该显示")
325
+ default_logger.error("ERROR消息 - 应该显示")
326
+
327
+ # 检查日志文件
328
+ print("5. 检查日志文件...")
329
+ if os.path.exists(log_file):
330
+ with open(log_file, 'r', encoding='utf-8') as f:
331
+ lines = f.readlines()
332
+ print(f" 日志文件行数: {len(lines)}")
333
+ for i, line in enumerate(lines):
334
+ print(f" {i+1}: {line.strip()}")
335
+ else:
336
+ print(" 日志文件不存在!")
337
+
338
+ finally:
339
+ # 清理
340
+ shutil.rmtree(temp_dir, ignore_errors=True)
341
+
342
+
343
+ def main():
344
+ """主测试函数"""
345
+ print("开始测试Crawlo框架日志系统功能...")
346
+ print("=" * 60)
347
+
348
+ try:
349
+ # 运行所有测试
350
+ test_log_configuration()
351
+ test_log_performance_monitoring()
352
+ test_log_sampling()
353
+ test_async_logging()
354
+ test_module_specific_levels()
355
+
356
+ print("\n" + "=" * 60)
357
+ print("所有日志系统功能测试完成!")
358
+ print("\n测试总结:")
359
+ print("✅ 日志配置功能")
360
+ print("✅ 日志性能监控")
361
+ print("✅ 日志采样")
362
+ print("✅ 异步日志处理")
363
+ print("✅ 模块特定日志级别")
364
+
365
+ except Exception as e:
366
+ print(f"\n测试过程中出现错误: {e}")
367
+ import traceback
368
+ traceback.print_exc()
369
+ return 1
370
+
371
+ return 0
372
+
373
+
374
+ if __name__ == '__main__':
375
375
  sys.exit(main())