crawlo 1.3.6__py3-none-any.whl → 1.3.7__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 (291) hide show
  1. crawlo/__init__.py +87 -87
  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 +151 -151
  7. crawlo/commands/help.py +138 -138
  8. crawlo/commands/list.py +155 -155
  9. crawlo/commands/run.py +341 -341
  10. crawlo/commands/startproject.py +436 -436
  11. crawlo/commands/stats.py +187 -187
  12. crawlo/commands/utils.py +196 -196
  13. crawlo/config.py +312 -312
  14. crawlo/config_validator.py +277 -277
  15. crawlo/core/__init__.py +45 -45
  16. crawlo/core/engine.py +439 -439
  17. crawlo/core/processor.py +40 -40
  18. crawlo/core/scheduler.py +257 -257
  19. crawlo/crawler.py +638 -638
  20. crawlo/data/__init__.py +5 -5
  21. crawlo/data/user_agents.py +194 -194
  22. crawlo/downloader/__init__.py +273 -273
  23. crawlo/downloader/aiohttp_downloader.py +228 -228
  24. crawlo/downloader/cffi_downloader.py +245 -245
  25. crawlo/downloader/httpx_downloader.py +259 -259
  26. crawlo/downloader/hybrid_downloader.py +212 -212
  27. crawlo/downloader/playwright_downloader.py +402 -402
  28. crawlo/downloader/selenium_downloader.py +472 -472
  29. crawlo/event.py +11 -11
  30. crawlo/exceptions.py +81 -81
  31. crawlo/extension/__init__.py +39 -39
  32. crawlo/extension/health_check.py +141 -141
  33. crawlo/extension/log_interval.py +57 -57
  34. crawlo/extension/log_stats.py +81 -81
  35. crawlo/extension/logging_extension.py +61 -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 +103 -103
  42. crawlo/factories/registry.py +84 -84
  43. crawlo/filters/__init__.py +154 -154
  44. crawlo/filters/aioredis_filter.py +257 -257
  45. crawlo/filters/memory_filter.py +269 -269
  46. crawlo/framework.py +292 -292
  47. crawlo/initialization/__init__.py +39 -39
  48. crawlo/initialization/built_in.py +425 -425
  49. crawlo/initialization/context.py +141 -141
  50. crawlo/initialization/core.py +193 -193
  51. crawlo/initialization/phases.py +148 -148
  52. crawlo/initialization/registry.py +145 -145
  53. crawlo/items/__init__.py +23 -23
  54. crawlo/items/base.py +23 -23
  55. crawlo/items/fields.py +52 -52
  56. crawlo/items/items.py +104 -104
  57. crawlo/logging/__init__.py +37 -37
  58. crawlo/logging/config.py +96 -96
  59. crawlo/logging/factory.py +128 -128
  60. crawlo/logging/manager.py +111 -111
  61. crawlo/middleware/__init__.py +21 -21
  62. crawlo/middleware/default_header.py +132 -132
  63. crawlo/middleware/download_delay.py +104 -104
  64. crawlo/middleware/middleware_manager.py +135 -135
  65. crawlo/middleware/offsite.py +123 -123
  66. crawlo/middleware/proxy.py +386 -386
  67. crawlo/middleware/request_ignore.py +86 -86
  68. crawlo/middleware/response_code.py +163 -163
  69. crawlo/middleware/response_filter.py +136 -136
  70. crawlo/middleware/retry.py +124 -124
  71. crawlo/middleware/simple_proxy.py +65 -65
  72. crawlo/mode_manager.py +212 -212
  73. crawlo/network/__init__.py +21 -21
  74. crawlo/network/request.py +379 -379
  75. crawlo/network/response.py +359 -359
  76. crawlo/pipelines/__init__.py +21 -21
  77. crawlo/pipelines/bloom_dedup_pipeline.py +156 -156
  78. crawlo/pipelines/console_pipeline.py +39 -39
  79. crawlo/pipelines/csv_pipeline.py +316 -316
  80. crawlo/pipelines/database_dedup_pipeline.py +222 -222
  81. crawlo/pipelines/json_pipeline.py +218 -218
  82. crawlo/pipelines/memory_dedup_pipeline.py +115 -115
  83. crawlo/pipelines/mongo_pipeline.py +131 -131
  84. crawlo/pipelines/mysql_pipeline.py +325 -325
  85. crawlo/pipelines/pipeline_manager.py +76 -76
  86. crawlo/pipelines/redis_dedup_pipeline.py +166 -166
  87. crawlo/project.py +327 -327
  88. crawlo/queue/pqueue.py +42 -42
  89. crawlo/queue/queue_manager.py +522 -503
  90. crawlo/queue/redis_priority_queue.py +367 -326
  91. crawlo/settings/__init__.py +7 -7
  92. crawlo/settings/default_settings.py +321 -321
  93. crawlo/settings/setting_manager.py +214 -214
  94. crawlo/spider/__init__.py +657 -657
  95. crawlo/stats_collector.py +73 -73
  96. crawlo/subscriber.py +129 -129
  97. crawlo/task_manager.py +138 -138
  98. crawlo/templates/crawlo.cfg.tmpl +10 -10
  99. crawlo/templates/project/__init__.py.tmpl +3 -3
  100. crawlo/templates/project/items.py.tmpl +17 -17
  101. crawlo/templates/project/middlewares.py.tmpl +118 -118
  102. crawlo/templates/project/pipelines.py.tmpl +96 -96
  103. crawlo/templates/project/settings.py.tmpl +170 -167
  104. crawlo/templates/project/settings_distributed.py.tmpl +169 -166
  105. crawlo/templates/project/settings_gentle.py.tmpl +166 -166
  106. crawlo/templates/project/settings_high_performance.py.tmpl +167 -167
  107. crawlo/templates/project/settings_minimal.py.tmpl +65 -65
  108. crawlo/templates/project/settings_simple.py.tmpl +164 -164
  109. crawlo/templates/project/spiders/__init__.py.tmpl +9 -9
  110. crawlo/templates/run.py.tmpl +34 -34
  111. crawlo/templates/spider/spider.py.tmpl +143 -143
  112. crawlo/templates/spiders_init.py.tmpl +9 -9
  113. crawlo/tools/__init__.py +200 -200
  114. crawlo/tools/anti_crawler.py +268 -268
  115. crawlo/tools/authenticated_proxy.py +240 -240
  116. crawlo/tools/data_formatter.py +225 -225
  117. crawlo/tools/data_validator.py +180 -180
  118. crawlo/tools/date_tools.py +289 -289
  119. crawlo/tools/distributed_coordinator.py +388 -388
  120. crawlo/tools/encoding_converter.py +127 -127
  121. crawlo/tools/network_diagnostic.py +364 -364
  122. crawlo/tools/request_tools.py +82 -82
  123. crawlo/tools/retry_mechanism.py +224 -224
  124. crawlo/tools/scenario_adapter.py +262 -262
  125. crawlo/tools/text_cleaner.py +232 -232
  126. crawlo/utils/__init__.py +34 -34
  127. crawlo/utils/batch_processor.py +259 -259
  128. crawlo/utils/class_loader.py +25 -25
  129. crawlo/utils/controlled_spider_mixin.py +439 -439
  130. crawlo/utils/db_helper.py +343 -343
  131. crawlo/utils/enhanced_error_handler.py +356 -356
  132. crawlo/utils/env_config.py +142 -142
  133. crawlo/utils/error_handler.py +165 -165
  134. crawlo/utils/func_tools.py +82 -82
  135. crawlo/utils/large_scale_config.py +286 -286
  136. crawlo/utils/large_scale_helper.py +344 -344
  137. crawlo/utils/log.py +79 -79
  138. crawlo/utils/performance_monitor.py +285 -285
  139. crawlo/utils/queue_helper.py +175 -175
  140. crawlo/utils/redis_connection_pool.py +388 -388
  141. crawlo/utils/redis_key_validator.py +198 -198
  142. crawlo/utils/request.py +267 -267
  143. crawlo/utils/request_serializer.py +225 -225
  144. crawlo/utils/spider_loader.py +61 -61
  145. crawlo/utils/system.py +11 -11
  146. crawlo/utils/tools.py +4 -4
  147. crawlo/utils/url.py +39 -39
  148. {crawlo-1.3.6.dist-info → crawlo-1.3.7.dist-info}/METADATA +1199 -1126
  149. crawlo-1.3.7.dist-info/RECORD +292 -0
  150. examples/__init__.py +7 -7
  151. tests/__init__.py +7 -7
  152. tests/advanced_tools_example.py +275 -275
  153. tests/authenticated_proxy_example.py +106 -106
  154. tests/baidu_performance_test.py +108 -108
  155. tests/baidu_test.py +59 -59
  156. tests/cleaners_example.py +160 -160
  157. tests/comprehensive_framework_test.py +212 -212
  158. tests/comprehensive_test.py +81 -81
  159. tests/comprehensive_testing_summary.md +186 -186
  160. tests/config_validation_demo.py +142 -142
  161. tests/controlled_spider_example.py +205 -205
  162. tests/date_tools_example.py +180 -180
  163. tests/debug_configure.py +69 -69
  164. tests/debug_framework_logger.py +84 -84
  165. tests/debug_log_config.py +126 -126
  166. tests/debug_log_levels.py +63 -63
  167. tests/debug_pipelines.py +66 -66
  168. tests/detailed_log_test.py +233 -233
  169. tests/distributed_test.py +66 -66
  170. tests/distributed_test_debug.py +76 -76
  171. tests/dynamic_loading_example.py +523 -523
  172. tests/dynamic_loading_test.py +104 -104
  173. tests/env_config_example.py +133 -133
  174. tests/error_handling_example.py +171 -171
  175. tests/final_comprehensive_test.py +151 -151
  176. tests/final_log_test.py +260 -260
  177. tests/final_validation_test.py +182 -182
  178. tests/fix_log_test.py +142 -142
  179. tests/framework_performance_test.py +202 -202
  180. tests/log_buffering_test.py +111 -111
  181. tests/log_generation_timing_test.py +153 -153
  182. tests/optimized_performance_test.py +211 -211
  183. tests/performance_comparison.py +245 -245
  184. tests/queue_blocking_test.py +113 -113
  185. tests/queue_test.py +89 -89
  186. tests/redis_key_validation_demo.py +130 -130
  187. tests/request_params_example.py +150 -150
  188. tests/response_improvements_example.py +144 -144
  189. tests/scrapy_comparison/ofweek_scrapy.py +138 -138
  190. tests/scrapy_comparison/scrapy_test.py +133 -133
  191. tests/simple_command_test.py +119 -119
  192. tests/simple_crawlo_test.py +127 -127
  193. tests/simple_log_test.py +57 -57
  194. tests/simple_log_test2.py +137 -137
  195. tests/simple_optimization_test.py +128 -128
  196. tests/simple_queue_type_test.py +41 -41
  197. tests/simple_spider_test.py +49 -49
  198. tests/simple_test.py +47 -47
  199. tests/spider_log_timing_test.py +177 -177
  200. tests/test_advanced_tools.py +148 -148
  201. tests/test_all_commands.py +230 -230
  202. tests/test_all_redis_key_configs.py +145 -145
  203. tests/test_authenticated_proxy.py +141 -141
  204. tests/test_batch_processor.py +178 -178
  205. tests/test_cleaners.py +54 -54
  206. tests/test_component_factory.py +174 -174
  207. tests/test_comprehensive.py +146 -146
  208. tests/test_config_consistency.py +80 -80
  209. tests/test_config_merge.py +152 -152
  210. tests/test_config_validator.py +182 -182
  211. tests/test_controlled_spider_mixin.py +79 -79
  212. tests/test_crawlo_proxy_integration.py +108 -108
  213. tests/test_date_tools.py +123 -123
  214. tests/test_default_header_middleware.py +158 -158
  215. tests/test_distributed.py +65 -65
  216. tests/test_double_crawlo_fix.py +204 -207
  217. tests/test_double_crawlo_fix_simple.py +124 -124
  218. tests/test_download_delay_middleware.py +221 -221
  219. tests/test_downloader_proxy_compatibility.py +268 -268
  220. tests/test_dynamic_downloaders_proxy.py +124 -124
  221. tests/test_dynamic_proxy.py +92 -92
  222. tests/test_dynamic_proxy_config.py +146 -146
  223. tests/test_dynamic_proxy_real.py +109 -109
  224. tests/test_edge_cases.py +303 -303
  225. tests/test_enhanced_error_handler.py +270 -270
  226. tests/test_enhanced_error_handler_comprehensive.py +245 -245
  227. tests/test_env_config.py +121 -121
  228. tests/test_error_handler_compatibility.py +112 -112
  229. tests/test_factories.py +252 -252
  230. tests/test_final_validation.py +153 -153
  231. tests/test_framework_env_usage.py +103 -103
  232. tests/test_framework_logger.py +66 -66
  233. tests/test_framework_startup.py +64 -64
  234. tests/test_get_component_logger.py +83 -83
  235. tests/test_integration.py +169 -169
  236. tests/test_item_dedup_redis_key.py +122 -122
  237. tests/test_large_scale_config.py +112 -112
  238. tests/test_large_scale_helper.py +235 -235
  239. tests/test_logging_system.py +282 -282
  240. tests/test_mode_change.py +72 -72
  241. tests/test_mode_consistency.py +51 -51
  242. tests/test_offsite_middleware.py +221 -221
  243. tests/test_parsel.py +29 -29
  244. tests/test_performance.py +327 -327
  245. tests/test_performance_monitor.py +115 -115
  246. tests/test_proxy_api.py +264 -264
  247. tests/test_proxy_health_check.py +32 -32
  248. tests/test_proxy_middleware.py +121 -121
  249. tests/test_proxy_middleware_enhanced.py +216 -216
  250. tests/test_proxy_middleware_integration.py +136 -136
  251. tests/test_proxy_middleware_refactored.py +184 -184
  252. tests/test_proxy_providers.py +56 -56
  253. tests/test_proxy_stats.py +19 -19
  254. tests/test_proxy_strategies.py +59 -59
  255. tests/test_queue_empty_check.py +41 -41
  256. tests/test_queue_manager_double_crawlo.py +173 -173
  257. tests/test_queue_manager_redis_key.py +179 -176
  258. tests/test_queue_naming.py +155 -0
  259. tests/test_queue_type.py +106 -106
  260. tests/test_random_user_agent.py +72 -72
  261. tests/test_real_scenario_proxy.py +195 -195
  262. tests/test_redis_config.py +28 -28
  263. tests/test_redis_connection_pool.py +294 -294
  264. tests/test_redis_key_naming.py +181 -181
  265. tests/test_redis_key_validator.py +123 -123
  266. tests/test_redis_queue.py +224 -224
  267. tests/test_redis_queue_name_fix.py +176 -0
  268. tests/test_request_ignore_middleware.py +182 -182
  269. tests/test_request_params.py +111 -111
  270. tests/test_request_serialization.py +70 -70
  271. tests/test_response_code_middleware.py +349 -349
  272. tests/test_response_filter_middleware.py +427 -427
  273. tests/test_response_improvements.py +152 -152
  274. tests/test_retry_middleware.py +241 -241
  275. tests/test_scheduler.py +252 -252
  276. tests/test_scheduler_config_update.py +133 -133
  277. tests/test_simple_response.py +61 -61
  278. tests/test_telecom_spider_redis_key.py +205 -205
  279. tests/test_template_content.py +87 -87
  280. tests/test_template_redis_key.py +134 -134
  281. tests/test_tools.py +159 -159
  282. tests/test_user_agents.py +96 -96
  283. tests/tools_example.py +260 -260
  284. tests/untested_features_report.md +138 -138
  285. tests/verify_debug.py +51 -51
  286. tests/verify_distributed.py +117 -117
  287. tests/verify_log_fix.py +111 -111
  288. crawlo-1.3.6.dist-info/RECORD +0 -290
  289. {crawlo-1.3.6.dist-info → crawlo-1.3.7.dist-info}/WHEEL +0 -0
  290. {crawlo-1.3.6.dist-info → crawlo-1.3.7.dist-info}/entry_points.txt +0 -0
  291. {crawlo-1.3.6.dist-info → crawlo-1.3.7.dist-info}/top_level.txt +0 -0
@@ -1,283 +1,283 @@
1
- #!/usr/bin/env python
2
- # -*- coding: UTF-8 -*-
3
- """
4
- 全面测试Crawlo框架日志系统的所有功能
5
- """
6
-
7
- import os
8
- import sys
9
- import tempfile
10
- import shutil
11
- from pathlib import Path
12
-
13
- # 添加项目根目录到Python路径
14
- project_root = Path(__file__).parent.parent
15
- sys.path.insert(0, str(project_root))
16
-
17
- from crawlo.logging import configure_logging as configure, get_logger, LogManager
18
- from crawlo.logging.config import LogConfig
19
-
20
-
21
- def test_basic_logging():
22
- """测试基本日志功能"""
23
- print("=== 测试基本日志功能 ===")
24
-
25
- # 1. 测试默认配置
26
- print("1. 测试默认配置...")
27
- configure()
28
- config = LogManager().config
29
- print(f" 默认级别: {config.level}")
30
- print(f" 默认格式: {config.format}")
31
-
32
- # 2. 测试获取logger
33
- logger = get_logger('test.basic')
34
- print(f" Logger名称: {logger.name}")
35
- print(f" Handlers数量: {len(logger.handlers)}")
36
-
37
- # 3. 测试日志输出
38
- logger.info("这是一条INFO级别日志")
39
- logger.warning("这是一条WARNING级别日志")
40
- logger.error("这是一条ERROR级别日志")
41
- print(" 日志输出完成")
42
-
43
-
44
- def test_file_logging():
45
- """测试文件日志功能"""
46
- print("\n=== 测试文件日志功能 ===")
47
-
48
- # 创建临时目录
49
- temp_dir = tempfile.mkdtemp()
50
- log_file = os.path.join(temp_dir, 'test.log')
51
-
52
- try:
53
- # 1. 配置文件日志
54
- print("1. 配置文件日志...")
55
- configure(
56
- LOG_LEVEL='DEBUG',
57
- LOG_FILE=log_file,
58
- LOG_MAX_BYTES=1024, # 小的轮转大小用于测试
59
- LOG_BACKUP_COUNT=3
60
- )
61
-
62
- config = LogManager().config
63
- print(f" 日志文件: {config.file_path}")
64
- print(f" 轮转大小: {config.max_bytes}")
65
- print(f" 备份数量: {config.backup_count}")
66
-
67
- # 2. 获取logger并测试输出
68
- logger = get_logger('test.file')
69
- print("2. 测试日志输出...")
70
-
71
- # 输出多条日志以触发轮转
72
- for i in range(50):
73
- logger.debug(f"这是第{i+1}条DEBUG日志,用于测试轮转功能")
74
- logger.info(f"这是第{i+1}条INFO日志,用于测试轮转功能")
75
-
76
- # 3. 检查日志文件
77
- print("3. 检查日志文件...")
78
- if os.path.exists(log_file):
79
- file_size = os.path.getsize(log_file)
80
- print(f" 主日志文件大小: {file_size} 字节")
81
- else:
82
- print(" 主日志文件不存在!")
83
-
84
- # 检查备份文件
85
- for i in range(1, 4):
86
- backup_file = f"{log_file}.{i}"
87
- if os.path.exists(backup_file):
88
- backup_size = os.path.getsize(backup_file)
89
- print(f" 备份文件 {i} 大小: {backup_size} 字节")
90
- else:
91
- print(f" 备份文件 {i} 不存在")
92
-
93
- finally:
94
- # 清理临时目录
95
- shutil.rmtree(temp_dir, ignore_errors=True)
96
-
97
-
98
- def test_console_logging():
99
- """测试控制台日志功能"""
100
- print("\n=== 测试控制台日志功能 ===")
101
-
102
- # 1. 配置仅控制台日志
103
- print("1. 配置仅控制台日志...")
104
- configure(
105
- LOG_LEVEL='INFO',
106
- LOG_CONSOLE_ENABLED=True,
107
- LOG_FILE_ENABLED=False
108
- )
109
-
110
- config = LogManager().config
111
- print(f" 控制台启用: {config.console_enabled}")
112
- print(f" 文件日志启用: {config.file_enabled}")
113
-
114
- # 2. 测试日志输出
115
- logger = get_logger('test.console')
116
- print("2. 测试日志输出...")
117
- logger.info("这条日志应该只在控制台显示")
118
- logger.debug("这条DEBUG日志不应该显示(级别低于INFO)")
119
-
120
-
121
- def test_module_level_logging():
122
- """测试模块级别日志功能"""
123
- print("\n=== 测试模块级别日志功能 ===")
124
-
125
- # 1. 配置模块级别
126
- print("1. 配置模块级别...")
127
- configure(
128
- LOG_LEVEL='WARNING',
129
- LOG_LEVELS={
130
- 'test.module.high': 'DEBUG',
131
- 'test.module.low': 'ERROR'
132
- }
133
- )
134
-
135
- config = LogManager().config
136
- print(f" 默认级别: {config.level}")
137
- print(f" 模块级别配置: {config.module_levels}")
138
-
139
- # 2. 测试不同模块的日志级别
140
- print("2. 测试不同模块的日志级别...")
141
-
142
- # 高级别模块(DEBUG)
143
- high_logger = get_logger('test.module.high')
144
- print(f" high_logger 级别: {logging.getLevelName(high_logger.handlers[0].level)}")
145
- high_logger.debug("这条DEBUG日志应该显示")
146
- high_logger.info("这条INFO日志应该显示")
147
-
148
- # 低级别模块(ERROR)
149
- low_logger = get_logger('test.module.low')
150
- print(f" low_logger 级别: {logging.getLevelName(low_logger.handlers[0].level)}")
151
- low_logger.info("这条INFO日志不应该显示")
152
- low_logger.error("这条ERROR日志应该显示")
153
-
154
- # 默认级别模块(WARNING)
155
- default_logger = get_logger('test.module.default')
156
- print(f" default_logger 级别: {logging.getLevelName(default_logger.handlers[0].level)}")
157
- default_logger.info("这条INFO日志不应该显示")
158
- default_logger.warning("这条WARNING日志应该显示")
159
-
160
-
161
- def test_log_config_from_settings():
162
- """测试从settings对象创建配置"""
163
- print("\n=== 测试从settings对象创建配置 ===")
164
-
165
- # 1. 创建模拟settings对象
166
- class MockSettings:
167
- def __init__(self):
168
- self.LOG_LEVEL = 'DEBUG'
169
- self.LOG_FILE = 'test_settings.log'
170
- self.LOG_MAX_BYTES = 5 * 1024 * 1024
171
- self.LOG_BACKUP_COUNT = 2
172
- self.LOG_ENCODING = 'utf-8'
173
- self.LOG_CONSOLE_ENABLED = True
174
- self.LOG_FILE_ENABLED = True
175
- self.LOG_LEVELS = {'test.custom': 'INFO'}
176
-
177
- # 2. 从settings创建配置
178
- print("2. 从settings创建配置...")
179
- settings = MockSettings()
180
- config = LogConfig.from_settings(settings)
181
-
182
- print(f" 级别: {config.level}")
183
- print(f" 文件: {config.file_path}")
184
- print(f" 轮转大小: {config.max_bytes}")
185
- print(f" 备份数量: {config.backup_count}")
186
- print(f" 编码: {config.encoding}")
187
- print(f" 控制台启用: {config.console_enabled}")
188
- print(f" 文件启用: {config.file_enabled}")
189
- print(f" 模块级别: {config.module_levels}")
190
-
191
-
192
- def test_log_config_validation():
193
- """测试日志配置验证功能"""
194
- print("\n=== 测试日志配置验证功能 ===")
195
-
196
- # 1. 测试有效配置
197
- print("1. 测试有效配置...")
198
- valid_config = LogConfig(
199
- level='INFO',
200
- file_path='test_valid.log'
201
- )
202
- is_valid = valid_config.validate()
203
- print(f" 有效配置验证结果: {is_valid}")
204
-
205
- # 2. 测试无效级别
206
- print("2. 测试无效级别...")
207
- invalid_config = LogConfig(
208
- level='INVALID',
209
- file_path='test_invalid.log'
210
- )
211
- is_valid = invalid_config.validate()
212
- print(f" 无效级别验证结果: {is_valid}")
213
-
214
- # 3. 测试目录创建
215
- print("3. 测试目录创建...")
216
- temp_dir = tempfile.mkdtemp()
217
- nested_log_file = os.path.join(temp_dir, 'subdir', 'test.log')
218
-
219
- config = LogConfig(
220
- level='INFO',
221
- file_path=nested_log_file
222
- )
223
- is_valid = config.validate()
224
- print(f" 目录创建验证结果: {is_valid}")
225
-
226
- # 清理
227
- shutil.rmtree(temp_dir, ignore_errors=True)
228
-
229
-
230
- def test_logger_caching():
231
- """测试Logger缓存功能"""
232
- print("\n=== 测试Logger缓存功能 ===")
233
-
234
- # 重置配置以确保测试环境干净
235
- LogManager().reset()
236
-
237
- # 1. 获取同一个logger多次
238
- print("1. 测试缓存...")
239
- logger1 = get_logger('test.cache')
240
- logger2 = get_logger('test.cache')
241
-
242
- print(f" 第一次获取: {id(logger1)}")
243
- print(f" 第二次获取: {id(logger2)}")
244
- print(f" 是否为同一对象: {logger1 is logger2}")
245
-
246
- # 2. 测试配置更新后缓存刷新
247
- print("2. 测试配置更新...")
248
- configure(LOG_LEVEL='ERROR')
249
- logger3 = get_logger('test.cache')
250
- print(f" 更新配置后获取: {id(logger3)}")
251
- print(f" 是否与之前相同: {logger1 is logger3}")
252
-
253
-
254
- def main():
255
- """主测试函数"""
256
- print("开始全面测试Crawlo框架日志系统...")
257
-
258
- try:
259
- # 运行所有测试
260
- test_basic_logging()
261
- test_file_logging()
262
- test_console_logging()
263
- test_module_level_logging()
264
- test_log_config_from_settings()
265
- test_log_config_validation()
266
- test_logger_caching()
267
-
268
- print("\n=== 所有测试完成 ===")
269
-
270
- except Exception as e:
271
- print(f"\n测试过程中出现错误: {e}")
272
- import traceback
273
- traceback.print_exc()
274
- return 1
275
-
276
- return 0
277
-
278
-
279
- # 导入logging模块用于测试
280
- import logging
281
-
282
- if __name__ == '__main__':
1
+ #!/usr/bin/env python
2
+ # -*- coding: UTF-8 -*-
3
+ """
4
+ 全面测试Crawlo框架日志系统的所有功能
5
+ """
6
+
7
+ import os
8
+ import sys
9
+ import tempfile
10
+ import shutil
11
+ from pathlib import Path
12
+
13
+ # 添加项目根目录到Python路径
14
+ project_root = Path(__file__).parent.parent
15
+ sys.path.insert(0, str(project_root))
16
+
17
+ from crawlo.logging import configure_logging as configure, get_logger, LogManager
18
+ from crawlo.logging.config import LogConfig
19
+
20
+
21
+ def test_basic_logging():
22
+ """测试基本日志功能"""
23
+ print("=== 测试基本日志功能 ===")
24
+
25
+ # 1. 测试默认配置
26
+ print("1. 测试默认配置...")
27
+ configure()
28
+ config = LogManager().config
29
+ print(f" 默认级别: {config.level}")
30
+ print(f" 默认格式: {config.format}")
31
+
32
+ # 2. 测试获取logger
33
+ logger = get_logger('test.basic')
34
+ print(f" Logger名称: {logger.name}")
35
+ print(f" Handlers数量: {len(logger.handlers)}")
36
+
37
+ # 3. 测试日志输出
38
+ logger.info("这是一条INFO级别日志")
39
+ logger.warning("这是一条WARNING级别日志")
40
+ logger.error("这是一条ERROR级别日志")
41
+ print(" 日志输出完成")
42
+
43
+
44
+ def test_file_logging():
45
+ """测试文件日志功能"""
46
+ print("\n=== 测试文件日志功能 ===")
47
+
48
+ # 创建临时目录
49
+ temp_dir = tempfile.mkdtemp()
50
+ log_file = os.path.join(temp_dir, 'test.log')
51
+
52
+ try:
53
+ # 1. 配置文件日志
54
+ print("1. 配置文件日志...")
55
+ configure(
56
+ LOG_LEVEL='DEBUG',
57
+ LOG_FILE=log_file,
58
+ LOG_MAX_BYTES=1024, # 小的轮转大小用于测试
59
+ LOG_BACKUP_COUNT=3
60
+ )
61
+
62
+ config = LogManager().config
63
+ print(f" 日志文件: {config.file_path}")
64
+ print(f" 轮转大小: {config.max_bytes}")
65
+ print(f" 备份数量: {config.backup_count}")
66
+
67
+ # 2. 获取logger并测试输出
68
+ logger = get_logger('test.file')
69
+ print("2. 测试日志输出...")
70
+
71
+ # 输出多条日志以触发轮转
72
+ for i in range(50):
73
+ logger.debug(f"这是第{i+1}条DEBUG日志,用于测试轮转功能")
74
+ logger.info(f"这是第{i+1}条INFO日志,用于测试轮转功能")
75
+
76
+ # 3. 检查日志文件
77
+ print("3. 检查日志文件...")
78
+ if os.path.exists(log_file):
79
+ file_size = os.path.getsize(log_file)
80
+ print(f" 主日志文件大小: {file_size} 字节")
81
+ else:
82
+ print(" 主日志文件不存在!")
83
+
84
+ # 检查备份文件
85
+ for i in range(1, 4):
86
+ backup_file = f"{log_file}.{i}"
87
+ if os.path.exists(backup_file):
88
+ backup_size = os.path.getsize(backup_file)
89
+ print(f" 备份文件 {i} 大小: {backup_size} 字节")
90
+ else:
91
+ print(f" 备份文件 {i} 不存在")
92
+
93
+ finally:
94
+ # 清理临时目录
95
+ shutil.rmtree(temp_dir, ignore_errors=True)
96
+
97
+
98
+ def test_console_logging():
99
+ """测试控制台日志功能"""
100
+ print("\n=== 测试控制台日志功能 ===")
101
+
102
+ # 1. 配置仅控制台日志
103
+ print("1. 配置仅控制台日志...")
104
+ configure(
105
+ LOG_LEVEL='INFO',
106
+ LOG_CONSOLE_ENABLED=True,
107
+ LOG_FILE_ENABLED=False
108
+ )
109
+
110
+ config = LogManager().config
111
+ print(f" 控制台启用: {config.console_enabled}")
112
+ print(f" 文件日志启用: {config.file_enabled}")
113
+
114
+ # 2. 测试日志输出
115
+ logger = get_logger('test.console')
116
+ print("2. 测试日志输出...")
117
+ logger.info("这条日志应该只在控制台显示")
118
+ logger.debug("这条DEBUG日志不应该显示(级别低于INFO)")
119
+
120
+
121
+ def test_module_level_logging():
122
+ """测试模块级别日志功能"""
123
+ print("\n=== 测试模块级别日志功能 ===")
124
+
125
+ # 1. 配置模块级别
126
+ print("1. 配置模块级别...")
127
+ configure(
128
+ LOG_LEVEL='WARNING',
129
+ LOG_LEVELS={
130
+ 'test.module.high': 'DEBUG',
131
+ 'test.module.low': 'ERROR'
132
+ }
133
+ )
134
+
135
+ config = LogManager().config
136
+ print(f" 默认级别: {config.level}")
137
+ print(f" 模块级别配置: {config.module_levels}")
138
+
139
+ # 2. 测试不同模块的日志级别
140
+ print("2. 测试不同模块的日志级别...")
141
+
142
+ # 高级别模块(DEBUG)
143
+ high_logger = get_logger('test.module.high')
144
+ print(f" high_logger 级别: {logging.getLevelName(high_logger.handlers[0].level)}")
145
+ high_logger.debug("这条DEBUG日志应该显示")
146
+ high_logger.info("这条INFO日志应该显示")
147
+
148
+ # 低级别模块(ERROR)
149
+ low_logger = get_logger('test.module.low')
150
+ print(f" low_logger 级别: {logging.getLevelName(low_logger.handlers[0].level)}")
151
+ low_logger.info("这条INFO日志不应该显示")
152
+ low_logger.error("这条ERROR日志应该显示")
153
+
154
+ # 默认级别模块(WARNING)
155
+ default_logger = get_logger('test.module.default')
156
+ print(f" default_logger 级别: {logging.getLevelName(default_logger.handlers[0].level)}")
157
+ default_logger.info("这条INFO日志不应该显示")
158
+ default_logger.warning("这条WARNING日志应该显示")
159
+
160
+
161
+ def test_log_config_from_settings():
162
+ """测试从settings对象创建配置"""
163
+ print("\n=== 测试从settings对象创建配置 ===")
164
+
165
+ # 1. 创建模拟settings对象
166
+ class MockSettings:
167
+ def __init__(self):
168
+ self.LOG_LEVEL = 'DEBUG'
169
+ self.LOG_FILE = 'test_settings.log'
170
+ self.LOG_MAX_BYTES = 5 * 1024 * 1024
171
+ self.LOG_BACKUP_COUNT = 2
172
+ self.LOG_ENCODING = 'utf-8'
173
+ self.LOG_CONSOLE_ENABLED = True
174
+ self.LOG_FILE_ENABLED = True
175
+ self.LOG_LEVELS = {'test.custom': 'INFO'}
176
+
177
+ # 2. 从settings创建配置
178
+ print("2. 从settings创建配置...")
179
+ settings = MockSettings()
180
+ config = LogConfig.from_settings(settings)
181
+
182
+ print(f" 级别: {config.level}")
183
+ print(f" 文件: {config.file_path}")
184
+ print(f" 轮转大小: {config.max_bytes}")
185
+ print(f" 备份数量: {config.backup_count}")
186
+ print(f" 编码: {config.encoding}")
187
+ print(f" 控制台启用: {config.console_enabled}")
188
+ print(f" 文件启用: {config.file_enabled}")
189
+ print(f" 模块级别: {config.module_levels}")
190
+
191
+
192
+ def test_log_config_validation():
193
+ """测试日志配置验证功能"""
194
+ print("\n=== 测试日志配置验证功能 ===")
195
+
196
+ # 1. 测试有效配置
197
+ print("1. 测试有效配置...")
198
+ valid_config = LogConfig(
199
+ level='INFO',
200
+ file_path='test_valid.log'
201
+ )
202
+ is_valid = valid_config.validate()
203
+ print(f" 有效配置验证结果: {is_valid}")
204
+
205
+ # 2. 测试无效级别
206
+ print("2. 测试无效级别...")
207
+ invalid_config = LogConfig(
208
+ level='INVALID',
209
+ file_path='test_invalid.log'
210
+ )
211
+ is_valid = invalid_config.validate()
212
+ print(f" 无效级别验证结果: {is_valid}")
213
+
214
+ # 3. 测试目录创建
215
+ print("3. 测试目录创建...")
216
+ temp_dir = tempfile.mkdtemp()
217
+ nested_log_file = os.path.join(temp_dir, 'subdir', 'test.log')
218
+
219
+ config = LogConfig(
220
+ level='INFO',
221
+ file_path=nested_log_file
222
+ )
223
+ is_valid = config.validate()
224
+ print(f" 目录创建验证结果: {is_valid}")
225
+
226
+ # 清理
227
+ shutil.rmtree(temp_dir, ignore_errors=True)
228
+
229
+
230
+ def test_logger_caching():
231
+ """测试Logger缓存功能"""
232
+ print("\n=== 测试Logger缓存功能 ===")
233
+
234
+ # 重置配置以确保测试环境干净
235
+ LogManager().reset()
236
+
237
+ # 1. 获取同一个logger多次
238
+ print("1. 测试缓存...")
239
+ logger1 = get_logger('test.cache')
240
+ logger2 = get_logger('test.cache')
241
+
242
+ print(f" 第一次获取: {id(logger1)}")
243
+ print(f" 第二次获取: {id(logger2)}")
244
+ print(f" 是否为同一对象: {logger1 is logger2}")
245
+
246
+ # 2. 测试配置更新后缓存刷新
247
+ print("2. 测试配置更新...")
248
+ configure(LOG_LEVEL='ERROR')
249
+ logger3 = get_logger('test.cache')
250
+ print(f" 更新配置后获取: {id(logger3)}")
251
+ print(f" 是否与之前相同: {logger1 is logger3}")
252
+
253
+
254
+ def main():
255
+ """主测试函数"""
256
+ print("开始全面测试Crawlo框架日志系统...")
257
+
258
+ try:
259
+ # 运行所有测试
260
+ test_basic_logging()
261
+ test_file_logging()
262
+ test_console_logging()
263
+ test_module_level_logging()
264
+ test_log_config_from_settings()
265
+ test_log_config_validation()
266
+ test_logger_caching()
267
+
268
+ print("\n=== 所有测试完成 ===")
269
+
270
+ except Exception as e:
271
+ print(f"\n测试过程中出现错误: {e}")
272
+ import traceback
273
+ traceback.print_exc()
274
+ return 1
275
+
276
+ return 0
277
+
278
+
279
+ # 导入logging模块用于测试
280
+ import logging
281
+
282
+ if __name__ == '__main__':
283
283
  sys.exit(main())