crawlo 1.4.1__py3-none-any.whl → 1.4.3__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 (326) hide show
  1. crawlo/__init__.py +93 -93
  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 +52 -52
  16. crawlo/core/engine.py +438 -439
  17. crawlo/core/processor.py +47 -47
  18. crawlo/core/scheduler.py +291 -257
  19. crawlo/crawler.py +650 -650
  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 +233 -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 +63 -63
  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 +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 +44 -44
  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 +45 -37
  58. crawlo/logging/async_handler.py +181 -0
  59. crawlo/logging/config.py +196 -96
  60. crawlo/logging/factory.py +171 -128
  61. crawlo/logging/manager.py +111 -111
  62. crawlo/logging/monitor.py +153 -0
  63. crawlo/logging/sampler.py +167 -0
  64. crawlo/middleware/__init__.py +21 -21
  65. crawlo/middleware/default_header.py +132 -132
  66. crawlo/middleware/download_delay.py +104 -104
  67. crawlo/middleware/middleware_manager.py +135 -135
  68. crawlo/middleware/offsite.py +123 -123
  69. crawlo/middleware/proxy.py +386 -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/middleware/simple_proxy.py +65 -65
  75. crawlo/mode_manager.py +219 -219
  76. crawlo/network/__init__.py +21 -21
  77. crawlo/network/request.py +379 -379
  78. crawlo/network/response.py +359 -359
  79. crawlo/pipelines/__init__.py +21 -21
  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 +131 -131
  87. crawlo/pipelines/mysql_pipeline.py +325 -325
  88. crawlo/pipelines/pipeline_manager.py +100 -84
  89. crawlo/pipelines/redis_dedup_pipeline.py +156 -156
  90. crawlo/project.py +349 -338
  91. crawlo/queue/pqueue.py +42 -42
  92. crawlo/queue/queue_manager.py +526 -522
  93. crawlo/queue/redis_priority_queue.py +370 -367
  94. crawlo/settings/__init__.py +7 -7
  95. crawlo/settings/default_settings.py +284 -284
  96. crawlo/settings/setting_manager.py +219 -219
  97. crawlo/spider/__init__.py +657 -657
  98. crawlo/stats_collector.py +73 -73
  99. crawlo/subscriber.py +129 -129
  100. crawlo/task_manager.py +138 -138
  101. crawlo/templates/crawlo.cfg.tmpl +10 -10
  102. crawlo/templates/project/__init__.py.tmpl +3 -3
  103. crawlo/templates/project/items.py.tmpl +17 -17
  104. crawlo/templates/project/middlewares.py.tmpl +118 -118
  105. crawlo/templates/project/pipelines.py.tmpl +96 -96
  106. crawlo/templates/project/settings.py.tmpl +170 -170
  107. crawlo/templates/project/settings_distributed.py.tmpl +169 -169
  108. crawlo/templates/project/settings_gentle.py.tmpl +166 -166
  109. crawlo/templates/project/settings_high_performance.py.tmpl +167 -167
  110. crawlo/templates/project/settings_minimal.py.tmpl +65 -65
  111. crawlo/templates/project/settings_simple.py.tmpl +164 -164
  112. crawlo/templates/project/spiders/__init__.py.tmpl +9 -9
  113. crawlo/templates/run.py.tmpl +34 -34
  114. crawlo/templates/spider/spider.py.tmpl +143 -143
  115. crawlo/templates/spiders_init.py.tmpl +9 -9
  116. crawlo/tools/__init__.py +200 -200
  117. crawlo/tools/anti_crawler.py +268 -268
  118. crawlo/tools/authenticated_proxy.py +240 -240
  119. crawlo/tools/data_formatter.py +225 -225
  120. crawlo/tools/data_validator.py +180 -180
  121. crawlo/tools/date_tools.py +289 -289
  122. crawlo/tools/distributed_coordinator.py +384 -384
  123. crawlo/tools/encoding_converter.py +127 -127
  124. crawlo/tools/network_diagnostic.py +364 -364
  125. crawlo/tools/request_tools.py +82 -82
  126. crawlo/tools/retry_mechanism.py +224 -224
  127. crawlo/tools/scenario_adapter.py +262 -262
  128. crawlo/tools/text_cleaner.py +232 -232
  129. crawlo/utils/__init__.py +34 -34
  130. crawlo/utils/batch_processor.py +259 -259
  131. crawlo/utils/class_loader.py +25 -25
  132. crawlo/utils/controlled_spider_mixin.py +439 -439
  133. crawlo/utils/db_helper.py +343 -343
  134. crawlo/utils/enhanced_error_handler.py +356 -356
  135. crawlo/utils/env_config.py +142 -142
  136. crawlo/utils/error_handler.py +165 -165
  137. crawlo/utils/fingerprint.py +122 -122
  138. crawlo/utils/func_tools.py +82 -82
  139. crawlo/utils/large_scale_config.py +286 -286
  140. crawlo/utils/large_scale_helper.py +344 -344
  141. crawlo/utils/log.py +79 -79
  142. crawlo/utils/performance_monitor.py +285 -285
  143. crawlo/utils/queue_helper.py +175 -175
  144. crawlo/utils/redis_connection_pool.py +388 -388
  145. crawlo/utils/redis_key_validator.py +198 -198
  146. crawlo/utils/request.py +267 -267
  147. crawlo/utils/request_serializer.py +225 -225
  148. crawlo/utils/spider_loader.py +61 -61
  149. crawlo/utils/system.py +11 -11
  150. crawlo/utils/tools.py +4 -4
  151. crawlo/utils/url.py +39 -39
  152. crawlo-1.4.3.dist-info/METADATA +190 -0
  153. crawlo-1.4.3.dist-info/RECORD +326 -0
  154. examples/__init__.py +7 -7
  155. examples/test_project/__init__.py +7 -7
  156. examples/test_project/run.py +34 -34
  157. examples/test_project/test_project/__init__.py +3 -3
  158. examples/test_project/test_project/items.py +17 -17
  159. examples/test_project/test_project/middlewares.py +118 -118
  160. examples/test_project/test_project/pipelines.py +96 -96
  161. examples/test_project/test_project/settings.py +169 -169
  162. examples/test_project/test_project/spiders/__init__.py +9 -9
  163. examples/test_project/test_project/spiders/of_week_dis.py +143 -143
  164. tests/__init__.py +7 -7
  165. tests/advanced_tools_example.py +275 -275
  166. tests/authenticated_proxy_example.py +106 -106
  167. tests/baidu_performance_test.py +108 -108
  168. tests/baidu_test.py +59 -59
  169. tests/cleaners_example.py +160 -160
  170. tests/comprehensive_framework_test.py +212 -212
  171. tests/comprehensive_test.py +81 -81
  172. tests/comprehensive_testing_summary.md +186 -186
  173. tests/config_validation_demo.py +142 -142
  174. tests/controlled_spider_example.py +205 -205
  175. tests/date_tools_example.py +180 -180
  176. tests/debug_configure.py +69 -69
  177. tests/debug_framework_logger.py +84 -84
  178. tests/debug_log_config.py +126 -126
  179. tests/debug_log_levels.py +63 -63
  180. tests/debug_pipelines.py +66 -66
  181. tests/detailed_log_test.py +233 -233
  182. tests/distributed_test.py +66 -66
  183. tests/distributed_test_debug.py +76 -76
  184. tests/dynamic_loading_example.py +523 -523
  185. tests/dynamic_loading_test.py +104 -104
  186. tests/env_config_example.py +133 -133
  187. tests/error_handling_example.py +171 -171
  188. tests/final_comprehensive_test.py +151 -151
  189. tests/final_log_test.py +260 -260
  190. tests/final_validation_test.py +182 -182
  191. tests/fix_log_test.py +142 -142
  192. tests/framework_performance_test.py +202 -202
  193. tests/log_buffering_test.py +111 -111
  194. tests/log_generation_timing_test.py +153 -153
  195. tests/optimized_performance_test.py +211 -211
  196. tests/performance_comparison.py +245 -245
  197. tests/queue_blocking_test.py +113 -113
  198. tests/queue_test.py +89 -89
  199. tests/redis_key_validation_demo.py +130 -130
  200. tests/request_params_example.py +150 -150
  201. tests/response_improvements_example.py +144 -144
  202. tests/scrapy_comparison/ofweek_scrapy.py +138 -138
  203. tests/scrapy_comparison/scrapy_test.py +133 -133
  204. tests/simple_command_test.py +119 -119
  205. tests/simple_crawlo_test.py +127 -127
  206. tests/simple_log_test.py +57 -57
  207. tests/simple_log_test2.py +137 -137
  208. tests/simple_optimization_test.py +128 -128
  209. tests/simple_queue_type_test.py +41 -41
  210. tests/simple_spider_test.py +49 -49
  211. tests/simple_test.py +47 -47
  212. tests/spider_log_timing_test.py +177 -177
  213. tests/test_advanced_tools.py +148 -148
  214. tests/test_all_commands.py +230 -230
  215. tests/test_all_pipeline_fingerprints.py +133 -133
  216. tests/test_all_redis_key_configs.py +145 -145
  217. tests/test_authenticated_proxy.py +141 -141
  218. tests/test_batch_processor.py +178 -178
  219. tests/test_cleaners.py +54 -54
  220. tests/test_component_factory.py +174 -174
  221. tests/test_comprehensive.py +146 -146
  222. tests/test_config_consistency.py +80 -80
  223. tests/test_config_merge.py +152 -152
  224. tests/test_config_validator.py +182 -182
  225. tests/test_controlled_spider_mixin.py +79 -79
  226. tests/test_crawlo_proxy_integration.py +108 -108
  227. tests/test_date_tools.py +123 -123
  228. tests/test_dedup_fix.py +220 -220
  229. tests/test_dedup_pipeline_consistency.py +125 -0
  230. tests/test_default_header_middleware.py +313 -313
  231. tests/test_distributed.py +65 -65
  232. tests/test_double_crawlo_fix.py +204 -204
  233. tests/test_double_crawlo_fix_simple.py +124 -124
  234. tests/test_download_delay_middleware.py +221 -221
  235. tests/test_downloader_proxy_compatibility.py +268 -268
  236. tests/test_dynamic_downloaders_proxy.py +124 -124
  237. tests/test_dynamic_proxy.py +92 -92
  238. tests/test_dynamic_proxy_config.py +146 -146
  239. tests/test_dynamic_proxy_real.py +109 -109
  240. tests/test_edge_cases.py +303 -303
  241. tests/test_enhanced_error_handler.py +270 -270
  242. tests/test_enhanced_error_handler_comprehensive.py +245 -245
  243. tests/test_env_config.py +121 -121
  244. tests/test_error_handler_compatibility.py +112 -112
  245. tests/test_factories.py +252 -252
  246. tests/test_final_validation.py +153 -153
  247. tests/test_fingerprint_consistency.py +135 -135
  248. tests/test_fingerprint_simple.py +51 -51
  249. tests/test_framework_env_usage.py +103 -103
  250. tests/test_framework_logger.py +66 -66
  251. tests/test_framework_startup.py +64 -64
  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_config.py +112 -112
  257. tests/test_large_scale_helper.py +235 -235
  258. tests/test_logging_enhancements.py +375 -0
  259. tests/test_logging_final.py +185 -0
  260. tests/test_logging_integration.py +313 -0
  261. tests/test_logging_system.py +282 -282
  262. tests/test_middleware_debug.py +142 -0
  263. tests/test_mode_change.py +72 -72
  264. tests/test_mode_consistency.py +51 -51
  265. tests/test_offsite_middleware.py +244 -244
  266. tests/test_offsite_middleware_simple.py +203 -203
  267. tests/test_parsel.py +29 -29
  268. tests/test_performance.py +327 -327
  269. tests/test_performance_monitor.py +115 -115
  270. tests/test_pipeline_fingerprint_consistency.py +86 -86
  271. tests/test_priority_behavior.py +212 -0
  272. tests/test_priority_consistency.py +152 -0
  273. tests/test_priority_consistency_fixed.py +250 -0
  274. tests/test_proxy_api.py +264 -264
  275. tests/test_proxy_health_check.py +32 -32
  276. tests/test_proxy_middleware.py +121 -121
  277. tests/test_proxy_middleware_enhanced.py +216 -216
  278. tests/test_proxy_middleware_integration.py +136 -136
  279. tests/test_proxy_middleware_refactored.py +184 -184
  280. tests/test_proxy_providers.py +56 -56
  281. tests/test_proxy_stats.py +19 -19
  282. tests/test_proxy_strategies.py +59 -59
  283. tests/test_queue_empty_check.py +41 -41
  284. tests/test_queue_manager_double_crawlo.py +173 -173
  285. tests/test_queue_manager_redis_key.py +179 -179
  286. tests/test_queue_naming.py +154 -154
  287. tests/test_queue_type.py +106 -106
  288. tests/test_queue_type_redis_config_consistency.py +131 -0
  289. tests/test_random_headers_default.py +323 -0
  290. tests/test_random_headers_necessity.py +309 -0
  291. tests/test_random_user_agent.py +72 -72
  292. tests/test_real_scenario_proxy.py +195 -195
  293. tests/test_redis_config.py +28 -28
  294. tests/test_redis_connection_pool.py +294 -294
  295. tests/test_redis_key_naming.py +181 -181
  296. tests/test_redis_key_validator.py +123 -123
  297. tests/test_redis_queue.py +224 -224
  298. tests/test_redis_queue_name_fix.py +175 -175
  299. tests/test_redis_queue_type_fallback.py +130 -0
  300. tests/test_request_ignore_middleware.py +182 -182
  301. tests/test_request_params.py +111 -111
  302. tests/test_request_serialization.py +70 -70
  303. tests/test_response_code_middleware.py +349 -349
  304. tests/test_response_filter_middleware.py +427 -427
  305. tests/test_response_improvements.py +152 -152
  306. tests/test_retry_middleware.py +334 -242
  307. tests/test_retry_middleware_realistic.py +274 -0
  308. tests/test_scheduler.py +252 -252
  309. tests/test_scheduler_config_update.py +133 -133
  310. tests/test_simple_response.py +61 -61
  311. tests/test_telecom_spider_redis_key.py +205 -205
  312. tests/test_template_content.py +87 -87
  313. tests/test_template_redis_key.py +134 -134
  314. tests/test_tools.py +159 -159
  315. tests/test_user_agent_randomness.py +177 -0
  316. tests/test_user_agents.py +96 -96
  317. tests/tools_example.py +260 -260
  318. tests/untested_features_report.md +138 -138
  319. tests/verify_debug.py +51 -51
  320. tests/verify_distributed.py +117 -117
  321. tests/verify_log_fix.py +111 -111
  322. crawlo-1.4.1.dist-info/METADATA +0 -1199
  323. crawlo-1.4.1.dist-info/RECORD +0 -309
  324. {crawlo-1.4.1.dist-info → crawlo-1.4.3.dist-info}/WHEEL +0 -0
  325. {crawlo-1.4.1.dist-info → crawlo-1.4.3.dist-info}/entry_points.txt +0 -0
  326. {crawlo-1.4.1.dist-info → crawlo-1.4.3.dist-info}/top_level.txt +0 -0
@@ -1,271 +1,271 @@
1
- #!/usr/bin/env python3
2
- # -*- coding: utf-8 -*-
3
- """
4
- 增强版错误处理工具测试
5
- """
6
- import sys
7
- import os
8
- import asyncio
9
- import traceback
10
-
11
- # 添加项目根目录到 Python 路径
12
- sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
13
-
14
- from crawlo.utils.enhanced_error_handler import EnhancedErrorHandler, ErrorContext, DetailedException, handle_exception
15
-
16
-
17
- def test_basic_error_handling():
18
- """测试基本错误处理"""
19
- print("1. 测试基本错误处理...")
20
-
21
- try:
22
- handler = EnhancedErrorHandler("test_logger")
23
-
24
- # 测试同步函数错误处理
25
- def failing_function():
26
- raise ValueError("测试错误")
27
-
28
- context = ErrorContext(context="测试同步函数", module="test_module", function="failing_function")
29
-
30
- try:
31
- handler.safe_call(failing_function, context=context)
32
- print(" 同步函数错误处理成功")
33
- except Exception as e:
34
- print(f" 同步函数错误处理失败: {e}")
35
- return False
36
-
37
- # 测试普通函数的错误处理(不是异步函数)
38
- def normal_function():
39
- return "正常返回值"
40
-
41
- context = ErrorContext(context="测试普通函数", module="test_module", function="normal_function")
42
-
43
- result = handler.safe_call(normal_function, context=context)
44
- if result == "正常返回值":
45
- print(" 普通函数处理成功")
46
- else:
47
- print(" 普通函数处理失败")
48
- return False
49
-
50
- return True
51
-
52
- except Exception as e:
53
- print(f" 基本错误处理测试失败: {e}")
54
- traceback.print_exc()
55
- return False
56
-
57
-
58
- def test_detailed_exception():
59
- """测试详细异常"""
60
- print("2. 测试详细异常...")
61
-
62
- try:
63
- # 创建错误上下文
64
- context = ErrorContext(
65
- context="数据库连接失败",
66
- module="database_module",
67
- function="connect_to_db"
68
- )
69
-
70
- # 创建详细异常
71
- exception = DetailedException(
72
- "无法连接到数据库",
73
- context=context,
74
- error_code="DB_CONN_001",
75
- host="localhost",
76
- port=5432,
77
- database="test_db"
78
- )
79
-
80
- # 验证异常信息
81
- assert "无法连接到数据库" in str(exception)
82
- assert "数据库连接失败" in str(exception)
83
-
84
- # 获取完整详情
85
- details = exception.get_full_details()
86
- assert details["error_code"] == "DB_CONN_001"
87
- assert details["exception_type"] == "DetailedException"
88
-
89
- print(" 详细异常测试成功")
90
- return True
91
-
92
- except Exception as e:
93
- print(f" 详细异常测试失败: {e}")
94
- traceback.print_exc()
95
- return False
96
-
97
-
98
- async def test_retry_decorator():
99
- """测试重试装饰器"""
100
- print("3. 测试重试装饰器...")
101
-
102
- try:
103
- handler = EnhancedErrorHandler("test_retry_logger")
104
-
105
- # 测试同步函数重试
106
- attempt_count = 0
107
-
108
- @handler.retry_on_failure(max_retries=2, delay=0.1)
109
- def failing_function():
110
- nonlocal attempt_count
111
- attempt_count += 1
112
- if attempt_count < 3:
113
- raise ValueError(f"尝试 {attempt_count} 失败")
114
- return "成功"
115
-
116
- # 第一次调用应该成功(第3次尝试)
117
- result = failing_function()
118
- assert result == "成功"
119
- assert attempt_count == 3
120
-
121
- print(" 同步函数重试测试成功")
122
-
123
- # 测试异步函数重试
124
- async_attempt_count = 0
125
-
126
- @handler.retry_on_failure(max_retries=2, delay=0.1)
127
- async def async_failing_function():
128
- nonlocal async_attempt_count
129
- async_attempt_count += 1
130
- if async_attempt_count < 3:
131
- raise RuntimeError(f"异步尝试 {async_attempt_count} 失败")
132
- return "异步成功"
133
-
134
- # 异步调用
135
- result = await async_failing_function()
136
- assert result == "异步成功"
137
- assert async_attempt_count == 3
138
-
139
- print(" 异步函数重试测试成功")
140
- return True
141
-
142
- except Exception as e:
143
- print(f" 重试装饰器测试失败: {e}")
144
- traceback.print_exc()
145
- return False
146
-
147
-
148
- async def test_exception_decorator():
149
- """测试异常装饰器"""
150
- print("4. 测试异常装饰器...")
151
-
152
- try:
153
- # 测试同步函数装饰器
154
- @handle_exception(context="测试装饰器", module="test_module", function="decorated_function", raise_error=False)
155
- def decorated_function():
156
- raise ValueError("装饰器测试错误")
157
-
158
- # 调用应该捕获异常但不抛出
159
- try:
160
- decorated_function()
161
- print(" 同步函数装饰器测试成功")
162
- except Exception:
163
- print(" 同步函数装饰器测试失败:异常未被捕获")
164
- return False
165
-
166
- # 测试异步函数装饰器
167
- @handle_exception(context="异步测试装饰器", module="test_module", function="async_decorated_function", raise_error=False)
168
- async def async_decorated_function():
169
- raise RuntimeError("异步装饰器测试错误")
170
-
171
- # 异步调用
172
- try:
173
- await async_decorated_function()
174
- print(" 异步函数装饰器测试成功")
175
- except Exception:
176
- print(" 异步函数装饰器测试失败:异常未被捕获")
177
- return False
178
-
179
- return True
180
-
181
- except Exception as e:
182
- print(f" 异常装饰器测试失败: {e}")
183
- traceback.print_exc()
184
- return False
185
-
186
-
187
- def test_error_history():
188
- """测试错误历史记录"""
189
- print("5. 测试错误历史记录...")
190
-
191
- try:
192
- handler = EnhancedErrorHandler("history_test_logger")
193
-
194
- # 产生一些错误
195
- def error_function():
196
- raise ValueError("历史记录测试错误")
197
-
198
- context = ErrorContext(context="测试历史记录", module="history_module")
199
-
200
- # 记录几个错误
201
- for i in range(3):
202
- try:
203
- handler.safe_call(error_function, context=context)
204
- except:
205
- pass # 忽略异常
206
-
207
- # 检查历史记录
208
- history = handler.get_error_history()
209
- assert len(history) == 3
210
-
211
- # 检查历史记录内容
212
- for record in history:
213
- assert "历史记录测试错误" in record["message"]
214
- assert record["exception_type"] == "ValueError"
215
-
216
- print(" 错误历史记录测试成功")
217
- return True
218
-
219
- except Exception as e:
220
- print(f" 错误历史记录测试失败: {e}")
221
- traceback.print_exc()
222
- return False
223
-
224
-
225
- async def main():
226
- """主测试函数"""
227
- print("开始增强版错误处理工具测试...")
228
- print("=" * 50)
229
-
230
- tests = [
231
- test_basic_error_handling,
232
- test_detailed_exception,
233
- test_retry_decorator,
234
- test_exception_decorator,
235
- test_error_history
236
- ]
237
-
238
- passed = 0
239
- total = len(tests)
240
-
241
- for test_func in tests:
242
- try:
243
- if asyncio.iscoroutinefunction(test_func):
244
- result = await test_func()
245
- else:
246
- result = test_func()
247
-
248
- if result:
249
- passed += 1
250
- print(f"{test_func.__name__} 通过")
251
- else:
252
- print(f"{test_func.__name__} 失败")
253
- except Exception as e:
254
- print(f"{test_func.__name__} 异常: {e}")
255
- traceback.print_exc()
256
- print()
257
-
258
- print("=" * 50)
259
- print(f"测试结果: {passed}/{total} 通过")
260
-
261
- if passed == total:
262
- print("所有测试通过!增强版错误处理工具工作正常")
263
- return 0
264
- else:
265
- print("部分测试失败,请检查实现")
266
- return 1
267
-
268
-
269
- if __name__ == "__main__":
270
- exit_code = asyncio.run(main())
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ 增强版错误处理工具测试
5
+ """
6
+ import sys
7
+ import os
8
+ import asyncio
9
+ import traceback
10
+
11
+ # 添加项目根目录到 Python 路径
12
+ sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
13
+
14
+ from crawlo.utils.enhanced_error_handler import EnhancedErrorHandler, ErrorContext, DetailedException, handle_exception
15
+
16
+
17
+ def test_basic_error_handling():
18
+ """测试基本错误处理"""
19
+ print("1. 测试基本错误处理...")
20
+
21
+ try:
22
+ handler = EnhancedErrorHandler("test_logger")
23
+
24
+ # 测试同步函数错误处理
25
+ def failing_function():
26
+ raise ValueError("测试错误")
27
+
28
+ context = ErrorContext(context="测试同步函数", module="test_module", function="failing_function")
29
+
30
+ try:
31
+ handler.safe_call(failing_function, context=context)
32
+ print(" 同步函数错误处理成功")
33
+ except Exception as e:
34
+ print(f" 同步函数错误处理失败: {e}")
35
+ return False
36
+
37
+ # 测试普通函数的错误处理(不是异步函数)
38
+ def normal_function():
39
+ return "正常返回值"
40
+
41
+ context = ErrorContext(context="测试普通函数", module="test_module", function="normal_function")
42
+
43
+ result = handler.safe_call(normal_function, context=context)
44
+ if result == "正常返回值":
45
+ print(" 普通函数处理成功")
46
+ else:
47
+ print(" 普通函数处理失败")
48
+ return False
49
+
50
+ return True
51
+
52
+ except Exception as e:
53
+ print(f" 基本错误处理测试失败: {e}")
54
+ traceback.print_exc()
55
+ return False
56
+
57
+
58
+ def test_detailed_exception():
59
+ """测试详细异常"""
60
+ print("2. 测试详细异常...")
61
+
62
+ try:
63
+ # 创建错误上下文
64
+ context = ErrorContext(
65
+ context="数据库连接失败",
66
+ module="database_module",
67
+ function="connect_to_db"
68
+ )
69
+
70
+ # 创建详细异常
71
+ exception = DetailedException(
72
+ "无法连接到数据库",
73
+ context=context,
74
+ error_code="DB_CONN_001",
75
+ host="localhost",
76
+ port=5432,
77
+ database="test_db"
78
+ )
79
+
80
+ # 验证异常信息
81
+ assert "无法连接到数据库" in str(exception)
82
+ assert "数据库连接失败" in str(exception)
83
+
84
+ # 获取完整详情
85
+ details = exception.get_full_details()
86
+ assert details["error_code"] == "DB_CONN_001"
87
+ assert details["exception_type"] == "DetailedException"
88
+
89
+ print(" 详细异常测试成功")
90
+ return True
91
+
92
+ except Exception as e:
93
+ print(f" 详细异常测试失败: {e}")
94
+ traceback.print_exc()
95
+ return False
96
+
97
+
98
+ async def test_retry_decorator():
99
+ """测试重试装饰器"""
100
+ print("3. 测试重试装饰器...")
101
+
102
+ try:
103
+ handler = EnhancedErrorHandler("test_retry_logger")
104
+
105
+ # 测试同步函数重试
106
+ attempt_count = 0
107
+
108
+ @handler.retry_on_failure(max_retries=2, delay=0.1)
109
+ def failing_function():
110
+ nonlocal attempt_count
111
+ attempt_count += 1
112
+ if attempt_count < 3:
113
+ raise ValueError(f"尝试 {attempt_count} 失败")
114
+ return "成功"
115
+
116
+ # 第一次调用应该成功(第3次尝试)
117
+ result = failing_function()
118
+ assert result == "成功"
119
+ assert attempt_count == 3
120
+
121
+ print(" 同步函数重试测试成功")
122
+
123
+ # 测试异步函数重试
124
+ async_attempt_count = 0
125
+
126
+ @handler.retry_on_failure(max_retries=2, delay=0.1)
127
+ async def async_failing_function():
128
+ nonlocal async_attempt_count
129
+ async_attempt_count += 1
130
+ if async_attempt_count < 3:
131
+ raise RuntimeError(f"异步尝试 {async_attempt_count} 失败")
132
+ return "异步成功"
133
+
134
+ # 异步调用
135
+ result = await async_failing_function()
136
+ assert result == "异步成功"
137
+ assert async_attempt_count == 3
138
+
139
+ print(" 异步函数重试测试成功")
140
+ return True
141
+
142
+ except Exception as e:
143
+ print(f" 重试装饰器测试失败: {e}")
144
+ traceback.print_exc()
145
+ return False
146
+
147
+
148
+ async def test_exception_decorator():
149
+ """测试异常装饰器"""
150
+ print("4. 测试异常装饰器...")
151
+
152
+ try:
153
+ # 测试同步函数装饰器
154
+ @handle_exception(context="测试装饰器", module="test_module", function="decorated_function", raise_error=False)
155
+ def decorated_function():
156
+ raise ValueError("装饰器测试错误")
157
+
158
+ # 调用应该捕获异常但不抛出
159
+ try:
160
+ decorated_function()
161
+ print(" 同步函数装饰器测试成功")
162
+ except Exception:
163
+ print(" 同步函数装饰器测试失败:异常未被捕获")
164
+ return False
165
+
166
+ # 测试异步函数装饰器
167
+ @handle_exception(context="异步测试装饰器", module="test_module", function="async_decorated_function", raise_error=False)
168
+ async def async_decorated_function():
169
+ raise RuntimeError("异步装饰器测试错误")
170
+
171
+ # 异步调用
172
+ try:
173
+ await async_decorated_function()
174
+ print(" 异步函数装饰器测试成功")
175
+ except Exception:
176
+ print(" 异步函数装饰器测试失败:异常未被捕获")
177
+ return False
178
+
179
+ return True
180
+
181
+ except Exception as e:
182
+ print(f" 异常装饰器测试失败: {e}")
183
+ traceback.print_exc()
184
+ return False
185
+
186
+
187
+ def test_error_history():
188
+ """测试错误历史记录"""
189
+ print("5. 测试错误历史记录...")
190
+
191
+ try:
192
+ handler = EnhancedErrorHandler("history_test_logger")
193
+
194
+ # 产生一些错误
195
+ def error_function():
196
+ raise ValueError("历史记录测试错误")
197
+
198
+ context = ErrorContext(context="测试历史记录", module="history_module")
199
+
200
+ # 记录几个错误
201
+ for i in range(3):
202
+ try:
203
+ handler.safe_call(error_function, context=context)
204
+ except:
205
+ pass # 忽略异常
206
+
207
+ # 检查历史记录
208
+ history = handler.get_error_history()
209
+ assert len(history) == 3
210
+
211
+ # 检查历史记录内容
212
+ for record in history:
213
+ assert "历史记录测试错误" in record["message"]
214
+ assert record["exception_type"] == "ValueError"
215
+
216
+ print(" 错误历史记录测试成功")
217
+ return True
218
+
219
+ except Exception as e:
220
+ print(f" 错误历史记录测试失败: {e}")
221
+ traceback.print_exc()
222
+ return False
223
+
224
+
225
+ async def main():
226
+ """主测试函数"""
227
+ print("开始增强版错误处理工具测试...")
228
+ print("=" * 50)
229
+
230
+ tests = [
231
+ test_basic_error_handling,
232
+ test_detailed_exception,
233
+ test_retry_decorator,
234
+ test_exception_decorator,
235
+ test_error_history
236
+ ]
237
+
238
+ passed = 0
239
+ total = len(tests)
240
+
241
+ for test_func in tests:
242
+ try:
243
+ if asyncio.iscoroutinefunction(test_func):
244
+ result = await test_func()
245
+ else:
246
+ result = test_func()
247
+
248
+ if result:
249
+ passed += 1
250
+ print(f"{test_func.__name__} 通过")
251
+ else:
252
+ print(f"{test_func.__name__} 失败")
253
+ except Exception as e:
254
+ print(f"{test_func.__name__} 异常: {e}")
255
+ traceback.print_exc()
256
+ print()
257
+
258
+ print("=" * 50)
259
+ print(f"测试结果: {passed}/{total} 通过")
260
+
261
+ if passed == total:
262
+ print("所有测试通过!增强版错误处理工具工作正常")
263
+ return 0
264
+ else:
265
+ print("部分测试失败,请检查实现")
266
+ return 1
267
+
268
+
269
+ if __name__ == "__main__":
270
+ exit_code = asyncio.run(main())
271
271
  exit(exit_code)