crawlo 1.3.4__py3-none-any.whl → 1.3.6__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 (289) 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 -291
  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 -318
  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 +503 -503
  90. crawlo/queue/redis_priority_queue.py +326 -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 +167 -167
  104. crawlo/templates/project/settings_distributed.py.tmpl +166 -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 +80 -44
  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-1.3.4.dist-info → crawlo-1.3.6.dist-info}/METADATA +1126 -1126
  149. crawlo-1.3.6.dist-info/RECORD +290 -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 +127 -0
  166. tests/debug_log_levels.py +63 -63
  167. tests/debug_pipelines.py +66 -66
  168. tests/detailed_log_test.py +234 -0
  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 +261 -0
  177. tests/final_validation_test.py +182 -182
  178. tests/fix_log_test.py +143 -0
  179. tests/framework_performance_test.py +202 -202
  180. tests/log_buffering_test.py +112 -0
  181. tests/log_generation_timing_test.py +154 -0
  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 +138 -0
  195. tests/simple_optimization_test.py +128 -128
  196. tests/simple_queue_type_test.py +42 -0
  197. tests/simple_spider_test.py +49 -49
  198. tests/simple_test.py +47 -47
  199. tests/spider_log_timing_test.py +178 -0
  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 +207 -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 +84 -0
  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 +283 -0
  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 +176 -176
  258. tests/test_queue_type.py +107 -0
  259. tests/test_random_user_agent.py +72 -72
  260. tests/test_real_scenario_proxy.py +195 -195
  261. tests/test_redis_config.py +28 -28
  262. tests/test_redis_connection_pool.py +294 -294
  263. tests/test_redis_key_naming.py +181 -181
  264. tests/test_redis_key_validator.py +123 -123
  265. tests/test_redis_queue.py +224 -224
  266. tests/test_request_ignore_middleware.py +182 -182
  267. tests/test_request_params.py +111 -111
  268. tests/test_request_serialization.py +70 -70
  269. tests/test_response_code_middleware.py +349 -349
  270. tests/test_response_filter_middleware.py +427 -427
  271. tests/test_response_improvements.py +152 -152
  272. tests/test_retry_middleware.py +241 -241
  273. tests/test_scheduler.py +252 -252
  274. tests/test_scheduler_config_update.py +133 -133
  275. tests/test_simple_response.py +61 -61
  276. tests/test_telecom_spider_redis_key.py +205 -205
  277. tests/test_template_content.py +87 -87
  278. tests/test_template_redis_key.py +134 -134
  279. tests/test_tools.py +159 -159
  280. tests/test_user_agents.py +96 -96
  281. tests/tools_example.py +260 -260
  282. tests/untested_features_report.md +138 -138
  283. tests/verify_debug.py +51 -51
  284. tests/verify_distributed.py +117 -117
  285. tests/verify_log_fix.py +111 -111
  286. crawlo-1.3.4.dist-info/RECORD +0 -278
  287. {crawlo-1.3.4.dist-info → crawlo-1.3.6.dist-info}/WHEEL +0 -0
  288. {crawlo-1.3.4.dist-info → crawlo-1.3.6.dist-info}/entry_points.txt +0 -0
  289. {crawlo-1.3.4.dist-info → crawlo-1.3.6.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)