crawlo 1.3.3__py3-none-any.whl → 1.3.4__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 (279) hide show
  1. crawlo/__init__.py +87 -63
  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 -323
  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 +46 -2
  16. crawlo/core/engine.py +439 -365
  17. crawlo/core/processor.py +40 -40
  18. crawlo/core/scheduler.py +257 -256
  19. crawlo/crawler.py +639 -1167
  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 -226
  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 -52
  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 +28 -0
  40. crawlo/factories/base.py +69 -0
  41. crawlo/factories/crawler.py +104 -0
  42. crawlo/factories/registry.py +85 -0
  43. crawlo/filters/__init__.py +154 -154
  44. crawlo/filters/aioredis_filter.py +257 -234
  45. crawlo/filters/memory_filter.py +269 -269
  46. crawlo/framework.py +292 -0
  47. crawlo/initialization/__init__.py +40 -0
  48. crawlo/initialization/built_in.py +426 -0
  49. crawlo/initialization/context.py +142 -0
  50. crawlo/initialization/core.py +194 -0
  51. crawlo/initialization/phases.py +149 -0
  52. crawlo/initialization/registry.py +146 -0
  53. crawlo/items/__init__.py +23 -23
  54. crawlo/items/base.py +23 -22
  55. crawlo/items/fields.py +52 -52
  56. crawlo/items/items.py +104 -104
  57. crawlo/logging/__init__.py +38 -0
  58. crawlo/logging/config.py +97 -0
  59. crawlo/logging/factory.py +129 -0
  60. crawlo/logging/manager.py +112 -0
  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 -187
  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 +318 -318
  85. crawlo/pipelines/pipeline_manager.py +76 -75
  86. crawlo/pipelines/redis_dedup_pipeline.py +166 -166
  87. crawlo/project.py +327 -325
  88. crawlo/queue/pqueue.py +43 -37
  89. crawlo/queue/queue_manager.py +503 -379
  90. crawlo/queue/redis_priority_queue.py +326 -306
  91. crawlo/settings/__init__.py +7 -7
  92. crawlo/settings/default_settings.py +321 -225
  93. crawlo/settings/setting_manager.py +214 -198
  94. crawlo/spider/__init__.py +657 -639
  95. crawlo/stats_collector.py +73 -59
  96. crawlo/subscriber.py +129 -129
  97. crawlo/task_manager.py +139 -30
  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 +168 -267
  104. crawlo/templates/project/settings_distributed.py.tmpl +167 -180
  105. crawlo/templates/project/settings_gentle.py.tmpl +167 -61
  106. crawlo/templates/project/settings_high_performance.py.tmpl +168 -131
  107. crawlo/templates/project/settings_minimal.py.tmpl +66 -35
  108. crawlo/templates/project/settings_simple.py.tmpl +165 -102
  109. crawlo/templates/project/spiders/__init__.py.tmpl +10 -6
  110. crawlo/templates/run.py.tmpl +34 -38
  111. crawlo/templates/spider/spider.py.tmpl +143 -143
  112. crawlo/templates/spiders_init.py.tmpl +10 -0
  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 +365 -0
  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 +26 -0
  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 -124
  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 +44 -200
  138. crawlo/utils/performance_monitor.py +285 -285
  139. crawlo/utils/queue_helper.py +175 -175
  140. crawlo/utils/redis_connection_pool.py +388 -351
  141. crawlo/utils/redis_key_validator.py +198 -198
  142. crawlo/utils/request.py +267 -267
  143. crawlo/utils/request_serializer.py +225 -218
  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-1.3.3.dist-info → crawlo-1.3.4.dist-info}/METADATA +1126 -1020
  149. crawlo-1.3.4.dist-info/RECORD +278 -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 +107 -107
  154. tests/baidu_performance_test.py +109 -0
  155. tests/baidu_test.py +60 -0
  156. tests/cleaners_example.py +160 -160
  157. tests/comprehensive_framework_test.py +213 -0
  158. tests/comprehensive_test.py +82 -0
  159. tests/comprehensive_testing_summary.md +187 -0
  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 +70 -0
  164. tests/debug_framework_logger.py +85 -0
  165. tests/debug_log_levels.py +64 -0
  166. tests/debug_pipelines.py +66 -66
  167. tests/distributed_test.py +67 -0
  168. tests/distributed_test_debug.py +77 -0
  169. tests/dynamic_loading_example.py +523 -523
  170. tests/dynamic_loading_test.py +104 -104
  171. tests/env_config_example.py +133 -133
  172. tests/error_handling_example.py +171 -171
  173. tests/final_command_test_report.md +0 -0
  174. tests/final_comprehensive_test.py +152 -0
  175. tests/final_validation_test.py +183 -0
  176. tests/framework_performance_test.py +203 -0
  177. tests/optimized_performance_test.py +212 -0
  178. tests/performance_comparison.py +246 -0
  179. tests/queue_blocking_test.py +114 -0
  180. tests/queue_test.py +90 -0
  181. tests/redis_key_validation_demo.py +130 -130
  182. tests/request_params_example.py +150 -150
  183. tests/response_improvements_example.py +144 -144
  184. tests/scrapy_comparison/ofweek_scrapy.py +139 -0
  185. tests/scrapy_comparison/scrapy_test.py +134 -0
  186. tests/simple_command_test.py +120 -0
  187. tests/simple_crawlo_test.py +128 -0
  188. tests/simple_log_test.py +58 -0
  189. tests/simple_optimization_test.py +129 -0
  190. tests/simple_spider_test.py +50 -0
  191. tests/simple_test.py +48 -0
  192. tests/test_advanced_tools.py +148 -148
  193. tests/test_all_commands.py +231 -0
  194. tests/test_all_redis_key_configs.py +145 -145
  195. tests/test_authenticated_proxy.py +141 -141
  196. tests/test_batch_processor.py +179 -0
  197. tests/test_cleaners.py +54 -54
  198. tests/test_component_factory.py +175 -0
  199. tests/test_comprehensive.py +146 -146
  200. tests/test_config_consistency.py +80 -80
  201. tests/test_config_merge.py +152 -152
  202. tests/test_config_validator.py +182 -182
  203. tests/test_controlled_spider_mixin.py +80 -0
  204. tests/test_crawlo_proxy_integration.py +108 -108
  205. tests/test_date_tools.py +123 -123
  206. tests/test_default_header_middleware.py +158 -158
  207. tests/test_distributed.py +65 -65
  208. tests/test_double_crawlo_fix.py +207 -207
  209. tests/test_double_crawlo_fix_simple.py +124 -124
  210. tests/test_download_delay_middleware.py +221 -221
  211. tests/test_downloader_proxy_compatibility.py +268 -268
  212. tests/test_dynamic_downloaders_proxy.py +124 -124
  213. tests/test_dynamic_proxy.py +92 -92
  214. tests/test_dynamic_proxy_config.py +146 -146
  215. tests/test_dynamic_proxy_real.py +109 -109
  216. tests/test_edge_cases.py +303 -303
  217. tests/test_enhanced_error_handler.py +270 -270
  218. tests/test_enhanced_error_handler_comprehensive.py +246 -0
  219. tests/test_env_config.py +121 -121
  220. tests/test_error_handler_compatibility.py +112 -112
  221. tests/test_factories.py +253 -0
  222. tests/test_final_validation.py +153 -153
  223. tests/test_framework_env_usage.py +103 -103
  224. tests/test_framework_logger.py +67 -0
  225. tests/test_framework_startup.py +65 -0
  226. tests/test_integration.py +169 -169
  227. tests/test_item_dedup_redis_key.py +122 -122
  228. tests/test_large_scale_config.py +113 -0
  229. tests/test_large_scale_helper.py +236 -0
  230. tests/test_mode_change.py +73 -0
  231. tests/test_mode_consistency.py +51 -51
  232. tests/test_offsite_middleware.py +221 -221
  233. tests/test_parsel.py +29 -29
  234. tests/test_performance.py +327 -327
  235. tests/test_performance_monitor.py +116 -0
  236. tests/test_proxy_api.py +264 -264
  237. tests/test_proxy_health_check.py +32 -32
  238. tests/test_proxy_middleware.py +121 -121
  239. tests/test_proxy_middleware_enhanced.py +216 -216
  240. tests/test_proxy_middleware_integration.py +136 -136
  241. tests/test_proxy_middleware_refactored.py +184 -184
  242. tests/test_proxy_providers.py +56 -56
  243. tests/test_proxy_stats.py +19 -19
  244. tests/test_proxy_strategies.py +59 -59
  245. tests/test_queue_empty_check.py +42 -0
  246. tests/test_queue_manager_double_crawlo.py +173 -173
  247. tests/test_queue_manager_redis_key.py +176 -176
  248. tests/test_random_user_agent.py +72 -72
  249. tests/test_real_scenario_proxy.py +195 -195
  250. tests/test_redis_config.py +28 -28
  251. tests/test_redis_connection_pool.py +294 -294
  252. tests/test_redis_key_naming.py +181 -181
  253. tests/test_redis_key_validator.py +123 -123
  254. tests/test_redis_queue.py +224 -224
  255. tests/test_request_ignore_middleware.py +182 -182
  256. tests/test_request_params.py +111 -111
  257. tests/test_request_serialization.py +70 -70
  258. tests/test_response_code_middleware.py +349 -349
  259. tests/test_response_filter_middleware.py +427 -427
  260. tests/test_response_improvements.py +152 -152
  261. tests/test_retry_middleware.py +241 -241
  262. tests/test_scheduler.py +252 -252
  263. tests/test_scheduler_config_update.py +133 -133
  264. tests/test_simple_response.py +61 -61
  265. tests/test_telecom_spider_redis_key.py +205 -205
  266. tests/test_template_content.py +87 -87
  267. tests/test_template_redis_key.py +134 -134
  268. tests/test_tools.py +159 -159
  269. tests/test_user_agents.py +96 -96
  270. tests/tools_example.py +260 -260
  271. tests/untested_features_report.md +139 -0
  272. tests/verify_debug.py +52 -0
  273. tests/verify_distributed.py +117 -117
  274. tests/verify_log_fix.py +112 -0
  275. crawlo-1.3.3.dist-info/RECORD +0 -219
  276. tests/DOUBLE_CRAWLO_PREFIX_FIX_REPORT.md +0 -82
  277. {crawlo-1.3.3.dist-info → crawlo-1.3.4.dist-info}/WHEEL +0 -0
  278. {crawlo-1.3.3.dist-info → crawlo-1.3.4.dist-info}/entry_points.txt +0 -0
  279. {crawlo-1.3.3.dist-info → crawlo-1.3.4.dist-info}/top_level.txt +0 -0
@@ -0,0 +1,246 @@
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ 增强错误处理工具综合测试
5
+ 测试 ErrorContext, DetailedException, EnhancedErrorHandler 的更多功能
6
+ """
7
+ import sys
8
+ import os
9
+ import unittest
10
+ from unittest.mock import Mock, patch, MagicMock
11
+ import asyncio
12
+ import traceback
13
+
14
+ # 添加项目根目录到 Python 路径
15
+ sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
16
+
17
+ from crawlo.utils.enhanced_error_handler import (
18
+ EnhancedErrorHandler,
19
+ ErrorContext,
20
+ DetailedException,
21
+ handle_exception
22
+ )
23
+
24
+
25
+ class TestErrorContext(unittest.TestCase):
26
+ """错误上下文测试"""
27
+
28
+ def test_error_context_initialization(self):
29
+ """测试错误上下文初始化"""
30
+ context = ErrorContext(
31
+ context="测试上下文",
32
+ module="test_module",
33
+ function="test_function"
34
+ )
35
+
36
+ self.assertEqual(context.context, "测试上下文")
37
+ self.assertEqual(context.module, "test_module")
38
+ self.assertEqual(context.function, "test_function")
39
+ self.assertIsNotNone(context.timestamp)
40
+
41
+ def test_error_context_string_representation(self):
42
+ """测试错误上下文字符串表示"""
43
+ context = ErrorContext(
44
+ context="测试上下文",
45
+ module="test_module",
46
+ function="test_function"
47
+ )
48
+
49
+ context_str = str(context)
50
+ self.assertIn("Module: test_module", context_str)
51
+ self.assertIn("Function: test_function", context_str)
52
+ self.assertIn("Context: 测试上下文", context_str)
53
+ self.assertIn("Time:", context_str)
54
+
55
+
56
+ class TestDetailedException(unittest.TestCase):
57
+ """详细异常测试"""
58
+
59
+ def test_detailed_exception_initialization(self):
60
+ """测试详细异常初始化"""
61
+ context = ErrorContext(context="测试上下文")
62
+
63
+ exception = DetailedException(
64
+ message="测试异常消息",
65
+ context=context,
66
+ error_code="TEST_001",
67
+ detail1="详细信息1",
68
+ detail2="详细信息2"
69
+ )
70
+
71
+ self.assertIn("测试异常消息", str(exception))
72
+ self.assertEqual(exception.context, context)
73
+ self.assertEqual(exception.error_code, "TEST_001")
74
+ self.assertEqual(exception.details["detail1"], "详细信息1")
75
+ self.assertEqual(exception.details["detail2"], "详细信息2")
76
+ self.assertIsNotNone(exception.timestamp)
77
+
78
+ def test_detailed_exception_without_context(self):
79
+ """测试没有上下文的详细异常"""
80
+ exception = DetailedException(
81
+ message="测试异常消息",
82
+ error_code="TEST_002"
83
+ )
84
+
85
+ self.assertEqual(str(exception), "测试异常消息")
86
+ self.assertIsNone(exception.context)
87
+ self.assertEqual(exception.error_code, "TEST_002")
88
+ self.assertIsNotNone(exception.timestamp)
89
+
90
+ def test_detailed_exception_string_with_context(self):
91
+ """测试带上下文的详细异常字符串表示"""
92
+ context = ErrorContext(context="测试上下文")
93
+ exception = DetailedException(
94
+ message="测试异常消息",
95
+ context=context
96
+ )
97
+
98
+ exception_str = str(exception)
99
+ self.assertIn("测试异常消息", exception_str)
100
+ self.assertIn("测试上下文", exception_str)
101
+
102
+ def test_get_full_details(self):
103
+ """测试获取完整详情"""
104
+ context = ErrorContext(context="测试上下文")
105
+ exception = DetailedException(
106
+ message="测试异常消息",
107
+ context=context,
108
+ error_code="TEST_003",
109
+ detail1="详细信息1"
110
+ )
111
+
112
+ details = exception.get_full_details()
113
+ self.assertIn("测试异常消息", details["message"])
114
+ self.assertEqual(details["error_code"], "TEST_003")
115
+ self.assertIn("测试上下文", details["context"])
116
+ self.assertEqual(details["details"]["detail1"], "详细信息1")
117
+ self.assertEqual(details["exception_type"], "DetailedException")
118
+
119
+
120
+ class TestEnhancedErrorHandler(unittest.TestCase):
121
+ """增强错误处理器测试"""
122
+
123
+ def setUp(self):
124
+ """测试前准备"""
125
+ self.handler = EnhancedErrorHandler("test_logger", "ERROR")
126
+
127
+ def test_handler_initialization(self):
128
+ """测试处理器初始化"""
129
+ self.assertEqual(len(self.handler.error_history), 0)
130
+ self.assertEqual(self.handler.max_history_size, 100)
131
+
132
+ def test_handle_error_without_raising(self):
133
+ """测试不抛出异常的错误处理"""
134
+ try:
135
+ raise ValueError("测试错误")
136
+ except Exception as e:
137
+ # 处理错误但不重新抛出
138
+ error_details = self.handler.handle_error(e, raise_error=False, log_error=False)
139
+
140
+ # 验证返回的错误详情
141
+ self.assertIsInstance(error_details, dict)
142
+ self.assertEqual(error_details["exception_type"], "ValueError")
143
+ self.assertEqual(error_details["message"], "测试错误")
144
+
145
+ def test_safe_call_success(self):
146
+ """测试安全调用成功"""
147
+ def normal_function(x, y):
148
+ return x + y
149
+
150
+ result = self.handler.safe_call(normal_function, 1, 2, default_return=0)
151
+ self.assertEqual(result, 3)
152
+
153
+ def test_safe_call_with_exception(self):
154
+ """测试安全调用异常"""
155
+ def failing_function():
156
+ raise RuntimeError("函数执行失败")
157
+
158
+ result = self.handler.safe_call(failing_function, default_return="默认值")
159
+ self.assertEqual(result, "默认值")
160
+
161
+ def test_get_and_clear_error_history(self):
162
+ """测试获取和清空错误历史"""
163
+ # 产生一些错误
164
+ try:
165
+ raise ValueError("错误1")
166
+ except Exception as e:
167
+ self.handler.handle_error(e, raise_error=False, log_error=False)
168
+
169
+ try:
170
+ raise RuntimeError("错误2")
171
+ except Exception as e:
172
+ self.handler.handle_error(e, raise_error=False, log_error=False)
173
+
174
+ # 检查历史记录
175
+ history = self.handler.get_error_history()
176
+ self.assertEqual(len(history), 2)
177
+
178
+ # 清空历史记录
179
+ self.handler.clear_error_history()
180
+ history = self.handler.get_error_history()
181
+ self.assertEqual(len(history), 0)
182
+
183
+ def test_error_history_size_limit(self):
184
+ """测试错误历史大小限制"""
185
+ # 产生超过100个错误
186
+ for i in range(110):
187
+ try:
188
+ raise ValueError(f"错误{i}")
189
+ except Exception as e:
190
+ self.handler.handle_error(e, raise_error=False, log_error=False)
191
+
192
+ # 检查历史记录大小限制
193
+ history = self.handler.get_error_history()
194
+ self.assertEqual(len(history), 100) # 应该限制在100个
195
+
196
+
197
+ class TestHandleExceptionDecorator(unittest.TestCase):
198
+ """异常处理装饰器测试"""
199
+
200
+ def test_handle_exception_decorator_sync(self):
201
+ """测试同步函数的异常处理装饰器"""
202
+ @handle_exception(context="测试装饰器", module="test_module", raise_error=False)
203
+ def failing_function():
204
+ raise ValueError("装饰器测试错误")
205
+
206
+ # 调用应该捕获异常但不抛出
207
+ try:
208
+ result = failing_function()
209
+ self.assertIsNone(result) # 默认返回None
210
+ except Exception:
211
+ self.fail("异常应该被捕获")
212
+
213
+ def test_handle_exception_decorator_async(self):
214
+ """测试异步函数的异常处理装饰器"""
215
+ @handle_exception(context="异步测试装饰器", module="test_module", raise_error=False)
216
+ async def async_failing_function():
217
+ raise RuntimeError("异步装饰器测试错误")
218
+
219
+ # 异步调用
220
+ async def test_async():
221
+ try:
222
+ result = await async_failing_function()
223
+ self.assertIsNone(result) # 默认返回None
224
+ except Exception:
225
+ self.fail("异常应该被捕获")
226
+
227
+ # 使用事件循环运行异步函数
228
+ asyncio.run(test_async())
229
+
230
+ def test_handle_exception_decorator_with_detailed_exception(self):
231
+ """测试装饰器处理详细异常"""
232
+ @handle_exception(context="详细异常测试", module="test_module", raise_error=False)
233
+ def function_with_detailed_exception():
234
+ context = ErrorContext(context="函数内部上下文")
235
+ raise DetailedException("详细异常消息", context=context, error_code="DETAIL_001")
236
+
237
+ # 调用应该捕获异常
238
+ try:
239
+ result = function_with_detailed_exception()
240
+ self.assertIsNone(result)
241
+ except Exception:
242
+ self.fail("详细异常应该被捕获")
243
+
244
+
245
+ if __name__ == '__main__':
246
+ unittest.main()
tests/test_env_config.py CHANGED
@@ -1,122 +1,122 @@
1
- #!/usr/bin/env python3
2
- # -*- coding: utf-8 -*-
3
- """
4
- 环境变量配置工具测试
5
- """
6
- import os
7
- import sys
8
- import unittest
9
- from unittest.mock import patch
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.env_config import EnvConfigManager, get_env_var, get_redis_config, get_runtime_config
15
-
16
-
17
- class TestEnvConfigManager(unittest.TestCase):
18
- """环境变量配置管理器测试"""
19
-
20
- def test_get_env_var_str(self):
21
- """测试字符串环境变量获取"""
22
- with patch.dict(os.environ, {'TEST_VAR': 'test_value'}):
23
- result = EnvConfigManager.get_env_var('TEST_VAR', 'default', str)
24
- self.assertEqual(result, 'test_value')
25
-
26
- # 测试默认值
27
- result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 'default', str)
28
- self.assertEqual(result, 'default')
29
-
30
- def test_get_env_var_int(self):
31
- """测试整数环境变量获取"""
32
- with patch.dict(os.environ, {'TEST_INT': '42'}):
33
- result = EnvConfigManager.get_env_var('TEST_INT', 0, int)
34
- self.assertEqual(result, 42)
35
-
36
- # 测试默认值
37
- result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 10, int)
38
- self.assertEqual(result, 10)
39
-
40
- # 测试无效值
41
- with patch.dict(os.environ, {'INVALID_INT': 'not_a_number'}):
42
- result = EnvConfigManager.get_env_var('INVALID_INT', 5, int)
43
- self.assertEqual(result, 5)
44
-
45
- def test_get_env_var_float(self):
46
- """测试浮点数环境变量获取"""
47
- with patch.dict(os.environ, {'TEST_FLOAT': '3.14'}):
48
- result = EnvConfigManager.get_env_var('TEST_FLOAT', 0.0, float)
49
- self.assertEqual(result, 3.14)
50
-
51
- # 测试默认值
52
- result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 2.5, float)
53
- self.assertEqual(result, 2.5)
54
-
55
- def test_get_env_var_bool(self):
56
- """测试布尔环境变量获取"""
57
- # 测试 True 值
58
- for true_val in ['1', 'true', 'True', 'TRUE', 'yes', 'on']:
59
- with patch.dict(os.environ, {'TEST_BOOL': true_val}):
60
- result = EnvConfigManager.get_env_var('TEST_BOOL', False, bool)
61
- self.assertTrue(result)
62
-
63
- # 测试 False 值
64
- for false_val in ['0', 'false', 'False', 'FALSE', 'no', 'off', '']:
65
- with patch.dict(os.environ, {'TEST_BOOL': false_val}):
66
- result = EnvConfigManager.get_env_var('TEST_BOOL', True, bool)
67
- self.assertFalse(result)
68
-
69
- # 测试默认值
70
- result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', True, bool)
71
- self.assertTrue(result)
72
-
73
- def test_get_redis_config(self):
74
- """测试 Redis 配置获取"""
75
- with patch.dict(os.environ, {
76
- 'REDIS_HOST': 'localhost',
77
- 'REDIS_PORT': '6380',
78
- 'REDIS_PASSWORD': 'secret',
79
- 'REDIS_DB': '1'
80
- }):
81
- config = get_redis_config()
82
- self.assertEqual(config['REDIS_HOST'], 'localhost')
83
- self.assertEqual(config['REDIS_PORT'], 6380)
84
- self.assertEqual(config['REDIS_PASSWORD'], 'secret')
85
- self.assertEqual(config['REDIS_DB'], 1)
86
-
87
- # 测试默认值
88
- with patch.dict(os.environ, {}):
89
- config = get_redis_config()
90
- self.assertEqual(config['REDIS_HOST'], '127.0.0.1')
91
- self.assertEqual(config['REDIS_PORT'], 6379)
92
- self.assertEqual(config['REDIS_PASSWORD'], '')
93
- self.assertEqual(config['REDIS_DB'], 0)
94
-
95
- def test_get_runtime_config(self):
96
- """测试运行时配置获取"""
97
- with patch.dict(os.environ, {
98
- 'CRAWLO_MODE': 'distributed',
99
- 'PROJECT_NAME': 'test_project',
100
- 'CONCURRENCY': '20'
101
- }):
102
- config = get_runtime_config()
103
- self.assertEqual(config['CRAWLO_MODE'], 'distributed')
104
- self.assertEqual(config['PROJECT_NAME'], 'test_project')
105
- self.assertEqual(config['CONCURRENCY'], 20)
106
-
107
- # 测试默认值
108
- with patch.dict(os.environ, {}):
109
- config = get_runtime_config()
110
- self.assertEqual(config['CRAWLO_MODE'], 'standalone')
111
- self.assertEqual(config['PROJECT_NAME'], 'crawlo')
112
- self.assertEqual(config['CONCURRENCY'], 8)
113
-
114
- def test_convenience_functions(self):
115
- """测试便捷函数"""
116
- with patch.dict(os.environ, {'TEST_CONVENIENCE': 'convenience_value'}):
117
- result = get_env_var('TEST_CONVENIENCE', 'default')
118
- self.assertEqual(result, 'convenience_value')
119
-
120
-
121
- if __name__ == '__main__':
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ 环境变量配置工具测试
5
+ """
6
+ import os
7
+ import sys
8
+ import unittest
9
+ from unittest.mock import patch
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.env_config import EnvConfigManager, get_env_var, get_redis_config, get_runtime_config
15
+
16
+
17
+ class TestEnvConfigManager(unittest.TestCase):
18
+ """环境变量配置管理器测试"""
19
+
20
+ def test_get_env_var_str(self):
21
+ """测试字符串环境变量获取"""
22
+ with patch.dict(os.environ, {'TEST_VAR': 'test_value'}):
23
+ result = EnvConfigManager.get_env_var('TEST_VAR', 'default', str)
24
+ self.assertEqual(result, 'test_value')
25
+
26
+ # 测试默认值
27
+ result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 'default', str)
28
+ self.assertEqual(result, 'default')
29
+
30
+ def test_get_env_var_int(self):
31
+ """测试整数环境变量获取"""
32
+ with patch.dict(os.environ, {'TEST_INT': '42'}):
33
+ result = EnvConfigManager.get_env_var('TEST_INT', 0, int)
34
+ self.assertEqual(result, 42)
35
+
36
+ # 测试默认值
37
+ result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 10, int)
38
+ self.assertEqual(result, 10)
39
+
40
+ # 测试无效值
41
+ with patch.dict(os.environ, {'INVALID_INT': 'not_a_number'}):
42
+ result = EnvConfigManager.get_env_var('INVALID_INT', 5, int)
43
+ self.assertEqual(result, 5)
44
+
45
+ def test_get_env_var_float(self):
46
+ """测试浮点数环境变量获取"""
47
+ with patch.dict(os.environ, {'TEST_FLOAT': '3.14'}):
48
+ result = EnvConfigManager.get_env_var('TEST_FLOAT', 0.0, float)
49
+ self.assertEqual(result, 3.14)
50
+
51
+ # 测试默认值
52
+ result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 2.5, float)
53
+ self.assertEqual(result, 2.5)
54
+
55
+ def test_get_env_var_bool(self):
56
+ """测试布尔环境变量获取"""
57
+ # 测试 True 值
58
+ for true_val in ['1', 'true', 'True', 'TRUE', 'yes', 'on']:
59
+ with patch.dict(os.environ, {'TEST_BOOL': true_val}):
60
+ result = EnvConfigManager.get_env_var('TEST_BOOL', False, bool)
61
+ self.assertTrue(result)
62
+
63
+ # 测试 False 值
64
+ for false_val in ['0', 'false', 'False', 'FALSE', 'no', 'off', '']:
65
+ with patch.dict(os.environ, {'TEST_BOOL': false_val}):
66
+ result = EnvConfigManager.get_env_var('TEST_BOOL', True, bool)
67
+ self.assertFalse(result)
68
+
69
+ # 测试默认值
70
+ result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', True, bool)
71
+ self.assertTrue(result)
72
+
73
+ def test_get_redis_config(self):
74
+ """测试 Redis 配置获取"""
75
+ with patch.dict(os.environ, {
76
+ 'REDIS_HOST': 'localhost',
77
+ 'REDIS_PORT': '6380',
78
+ 'REDIS_PASSWORD': 'secret',
79
+ 'REDIS_DB': '1'
80
+ }):
81
+ config = get_redis_config()
82
+ self.assertEqual(config['REDIS_HOST'], 'localhost')
83
+ self.assertEqual(config['REDIS_PORT'], 6380)
84
+ self.assertEqual(config['REDIS_PASSWORD'], 'secret')
85
+ self.assertEqual(config['REDIS_DB'], 1)
86
+
87
+ # 测试默认值
88
+ with patch.dict(os.environ, {}):
89
+ config = get_redis_config()
90
+ self.assertEqual(config['REDIS_HOST'], '127.0.0.1')
91
+ self.assertEqual(config['REDIS_PORT'], 6379)
92
+ self.assertEqual(config['REDIS_PASSWORD'], '')
93
+ self.assertEqual(config['REDIS_DB'], 0)
94
+
95
+ def test_get_runtime_config(self):
96
+ """测试运行时配置获取"""
97
+ with patch.dict(os.environ, {
98
+ 'CRAWLO_MODE': 'distributed',
99
+ 'PROJECT_NAME': 'test_project',
100
+ 'CONCURRENCY': '20'
101
+ }):
102
+ config = get_runtime_config()
103
+ self.assertEqual(config['CRAWLO_MODE'], 'distributed')
104
+ self.assertEqual(config['PROJECT_NAME'], 'test_project')
105
+ self.assertEqual(config['CONCURRENCY'], 20)
106
+
107
+ # 测试默认值
108
+ with patch.dict(os.environ, {}):
109
+ config = get_runtime_config()
110
+ self.assertEqual(config['CRAWLO_MODE'], 'standalone')
111
+ self.assertEqual(config['PROJECT_NAME'], 'crawlo')
112
+ self.assertEqual(config['CONCURRENCY'], 8)
113
+
114
+ def test_convenience_functions(self):
115
+ """测试便捷函数"""
116
+ with patch.dict(os.environ, {'TEST_CONVENIENCE': 'convenience_value'}):
117
+ result = get_env_var('TEST_CONVENIENCE', 'default')
118
+ self.assertEqual(result, 'convenience_value')
119
+
120
+
121
+ if __name__ == '__main__':
122
122
  unittest.main()