crawlo 1.3.2__py3-none-any.whl → 1.3.3__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 (219) hide show
  1. crawlo/__init__.py +63 -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 +322 -314
  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 +2 -2
  16. crawlo/core/engine.py +365 -365
  17. crawlo/core/processor.py +40 -40
  18. crawlo/core/scheduler.py +256 -256
  19. crawlo/crawler.py +1166 -1168
  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 +226 -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 +52 -45
  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/filters/__init__.py +154 -154
  40. crawlo/filters/aioredis_filter.py +234 -234
  41. crawlo/filters/memory_filter.py +269 -269
  42. crawlo/items/__init__.py +23 -23
  43. crawlo/items/base.py +21 -21
  44. crawlo/items/fields.py +52 -52
  45. crawlo/items/items.py +104 -104
  46. crawlo/middleware/__init__.py +21 -21
  47. crawlo/middleware/default_header.py +132 -132
  48. crawlo/middleware/download_delay.py +104 -104
  49. crawlo/middleware/middleware_manager.py +135 -135
  50. crawlo/middleware/offsite.py +123 -123
  51. crawlo/middleware/proxy.py +386 -386
  52. crawlo/middleware/request_ignore.py +86 -86
  53. crawlo/middleware/response_code.py +163 -163
  54. crawlo/middleware/response_filter.py +136 -136
  55. crawlo/middleware/retry.py +124 -124
  56. crawlo/middleware/simple_proxy.py +65 -65
  57. crawlo/mode_manager.py +187 -187
  58. crawlo/network/__init__.py +21 -21
  59. crawlo/network/request.py +379 -379
  60. crawlo/network/response.py +359 -359
  61. crawlo/pipelines/__init__.py +21 -21
  62. crawlo/pipelines/bloom_dedup_pipeline.py +156 -156
  63. crawlo/pipelines/console_pipeline.py +39 -39
  64. crawlo/pipelines/csv_pipeline.py +316 -316
  65. crawlo/pipelines/database_dedup_pipeline.py +222 -222
  66. crawlo/pipelines/json_pipeline.py +218 -218
  67. crawlo/pipelines/memory_dedup_pipeline.py +115 -115
  68. crawlo/pipelines/mongo_pipeline.py +131 -131
  69. crawlo/pipelines/mysql_pipeline.py +318 -318
  70. crawlo/pipelines/pipeline_manager.py +75 -75
  71. crawlo/pipelines/redis_dedup_pipeline.py +166 -166
  72. crawlo/project.py +325 -297
  73. crawlo/queue/pqueue.py +37 -37
  74. crawlo/queue/queue_manager.py +379 -379
  75. crawlo/queue/redis_priority_queue.py +306 -306
  76. crawlo/settings/__init__.py +7 -7
  77. crawlo/settings/default_settings.py +225 -225
  78. crawlo/settings/setting_manager.py +198 -198
  79. crawlo/spider/__init__.py +639 -639
  80. crawlo/stats_collector.py +59 -59
  81. crawlo/subscriber.py +129 -129
  82. crawlo/task_manager.py +30 -30
  83. crawlo/templates/crawlo.cfg.tmpl +10 -10
  84. crawlo/templates/project/__init__.py.tmpl +3 -3
  85. crawlo/templates/project/items.py.tmpl +17 -17
  86. crawlo/templates/project/middlewares.py.tmpl +118 -118
  87. crawlo/templates/project/pipelines.py.tmpl +96 -96
  88. crawlo/templates/project/settings.py.tmpl +266 -266
  89. crawlo/templates/project/settings_distributed.py.tmpl +179 -179
  90. crawlo/templates/project/settings_gentle.py.tmpl +60 -60
  91. crawlo/templates/project/settings_high_performance.py.tmpl +130 -130
  92. crawlo/templates/project/settings_minimal.py.tmpl +34 -34
  93. crawlo/templates/project/settings_simple.py.tmpl +101 -101
  94. crawlo/templates/project/spiders/__init__.py.tmpl +5 -5
  95. crawlo/templates/run.py.tmpl +38 -38
  96. crawlo/templates/spider/spider.py.tmpl +143 -143
  97. crawlo/tools/__init__.py +200 -200
  98. crawlo/tools/anti_crawler.py +268 -268
  99. crawlo/tools/authenticated_proxy.py +240 -240
  100. crawlo/tools/data_formatter.py +225 -225
  101. crawlo/tools/data_validator.py +180 -180
  102. crawlo/tools/date_tools.py +289 -289
  103. crawlo/tools/distributed_coordinator.py +388 -388
  104. crawlo/tools/encoding_converter.py +127 -127
  105. crawlo/tools/request_tools.py +82 -82
  106. crawlo/tools/retry_mechanism.py +224 -224
  107. crawlo/tools/scenario_adapter.py +262 -262
  108. crawlo/tools/text_cleaner.py +232 -232
  109. crawlo/utils/__init__.py +34 -34
  110. crawlo/utils/batch_processor.py +259 -259
  111. crawlo/utils/controlled_spider_mixin.py +439 -439
  112. crawlo/utils/db_helper.py +343 -343
  113. crawlo/utils/enhanced_error_handler.py +356 -356
  114. crawlo/utils/env_config.py +142 -142
  115. crawlo/utils/error_handler.py +123 -123
  116. crawlo/utils/func_tools.py +82 -82
  117. crawlo/utils/large_scale_config.py +286 -286
  118. crawlo/utils/large_scale_helper.py +344 -344
  119. crawlo/utils/log.py +199 -146
  120. crawlo/utils/performance_monitor.py +285 -285
  121. crawlo/utils/queue_helper.py +175 -175
  122. crawlo/utils/redis_connection_pool.py +351 -351
  123. crawlo/utils/redis_key_validator.py +198 -198
  124. crawlo/utils/request.py +267 -267
  125. crawlo/utils/request_serializer.py +218 -218
  126. crawlo/utils/spider_loader.py +61 -61
  127. crawlo/utils/system.py +11 -11
  128. crawlo/utils/tools.py +4 -4
  129. crawlo/utils/url.py +39 -39
  130. {crawlo-1.3.2.dist-info → crawlo-1.3.3.dist-info}/METADATA +1020 -1020
  131. crawlo-1.3.3.dist-info/RECORD +219 -0
  132. examples/__init__.py +7 -7
  133. tests/DOUBLE_CRAWLO_PREFIX_FIX_REPORT.md +81 -81
  134. tests/__init__.py +7 -7
  135. tests/advanced_tools_example.py +275 -275
  136. tests/authenticated_proxy_example.py +107 -107
  137. tests/cleaners_example.py +160 -160
  138. tests/config_validation_demo.py +142 -142
  139. tests/controlled_spider_example.py +205 -205
  140. tests/date_tools_example.py +180 -180
  141. tests/debug_pipelines.py +66 -66
  142. tests/dynamic_loading_example.py +523 -523
  143. tests/dynamic_loading_test.py +104 -104
  144. tests/env_config_example.py +133 -133
  145. tests/error_handling_example.py +171 -171
  146. tests/redis_key_validation_demo.py +130 -130
  147. tests/request_params_example.py +150 -150
  148. tests/response_improvements_example.py +144 -144
  149. tests/test_advanced_tools.py +148 -148
  150. tests/test_all_redis_key_configs.py +145 -145
  151. tests/test_authenticated_proxy.py +141 -141
  152. tests/test_cleaners.py +54 -54
  153. tests/test_comprehensive.py +146 -146
  154. tests/test_config_consistency.py +80 -80
  155. tests/test_config_merge.py +152 -152
  156. tests/test_config_validator.py +182 -182
  157. tests/test_crawlo_proxy_integration.py +108 -108
  158. tests/test_date_tools.py +123 -123
  159. tests/test_default_header_middleware.py +158 -158
  160. tests/test_distributed.py +65 -65
  161. tests/test_double_crawlo_fix.py +207 -207
  162. tests/test_double_crawlo_fix_simple.py +124 -124
  163. tests/test_download_delay_middleware.py +221 -221
  164. tests/test_downloader_proxy_compatibility.py +268 -268
  165. tests/test_dynamic_downloaders_proxy.py +124 -124
  166. tests/test_dynamic_proxy.py +92 -92
  167. tests/test_dynamic_proxy_config.py +146 -146
  168. tests/test_dynamic_proxy_real.py +109 -109
  169. tests/test_edge_cases.py +303 -303
  170. tests/test_enhanced_error_handler.py +270 -270
  171. tests/test_env_config.py +121 -121
  172. tests/test_error_handler_compatibility.py +112 -112
  173. tests/test_final_validation.py +153 -153
  174. tests/test_framework_env_usage.py +103 -103
  175. tests/test_integration.py +169 -169
  176. tests/test_item_dedup_redis_key.py +122 -122
  177. tests/test_mode_consistency.py +51 -51
  178. tests/test_offsite_middleware.py +221 -221
  179. tests/test_parsel.py +29 -29
  180. tests/test_performance.py +327 -327
  181. tests/test_proxy_api.py +264 -264
  182. tests/test_proxy_health_check.py +32 -32
  183. tests/test_proxy_middleware.py +121 -121
  184. tests/test_proxy_middleware_enhanced.py +216 -216
  185. tests/test_proxy_middleware_integration.py +136 -136
  186. tests/test_proxy_middleware_refactored.py +184 -184
  187. tests/test_proxy_providers.py +56 -56
  188. tests/test_proxy_stats.py +19 -19
  189. tests/test_proxy_strategies.py +59 -59
  190. tests/test_queue_manager_double_crawlo.py +173 -173
  191. tests/test_queue_manager_redis_key.py +176 -176
  192. tests/test_random_user_agent.py +72 -72
  193. tests/test_real_scenario_proxy.py +195 -195
  194. tests/test_redis_config.py +28 -28
  195. tests/test_redis_connection_pool.py +294 -294
  196. tests/test_redis_key_naming.py +181 -181
  197. tests/test_redis_key_validator.py +123 -123
  198. tests/test_redis_queue.py +224 -224
  199. tests/test_request_ignore_middleware.py +182 -182
  200. tests/test_request_params.py +111 -111
  201. tests/test_request_serialization.py +70 -70
  202. tests/test_response_code_middleware.py +349 -349
  203. tests/test_response_filter_middleware.py +427 -427
  204. tests/test_response_improvements.py +152 -152
  205. tests/test_retry_middleware.py +241 -241
  206. tests/test_scheduler.py +252 -252
  207. tests/test_scheduler_config_update.py +133 -133
  208. tests/test_simple_response.py +61 -61
  209. tests/test_telecom_spider_redis_key.py +205 -205
  210. tests/test_template_content.py +87 -87
  211. tests/test_template_redis_key.py +134 -134
  212. tests/test_tools.py +159 -159
  213. tests/test_user_agents.py +96 -96
  214. tests/tools_example.py +260 -260
  215. tests/verify_distributed.py +117 -117
  216. crawlo-1.3.2.dist-info/RECORD +0 -219
  217. {crawlo-1.3.2.dist-info → crawlo-1.3.3.dist-info}/WHEEL +0 -0
  218. {crawlo-1.3.2.dist-info → crawlo-1.3.3.dist-info}/entry_points.txt +0 -0
  219. {crawlo-1.3.2.dist-info → crawlo-1.3.3.dist-info}/top_level.txt +0 -0
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()
@@ -1,113 +1,113 @@
1
- #!/usr/bin/env python3
2
- # -*- coding: utf-8 -*-
3
- """
4
- 错误处理模块兼容性测试
5
- 验证简化版 ErrorHandler 与增强版 EnhancedErrorHandler 的兼容性
6
- """
7
- import sys
8
- import os
9
- import unittest
10
- from unittest.mock import patch, MagicMock
11
-
12
- # 添加项目根目录到Python路径
13
- sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
14
-
15
- from crawlo.utils.error_handler import ErrorHandler, default_error_handler, handle_exception
16
- from crawlo.utils.enhanced_error_handler import EnhancedErrorHandler
17
-
18
-
19
- class TestErrorHandlerCompatibility(unittest.TestCase):
20
- """错误处理模块兼容性测试"""
21
-
22
- def setUp(self):
23
- """测试前准备"""
24
- self.error_handler = ErrorHandler("test_logger")
25
- # 禁用实际日志输出
26
- self.error_handler.logger.error = MagicMock()
27
- self.error_handler.logger.debug = MagicMock()
28
- self.error_handler._enhanced_handler.logger.error = MagicMock()
29
- self.error_handler._enhanced_handler.logger.debug = MagicMock()
30
-
31
- def test_init(self):
32
- """测试初始化"""
33
- self.assertIsInstance(self.error_handler._enhanced_handler, EnhancedErrorHandler)
34
- self.assertEqual(self.error_handler._enhanced_handler.logger.name, "test_logger")
35
-
36
- def test_handle_error(self):
37
- """测试错误处理"""
38
- test_exception = ValueError("Test error")
39
-
40
- # 测试不抛出异常的情况
41
- try:
42
- self.error_handler.handle_error(test_exception, "Test context", raise_error=False)
43
- except Exception:
44
- self.fail("handle_error should not raise exception when raise_error=False")
45
-
46
- # 验证增强版处理器被调用(通过检查日志是否被调用)
47
- self.assertTrue(self.error_handler._enhanced_handler.logger.error.called)
48
-
49
- def test_safe_call(self):
50
- """测试安全调用"""
51
- def failing_function():
52
- raise RuntimeError("Function failed")
53
-
54
- # 测试函数失败时返回默认值
55
- result = self.error_handler.safe_call(failing_function, default_return="default")
56
- self.assertEqual(result, "default")
57
-
58
- # 测试正常函数
59
- def normal_function(x, y):
60
- return x + y
61
-
62
- result = self.error_handler.safe_call(normal_function, 1, 2, default_return=0)
63
- self.assertEqual(result, 3)
64
-
65
- def test_retry_on_failure(self):
66
- """测试失败重试装饰器"""
67
- call_count = 0
68
-
69
- @self.error_handler.retry_on_failure(max_retries=2, delay=0.01)
70
- def failing_function():
71
- nonlocal call_count
72
- call_count += 1
73
- if call_count < 3:
74
- raise ValueError("Temporary failure")
75
- return "success"
76
-
77
- # 第三次调用应该成功
78
- result = failing_function()
79
- self.assertEqual(result, "success")
80
- self.assertEqual(call_count, 3)
81
-
82
- def test_handle_exception_decorator(self):
83
- """测试异常处理装饰器"""
84
- @handle_exception(context="Test function", raise_error=False)
85
- def decorated_function():
86
- raise ValueError("Decorated function error")
87
-
88
- # 调用装饰后的函数不应该抛出异常
89
- try:
90
- decorated_function()
91
- except Exception:
92
- self.fail("Decorated function should not raise exception when raise_error=False")
93
-
94
- def test_async_handle_exception_decorator(self):
95
- """测试异步异常处理装饰器"""
96
- import asyncio
97
-
98
- @handle_exception(context="Async test function", raise_error=False)
99
- async def async_decorated_function():
100
- raise ValueError("Async decorated function error")
101
-
102
- # 调用装饰后的异步函数不应该抛出异常
103
- async def test_async():
104
- try:
105
- await async_decorated_function()
106
- except Exception:
107
- self.fail("Async decorated function should not raise exception when raise_error=False")
108
-
109
- asyncio.run(test_async())
110
-
111
-
112
- if __name__ == '__main__':
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ 错误处理模块兼容性测试
5
+ 验证简化版 ErrorHandler 与增强版 EnhancedErrorHandler 的兼容性
6
+ """
7
+ import sys
8
+ import os
9
+ import unittest
10
+ from unittest.mock import patch, MagicMock
11
+
12
+ # 添加项目根目录到Python路径
13
+ sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
14
+
15
+ from crawlo.utils.error_handler import ErrorHandler, default_error_handler, handle_exception
16
+ from crawlo.utils.enhanced_error_handler import EnhancedErrorHandler
17
+
18
+
19
+ class TestErrorHandlerCompatibility(unittest.TestCase):
20
+ """错误处理模块兼容性测试"""
21
+
22
+ def setUp(self):
23
+ """测试前准备"""
24
+ self.error_handler = ErrorHandler("test_logger")
25
+ # 禁用实际日志输出
26
+ self.error_handler.logger.error = MagicMock()
27
+ self.error_handler.logger.debug = MagicMock()
28
+ self.error_handler._enhanced_handler.logger.error = MagicMock()
29
+ self.error_handler._enhanced_handler.logger.debug = MagicMock()
30
+
31
+ def test_init(self):
32
+ """测试初始化"""
33
+ self.assertIsInstance(self.error_handler._enhanced_handler, EnhancedErrorHandler)
34
+ self.assertEqual(self.error_handler._enhanced_handler.logger.name, "test_logger")
35
+
36
+ def test_handle_error(self):
37
+ """测试错误处理"""
38
+ test_exception = ValueError("Test error")
39
+
40
+ # 测试不抛出异常的情况
41
+ try:
42
+ self.error_handler.handle_error(test_exception, "Test context", raise_error=False)
43
+ except Exception:
44
+ self.fail("handle_error should not raise exception when raise_error=False")
45
+
46
+ # 验证增强版处理器被调用(通过检查日志是否被调用)
47
+ self.assertTrue(self.error_handler._enhanced_handler.logger.error.called)
48
+
49
+ def test_safe_call(self):
50
+ """测试安全调用"""
51
+ def failing_function():
52
+ raise RuntimeError("Function failed")
53
+
54
+ # 测试函数失败时返回默认值
55
+ result = self.error_handler.safe_call(failing_function, default_return="default")
56
+ self.assertEqual(result, "default")
57
+
58
+ # 测试正常函数
59
+ def normal_function(x, y):
60
+ return x + y
61
+
62
+ result = self.error_handler.safe_call(normal_function, 1, 2, default_return=0)
63
+ self.assertEqual(result, 3)
64
+
65
+ def test_retry_on_failure(self):
66
+ """测试失败重试装饰器"""
67
+ call_count = 0
68
+
69
+ @self.error_handler.retry_on_failure(max_retries=2, delay=0.01)
70
+ def failing_function():
71
+ nonlocal call_count
72
+ call_count += 1
73
+ if call_count < 3:
74
+ raise ValueError("Temporary failure")
75
+ return "success"
76
+
77
+ # 第三次调用应该成功
78
+ result = failing_function()
79
+ self.assertEqual(result, "success")
80
+ self.assertEqual(call_count, 3)
81
+
82
+ def test_handle_exception_decorator(self):
83
+ """测试异常处理装饰器"""
84
+ @handle_exception(context="Test function", raise_error=False)
85
+ def decorated_function():
86
+ raise ValueError("Decorated function error")
87
+
88
+ # 调用装饰后的函数不应该抛出异常
89
+ try:
90
+ decorated_function()
91
+ except Exception:
92
+ self.fail("Decorated function should not raise exception when raise_error=False")
93
+
94
+ def test_async_handle_exception_decorator(self):
95
+ """测试异步异常处理装饰器"""
96
+ import asyncio
97
+
98
+ @handle_exception(context="Async test function", raise_error=False)
99
+ async def async_decorated_function():
100
+ raise ValueError("Async decorated function error")
101
+
102
+ # 调用装饰后的异步函数不应该抛出异常
103
+ async def test_async():
104
+ try:
105
+ await async_decorated_function()
106
+ except Exception:
107
+ self.fail("Async decorated function should not raise exception when raise_error=False")
108
+
109
+ asyncio.run(test_async())
110
+
111
+
112
+ if __name__ == '__main__':
113
113
  unittest.main()