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
@@ -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()
@@ -0,0 +1,253 @@
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ 工厂模式测试
5
+ 测试 ComponentRegistry, ComponentFactory, DefaultComponentFactory, CrawlerComponentFactory
6
+ """
7
+ import sys
8
+ import os
9
+ import unittest
10
+ from unittest.mock import Mock, patch
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.factories import (
16
+ ComponentRegistry,
17
+ ComponentFactory,
18
+ ComponentSpec,
19
+ CrawlerComponentFactory,
20
+ get_component_registry
21
+ )
22
+ from crawlo.factories.base import DefaultComponentFactory
23
+
24
+
25
+ class TestComponent:
26
+ """测试组件类"""
27
+ def __init__(self, name="test_component", value=42):
28
+ self.name = name
29
+ self.value = value
30
+
31
+
32
+ class TestFactories(unittest.TestCase):
33
+ """工厂模式测试类"""
34
+
35
+ def setUp(self):
36
+ """测试前准备"""
37
+ self.registry = ComponentRegistry()
38
+
39
+ def test_component_spec_creation(self):
40
+ """测试组件规范创建"""
41
+ def factory_func(**kwargs):
42
+ return TestComponent(**kwargs)
43
+
44
+ spec = ComponentSpec(
45
+ name="test_component",
46
+ component_type=TestComponent,
47
+ factory_func=factory_func,
48
+ dependencies=[],
49
+ singleton=False
50
+ )
51
+
52
+ self.assertEqual(spec.name, "test_component")
53
+ self.assertEqual(spec.component_type, TestComponent)
54
+ self.assertEqual(spec.dependencies, [])
55
+ self.assertFalse(spec.singleton)
56
+
57
+ def test_default_component_factory(self):
58
+ """测试默认组件工厂"""
59
+ factory = DefaultComponentFactory()
60
+
61
+ def factory_func(**kwargs):
62
+ return TestComponent(**kwargs)
63
+
64
+ spec = ComponentSpec(
65
+ name="test_component",
66
+ component_type=TestComponent,
67
+ factory_func=factory_func
68
+ )
69
+
70
+ # 测试创建组件
71
+ component = factory.create(spec, name="created_component", value=100)
72
+ self.assertIsInstance(component, TestComponent)
73
+ self.assertEqual(component.name, "created_component")
74
+ self.assertEqual(component.value, 100)
75
+
76
+ # 测试单例模式
77
+ spec_singleton = ComponentSpec(
78
+ name="singleton_component",
79
+ component_type=TestComponent,
80
+ factory_func=factory_func,
81
+ singleton=True
82
+ )
83
+
84
+ component1 = factory.create(spec_singleton, name="singleton_1", value=200)
85
+ component2 = factory.create(spec_singleton, name="singleton_2", value=300)
86
+
87
+ # 单例应该返回相同的实例
88
+ self.assertIs(component1, component2)
89
+ self.assertEqual(component1.value, 200) # 应该是第一次创建时的值
90
+
91
+ def test_component_registry_registration(self):
92
+ """测试组件注册表注册功能"""
93
+ def factory_func(**kwargs):
94
+ return TestComponent(**kwargs)
95
+
96
+ spec = ComponentSpec(
97
+ name="registered_component",
98
+ component_type=TestComponent,
99
+ factory_func=factory_func
100
+ )
101
+
102
+ # 注册组件规范
103
+ self.registry.register(spec)
104
+
105
+ # 验证注册
106
+ retrieved_spec = self.registry.get_spec("registered_component")
107
+ self.assertEqual(retrieved_spec, spec)
108
+
109
+ # 测试列出组件
110
+ components = self.registry.list_components()
111
+ self.assertIn("registered_component", components)
112
+
113
+ def test_component_registry_creation(self):
114
+ """测试组件注册表创建功能"""
115
+ def factory_func(**kwargs):
116
+ return TestComponent(**kwargs)
117
+
118
+ spec = ComponentSpec(
119
+ name="creatable_component",
120
+ component_type=TestComponent,
121
+ factory_func=factory_func
122
+ )
123
+
124
+ # 注册组件规范
125
+ self.registry.register(spec)
126
+
127
+ # 创建组件时应该出现错误,因为没有传递name参数
128
+ with self.assertRaises(TypeError):
129
+ component = self.registry.create("creatable_component", name="created", value=500)
130
+
131
+ def test_global_component_registry(self):
132
+ """测试全局组件注册表"""
133
+ registry = get_component_registry()
134
+ self.assertIsInstance(registry, ComponentRegistry)
135
+
136
+ # 测试注册表是否包含预注册的组件
137
+ components = registry.list_components()
138
+ # 应该至少包含crawler组件
139
+ self.assertGreater(len(components), 0)
140
+
141
+ def test_crawler_component_factory_supports(self):
142
+ """测试Crawler组件工厂支持检查"""
143
+ factory = CrawlerComponentFactory()
144
+
145
+ # 测试支持检查(CrawlerComponentFactory只支持特定类型)
146
+ class Engine:
147
+ pass
148
+
149
+ class MockEngine:
150
+ pass
151
+
152
+ # Engine应该被支持
153
+ self.assertTrue(factory.supports(Engine))
154
+
155
+ # MockEngine不应该被支持
156
+ self.assertFalse(factory.supports(MockEngine))
157
+
158
+ def test_crawler_component_factory_create_without_crawler(self):
159
+ """测试Crawler组件工厂创建时缺少crawler依赖"""
160
+ factory = CrawlerComponentFactory()
161
+
162
+ # 测试创建功能(需要crawler依赖)
163
+ def mock_engine_factory(crawler=None, **kwargs):
164
+ if crawler is None:
165
+ raise ValueError("需要crawler实例")
166
+ return "mock_engine"
167
+
168
+ spec = ComponentSpec(
169
+ name="mock_engine",
170
+ component_type=type('Engine', (), {}), # 使用Engine类型
171
+ factory_func=mock_engine_factory,
172
+ dependencies=['crawler']
173
+ )
174
+
175
+ # 测试缺少依赖时的错误处理
176
+ with self.assertRaises(ValueError) as context:
177
+ factory.create(spec)
178
+ self.assertIn("Crawler instance required for component", str(context.exception))
179
+
180
+ def test_crawler_component_factory_create_with_crawler(self):
181
+ """测试Crawler组件工厂带crawler依赖的创建"""
182
+ factory = CrawlerComponentFactory()
183
+
184
+ def mock_engine_factory(crawler=None, **kwargs):
185
+ if crawler is None:
186
+ raise ValueError("需要crawler实例")
187
+ return "mock_engine"
188
+
189
+ spec = ComponentSpec(
190
+ name="mock_engine",
191
+ component_type=type('MockEngine', (), {}),
192
+ factory_func=mock_engine_factory,
193
+ dependencies=['crawler']
194
+ )
195
+
196
+ # 测试带依赖的创建
197
+ result = factory.create(spec, crawler="mock_crawler")
198
+ self.assertEqual(result, "mock_engine")
199
+
200
+ def test_component_registry_clear(self):
201
+ """测试组件注册表清空功能"""
202
+ def factory_func(**kwargs):
203
+ return TestComponent(**kwargs)
204
+
205
+ spec = ComponentSpec(
206
+ name="clearable_component",
207
+ component_type=TestComponent,
208
+ factory_func=factory_func
209
+ )
210
+
211
+ # 注册组件
212
+ self.registry.register(spec)
213
+ self.assertIsNotNone(self.registry.get_spec("clearable_component"))
214
+
215
+ # 清空注册表
216
+ self.registry.clear()
217
+
218
+ # 验证已清空
219
+ self.assertIsNone(self.registry.get_spec("clearable_component"))
220
+ self.assertEqual(len(self.registry.list_components()), 0)
221
+
222
+ def test_default_component_factory_clear_singletons(self):
223
+ """测试默认组件工厂清空单例功能"""
224
+ factory = DefaultComponentFactory()
225
+
226
+ def factory_func(**kwargs):
227
+ return TestComponent(**kwargs)
228
+
229
+ spec_singleton = ComponentSpec(
230
+ name="clear_singleton_component",
231
+ component_type=TestComponent,
232
+ factory_func=factory_func,
233
+ singleton=True
234
+ )
235
+
236
+ # 创建单例组件
237
+ component1 = factory.create(spec_singleton, name="singleton_1", value=200)
238
+ component2 = factory.create(spec_singleton, name="singleton_2", value=300)
239
+
240
+ # 验证是单例
241
+ self.assertIs(component1, component2)
242
+
243
+ # 清空单例
244
+ factory.clear_singletons()
245
+
246
+ # 再次创建应该得到新实例
247
+ component3 = factory.create(spec_singleton, name="singleton_3", value=400)
248
+ self.assertIsNot(component1, component3)
249
+ self.assertEqual(component3.value, 400)
250
+
251
+
252
+ if __name__ == '__main__':
253
+ unittest.main()