crawlo 1.4.1__py3-none-any.whl → 1.4.3__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.


This version of crawlo might be problematic. Click here for more details.

Files changed (326) hide show
  1. crawlo/__init__.py +93 -93
  2. crawlo/__version__.py +1 -1
  3. crawlo/cli.py +75 -75
  4. crawlo/commands/__init__.py +14 -14
  5. crawlo/commands/check.py +594 -594
  6. crawlo/commands/genspider.py +151 -151
  7. crawlo/commands/help.py +138 -138
  8. crawlo/commands/list.py +155 -155
  9. crawlo/commands/run.py +341 -341
  10. crawlo/commands/startproject.py +436 -436
  11. crawlo/commands/stats.py +187 -187
  12. crawlo/commands/utils.py +196 -196
  13. crawlo/config.py +312 -312
  14. crawlo/config_validator.py +277 -277
  15. crawlo/core/__init__.py +52 -52
  16. crawlo/core/engine.py +438 -439
  17. crawlo/core/processor.py +47 -47
  18. crawlo/core/scheduler.py +291 -257
  19. crawlo/crawler.py +650 -650
  20. crawlo/data/__init__.py +5 -5
  21. crawlo/data/user_agents.py +194 -194
  22. crawlo/downloader/__init__.py +273 -273
  23. crawlo/downloader/aiohttp_downloader.py +233 -228
  24. crawlo/downloader/cffi_downloader.py +245 -245
  25. crawlo/downloader/httpx_downloader.py +259 -259
  26. crawlo/downloader/hybrid_downloader.py +212 -212
  27. crawlo/downloader/playwright_downloader.py +402 -402
  28. crawlo/downloader/selenium_downloader.py +472 -472
  29. crawlo/event.py +11 -11
  30. crawlo/exceptions.py +81 -81
  31. crawlo/extension/__init__.py +63 -63
  32. crawlo/extension/health_check.py +141 -141
  33. crawlo/extension/log_interval.py +94 -94
  34. crawlo/extension/log_stats.py +70 -70
  35. crawlo/extension/logging_extension.py +61 -61
  36. crawlo/extension/memory_monitor.py +104 -104
  37. crawlo/extension/performance_profiler.py +133 -133
  38. crawlo/extension/request_recorder.py +107 -107
  39. crawlo/factories/__init__.py +27 -27
  40. crawlo/factories/base.py +68 -68
  41. crawlo/factories/crawler.py +103 -103
  42. crawlo/factories/registry.py +84 -84
  43. crawlo/filters/__init__.py +154 -154
  44. crawlo/filters/aioredis_filter.py +257 -257
  45. crawlo/filters/memory_filter.py +269 -269
  46. crawlo/framework.py +292 -292
  47. crawlo/initialization/__init__.py +44 -44
  48. crawlo/initialization/built_in.py +425 -425
  49. crawlo/initialization/context.py +141 -141
  50. crawlo/initialization/core.py +193 -193
  51. crawlo/initialization/phases.py +148 -148
  52. crawlo/initialization/registry.py +145 -145
  53. crawlo/items/__init__.py +23 -23
  54. crawlo/items/base.py +23 -23
  55. crawlo/items/fields.py +52 -52
  56. crawlo/items/items.py +104 -104
  57. crawlo/logging/__init__.py +45 -37
  58. crawlo/logging/async_handler.py +181 -0
  59. crawlo/logging/config.py +196 -96
  60. crawlo/logging/factory.py +171 -128
  61. crawlo/logging/manager.py +111 -111
  62. crawlo/logging/monitor.py +153 -0
  63. crawlo/logging/sampler.py +167 -0
  64. crawlo/middleware/__init__.py +21 -21
  65. crawlo/middleware/default_header.py +132 -132
  66. crawlo/middleware/download_delay.py +104 -104
  67. crawlo/middleware/middleware_manager.py +135 -135
  68. crawlo/middleware/offsite.py +123 -123
  69. crawlo/middleware/proxy.py +386 -386
  70. crawlo/middleware/request_ignore.py +86 -86
  71. crawlo/middleware/response_code.py +150 -150
  72. crawlo/middleware/response_filter.py +136 -136
  73. crawlo/middleware/retry.py +124 -124
  74. crawlo/middleware/simple_proxy.py +65 -65
  75. crawlo/mode_manager.py +219 -219
  76. crawlo/network/__init__.py +21 -21
  77. crawlo/network/request.py +379 -379
  78. crawlo/network/response.py +359 -359
  79. crawlo/pipelines/__init__.py +21 -21
  80. crawlo/pipelines/bloom_dedup_pipeline.py +146 -146
  81. crawlo/pipelines/console_pipeline.py +39 -39
  82. crawlo/pipelines/csv_pipeline.py +316 -316
  83. crawlo/pipelines/database_dedup_pipeline.py +197 -197
  84. crawlo/pipelines/json_pipeline.py +218 -218
  85. crawlo/pipelines/memory_dedup_pipeline.py +105 -105
  86. crawlo/pipelines/mongo_pipeline.py +131 -131
  87. crawlo/pipelines/mysql_pipeline.py +325 -325
  88. crawlo/pipelines/pipeline_manager.py +100 -84
  89. crawlo/pipelines/redis_dedup_pipeline.py +156 -156
  90. crawlo/project.py +349 -338
  91. crawlo/queue/pqueue.py +42 -42
  92. crawlo/queue/queue_manager.py +526 -522
  93. crawlo/queue/redis_priority_queue.py +370 -367
  94. crawlo/settings/__init__.py +7 -7
  95. crawlo/settings/default_settings.py +284 -284
  96. crawlo/settings/setting_manager.py +219 -219
  97. crawlo/spider/__init__.py +657 -657
  98. crawlo/stats_collector.py +73 -73
  99. crawlo/subscriber.py +129 -129
  100. crawlo/task_manager.py +138 -138
  101. crawlo/templates/crawlo.cfg.tmpl +10 -10
  102. crawlo/templates/project/__init__.py.tmpl +3 -3
  103. crawlo/templates/project/items.py.tmpl +17 -17
  104. crawlo/templates/project/middlewares.py.tmpl +118 -118
  105. crawlo/templates/project/pipelines.py.tmpl +96 -96
  106. crawlo/templates/project/settings.py.tmpl +170 -170
  107. crawlo/templates/project/settings_distributed.py.tmpl +169 -169
  108. crawlo/templates/project/settings_gentle.py.tmpl +166 -166
  109. crawlo/templates/project/settings_high_performance.py.tmpl +167 -167
  110. crawlo/templates/project/settings_minimal.py.tmpl +65 -65
  111. crawlo/templates/project/settings_simple.py.tmpl +164 -164
  112. crawlo/templates/project/spiders/__init__.py.tmpl +9 -9
  113. crawlo/templates/run.py.tmpl +34 -34
  114. crawlo/templates/spider/spider.py.tmpl +143 -143
  115. crawlo/templates/spiders_init.py.tmpl +9 -9
  116. crawlo/tools/__init__.py +200 -200
  117. crawlo/tools/anti_crawler.py +268 -268
  118. crawlo/tools/authenticated_proxy.py +240 -240
  119. crawlo/tools/data_formatter.py +225 -225
  120. crawlo/tools/data_validator.py +180 -180
  121. crawlo/tools/date_tools.py +289 -289
  122. crawlo/tools/distributed_coordinator.py +384 -384
  123. crawlo/tools/encoding_converter.py +127 -127
  124. crawlo/tools/network_diagnostic.py +364 -364
  125. crawlo/tools/request_tools.py +82 -82
  126. crawlo/tools/retry_mechanism.py +224 -224
  127. crawlo/tools/scenario_adapter.py +262 -262
  128. crawlo/tools/text_cleaner.py +232 -232
  129. crawlo/utils/__init__.py +34 -34
  130. crawlo/utils/batch_processor.py +259 -259
  131. crawlo/utils/class_loader.py +25 -25
  132. crawlo/utils/controlled_spider_mixin.py +439 -439
  133. crawlo/utils/db_helper.py +343 -343
  134. crawlo/utils/enhanced_error_handler.py +356 -356
  135. crawlo/utils/env_config.py +142 -142
  136. crawlo/utils/error_handler.py +165 -165
  137. crawlo/utils/fingerprint.py +122 -122
  138. crawlo/utils/func_tools.py +82 -82
  139. crawlo/utils/large_scale_config.py +286 -286
  140. crawlo/utils/large_scale_helper.py +344 -344
  141. crawlo/utils/log.py +79 -79
  142. crawlo/utils/performance_monitor.py +285 -285
  143. crawlo/utils/queue_helper.py +175 -175
  144. crawlo/utils/redis_connection_pool.py +388 -388
  145. crawlo/utils/redis_key_validator.py +198 -198
  146. crawlo/utils/request.py +267 -267
  147. crawlo/utils/request_serializer.py +225 -225
  148. crawlo/utils/spider_loader.py +61 -61
  149. crawlo/utils/system.py +11 -11
  150. crawlo/utils/tools.py +4 -4
  151. crawlo/utils/url.py +39 -39
  152. crawlo-1.4.3.dist-info/METADATA +190 -0
  153. crawlo-1.4.3.dist-info/RECORD +326 -0
  154. examples/__init__.py +7 -7
  155. examples/test_project/__init__.py +7 -7
  156. examples/test_project/run.py +34 -34
  157. examples/test_project/test_project/__init__.py +3 -3
  158. examples/test_project/test_project/items.py +17 -17
  159. examples/test_project/test_project/middlewares.py +118 -118
  160. examples/test_project/test_project/pipelines.py +96 -96
  161. examples/test_project/test_project/settings.py +169 -169
  162. examples/test_project/test_project/spiders/__init__.py +9 -9
  163. examples/test_project/test_project/spiders/of_week_dis.py +143 -143
  164. tests/__init__.py +7 -7
  165. tests/advanced_tools_example.py +275 -275
  166. tests/authenticated_proxy_example.py +106 -106
  167. tests/baidu_performance_test.py +108 -108
  168. tests/baidu_test.py +59 -59
  169. tests/cleaners_example.py +160 -160
  170. tests/comprehensive_framework_test.py +212 -212
  171. tests/comprehensive_test.py +81 -81
  172. tests/comprehensive_testing_summary.md +186 -186
  173. tests/config_validation_demo.py +142 -142
  174. tests/controlled_spider_example.py +205 -205
  175. tests/date_tools_example.py +180 -180
  176. tests/debug_configure.py +69 -69
  177. tests/debug_framework_logger.py +84 -84
  178. tests/debug_log_config.py +126 -126
  179. tests/debug_log_levels.py +63 -63
  180. tests/debug_pipelines.py +66 -66
  181. tests/detailed_log_test.py +233 -233
  182. tests/distributed_test.py +66 -66
  183. tests/distributed_test_debug.py +76 -76
  184. tests/dynamic_loading_example.py +523 -523
  185. tests/dynamic_loading_test.py +104 -104
  186. tests/env_config_example.py +133 -133
  187. tests/error_handling_example.py +171 -171
  188. tests/final_comprehensive_test.py +151 -151
  189. tests/final_log_test.py +260 -260
  190. tests/final_validation_test.py +182 -182
  191. tests/fix_log_test.py +142 -142
  192. tests/framework_performance_test.py +202 -202
  193. tests/log_buffering_test.py +111 -111
  194. tests/log_generation_timing_test.py +153 -153
  195. tests/optimized_performance_test.py +211 -211
  196. tests/performance_comparison.py +245 -245
  197. tests/queue_blocking_test.py +113 -113
  198. tests/queue_test.py +89 -89
  199. tests/redis_key_validation_demo.py +130 -130
  200. tests/request_params_example.py +150 -150
  201. tests/response_improvements_example.py +144 -144
  202. tests/scrapy_comparison/ofweek_scrapy.py +138 -138
  203. tests/scrapy_comparison/scrapy_test.py +133 -133
  204. tests/simple_command_test.py +119 -119
  205. tests/simple_crawlo_test.py +127 -127
  206. tests/simple_log_test.py +57 -57
  207. tests/simple_log_test2.py +137 -137
  208. tests/simple_optimization_test.py +128 -128
  209. tests/simple_queue_type_test.py +41 -41
  210. tests/simple_spider_test.py +49 -49
  211. tests/simple_test.py +47 -47
  212. tests/spider_log_timing_test.py +177 -177
  213. tests/test_advanced_tools.py +148 -148
  214. tests/test_all_commands.py +230 -230
  215. tests/test_all_pipeline_fingerprints.py +133 -133
  216. tests/test_all_redis_key_configs.py +145 -145
  217. tests/test_authenticated_proxy.py +141 -141
  218. tests/test_batch_processor.py +178 -178
  219. tests/test_cleaners.py +54 -54
  220. tests/test_component_factory.py +174 -174
  221. tests/test_comprehensive.py +146 -146
  222. tests/test_config_consistency.py +80 -80
  223. tests/test_config_merge.py +152 -152
  224. tests/test_config_validator.py +182 -182
  225. tests/test_controlled_spider_mixin.py +79 -79
  226. tests/test_crawlo_proxy_integration.py +108 -108
  227. tests/test_date_tools.py +123 -123
  228. tests/test_dedup_fix.py +220 -220
  229. tests/test_dedup_pipeline_consistency.py +125 -0
  230. tests/test_default_header_middleware.py +313 -313
  231. tests/test_distributed.py +65 -65
  232. tests/test_double_crawlo_fix.py +204 -204
  233. tests/test_double_crawlo_fix_simple.py +124 -124
  234. tests/test_download_delay_middleware.py +221 -221
  235. tests/test_downloader_proxy_compatibility.py +268 -268
  236. tests/test_dynamic_downloaders_proxy.py +124 -124
  237. tests/test_dynamic_proxy.py +92 -92
  238. tests/test_dynamic_proxy_config.py +146 -146
  239. tests/test_dynamic_proxy_real.py +109 -109
  240. tests/test_edge_cases.py +303 -303
  241. tests/test_enhanced_error_handler.py +270 -270
  242. tests/test_enhanced_error_handler_comprehensive.py +245 -245
  243. tests/test_env_config.py +121 -121
  244. tests/test_error_handler_compatibility.py +112 -112
  245. tests/test_factories.py +252 -252
  246. tests/test_final_validation.py +153 -153
  247. tests/test_fingerprint_consistency.py +135 -135
  248. tests/test_fingerprint_simple.py +51 -51
  249. tests/test_framework_env_usage.py +103 -103
  250. tests/test_framework_logger.py +66 -66
  251. tests/test_framework_startup.py +64 -64
  252. tests/test_get_component_logger.py +83 -83
  253. tests/test_hash_performance.py +99 -99
  254. tests/test_integration.py +169 -169
  255. tests/test_item_dedup_redis_key.py +122 -122
  256. tests/test_large_scale_config.py +112 -112
  257. tests/test_large_scale_helper.py +235 -235
  258. tests/test_logging_enhancements.py +375 -0
  259. tests/test_logging_final.py +185 -0
  260. tests/test_logging_integration.py +313 -0
  261. tests/test_logging_system.py +282 -282
  262. tests/test_middleware_debug.py +142 -0
  263. tests/test_mode_change.py +72 -72
  264. tests/test_mode_consistency.py +51 -51
  265. tests/test_offsite_middleware.py +244 -244
  266. tests/test_offsite_middleware_simple.py +203 -203
  267. tests/test_parsel.py +29 -29
  268. tests/test_performance.py +327 -327
  269. tests/test_performance_monitor.py +115 -115
  270. tests/test_pipeline_fingerprint_consistency.py +86 -86
  271. tests/test_priority_behavior.py +212 -0
  272. tests/test_priority_consistency.py +152 -0
  273. tests/test_priority_consistency_fixed.py +250 -0
  274. tests/test_proxy_api.py +264 -264
  275. tests/test_proxy_health_check.py +32 -32
  276. tests/test_proxy_middleware.py +121 -121
  277. tests/test_proxy_middleware_enhanced.py +216 -216
  278. tests/test_proxy_middleware_integration.py +136 -136
  279. tests/test_proxy_middleware_refactored.py +184 -184
  280. tests/test_proxy_providers.py +56 -56
  281. tests/test_proxy_stats.py +19 -19
  282. tests/test_proxy_strategies.py +59 -59
  283. tests/test_queue_empty_check.py +41 -41
  284. tests/test_queue_manager_double_crawlo.py +173 -173
  285. tests/test_queue_manager_redis_key.py +179 -179
  286. tests/test_queue_naming.py +154 -154
  287. tests/test_queue_type.py +106 -106
  288. tests/test_queue_type_redis_config_consistency.py +131 -0
  289. tests/test_random_headers_default.py +323 -0
  290. tests/test_random_headers_necessity.py +309 -0
  291. tests/test_random_user_agent.py +72 -72
  292. tests/test_real_scenario_proxy.py +195 -195
  293. tests/test_redis_config.py +28 -28
  294. tests/test_redis_connection_pool.py +294 -294
  295. tests/test_redis_key_naming.py +181 -181
  296. tests/test_redis_key_validator.py +123 -123
  297. tests/test_redis_queue.py +224 -224
  298. tests/test_redis_queue_name_fix.py +175 -175
  299. tests/test_redis_queue_type_fallback.py +130 -0
  300. tests/test_request_ignore_middleware.py +182 -182
  301. tests/test_request_params.py +111 -111
  302. tests/test_request_serialization.py +70 -70
  303. tests/test_response_code_middleware.py +349 -349
  304. tests/test_response_filter_middleware.py +427 -427
  305. tests/test_response_improvements.py +152 -152
  306. tests/test_retry_middleware.py +334 -242
  307. tests/test_retry_middleware_realistic.py +274 -0
  308. tests/test_scheduler.py +252 -252
  309. tests/test_scheduler_config_update.py +133 -133
  310. tests/test_simple_response.py +61 -61
  311. tests/test_telecom_spider_redis_key.py +205 -205
  312. tests/test_template_content.py +87 -87
  313. tests/test_template_redis_key.py +134 -134
  314. tests/test_tools.py +159 -159
  315. tests/test_user_agent_randomness.py +177 -0
  316. tests/test_user_agents.py +96 -96
  317. tests/tools_example.py +260 -260
  318. tests/untested_features_report.md +138 -138
  319. tests/verify_debug.py +51 -51
  320. tests/verify_distributed.py +117 -117
  321. tests/verify_log_fix.py +111 -111
  322. crawlo-1.4.1.dist-info/METADATA +0 -1199
  323. crawlo-1.4.1.dist-info/RECORD +0 -309
  324. {crawlo-1.4.1.dist-info → crawlo-1.4.3.dist-info}/WHEEL +0 -0
  325. {crawlo-1.4.1.dist-info → crawlo-1.4.3.dist-info}/entry_points.txt +0 -0
  326. {crawlo-1.4.1.dist-info → crawlo-1.4.3.dist-info}/top_level.txt +0 -0
@@ -1,176 +1,176 @@
1
- #!/usr/bin/env python3
2
- # -*- coding: utf-8 -*-
3
- """
4
- Redis队列名称修复测试脚本
5
- 用于验证RedisPriorityQueue中队列名称处理的修复
6
- """
7
- import sys
8
- import os
9
-
10
- # 添加项目根目录到路径
11
- sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
12
-
13
- # 导入相关模块
14
- from crawlo.queue.redis_priority_queue import RedisPriorityQueue
15
-
16
-
17
- def test_normalize_queue_name():
18
- """测试队列名称规范化函数"""
19
- print("开始测试RedisPriorityQueue队列名称规范化...")
20
- print("=" * 50)
21
-
22
- # 创建一个RedisPriorityQueue实例用于测试
23
- queue = RedisPriorityQueue(redis_url="redis://127.0.0.1:6379/15")
24
-
25
- test_cases = [
26
- {
27
- "name": "已经规范化的名称",
28
- "input": "crawlo:test_project:queue:requests",
29
- "expected": "crawlo:test_project:queue:requests"
30
- },
31
- {
32
- "name": "双重 crawlo 前缀",
33
- "input": "crawlo:crawlo:queue:requests",
34
- "expected": "crawlo:queue:requests"
35
- },
36
- {
37
- "name": "三重 crawlo 前缀",
38
- "input": "crawlo:crawlo:crawlo:queue:requests",
39
- "expected": "crawlo:queue:requests"
40
- },
41
- {
42
- "name": "无 crawlo 前缀",
43
- "input": "test_project:queue:requests",
44
- "expected": "crawlo:test_project:queue:requests"
45
- },
46
- {
47
- "name": "空队列名称",
48
- "input": "",
49
- "expected": "crawlo:requests"
50
- }
51
- ]
52
-
53
- all_passed = True
54
-
55
- for i, test_case in enumerate(test_cases, 1):
56
- print(f"测试 {i}: {test_case['name']}")
57
- print(f" 输入: {test_case['input']}")
58
-
59
- # 测试规范化函数
60
- result = queue._normalize_queue_name(test_case['input'])
61
- print(f" 输出: {result}")
62
- print(f" 期望: {test_case['expected']}")
63
-
64
- # 验证结果
65
- if result == test_case['expected']:
66
- print(" ✓ 测试通过")
67
- else:
68
- print(" ✗ 测试失败")
69
- all_passed = False
70
-
71
- print()
72
-
73
- print("=" * 50)
74
- if all_passed:
75
- print("所有测试通过!队列名称规范化修复成功")
76
- return True
77
- else:
78
- print("部分测试失败,请检查实现")
79
- return False
80
-
81
-
82
- def test_queue_initialization():
83
- """测试队列初始化时的名称处理"""
84
- print("开始测试RedisPriorityQueue初始化时的名称处理...")
85
- print("=" * 50)
86
-
87
- test_cases = [
88
- {
89
- "name": "正常命名",
90
- "queue_name": "crawlo:test_project:queue:requests",
91
- "expected_queue": "crawlo:test_project:queue:requests",
92
- "expected_processing": "crawlo:test_project:queue:processing",
93
- "expected_failed": "crawlo:test_project:queue:failed"
94
- },
95
- {
96
- "name": "双重 crawlo 前缀",
97
- "queue_name": "crawlo:crawlo:queue:requests",
98
- "expected_queue": "crawlo:queue:requests",
99
- "expected_processing": "crawlo:queue:processing",
100
- "expected_failed": "crawlo:queue:failed"
101
- }
102
- ]
103
-
104
- all_passed = True
105
-
106
- for i, test_case in enumerate(test_cases, 1):
107
- print(f"测试 {i}: {test_case['name']}")
108
- print(f" 输入队列名称: {test_case['queue_name']}")
109
-
110
- try:
111
- # 创建RedisPriorityQueue实例
112
- queue = RedisPriorityQueue(
113
- redis_url="redis://127.0.0.1:6379/15",
114
- queue_name=test_case['queue_name'],
115
- module_name="test_project"
116
- )
117
-
118
- print(f" 实际队列名称: {queue.queue_name}")
119
- print(f" 实际处理队列: {queue.processing_queue}")
120
- print(f" 实际失败队列: {queue.failed_queue}")
121
-
122
- print(f" 期望队列名称: {test_case['expected_queue']}")
123
- print(f" 期望处理队列: {test_case['expected_processing']}")
124
- print(f" 期望失败队列: {test_case['expected_failed']}")
125
-
126
- # 验证结果
127
- queue_name_ok = queue.queue_name == test_case['expected_queue']
128
- processing_queue_ok = queue.processing_queue == test_case['expected_processing']
129
- failed_queue_ok = queue.failed_queue == test_case['expected_failed']
130
-
131
- if queue_name_ok and processing_queue_ok and failed_queue_ok:
132
- print(" ✓ 测试通过")
133
- else:
134
- print(" ✗ 测试失败")
135
- all_passed = False
136
-
137
- except Exception as e:
138
- print(f" ✗ 测试异常: {e}")
139
- all_passed = False
140
-
141
- print()
142
-
143
- print("=" * 50)
144
- if all_passed:
145
- print("队列初始化测试通过!")
146
- return True
147
- else:
148
- print("队列初始化测试失败!")
149
- return False
150
-
151
-
152
- def main():
153
- """主测试函数"""
154
- print("开始Redis队列名称修复测试...")
155
- print("=" * 50)
156
-
157
- # 测试队列名称规范化函数
158
- normalize_test_passed = test_normalize_queue_name()
159
- print()
160
-
161
- # 测试队列初始化
162
- init_test_passed = test_queue_initialization()
163
- print()
164
-
165
- print("=" * 50)
166
- if normalize_test_passed and init_test_passed:
167
- print("所有测试通过!Redis队列名称修复完成")
168
- return 0
169
- else:
170
- print("部分测试失败,请检查实现")
171
- return 1
172
-
173
-
174
- if __name__ == "__main__":
175
- exit_code = main()
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ Redis队列名称修复测试脚本
5
+ 用于验证RedisPriorityQueue中队列名称处理的修复
6
+ """
7
+ import sys
8
+ import os
9
+
10
+ # 添加项目根目录到路径
11
+ sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
12
+
13
+ # 导入相关模块
14
+ from crawlo.queue.redis_priority_queue import RedisPriorityQueue
15
+
16
+
17
+ def test_normalize_queue_name():
18
+ """测试队列名称规范化函数"""
19
+ print("开始测试RedisPriorityQueue队列名称规范化...")
20
+ print("=" * 50)
21
+
22
+ # 创建一个RedisPriorityQueue实例用于测试
23
+ queue = RedisPriorityQueue(redis_url="redis://127.0.0.1:6379/15")
24
+
25
+ test_cases = [
26
+ {
27
+ "name": "已经规范化的名称",
28
+ "input": "crawlo:test_project:queue:requests",
29
+ "expected": "crawlo:test_project:queue:requests"
30
+ },
31
+ {
32
+ "name": "双重 crawlo 前缀",
33
+ "input": "crawlo:crawlo:queue:requests",
34
+ "expected": "crawlo:queue:requests"
35
+ },
36
+ {
37
+ "name": "三重 crawlo 前缀",
38
+ "input": "crawlo:crawlo:crawlo:queue:requests",
39
+ "expected": "crawlo:queue:requests"
40
+ },
41
+ {
42
+ "name": "无 crawlo 前缀",
43
+ "input": "test_project:queue:requests",
44
+ "expected": "crawlo:test_project:queue:requests"
45
+ },
46
+ {
47
+ "name": "空队列名称",
48
+ "input": "",
49
+ "expected": "crawlo:requests"
50
+ }
51
+ ]
52
+
53
+ all_passed = True
54
+
55
+ for i, test_case in enumerate(test_cases, 1):
56
+ print(f"测试 {i}: {test_case['name']}")
57
+ print(f" 输入: {test_case['input']}")
58
+
59
+ # 测试规范化函数
60
+ result = queue._normalize_queue_name(test_case['input'])
61
+ print(f" 输出: {result}")
62
+ print(f" 期望: {test_case['expected']}")
63
+
64
+ # 验证结果
65
+ if result == test_case['expected']:
66
+ print(" ✓ 测试通过")
67
+ else:
68
+ print(" ✗ 测试失败")
69
+ all_passed = False
70
+
71
+ print()
72
+
73
+ print("=" * 50)
74
+ if all_passed:
75
+ print("所有测试通过!队列名称规范化修复成功")
76
+ return True
77
+ else:
78
+ print("部分测试失败,请检查实现")
79
+ return False
80
+
81
+
82
+ def test_queue_initialization():
83
+ """测试队列初始化时的名称处理"""
84
+ print("开始测试RedisPriorityQueue初始化时的名称处理...")
85
+ print("=" * 50)
86
+
87
+ test_cases = [
88
+ {
89
+ "name": "正常命名",
90
+ "queue_name": "crawlo:test_project:queue:requests",
91
+ "expected_queue": "crawlo:test_project:queue:requests",
92
+ "expected_processing": "crawlo:test_project:queue:processing",
93
+ "expected_failed": "crawlo:test_project:queue:failed"
94
+ },
95
+ {
96
+ "name": "双重 crawlo 前缀",
97
+ "queue_name": "crawlo:crawlo:queue:requests",
98
+ "expected_queue": "crawlo:queue:requests",
99
+ "expected_processing": "crawlo:queue:processing",
100
+ "expected_failed": "crawlo:queue:failed"
101
+ }
102
+ ]
103
+
104
+ all_passed = True
105
+
106
+ for i, test_case in enumerate(test_cases, 1):
107
+ print(f"测试 {i}: {test_case['name']}")
108
+ print(f" 输入队列名称: {test_case['queue_name']}")
109
+
110
+ try:
111
+ # 创建RedisPriorityQueue实例
112
+ queue = RedisPriorityQueue(
113
+ redis_url="redis://127.0.0.1:6379/15",
114
+ queue_name=test_case['queue_name'],
115
+ module_name="test_project"
116
+ )
117
+
118
+ print(f" 实际队列名称: {queue.queue_name}")
119
+ print(f" 实际处理队列: {queue.processing_queue}")
120
+ print(f" 实际失败队列: {queue.failed_queue}")
121
+
122
+ print(f" 期望队列名称: {test_case['expected_queue']}")
123
+ print(f" 期望处理队列: {test_case['expected_processing']}")
124
+ print(f" 期望失败队列: {test_case['expected_failed']}")
125
+
126
+ # 验证结果
127
+ queue_name_ok = queue.queue_name == test_case['expected_queue']
128
+ processing_queue_ok = queue.processing_queue == test_case['expected_processing']
129
+ failed_queue_ok = queue.failed_queue == test_case['expected_failed']
130
+
131
+ if queue_name_ok and processing_queue_ok and failed_queue_ok:
132
+ print(" ✓ 测试通过")
133
+ else:
134
+ print(" ✗ 测试失败")
135
+ all_passed = False
136
+
137
+ except Exception as e:
138
+ print(f" ✗ 测试异常: {e}")
139
+ all_passed = False
140
+
141
+ print()
142
+
143
+ print("=" * 50)
144
+ if all_passed:
145
+ print("队列初始化测试通过!")
146
+ return True
147
+ else:
148
+ print("队列初始化测试失败!")
149
+ return False
150
+
151
+
152
+ def main():
153
+ """主测试函数"""
154
+ print("开始Redis队列名称修复测试...")
155
+ print("=" * 50)
156
+
157
+ # 测试队列名称规范化函数
158
+ normalize_test_passed = test_normalize_queue_name()
159
+ print()
160
+
161
+ # 测试队列初始化
162
+ init_test_passed = test_queue_initialization()
163
+ print()
164
+
165
+ print("=" * 50)
166
+ if normalize_test_passed and init_test_passed:
167
+ print("所有测试通过!Redis队列名称修复完成")
168
+ return 0
169
+ else:
170
+ print("部分测试失败,请检查实现")
171
+ return 1
172
+
173
+
174
+ if __name__ == "__main__":
175
+ exit_code = main()
176
176
  sys.exit(exit_code)
@@ -0,0 +1,130 @@
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ 测试 QUEUE_TYPE = 'redis' 时的行为,验证其等同于 'auto' 模式
5
+ 即:当 Redis 不可用时应该回退到内存队列
6
+ """
7
+
8
+ import sys
9
+ import os
10
+ import asyncio
11
+
12
+ # 添加项目根目录到路径
13
+ sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
14
+
15
+ from crawlo.config import CrawloConfig
16
+ from crawlo.queue.queue_manager import QueueManager, QueueConfig, QueueType
17
+
18
+
19
+ async def test_redis_queue_type_fallback():
20
+ """测试 QUEUE_TYPE = 'redis' 时的回退行为"""
21
+ print("=== 测试 QUEUE_TYPE = 'redis' 时的回退行为 ===")
22
+
23
+ # 创建一个 Redis 不可用的配置(使用一个不存在的 Redis 地址)
24
+ config = QueueConfig(
25
+ queue_type=QueueType.REDIS,
26
+ redis_url="redis://127.0.0.1:6380/9", # 一个不存在的 Redis 实例
27
+ queue_name="test:queue:requests"
28
+ )
29
+
30
+ # 创建队列管理器
31
+ queue_manager = QueueManager(config)
32
+
33
+ # 初始化队列
34
+ print("正在初始化队列管理器...")
35
+ try:
36
+ needs_config_update = await queue_manager.initialize()
37
+ print(f"初始化完成,needs_config_update: {needs_config_update}")
38
+
39
+ # 获取队列状态
40
+ status = queue_manager.get_status()
41
+ print(f"队列类型: {status['type']}")
42
+ print(f"队列健康状态: {status['health']}")
43
+
44
+ # 验证队列类型应该是 memory(因为 Redis 不可用,应该回退)
45
+ assert status['type'] == 'memory', f"期望队列类型为 'memory',实际得到 '{status['type']}'"
46
+ print("✅ 队列类型正确回退到 memory")
47
+
48
+ # 验证健康状态应该是 healthy
49
+ assert status['health'] == 'healthy', f"期望健康状态为 'healthy',实际得到 '{status['health']}'"
50
+ print("✅ 队列健康状态正常")
51
+
52
+ except Exception as e:
53
+ print(f"初始化队列时发生错误: {e}")
54
+ import traceback
55
+ traceback.print_exc()
56
+ raise
57
+
58
+ finally:
59
+ # 清理资源
60
+ if queue_manager:
61
+ try:
62
+ await queue_manager.close()
63
+ except:
64
+ pass
65
+
66
+ print("✅ Redis 队列类型回退测试通过")
67
+
68
+
69
+ async def test_redis_queue_type_with_valid_redis():
70
+ """测试 QUEUE_TYPE = 'redis' 时,当 Redis 可用的情况"""
71
+ print("\n=== 测试 QUEUE_TYPE = 'redis' 时 Redis 可用的情况 ===")
72
+
73
+ # 创建一个 Redis 可用的配置(使用默认的本地 Redis)
74
+ config = QueueConfig(
75
+ queue_type=QueueType.REDIS,
76
+ redis_url="redis://127.0.0.1:6379/2", # 默认的本地 Redis 实例
77
+ queue_name="test:queue:requests"
78
+ )
79
+
80
+ # 创建队列管理器
81
+ queue_manager = QueueManager(config)
82
+
83
+ # 初始化队列
84
+ print("正在初始化队列管理器...")
85
+ try:
86
+ needs_config_update = await queue_manager.initialize()
87
+ print(f"初始化完成,needs_config_update: {needs_config_update}")
88
+
89
+ # 获取队列状态
90
+ status = queue_manager.get_status()
91
+ print(f"队列类型: {status['type']}")
92
+ print(f"队列健康状态: {status['health']}")
93
+
94
+ # 验证队列类型应该是 redis(因为 Redis 可用)
95
+ # 注意:这取决于本地是否真的有 Redis 服务运行
96
+ print(f"队列类型: {status['type']} (期望为 'redis' 如果本地 Redis 可用)")
97
+
98
+ # 验证健康状态应该是 healthy
99
+ assert status['health'] == 'healthy', f"期望健康状态为 'healthy',实际得到 '{status['health']}'"
100
+ print("✅ 队列健康状态正常")
101
+
102
+ except Exception as e:
103
+ print(f"初始化队列时发生错误: {e}")
104
+ print("这可能是因为本地没有运行 Redis 服务,这是正常的")
105
+
106
+ finally:
107
+ # 清理资源
108
+ if queue_manager:
109
+ try:
110
+ await queue_manager.close()
111
+ except:
112
+ pass
113
+
114
+ print("✅ Redis 队列类型可用性测试完成")
115
+
116
+
117
+ if __name__ == "__main__":
118
+ print("开始测试 QUEUE_TYPE = 'redis' 的行为...")
119
+
120
+ try:
121
+ # 运行异步测试
122
+ asyncio.run(test_redis_queue_type_fallback())
123
+ asyncio.run(test_redis_queue_type_with_valid_redis())
124
+
125
+ print("\n🎉 所有测试通过!QUEUE_TYPE = 'redis' 的行为等同于 'auto' 模式。")
126
+
127
+ except Exception as e:
128
+ print(f"\n❌ 测试失败: {e}")
129
+ import traceback
130
+ traceback.print_exc()