crawlo 1.3.5__py3-none-any.whl → 1.3.6__py3-none-any.whl

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

Potentially problematic release.


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

Files changed (289) hide show
  1. crawlo/__init__.py +87 -87
  2. crawlo/__version__.py +1 -1
  3. crawlo/cli.py +75 -75
  4. crawlo/commands/__init__.py +14 -14
  5. crawlo/commands/check.py +594 -594
  6. crawlo/commands/genspider.py +151 -151
  7. crawlo/commands/help.py +138 -138
  8. crawlo/commands/list.py +155 -155
  9. crawlo/commands/run.py +341 -341
  10. crawlo/commands/startproject.py +436 -436
  11. crawlo/commands/stats.py +187 -187
  12. crawlo/commands/utils.py +196 -196
  13. crawlo/config.py +312 -312
  14. crawlo/config_validator.py +277 -277
  15. crawlo/core/__init__.py +45 -45
  16. crawlo/core/engine.py +439 -439
  17. crawlo/core/processor.py +40 -40
  18. crawlo/core/scheduler.py +257 -257
  19. crawlo/crawler.py +638 -638
  20. crawlo/data/__init__.py +5 -5
  21. crawlo/data/user_agents.py +194 -194
  22. crawlo/downloader/__init__.py +273 -273
  23. crawlo/downloader/aiohttp_downloader.py +228 -228
  24. crawlo/downloader/cffi_downloader.py +245 -245
  25. crawlo/downloader/httpx_downloader.py +259 -259
  26. crawlo/downloader/hybrid_downloader.py +212 -212
  27. crawlo/downloader/playwright_downloader.py +402 -402
  28. crawlo/downloader/selenium_downloader.py +472 -472
  29. crawlo/event.py +11 -11
  30. crawlo/exceptions.py +81 -81
  31. crawlo/extension/__init__.py +39 -39
  32. crawlo/extension/health_check.py +141 -141
  33. crawlo/extension/log_interval.py +57 -57
  34. crawlo/extension/log_stats.py +81 -81
  35. crawlo/extension/logging_extension.py +61 -61
  36. crawlo/extension/memory_monitor.py +104 -104
  37. crawlo/extension/performance_profiler.py +133 -133
  38. crawlo/extension/request_recorder.py +107 -107
  39. crawlo/factories/__init__.py +27 -27
  40. crawlo/factories/base.py +68 -68
  41. crawlo/factories/crawler.py +103 -103
  42. crawlo/factories/registry.py +84 -84
  43. crawlo/filters/__init__.py +154 -154
  44. crawlo/filters/aioredis_filter.py +257 -257
  45. crawlo/filters/memory_filter.py +269 -269
  46. crawlo/framework.py +292 -291
  47. crawlo/initialization/__init__.py +39 -39
  48. crawlo/initialization/built_in.py +425 -425
  49. crawlo/initialization/context.py +141 -141
  50. crawlo/initialization/core.py +193 -193
  51. crawlo/initialization/phases.py +148 -148
  52. crawlo/initialization/registry.py +145 -145
  53. crawlo/items/__init__.py +23 -23
  54. crawlo/items/base.py +23 -23
  55. crawlo/items/fields.py +52 -52
  56. crawlo/items/items.py +104 -104
  57. crawlo/logging/__init__.py +37 -37
  58. crawlo/logging/config.py +96 -96
  59. crawlo/logging/factory.py +128 -128
  60. crawlo/logging/manager.py +111 -111
  61. crawlo/middleware/__init__.py +21 -21
  62. crawlo/middleware/default_header.py +132 -132
  63. crawlo/middleware/download_delay.py +104 -104
  64. crawlo/middleware/middleware_manager.py +135 -135
  65. crawlo/middleware/offsite.py +123 -123
  66. crawlo/middleware/proxy.py +386 -386
  67. crawlo/middleware/request_ignore.py +86 -86
  68. crawlo/middleware/response_code.py +163 -163
  69. crawlo/middleware/response_filter.py +136 -136
  70. crawlo/middleware/retry.py +124 -124
  71. crawlo/middleware/simple_proxy.py +65 -65
  72. crawlo/mode_manager.py +212 -212
  73. crawlo/network/__init__.py +21 -21
  74. crawlo/network/request.py +379 -379
  75. crawlo/network/response.py +359 -359
  76. crawlo/pipelines/__init__.py +21 -21
  77. crawlo/pipelines/bloom_dedup_pipeline.py +156 -156
  78. crawlo/pipelines/console_pipeline.py +39 -39
  79. crawlo/pipelines/csv_pipeline.py +316 -316
  80. crawlo/pipelines/database_dedup_pipeline.py +222 -222
  81. crawlo/pipelines/json_pipeline.py +218 -218
  82. crawlo/pipelines/memory_dedup_pipeline.py +115 -115
  83. crawlo/pipelines/mongo_pipeline.py +131 -131
  84. crawlo/pipelines/mysql_pipeline.py +325 -325
  85. crawlo/pipelines/pipeline_manager.py +76 -76
  86. crawlo/pipelines/redis_dedup_pipeline.py +166 -166
  87. crawlo/project.py +327 -327
  88. crawlo/queue/pqueue.py +42 -42
  89. crawlo/queue/queue_manager.py +503 -503
  90. crawlo/queue/redis_priority_queue.py +326 -326
  91. crawlo/settings/__init__.py +7 -7
  92. crawlo/settings/default_settings.py +321 -321
  93. crawlo/settings/setting_manager.py +214 -214
  94. crawlo/spider/__init__.py +657 -657
  95. crawlo/stats_collector.py +73 -73
  96. crawlo/subscriber.py +129 -129
  97. crawlo/task_manager.py +138 -138
  98. crawlo/templates/crawlo.cfg.tmpl +10 -10
  99. crawlo/templates/project/__init__.py.tmpl +3 -3
  100. crawlo/templates/project/items.py.tmpl +17 -17
  101. crawlo/templates/project/middlewares.py.tmpl +118 -118
  102. crawlo/templates/project/pipelines.py.tmpl +96 -96
  103. crawlo/templates/project/settings.py.tmpl +167 -167
  104. crawlo/templates/project/settings_distributed.py.tmpl +166 -166
  105. crawlo/templates/project/settings_gentle.py.tmpl +166 -166
  106. crawlo/templates/project/settings_high_performance.py.tmpl +167 -167
  107. crawlo/templates/project/settings_minimal.py.tmpl +65 -65
  108. crawlo/templates/project/settings_simple.py.tmpl +164 -164
  109. crawlo/templates/project/spiders/__init__.py.tmpl +9 -9
  110. crawlo/templates/run.py.tmpl +34 -34
  111. crawlo/templates/spider/spider.py.tmpl +143 -143
  112. crawlo/templates/spiders_init.py.tmpl +9 -9
  113. crawlo/tools/__init__.py +200 -200
  114. crawlo/tools/anti_crawler.py +268 -268
  115. crawlo/tools/authenticated_proxy.py +240 -240
  116. crawlo/tools/data_formatter.py +225 -225
  117. crawlo/tools/data_validator.py +180 -180
  118. crawlo/tools/date_tools.py +289 -289
  119. crawlo/tools/distributed_coordinator.py +388 -388
  120. crawlo/tools/encoding_converter.py +127 -127
  121. crawlo/tools/network_diagnostic.py +364 -364
  122. crawlo/tools/request_tools.py +82 -82
  123. crawlo/tools/retry_mechanism.py +224 -224
  124. crawlo/tools/scenario_adapter.py +262 -262
  125. crawlo/tools/text_cleaner.py +232 -232
  126. crawlo/utils/__init__.py +34 -34
  127. crawlo/utils/batch_processor.py +259 -259
  128. crawlo/utils/class_loader.py +25 -25
  129. crawlo/utils/controlled_spider_mixin.py +439 -439
  130. crawlo/utils/db_helper.py +343 -343
  131. crawlo/utils/enhanced_error_handler.py +356 -356
  132. crawlo/utils/env_config.py +142 -142
  133. crawlo/utils/error_handler.py +165 -165
  134. crawlo/utils/func_tools.py +82 -82
  135. crawlo/utils/large_scale_config.py +286 -286
  136. crawlo/utils/large_scale_helper.py +344 -344
  137. crawlo/utils/log.py +79 -79
  138. crawlo/utils/performance_monitor.py +285 -285
  139. crawlo/utils/queue_helper.py +175 -175
  140. crawlo/utils/redis_connection_pool.py +388 -388
  141. crawlo/utils/redis_key_validator.py +198 -198
  142. crawlo/utils/request.py +267 -267
  143. crawlo/utils/request_serializer.py +225 -225
  144. crawlo/utils/spider_loader.py +61 -61
  145. crawlo/utils/system.py +11 -11
  146. crawlo/utils/tools.py +4 -4
  147. crawlo/utils/url.py +39 -39
  148. {crawlo-1.3.5.dist-info → crawlo-1.3.6.dist-info}/METADATA +1126 -1126
  149. crawlo-1.3.6.dist-info/RECORD +290 -0
  150. examples/__init__.py +7 -7
  151. tests/__init__.py +7 -7
  152. tests/advanced_tools_example.py +275 -275
  153. tests/authenticated_proxy_example.py +106 -106
  154. tests/baidu_performance_test.py +108 -108
  155. tests/baidu_test.py +59 -59
  156. tests/cleaners_example.py +160 -160
  157. tests/comprehensive_framework_test.py +212 -212
  158. tests/comprehensive_test.py +81 -81
  159. tests/comprehensive_testing_summary.md +186 -186
  160. tests/config_validation_demo.py +142 -142
  161. tests/controlled_spider_example.py +205 -205
  162. tests/date_tools_example.py +180 -180
  163. tests/debug_configure.py +69 -69
  164. tests/debug_framework_logger.py +84 -84
  165. tests/debug_log_config.py +126 -126
  166. tests/debug_log_levels.py +63 -63
  167. tests/debug_pipelines.py +66 -66
  168. tests/detailed_log_test.py +233 -233
  169. tests/distributed_test.py +66 -66
  170. tests/distributed_test_debug.py +76 -76
  171. tests/dynamic_loading_example.py +523 -523
  172. tests/dynamic_loading_test.py +104 -104
  173. tests/env_config_example.py +133 -133
  174. tests/error_handling_example.py +171 -171
  175. tests/final_comprehensive_test.py +151 -151
  176. tests/final_log_test.py +260 -260
  177. tests/final_validation_test.py +182 -182
  178. tests/fix_log_test.py +142 -142
  179. tests/framework_performance_test.py +202 -202
  180. tests/log_buffering_test.py +111 -111
  181. tests/log_generation_timing_test.py +153 -153
  182. tests/optimized_performance_test.py +211 -211
  183. tests/performance_comparison.py +245 -245
  184. tests/queue_blocking_test.py +113 -113
  185. tests/queue_test.py +89 -89
  186. tests/redis_key_validation_demo.py +130 -130
  187. tests/request_params_example.py +150 -150
  188. tests/response_improvements_example.py +144 -144
  189. tests/scrapy_comparison/ofweek_scrapy.py +138 -138
  190. tests/scrapy_comparison/scrapy_test.py +133 -133
  191. tests/simple_command_test.py +119 -119
  192. tests/simple_crawlo_test.py +127 -127
  193. tests/simple_log_test.py +57 -57
  194. tests/simple_log_test2.py +137 -137
  195. tests/simple_optimization_test.py +128 -128
  196. tests/simple_queue_type_test.py +42 -0
  197. tests/simple_spider_test.py +49 -49
  198. tests/simple_test.py +47 -47
  199. tests/spider_log_timing_test.py +177 -177
  200. tests/test_advanced_tools.py +148 -148
  201. tests/test_all_commands.py +230 -230
  202. tests/test_all_redis_key_configs.py +145 -145
  203. tests/test_authenticated_proxy.py +141 -141
  204. tests/test_batch_processor.py +178 -178
  205. tests/test_cleaners.py +54 -54
  206. tests/test_component_factory.py +174 -174
  207. tests/test_comprehensive.py +146 -146
  208. tests/test_config_consistency.py +80 -80
  209. tests/test_config_merge.py +152 -152
  210. tests/test_config_validator.py +182 -182
  211. tests/test_controlled_spider_mixin.py +79 -79
  212. tests/test_crawlo_proxy_integration.py +108 -108
  213. tests/test_date_tools.py +123 -123
  214. tests/test_default_header_middleware.py +158 -158
  215. tests/test_distributed.py +65 -65
  216. tests/test_double_crawlo_fix.py +207 -207
  217. tests/test_double_crawlo_fix_simple.py +124 -124
  218. tests/test_download_delay_middleware.py +221 -221
  219. tests/test_downloader_proxy_compatibility.py +268 -268
  220. tests/test_dynamic_downloaders_proxy.py +124 -124
  221. tests/test_dynamic_proxy.py +92 -92
  222. tests/test_dynamic_proxy_config.py +146 -146
  223. tests/test_dynamic_proxy_real.py +109 -109
  224. tests/test_edge_cases.py +303 -303
  225. tests/test_enhanced_error_handler.py +270 -270
  226. tests/test_enhanced_error_handler_comprehensive.py +245 -245
  227. tests/test_env_config.py +121 -121
  228. tests/test_error_handler_compatibility.py +112 -112
  229. tests/test_factories.py +252 -252
  230. tests/test_final_validation.py +153 -153
  231. tests/test_framework_env_usage.py +103 -103
  232. tests/test_framework_logger.py +66 -66
  233. tests/test_framework_startup.py +64 -64
  234. tests/test_get_component_logger.py +83 -83
  235. tests/test_integration.py +169 -169
  236. tests/test_item_dedup_redis_key.py +122 -122
  237. tests/test_large_scale_config.py +112 -112
  238. tests/test_large_scale_helper.py +235 -235
  239. tests/test_logging_system.py +282 -282
  240. tests/test_mode_change.py +72 -72
  241. tests/test_mode_consistency.py +51 -51
  242. tests/test_offsite_middleware.py +221 -221
  243. tests/test_parsel.py +29 -29
  244. tests/test_performance.py +327 -327
  245. tests/test_performance_monitor.py +115 -115
  246. tests/test_proxy_api.py +264 -264
  247. tests/test_proxy_health_check.py +32 -32
  248. tests/test_proxy_middleware.py +121 -121
  249. tests/test_proxy_middleware_enhanced.py +216 -216
  250. tests/test_proxy_middleware_integration.py +136 -136
  251. tests/test_proxy_middleware_refactored.py +184 -184
  252. tests/test_proxy_providers.py +56 -56
  253. tests/test_proxy_stats.py +19 -19
  254. tests/test_proxy_strategies.py +59 -59
  255. tests/test_queue_empty_check.py +41 -41
  256. tests/test_queue_manager_double_crawlo.py +173 -173
  257. tests/test_queue_manager_redis_key.py +176 -176
  258. tests/test_queue_type.py +107 -0
  259. tests/test_random_user_agent.py +72 -72
  260. tests/test_real_scenario_proxy.py +195 -195
  261. tests/test_redis_config.py +28 -28
  262. tests/test_redis_connection_pool.py +294 -294
  263. tests/test_redis_key_naming.py +181 -181
  264. tests/test_redis_key_validator.py +123 -123
  265. tests/test_redis_queue.py +224 -224
  266. tests/test_request_ignore_middleware.py +182 -182
  267. tests/test_request_params.py +111 -111
  268. tests/test_request_serialization.py +70 -70
  269. tests/test_response_code_middleware.py +349 -349
  270. tests/test_response_filter_middleware.py +427 -427
  271. tests/test_response_improvements.py +152 -152
  272. tests/test_retry_middleware.py +241 -241
  273. tests/test_scheduler.py +252 -252
  274. tests/test_scheduler_config_update.py +133 -133
  275. tests/test_simple_response.py +61 -61
  276. tests/test_telecom_spider_redis_key.py +205 -205
  277. tests/test_template_content.py +87 -87
  278. tests/test_template_redis_key.py +134 -134
  279. tests/test_tools.py +159 -159
  280. tests/test_user_agents.py +96 -96
  281. tests/tools_example.py +260 -260
  282. tests/untested_features_report.md +138 -138
  283. tests/verify_debug.py +51 -51
  284. tests/verify_distributed.py +117 -117
  285. tests/verify_log_fix.py +111 -111
  286. crawlo-1.3.5.dist-info/RECORD +0 -288
  287. {crawlo-1.3.5.dist-info → crawlo-1.3.6.dist-info}/WHEEL +0 -0
  288. {crawlo-1.3.5.dist-info → crawlo-1.3.6.dist-info}/entry_points.txt +0 -0
  289. {crawlo-1.3.5.dist-info → crawlo-1.3.6.dist-info}/top_level.txt +0 -0
@@ -1,208 +1,208 @@
1
- #!/usr/bin/env python3
2
- # -*- coding: utf-8 -*-
3
- """
4
- 双重 crawlo 前缀问题修复测试脚本
5
- 用于验证 Redis 队列名称中双重 crawlo 前缀问题的修复
6
- """
7
- import sys
8
- import os
9
- import asyncio
10
- import traceback
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
- # 导入相关模块
16
- from crawlo.queue.queue_manager import QueueManager, QueueConfig, QueueType
17
- from crawlo.queue.redis_priority_queue import RedisPriorityQueue
18
-
19
-
20
- async def test_redis_queue_naming():
21
- """测试 Redis 队列命名修复"""
22
- print("开始测试 Redis 队列命名修复...")
23
- print("=" * 50)
24
-
25
- test_cases = [
26
- {
27
- "name": "正常命名",
28
- "queue_name": "crawlo:test_project:queue:requests",
29
- "expected_module": "test_project",
30
- "expected_queue": "crawlo:test_project:queue:requests",
31
- "expected_processing": "crawlo:test_project:queue:processing",
32
- "expected_failed": "crawlo:test_project:queue:failed"
33
- },
34
- {
35
- "name": "双重 crawlo 前缀",
36
- "queue_name": "crawlo:crawlo:queue:requests",
37
- "expected_module": "crawlo",
38
- "expected_queue": "crawlo:crawlo:queue:requests",
39
- "expected_processing": "crawlo:crawlo:queue:processing",
40
- "expected_failed": "crawlo:crawlo:queue:failed"
41
- },
42
- {
43
- "name": "三重 crawlo 前缀",
44
- "queue_name": "crawlo:crawlo:crawlo:queue:requests",
45
- "expected_module": "crawlo",
46
- "expected_queue": "crawlo:crawlo:queue:requests",
47
- "expected_processing": "crawlo:crawlo:queue:processing",
48
- "expected_failed": "crawlo:crawlo:queue:failed"
49
- },
50
- {
51
- "name": "无 crawlo 前缀",
52
- "queue_name": "test_project:queue:requests",
53
- "expected_module": "test_project",
54
- "expected_queue": "crawlo:test_project:queue:requests",
55
- "expected_processing": "crawlo:test_project:queue:processing",
56
- "expected_failed": "crawlo:test_project:queue:failed"
57
- }
58
- ]
59
-
60
- try:
61
- for i, test_case in enumerate(test_cases, 1):
62
- print(f"测试 {i}: {test_case['name']}")
63
- print(f" 输入队列名称: {test_case['queue_name']}")
64
-
65
- # 测试 RedisPriorityQueue 直接初始化
66
- try:
67
- queue = RedisPriorityQueue(
68
- redis_url="redis://127.0.0.1:6379/15",
69
- queue_name=test_case['queue_name'],
70
- module_name=test_case['expected_module']
71
- )
72
-
73
- print(f" 修复后队列名称: {queue.queue_name}")
74
- print(f" 修复后处理队列: {queue.processing_queue}")
75
- print(f" 修复后失败队列: {queue.failed_queue}")
76
-
77
- # 验证结果
78
- assert queue.queue_name == test_case['expected_queue'], \
79
- f"队列名称不匹配: {queue.queue_name} != {test_case['expected_queue']}"
80
- assert queue.processing_queue == test_case['expected_processing'], \
81
- f"处理队列名称不匹配: {queue.processing_queue} != {test_case['expected_processing']}"
82
- assert queue.failed_queue == test_case['expected_failed'], \
83
- f"失败队列名称不匹配: {queue.failed_queue} != {test_case['expected_failed']}"
84
-
85
- print(" 测试通过")
86
- except Exception as e:
87
- print(f" 测试失败: {e}")
88
- traceback.print_exc()
89
- return False
90
-
91
- print()
92
-
93
- print("Redis 队列命名修复测试通过!")
94
- return True
95
-
96
- except Exception as e:
97
- print(f"Redis 队列命名修复测试失败: {e}")
98
- traceback.print_exc()
99
- return False
100
-
101
-
102
- async def test_queue_manager_naming():
103
- """测试队列管理器中的项目名称提取"""
104
- print("开始测试队列管理器项目名称提取...")
105
- print("=" * 50)
106
-
107
- test_cases = [
108
- {
109
- "name": "正常命名",
110
- "queue_name": "crawlo:test_project:queue:requests",
111
- "expected_module": "test_project"
112
- },
113
- {
114
- "name": "双重 crawlo 前缀",
115
- "queue_name": "crawlo:crawlo:queue:requests",
116
- "expected_module": "crawlo"
117
- },
118
- {
119
- "name": "三重 crawlo 前缀",
120
- "queue_name": "crawlo:crawlo:crawlo:queue:requests",
121
- "expected_module": "crawlo"
122
- },
123
- {
124
- "name": "无 crawlo 前缀",
125
- "queue_name": "test_project:queue:requests",
126
- "expected_module": "test_project"
127
- }
128
- ]
129
-
130
- try:
131
- for i, test_case in enumerate(test_cases, 1):
132
- print(f"测试 {i}: {test_case['name']}")
133
- print(f" 输入队列名称: {test_case['queue_name']}")
134
-
135
- # 测试队列管理器中的项目名称提取逻辑
136
- if ':' in test_case['queue_name']:
137
- parts = test_case['queue_name'].split(':')
138
- if len(parts) >= 2:
139
- # 处理可能的双重 crawlo 前缀
140
- if parts[0] == "crawlo" and parts[1] == "crawlo":
141
- # 双重 crawlo 前缀,取第三个部分作为项目名称
142
- if len(parts) >= 3:
143
- project_name = parts[2]
144
- else:
145
- project_name = "default"
146
- elif parts[0] == "crawlo":
147
- # 正常的 crawlo 前缀,取第二个部分作为项目名称
148
- project_name = parts[1]
149
- else:
150
- # 没有 crawlo 前缀,使用第一个部分作为项目名称
151
- project_name = parts[0]
152
- else:
153
- project_name = test_case['queue_name'] or "default"
154
- else:
155
- project_name = test_case['queue_name'] or "default"
156
-
157
- print(f" 提取的项目名称: {project_name}")
158
- print(f" 期望的项目名称: {test_case['expected_module']}")
159
-
160
- # 验证结果
161
- assert project_name == test_case['expected_module'], \
162
- f"项目名称不匹配: {project_name} != {test_case['expected_module']}"
163
-
164
- print(" 测试通过")
165
- print()
166
-
167
- print("队列管理器项目名称提取测试通过!")
168
- return True
169
-
170
- except Exception as e:
171
- print(f"队列管理器项目名称提取测试失败: {e}")
172
- traceback.print_exc()
173
- return False
174
-
175
-
176
- async def main():
177
- """主测试函数"""
178
- print("开始双重 crawlo 前缀问题修复测试...")
179
- print("=" * 50)
180
-
181
- try:
182
- # 测试 Redis 队列命名修复
183
- redis_test_success = await test_redis_queue_naming()
184
- print()
185
-
186
- # 测试队列管理器项目名称提取
187
- manager_test_success = await test_queue_manager_naming()
188
- print()
189
-
190
- print("=" * 50)
191
- if redis_test_success and manager_test_success:
192
- print("所有测试通过!双重 crawlo 前缀问题已修复")
193
- else:
194
- print("部分测试失败,请检查实现")
195
- return 1
196
-
197
- except Exception as e:
198
- print("=" * 50)
199
- print(f"测试过程中发生异常: {e}")
200
- traceback.print_exc()
201
- return 1
202
-
203
- return 0
204
-
205
-
206
- if __name__ == "__main__":
207
- exit_code = asyncio.run(main())
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ 双重 crawlo 前缀问题修复测试脚本
5
+ 用于验证 Redis 队列名称中双重 crawlo 前缀问题的修复
6
+ """
7
+ import sys
8
+ import os
9
+ import asyncio
10
+ import traceback
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
+ # 导入相关模块
16
+ from crawlo.queue.queue_manager import QueueManager, QueueConfig, QueueType
17
+ from crawlo.queue.redis_priority_queue import RedisPriorityQueue
18
+
19
+
20
+ async def test_redis_queue_naming():
21
+ """测试 Redis 队列命名修复"""
22
+ print("开始测试 Redis 队列命名修复...")
23
+ print("=" * 50)
24
+
25
+ test_cases = [
26
+ {
27
+ "name": "正常命名",
28
+ "queue_name": "crawlo:test_project:queue:requests",
29
+ "expected_module": "test_project",
30
+ "expected_queue": "crawlo:test_project:queue:requests",
31
+ "expected_processing": "crawlo:test_project:queue:processing",
32
+ "expected_failed": "crawlo:test_project:queue:failed"
33
+ },
34
+ {
35
+ "name": "双重 crawlo 前缀",
36
+ "queue_name": "crawlo:crawlo:queue:requests",
37
+ "expected_module": "crawlo",
38
+ "expected_queue": "crawlo:crawlo:queue:requests",
39
+ "expected_processing": "crawlo:crawlo:queue:processing",
40
+ "expected_failed": "crawlo:crawlo:queue:failed"
41
+ },
42
+ {
43
+ "name": "三重 crawlo 前缀",
44
+ "queue_name": "crawlo:crawlo:crawlo:queue:requests",
45
+ "expected_module": "crawlo",
46
+ "expected_queue": "crawlo:crawlo:queue:requests",
47
+ "expected_processing": "crawlo:crawlo:queue:processing",
48
+ "expected_failed": "crawlo:crawlo:queue:failed"
49
+ },
50
+ {
51
+ "name": "无 crawlo 前缀",
52
+ "queue_name": "test_project:queue:requests",
53
+ "expected_module": "test_project",
54
+ "expected_queue": "crawlo:test_project:queue:requests",
55
+ "expected_processing": "crawlo:test_project:queue:processing",
56
+ "expected_failed": "crawlo:test_project:queue:failed"
57
+ }
58
+ ]
59
+
60
+ try:
61
+ for i, test_case in enumerate(test_cases, 1):
62
+ print(f"测试 {i}: {test_case['name']}")
63
+ print(f" 输入队列名称: {test_case['queue_name']}")
64
+
65
+ # 测试 RedisPriorityQueue 直接初始化
66
+ try:
67
+ queue = RedisPriorityQueue(
68
+ redis_url="redis://127.0.0.1:6379/15",
69
+ queue_name=test_case['queue_name'],
70
+ module_name=test_case['expected_module']
71
+ )
72
+
73
+ print(f" 修复后队列名称: {queue.queue_name}")
74
+ print(f" 修复后处理队列: {queue.processing_queue}")
75
+ print(f" 修复后失败队列: {queue.failed_queue}")
76
+
77
+ # 验证结果
78
+ assert queue.queue_name == test_case['expected_queue'], \
79
+ f"队列名称不匹配: {queue.queue_name} != {test_case['expected_queue']}"
80
+ assert queue.processing_queue == test_case['expected_processing'], \
81
+ f"处理队列名称不匹配: {queue.processing_queue} != {test_case['expected_processing']}"
82
+ assert queue.failed_queue == test_case['expected_failed'], \
83
+ f"失败队列名称不匹配: {queue.failed_queue} != {test_case['expected_failed']}"
84
+
85
+ print(" 测试通过")
86
+ except Exception as e:
87
+ print(f" 测试失败: {e}")
88
+ traceback.print_exc()
89
+ return False
90
+
91
+ print()
92
+
93
+ print("Redis 队列命名修复测试通过!")
94
+ return True
95
+
96
+ except Exception as e:
97
+ print(f"Redis 队列命名修复测试失败: {e}")
98
+ traceback.print_exc()
99
+ return False
100
+
101
+
102
+ async def test_queue_manager_naming():
103
+ """测试队列管理器中的项目名称提取"""
104
+ print("开始测试队列管理器项目名称提取...")
105
+ print("=" * 50)
106
+
107
+ test_cases = [
108
+ {
109
+ "name": "正常命名",
110
+ "queue_name": "crawlo:test_project:queue:requests",
111
+ "expected_module": "test_project"
112
+ },
113
+ {
114
+ "name": "双重 crawlo 前缀",
115
+ "queue_name": "crawlo:crawlo:queue:requests",
116
+ "expected_module": "crawlo"
117
+ },
118
+ {
119
+ "name": "三重 crawlo 前缀",
120
+ "queue_name": "crawlo:crawlo:crawlo:queue:requests",
121
+ "expected_module": "crawlo"
122
+ },
123
+ {
124
+ "name": "无 crawlo 前缀",
125
+ "queue_name": "test_project:queue:requests",
126
+ "expected_module": "test_project"
127
+ }
128
+ ]
129
+
130
+ try:
131
+ for i, test_case in enumerate(test_cases, 1):
132
+ print(f"测试 {i}: {test_case['name']}")
133
+ print(f" 输入队列名称: {test_case['queue_name']}")
134
+
135
+ # 测试队列管理器中的项目名称提取逻辑
136
+ if ':' in test_case['queue_name']:
137
+ parts = test_case['queue_name'].split(':')
138
+ if len(parts) >= 2:
139
+ # 处理可能的双重 crawlo 前缀
140
+ if parts[0] == "crawlo" and parts[1] == "crawlo":
141
+ # 双重 crawlo 前缀,取第三个部分作为项目名称
142
+ if len(parts) >= 3:
143
+ project_name = parts[2]
144
+ else:
145
+ project_name = "default"
146
+ elif parts[0] == "crawlo":
147
+ # 正常的 crawlo 前缀,取第二个部分作为项目名称
148
+ project_name = parts[1]
149
+ else:
150
+ # 没有 crawlo 前缀,使用第一个部分作为项目名称
151
+ project_name = parts[0]
152
+ else:
153
+ project_name = test_case['queue_name'] or "default"
154
+ else:
155
+ project_name = test_case['queue_name'] or "default"
156
+
157
+ print(f" 提取的项目名称: {project_name}")
158
+ print(f" 期望的项目名称: {test_case['expected_module']}")
159
+
160
+ # 验证结果
161
+ assert project_name == test_case['expected_module'], \
162
+ f"项目名称不匹配: {project_name} != {test_case['expected_module']}"
163
+
164
+ print(" 测试通过")
165
+ print()
166
+
167
+ print("队列管理器项目名称提取测试通过!")
168
+ return True
169
+
170
+ except Exception as e:
171
+ print(f"队列管理器项目名称提取测试失败: {e}")
172
+ traceback.print_exc()
173
+ return False
174
+
175
+
176
+ async def main():
177
+ """主测试函数"""
178
+ print("开始双重 crawlo 前缀问题修复测试...")
179
+ print("=" * 50)
180
+
181
+ try:
182
+ # 测试 Redis 队列命名修复
183
+ redis_test_success = await test_redis_queue_naming()
184
+ print()
185
+
186
+ # 测试队列管理器项目名称提取
187
+ manager_test_success = await test_queue_manager_naming()
188
+ print()
189
+
190
+ print("=" * 50)
191
+ if redis_test_success and manager_test_success:
192
+ print("所有测试通过!双重 crawlo 前缀问题已修复")
193
+ else:
194
+ print("部分测试失败,请检查实现")
195
+ return 1
196
+
197
+ except Exception as e:
198
+ print("=" * 50)
199
+ print(f"测试过程中发生异常: {e}")
200
+ traceback.print_exc()
201
+ return 1
202
+
203
+ return 0
204
+
205
+
206
+ if __name__ == "__main__":
207
+ exit_code = asyncio.run(main())
208
208
  sys.exit(exit_code)