crawlo 1.2.6__py3-none-any.whl → 1.2.8__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 (209) hide show
  1. crawlo/__init__.py +61 -61
  2. crawlo/__version__.py +1 -1
  3. crawlo/cleaners/__init__.py +60 -60
  4. crawlo/cleaners/data_formatter.py +225 -225
  5. crawlo/cleaners/encoding_converter.py +125 -125
  6. crawlo/cleaners/text_cleaner.py +232 -232
  7. crawlo/cli.py +75 -88
  8. crawlo/commands/__init__.py +14 -14
  9. crawlo/commands/check.py +594 -594
  10. crawlo/commands/genspider.py +151 -151
  11. crawlo/commands/help.py +138 -144
  12. crawlo/commands/list.py +155 -155
  13. crawlo/commands/run.py +323 -323
  14. crawlo/commands/startproject.py +436 -436
  15. crawlo/commands/stats.py +187 -187
  16. crawlo/commands/utils.py +186 -186
  17. crawlo/config.py +312 -312
  18. crawlo/config_validator.py +251 -251
  19. crawlo/core/__init__.py +2 -2
  20. crawlo/core/engine.py +365 -356
  21. crawlo/core/processor.py +40 -40
  22. crawlo/core/scheduler.py +251 -239
  23. crawlo/crawler.py +1099 -1110
  24. crawlo/data/__init__.py +5 -5
  25. crawlo/data/user_agents.py +107 -107
  26. crawlo/downloader/__init__.py +266 -266
  27. crawlo/downloader/aiohttp_downloader.py +228 -221
  28. crawlo/downloader/cffi_downloader.py +256 -256
  29. crawlo/downloader/httpx_downloader.py +259 -259
  30. crawlo/downloader/hybrid_downloader.py +212 -212
  31. crawlo/downloader/playwright_downloader.py +402 -402
  32. crawlo/downloader/selenium_downloader.py +472 -472
  33. crawlo/event.py +11 -11
  34. crawlo/exceptions.py +81 -81
  35. crawlo/extension/__init__.py +39 -38
  36. crawlo/extension/health_check.py +141 -141
  37. crawlo/extension/log_interval.py +57 -57
  38. crawlo/extension/log_stats.py +81 -81
  39. crawlo/extension/logging_extension.py +43 -43
  40. crawlo/extension/memory_monitor.py +104 -104
  41. crawlo/extension/performance_profiler.py +133 -133
  42. crawlo/extension/request_recorder.py +107 -107
  43. crawlo/filters/__init__.py +154 -154
  44. crawlo/filters/aioredis_filter.py +234 -234
  45. crawlo/filters/memory_filter.py +269 -269
  46. crawlo/items/__init__.py +23 -23
  47. crawlo/items/base.py +21 -21
  48. crawlo/items/fields.py +52 -52
  49. crawlo/items/items.py +104 -104
  50. crawlo/middleware/__init__.py +21 -21
  51. crawlo/middleware/default_header.py +131 -131
  52. crawlo/middleware/download_delay.py +104 -104
  53. crawlo/middleware/middleware_manager.py +136 -135
  54. crawlo/middleware/offsite.py +114 -114
  55. crawlo/middleware/proxy.py +367 -367
  56. crawlo/middleware/request_ignore.py +86 -86
  57. crawlo/middleware/response_code.py +163 -163
  58. crawlo/middleware/response_filter.py +136 -136
  59. crawlo/middleware/retry.py +124 -124
  60. crawlo/mode_manager.py +211 -211
  61. crawlo/network/__init__.py +21 -21
  62. crawlo/network/request.py +338 -338
  63. crawlo/network/response.py +359 -359
  64. crawlo/pipelines/__init__.py +21 -21
  65. crawlo/pipelines/bloom_dedup_pipeline.py +156 -156
  66. crawlo/pipelines/console_pipeline.py +39 -39
  67. crawlo/pipelines/csv_pipeline.py +316 -316
  68. crawlo/pipelines/database_dedup_pipeline.py +222 -222
  69. crawlo/pipelines/json_pipeline.py +218 -218
  70. crawlo/pipelines/memory_dedup_pipeline.py +115 -115
  71. crawlo/pipelines/mongo_pipeline.py +131 -131
  72. crawlo/pipelines/mysql_pipeline.py +317 -317
  73. crawlo/pipelines/pipeline_manager.py +62 -61
  74. crawlo/pipelines/redis_dedup_pipeline.py +166 -165
  75. crawlo/project.py +314 -279
  76. crawlo/queue/pqueue.py +37 -37
  77. crawlo/queue/queue_manager.py +377 -376
  78. crawlo/queue/redis_priority_queue.py +306 -306
  79. crawlo/settings/__init__.py +7 -7
  80. crawlo/settings/default_settings.py +219 -215
  81. crawlo/settings/setting_manager.py +122 -122
  82. crawlo/spider/__init__.py +639 -639
  83. crawlo/stats_collector.py +59 -59
  84. crawlo/subscriber.py +129 -129
  85. crawlo/task_manager.py +30 -30
  86. crawlo/templates/crawlo.cfg.tmpl +10 -10
  87. crawlo/templates/project/__init__.py.tmpl +3 -3
  88. crawlo/templates/project/items.py.tmpl +17 -17
  89. crawlo/templates/project/middlewares.py.tmpl +118 -118
  90. crawlo/templates/project/pipelines.py.tmpl +96 -96
  91. crawlo/templates/project/settings.py.tmpl +288 -288
  92. crawlo/templates/project/settings_distributed.py.tmpl +157 -157
  93. crawlo/templates/project/settings_gentle.py.tmpl +100 -100
  94. crawlo/templates/project/settings_high_performance.py.tmpl +134 -134
  95. crawlo/templates/project/settings_simple.py.tmpl +98 -98
  96. crawlo/templates/project/spiders/__init__.py.tmpl +5 -5
  97. crawlo/templates/run.py.tmpl +47 -45
  98. crawlo/templates/spider/spider.py.tmpl +143 -143
  99. crawlo/tools/__init__.py +182 -182
  100. crawlo/tools/anti_crawler.py +268 -268
  101. crawlo/tools/authenticated_proxy.py +240 -240
  102. crawlo/tools/data_validator.py +180 -180
  103. crawlo/tools/date_tools.py +35 -35
  104. crawlo/tools/distributed_coordinator.py +386 -386
  105. crawlo/tools/retry_mechanism.py +220 -220
  106. crawlo/tools/scenario_adapter.py +262 -262
  107. crawlo/utils/__init__.py +35 -35
  108. crawlo/utils/batch_processor.py +259 -259
  109. crawlo/utils/controlled_spider_mixin.py +439 -439
  110. crawlo/utils/date_tools.py +290 -290
  111. crawlo/utils/db_helper.py +343 -343
  112. crawlo/utils/enhanced_error_handler.py +356 -356
  113. crawlo/utils/env_config.py +143 -106
  114. crawlo/utils/error_handler.py +123 -123
  115. crawlo/utils/func_tools.py +82 -82
  116. crawlo/utils/large_scale_config.py +286 -286
  117. crawlo/utils/large_scale_helper.py +344 -344
  118. crawlo/utils/log.py +128 -128
  119. crawlo/utils/performance_monitor.py +285 -285
  120. crawlo/utils/queue_helper.py +175 -175
  121. crawlo/utils/redis_connection_pool.py +351 -351
  122. crawlo/utils/redis_key_validator.py +198 -198
  123. crawlo/utils/request.py +267 -267
  124. crawlo/utils/request_serializer.py +218 -218
  125. crawlo/utils/spider_loader.py +61 -61
  126. crawlo/utils/system.py +11 -11
  127. crawlo/utils/tools.py +4 -4
  128. crawlo/utils/url.py +39 -39
  129. {crawlo-1.2.6.dist-info → crawlo-1.2.8.dist-info}/METADATA +764 -764
  130. crawlo-1.2.8.dist-info/RECORD +209 -0
  131. examples/__init__.py +7 -7
  132. tests/DOUBLE_CRAWLO_PREFIX_FIX_REPORT.md +81 -81
  133. tests/__init__.py +7 -7
  134. tests/advanced_tools_example.py +275 -275
  135. tests/authenticated_proxy_example.py +236 -236
  136. tests/cleaners_example.py +160 -160
  137. tests/config_validation_demo.py +102 -102
  138. tests/controlled_spider_example.py +205 -205
  139. tests/date_tools_example.py +180 -180
  140. tests/dynamic_loading_example.py +523 -523
  141. tests/dynamic_loading_test.py +104 -104
  142. tests/env_config_example.py +133 -133
  143. tests/error_handling_example.py +171 -171
  144. tests/redis_key_validation_demo.py +130 -130
  145. tests/response_improvements_example.py +144 -144
  146. tests/test_advanced_tools.py +148 -148
  147. tests/test_all_redis_key_configs.py +145 -145
  148. tests/test_authenticated_proxy.py +141 -141
  149. tests/test_cleaners.py +54 -54
  150. tests/test_comprehensive.py +146 -146
  151. tests/test_config_consistency.py +81 -0
  152. tests/test_config_validator.py +193 -193
  153. tests/test_crawlo_proxy_integration.py +172 -172
  154. tests/test_date_tools.py +123 -123
  155. tests/test_default_header_middleware.py +158 -158
  156. tests/test_double_crawlo_fix.py +207 -207
  157. tests/test_double_crawlo_fix_simple.py +124 -124
  158. tests/test_download_delay_middleware.py +221 -221
  159. tests/test_downloader_proxy_compatibility.py +268 -268
  160. tests/test_dynamic_downloaders_proxy.py +124 -124
  161. tests/test_dynamic_proxy.py +92 -92
  162. tests/test_dynamic_proxy_config.py +146 -146
  163. tests/test_dynamic_proxy_real.py +109 -109
  164. tests/test_edge_cases.py +303 -303
  165. tests/test_enhanced_error_handler.py +270 -270
  166. tests/test_env_config.py +121 -121
  167. tests/test_error_handler_compatibility.py +112 -112
  168. tests/test_final_validation.py +153 -153
  169. tests/test_framework_env_usage.py +103 -103
  170. tests/test_integration.py +356 -356
  171. tests/test_item_dedup_redis_key.py +122 -122
  172. tests/test_mode_consistency.py +52 -0
  173. tests/test_offsite_middleware.py +221 -221
  174. tests/test_parsel.py +29 -29
  175. tests/test_performance.py +327 -327
  176. tests/test_proxy_api.py +264 -264
  177. tests/test_proxy_health_check.py +32 -32
  178. tests/test_proxy_middleware.py +121 -121
  179. tests/test_proxy_middleware_enhanced.py +216 -216
  180. tests/test_proxy_middleware_integration.py +136 -136
  181. tests/test_proxy_providers.py +56 -56
  182. tests/test_proxy_stats.py +19 -19
  183. tests/test_proxy_strategies.py +59 -59
  184. tests/test_queue_manager_double_crawlo.py +173 -173
  185. tests/test_queue_manager_redis_key.py +176 -176
  186. tests/test_real_scenario_proxy.py +195 -195
  187. tests/test_redis_config.py +28 -28
  188. tests/test_redis_connection_pool.py +294 -294
  189. tests/test_redis_key_naming.py +181 -181
  190. tests/test_redis_key_validator.py +123 -123
  191. tests/test_redis_queue.py +224 -224
  192. tests/test_request_ignore_middleware.py +182 -182
  193. tests/test_request_serialization.py +70 -70
  194. tests/test_response_code_middleware.py +349 -349
  195. tests/test_response_filter_middleware.py +427 -427
  196. tests/test_response_improvements.py +152 -152
  197. tests/test_retry_middleware.py +241 -241
  198. tests/test_scheduler.py +252 -241
  199. tests/test_scheduler_config_update.py +134 -0
  200. tests/test_simple_response.py +61 -61
  201. tests/test_telecom_spider_redis_key.py +205 -205
  202. tests/test_template_content.py +87 -87
  203. tests/test_template_redis_key.py +134 -134
  204. tests/test_tools.py +153 -153
  205. tests/tools_example.py +257 -257
  206. crawlo-1.2.6.dist-info/RECORD +0 -206
  207. {crawlo-1.2.6.dist-info → crawlo-1.2.8.dist-info}/WHEEL +0 -0
  208. {crawlo-1.2.6.dist-info → crawlo-1.2.8.dist-info}/entry_points.txt +0 -0
  209. {crawlo-1.2.6.dist-info → crawlo-1.2.8.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)