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
@@ -1,295 +1,295 @@
1
- #!/usr/bin/env python3
2
- # -*- coding: utf-8 -*-
3
- """
4
- Redis连接池优化工具测试
5
- """
6
- import asyncio
7
- import sys
8
- import os
9
- import traceback
10
-
11
- # 添加项目根目录到Python路径
12
- sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
13
-
14
- from crawlo.utils.redis_connection_pool import (
15
- OptimizedRedisConnectionPool,
16
- RedisBatchOperationHelper,
17
- get_redis_pool,
18
- close_all_pools
19
- )
20
-
21
-
22
- async def test_connection_pool():
23
- """测试连接池基本功能"""
24
- print("1. 测试连接池基本功能...")
25
-
26
- try:
27
- # 创建连接池
28
- pool = OptimizedRedisConnectionPool(
29
- "redis://127.0.0.1:6379/15", # 使用测试数据库
30
- max_connections=10,
31
- socket_connect_timeout=5,
32
- socket_timeout=30
33
- )
34
-
35
- # 获取连接
36
- redis_client = await pool.get_connection()
37
-
38
- # 测试连接
39
- await redis_client.ping()
40
- print(" Redis连接测试成功")
41
-
42
- # 获取统计信息
43
- stats = pool.get_stats()
44
- print(f" 连接池统计: {stats}")
45
-
46
- # 关闭连接池
47
- await pool.close()
48
- print(" 连接池关闭成功")
49
-
50
- return True
51
-
52
- except Exception as e:
53
- print(f" 连接池测试失败: {e}")
54
- traceback.print_exc()
55
- return False
56
-
57
-
58
- async def test_batch_operations():
59
- """测试批量操作功能"""
60
- print("2. 测试批量操作功能...")
61
-
62
- try:
63
- # 创建连接池和批量操作助手
64
- pool = get_redis_pool("redis://127.0.0.1:6379/15")
65
- redis_client = await pool.get_connection()
66
- helper = RedisBatchOperationHelper(redis_client, batch_size=5)
67
-
68
- # 准备测试数据
69
- test_key = "test_batch_key"
70
- await redis_client.delete(test_key) # 清理旧数据
71
-
72
- # 测试批量执行
73
- operations = [
74
- ("set", f"{test_key}:1", "value1"),
75
- ("set", f"{test_key}:2", "value2"),
76
- ("set", f"{test_key}:3", "value3"),
77
- ("set", f"{test_key}:4", "value4"),
78
- ("set", f"{test_key}:5", "value5"),
79
- ("set", f"{test_key}:6", "value6"),
80
- ("set", f"{test_key}:7", "value7"),
81
- ]
82
-
83
- results = await helper.batch_execute(operations)
84
- print(f" 批量执行完成,结果数量: {len(results)}")
85
-
86
- # 验证结果
87
- for i in range(1, 8):
88
- value = await redis_client.get(f"{test_key}:{i}")
89
- assert value == f"value{i}", f"值不匹配: {value} != value{i}"
90
-
91
- print(" 批量执行结果验证成功")
92
-
93
- # 测试批量Hash操作
94
- hash_key = "test_batch_hash"
95
- await redis_client.delete(hash_key) # 清理旧数据
96
-
97
- # 批量设置Hash字段
98
- hash_items = {
99
- "field1": "value1",
100
- "field2": "value2",
101
- "field3": "value3",
102
- "field4": "value4",
103
- "field5": "value5",
104
- "field6": "value6",
105
- }
106
-
107
- count = await helper.batch_set_hash(hash_key, hash_items)
108
- print(f" 批量设置Hash字段,设置数量: {count}")
109
-
110
- # 批量获取Hash字段
111
- fields = ["field1", "field3", "field5", "field7"] # 包含一个不存在的字段
112
- hash_values = await helper.batch_get_hash(hash_key, fields)
113
- print(f" 批量获取Hash字段,获取数量: {len(hash_values)}")
114
-
115
- # 验证获取的值
116
- assert hash_values["field1"] == "value1"
117
- assert hash_values["field3"] == "value3"
118
- assert hash_values["field5"] == "value5"
119
- assert "field7" not in hash_values # 不存在的字段
120
-
121
- print(" 批量Hash操作验证成功")
122
-
123
- # 清理测试数据
124
- await redis_client.delete(test_key, hash_key)
125
-
126
- return True
127
-
128
- except Exception as e:
129
- print(f" 批量操作测试失败: {e}")
130
- traceback.print_exc()
131
- return False
132
-
133
-
134
- async def test_connection_context():
135
- """测试连接上下文管理器"""
136
- print("3. 测试连接上下文管理器...")
137
-
138
- try:
139
- # 创建连接池
140
- pool = get_redis_pool("redis://127.0.0.1:6379/15")
141
-
142
- # 使用上下文管理器
143
- async with pool.connection_context() as redis_client:
144
- # 测试连接
145
- await redis_client.ping()
146
- print(" 连接上下文管理器测试成功")
147
-
148
- # 设置测试值
149
- await redis_client.set("context_test_key", "context_test_value")
150
-
151
- # 获取测试值
152
- value = await redis_client.get("context_test_key")
153
- assert value == "context_test_value"
154
-
155
- print(" 上下文管理器操作验证成功")
156
-
157
- # 检查统计信息
158
- stats = pool.get_stats()
159
- print(f" 连接池统计: {stats}")
160
-
161
- # 清理测试数据
162
- redis_client = await pool.get_connection()
163
- await redis_client.delete("context_test_key")
164
-
165
- return True
166
-
167
- except Exception as e:
168
- print(f" 连接上下文管理器测试失败: {e}")
169
- traceback.print_exc()
170
- return False
171
-
172
-
173
- async def test_singleton_pattern():
174
- """测试单例模式"""
175
- print("4. 测试单例模式...")
176
-
177
- try:
178
- # 获取同一个URL的连接池
179
- pool1 = get_redis_pool("redis://127.0.0.1:6379/15")
180
- pool2 = get_redis_pool("redis://127.0.0.1:6379/15")
181
-
182
- # 验证是否为同一个实例
183
- assert pool1 is pool2, "单例模式失败"
184
- print(" 单例模式测试成功")
185
-
186
- # 获取不同URL的连接池
187
- pool3 = get_redis_pool("redis://127.0.0.1:6379/14")
188
- assert pool1 is not pool3, "不同URL应该创建不同实例"
189
- print(" 不同URL创建不同实例测试成功")
190
-
191
- return True
192
-
193
- except Exception as e:
194
- print(f" 单例模式测试失败: {e}")
195
- traceback.print_exc()
196
- return False
197
-
198
-
199
- async def test_concurrent_access():
200
- """测试并发访问"""
201
- print("5. 测试并发访问...")
202
-
203
- try:
204
- # 创建连接池
205
- pool = get_redis_pool("redis://127.0.0.1:6379/15", max_connections=20)
206
-
207
- async def worker(worker_id: int):
208
- """工作协程"""
209
- try:
210
- # 获取连接
211
- redis_client = await pool.get_connection()
212
-
213
- # 执行一些操作
214
- key = f"concurrent_test:{worker_id}"
215
- await redis_client.set(key, f"value_{worker_id}")
216
- value = await redis_client.get(key)
217
-
218
- # 验证操作结果
219
- assert value == f"value_{worker_id}"
220
-
221
- # 等待一小段时间
222
- await asyncio.sleep(0.1)
223
-
224
- # 删除测试键
225
- await redis_client.delete(key)
226
-
227
- return True
228
- except Exception as e:
229
- print(f" 工作协程 {worker_id} 失败: {e}")
230
- return False
231
-
232
- # 创建多个并发任务
233
- tasks = [worker(i) for i in range(10)]
234
- results = await asyncio.gather(*tasks, return_exceptions=True)
235
-
236
- # 检查结果
237
- success_count = sum(1 for result in results if result is True)
238
- print(f" 并发访问测试完成,成功: {success_count}/10")
239
-
240
- # 检查连接池统计
241
- stats = pool.get_stats()
242
- print(f" 连接池统计: {stats}")
243
-
244
- return success_count >= 8 # 允许少量失败
245
-
246
- except Exception as e:
247
- print(f" 并发访问测试失败: {e}")
248
- traceback.print_exc()
249
- return False
250
-
251
-
252
- async def main():
253
- """主测试函数"""
254
- print("开始Redis连接池优化工具测试...")
255
- print("=" * 50)
256
-
257
- tests = [
258
- test_connection_pool,
259
- test_batch_operations,
260
- test_connection_context,
261
- test_singleton_pattern,
262
- test_concurrent_access
263
- ]
264
-
265
- passed = 0
266
- total = len(tests)
267
-
268
- for test_func in tests:
269
- try:
270
- if await test_func():
271
- passed += 1
272
- print(f"{test_func.__name__} 通过")
273
- else:
274
- print(f"{test_func.__name__} 失败")
275
- except Exception as e:
276
- print(f"{test_func.__name__} 异常: {e}")
277
- print()
278
-
279
- # 关闭所有连接池
280
- await close_all_pools()
281
-
282
- print("=" * 50)
283
- print(f"测试结果: {passed}/{total} 通过")
284
-
285
- if passed == total:
286
- print("所有测试通过!Redis连接池优化工具工作正常")
287
- return 0
288
- else:
289
- print("部分测试失败,请检查实现")
290
- return 1
291
-
292
-
293
- if __name__ == "__main__":
294
- exit_code = asyncio.run(main())
1
+ #!/usr/bin/env python3
2
+ # -*- coding: utf-8 -*-
3
+ """
4
+ Redis连接池优化工具测试
5
+ """
6
+ import asyncio
7
+ import sys
8
+ import os
9
+ import traceback
10
+
11
+ # 添加项目根目录到Python路径
12
+ sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
13
+
14
+ from crawlo.utils.redis_connection_pool import (
15
+ OptimizedRedisConnectionPool,
16
+ RedisBatchOperationHelper,
17
+ get_redis_pool,
18
+ close_all_pools
19
+ )
20
+
21
+
22
+ async def test_connection_pool():
23
+ """测试连接池基本功能"""
24
+ print("1. 测试连接池基本功能...")
25
+
26
+ try:
27
+ # 创建连接池
28
+ pool = OptimizedRedisConnectionPool(
29
+ "redis://127.0.0.1:6379/15", # 使用测试数据库
30
+ max_connections=10,
31
+ socket_connect_timeout=5,
32
+ socket_timeout=30
33
+ )
34
+
35
+ # 获取连接
36
+ redis_client = await pool.get_connection()
37
+
38
+ # 测试连接
39
+ await redis_client.ping()
40
+ print(" Redis连接测试成功")
41
+
42
+ # 获取统计信息
43
+ stats = pool.get_stats()
44
+ print(f" 连接池统计: {stats}")
45
+
46
+ # 关闭连接池
47
+ await pool.close()
48
+ print(" 连接池关闭成功")
49
+
50
+ return True
51
+
52
+ except Exception as e:
53
+ print(f" 连接池测试失败: {e}")
54
+ traceback.print_exc()
55
+ return False
56
+
57
+
58
+ async def test_batch_operations():
59
+ """测试批量操作功能"""
60
+ print("2. 测试批量操作功能...")
61
+
62
+ try:
63
+ # 创建连接池和批量操作助手
64
+ pool = get_redis_pool("redis://127.0.0.1:6379/15")
65
+ redis_client = await pool.get_connection()
66
+ helper = RedisBatchOperationHelper(redis_client, batch_size=5)
67
+
68
+ # 准备测试数据
69
+ test_key = "test_batch_key"
70
+ await redis_client.delete(test_key) # 清理旧数据
71
+
72
+ # 测试批量执行
73
+ operations = [
74
+ ("set", f"{test_key}:1", "value1"),
75
+ ("set", f"{test_key}:2", "value2"),
76
+ ("set", f"{test_key}:3", "value3"),
77
+ ("set", f"{test_key}:4", "value4"),
78
+ ("set", f"{test_key}:5", "value5"),
79
+ ("set", f"{test_key}:6", "value6"),
80
+ ("set", f"{test_key}:7", "value7"),
81
+ ]
82
+
83
+ results = await helper.batch_execute(operations)
84
+ print(f" 批量执行完成,结果数量: {len(results)}")
85
+
86
+ # 验证结果
87
+ for i in range(1, 8):
88
+ value = await redis_client.get(f"{test_key}:{i}")
89
+ assert value == f"value{i}", f"值不匹配: {value} != value{i}"
90
+
91
+ print(" 批量执行结果验证成功")
92
+
93
+ # 测试批量Hash操作
94
+ hash_key = "test_batch_hash"
95
+ await redis_client.delete(hash_key) # 清理旧数据
96
+
97
+ # 批量设置Hash字段
98
+ hash_items = {
99
+ "field1": "value1",
100
+ "field2": "value2",
101
+ "field3": "value3",
102
+ "field4": "value4",
103
+ "field5": "value5",
104
+ "field6": "value6",
105
+ }
106
+
107
+ count = await helper.batch_set_hash(hash_key, hash_items)
108
+ print(f" 批量设置Hash字段,设置数量: {count}")
109
+
110
+ # 批量获取Hash字段
111
+ fields = ["field1", "field3", "field5", "field7"] # 包含一个不存在的字段
112
+ hash_values = await helper.batch_get_hash(hash_key, fields)
113
+ print(f" 批量获取Hash字段,获取数量: {len(hash_values)}")
114
+
115
+ # 验证获取的值
116
+ assert hash_values["field1"] == "value1"
117
+ assert hash_values["field3"] == "value3"
118
+ assert hash_values["field5"] == "value5"
119
+ assert "field7" not in hash_values # 不存在的字段
120
+
121
+ print(" 批量Hash操作验证成功")
122
+
123
+ # 清理测试数据
124
+ await redis_client.delete(test_key, hash_key)
125
+
126
+ return True
127
+
128
+ except Exception as e:
129
+ print(f" 批量操作测试失败: {e}")
130
+ traceback.print_exc()
131
+ return False
132
+
133
+
134
+ async def test_connection_context():
135
+ """测试连接上下文管理器"""
136
+ print("3. 测试连接上下文管理器...")
137
+
138
+ try:
139
+ # 创建连接池
140
+ pool = get_redis_pool("redis://127.0.0.1:6379/15")
141
+
142
+ # 使用上下文管理器
143
+ async with pool.connection_context() as redis_client:
144
+ # 测试连接
145
+ await redis_client.ping()
146
+ print(" 连接上下文管理器测试成功")
147
+
148
+ # 设置测试值
149
+ await redis_client.set("context_test_key", "context_test_value")
150
+
151
+ # 获取测试值
152
+ value = await redis_client.get("context_test_key")
153
+ assert value == "context_test_value"
154
+
155
+ print(" 上下文管理器操作验证成功")
156
+
157
+ # 检查统计信息
158
+ stats = pool.get_stats()
159
+ print(f" 连接池统计: {stats}")
160
+
161
+ # 清理测试数据
162
+ redis_client = await pool.get_connection()
163
+ await redis_client.delete("context_test_key")
164
+
165
+ return True
166
+
167
+ except Exception as e:
168
+ print(f" 连接上下文管理器测试失败: {e}")
169
+ traceback.print_exc()
170
+ return False
171
+
172
+
173
+ async def test_singleton_pattern():
174
+ """测试单例模式"""
175
+ print("4. 测试单例模式...")
176
+
177
+ try:
178
+ # 获取同一个URL的连接池
179
+ pool1 = get_redis_pool("redis://127.0.0.1:6379/15")
180
+ pool2 = get_redis_pool("redis://127.0.0.1:6379/15")
181
+
182
+ # 验证是否为同一个实例
183
+ assert pool1 is pool2, "单例模式失败"
184
+ print(" 单例模式测试成功")
185
+
186
+ # 获取不同URL的连接池
187
+ pool3 = get_redis_pool("redis://127.0.0.1:6379/14")
188
+ assert pool1 is not pool3, "不同URL应该创建不同实例"
189
+ print(" 不同URL创建不同实例测试成功")
190
+
191
+ return True
192
+
193
+ except Exception as e:
194
+ print(f" 单例模式测试失败: {e}")
195
+ traceback.print_exc()
196
+ return False
197
+
198
+
199
+ async def test_concurrent_access():
200
+ """测试并发访问"""
201
+ print("5. 测试并发访问...")
202
+
203
+ try:
204
+ # 创建连接池
205
+ pool = get_redis_pool("redis://127.0.0.1:6379/15", max_connections=20)
206
+
207
+ async def worker(worker_id: int):
208
+ """工作协程"""
209
+ try:
210
+ # 获取连接
211
+ redis_client = await pool.get_connection()
212
+
213
+ # 执行一些操作
214
+ key = f"concurrent_test:{worker_id}"
215
+ await redis_client.set(key, f"value_{worker_id}")
216
+ value = await redis_client.get(key)
217
+
218
+ # 验证操作结果
219
+ assert value == f"value_{worker_id}"
220
+
221
+ # 等待一小段时间
222
+ await asyncio.sleep(0.1)
223
+
224
+ # 删除测试键
225
+ await redis_client.delete(key)
226
+
227
+ return True
228
+ except Exception as e:
229
+ print(f" 工作协程 {worker_id} 失败: {e}")
230
+ return False
231
+
232
+ # 创建多个并发任务
233
+ tasks = [worker(i) for i in range(10)]
234
+ results = await asyncio.gather(*tasks, return_exceptions=True)
235
+
236
+ # 检查结果
237
+ success_count = sum(1 for result in results if result is True)
238
+ print(f" 并发访问测试完成,成功: {success_count}/10")
239
+
240
+ # 检查连接池统计
241
+ stats = pool.get_stats()
242
+ print(f" 连接池统计: {stats}")
243
+
244
+ return success_count >= 8 # 允许少量失败
245
+
246
+ except Exception as e:
247
+ print(f" 并发访问测试失败: {e}")
248
+ traceback.print_exc()
249
+ return False
250
+
251
+
252
+ async def main():
253
+ """主测试函数"""
254
+ print("开始Redis连接池优化工具测试...")
255
+ print("=" * 50)
256
+
257
+ tests = [
258
+ test_connection_pool,
259
+ test_batch_operations,
260
+ test_connection_context,
261
+ test_singleton_pattern,
262
+ test_concurrent_access
263
+ ]
264
+
265
+ passed = 0
266
+ total = len(tests)
267
+
268
+ for test_func in tests:
269
+ try:
270
+ if await test_func():
271
+ passed += 1
272
+ print(f"{test_func.__name__} 通过")
273
+ else:
274
+ print(f"{test_func.__name__} 失败")
275
+ except Exception as e:
276
+ print(f"{test_func.__name__} 异常: {e}")
277
+ print()
278
+
279
+ # 关闭所有连接池
280
+ await close_all_pools()
281
+
282
+ print("=" * 50)
283
+ print(f"测试结果: {passed}/{total} 通过")
284
+
285
+ if passed == total:
286
+ print("所有测试通过!Redis连接池优化工具工作正常")
287
+ return 0
288
+ else:
289
+ print("部分测试失败,请检查实现")
290
+ return 1
291
+
292
+
293
+ if __name__ == "__main__":
294
+ exit_code = asyncio.run(main())
295
295
  exit(exit_code)