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
@@ -1,295 +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)