crawlo 1.4.6__py3-none-any.whl → 1.4.7__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 (374) hide show
  1. crawlo/__init__.py +90 -89
  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 +186 -186
  7. crawlo/commands/help.py +140 -138
  8. crawlo/commands/list.py +155 -155
  9. crawlo/commands/run.py +379 -341
  10. crawlo/commands/startproject.py +460 -460
  11. crawlo/commands/stats.py +187 -187
  12. crawlo/commands/utils.py +196 -196
  13. crawlo/config.py +320 -312
  14. crawlo/config_validator.py +277 -277
  15. crawlo/core/__init__.py +52 -52
  16. crawlo/core/engine.py +451 -438
  17. crawlo/core/processor.py +47 -47
  18. crawlo/core/scheduler.py +290 -291
  19. crawlo/crawler.py +698 -657
  20. crawlo/data/__init__.py +5 -5
  21. crawlo/data/user_agents.py +194 -194
  22. crawlo/downloader/__init__.py +280 -276
  23. crawlo/downloader/aiohttp_downloader.py +233 -233
  24. crawlo/downloader/cffi_downloader.py +250 -247
  25. crawlo/downloader/httpx_downloader.py +265 -259
  26. crawlo/downloader/hybrid_downloader.py +212 -212
  27. crawlo/downloader/playwright_downloader.py +425 -402
  28. crawlo/downloader/selenium_downloader.py +486 -472
  29. crawlo/event.py +45 -11
  30. crawlo/exceptions.py +215 -82
  31. crawlo/extension/__init__.py +65 -64
  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 +53 -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 +104 -103
  42. crawlo/factories/registry.py +84 -84
  43. crawlo/factories/utils.py +135 -0
  44. crawlo/filters/__init__.py +170 -153
  45. crawlo/filters/aioredis_filter.py +348 -264
  46. crawlo/filters/memory_filter.py +261 -276
  47. crawlo/framework.py +306 -292
  48. crawlo/initialization/__init__.py +44 -44
  49. crawlo/initialization/built_in.py +391 -434
  50. crawlo/initialization/context.py +141 -141
  51. crawlo/initialization/core.py +240 -194
  52. crawlo/initialization/phases.py +230 -149
  53. crawlo/initialization/registry.py +143 -145
  54. crawlo/initialization/utils.py +49 -0
  55. crawlo/interfaces.py +23 -23
  56. crawlo/items/__init__.py +23 -23
  57. crawlo/items/base.py +23 -23
  58. crawlo/items/fields.py +52 -52
  59. crawlo/items/items.py +104 -104
  60. crawlo/logging/__init__.py +42 -46
  61. crawlo/logging/config.py +277 -197
  62. crawlo/logging/factory.py +175 -171
  63. crawlo/logging/manager.py +104 -112
  64. crawlo/middleware/__init__.py +87 -24
  65. crawlo/middleware/default_header.py +132 -132
  66. crawlo/middleware/download_delay.py +104 -104
  67. crawlo/middleware/middleware_manager.py +142 -142
  68. crawlo/middleware/offsite.py +123 -123
  69. crawlo/middleware/proxy.py +209 -209
  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/mode_manager.py +287 -253
  75. crawlo/network/__init__.py +21 -21
  76. crawlo/network/request.py +375 -379
  77. crawlo/network/response.py +569 -664
  78. crawlo/pipelines/__init__.py +53 -22
  79. crawlo/pipelines/base_pipeline.py +452 -0
  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 +140 -132
  87. crawlo/pipelines/mysql_pipeline.py +469 -476
  88. crawlo/pipelines/pipeline_manager.py +100 -100
  89. crawlo/pipelines/redis_dedup_pipeline.py +155 -156
  90. crawlo/project.py +347 -347
  91. crawlo/queue/__init__.py +10 -0
  92. crawlo/queue/pqueue.py +38 -38
  93. crawlo/queue/queue_manager.py +591 -525
  94. crawlo/queue/redis_priority_queue.py +519 -370
  95. crawlo/settings/__init__.py +7 -7
  96. crawlo/settings/default_settings.py +284 -277
  97. crawlo/settings/setting_manager.py +219 -219
  98. crawlo/spider/__init__.py +657 -657
  99. crawlo/stats_collector.py +81 -81
  100. crawlo/subscriber.py +129 -129
  101. crawlo/task_manager.py +138 -138
  102. crawlo/templates/crawlo.cfg.tmpl +10 -10
  103. crawlo/templates/project/__init__.py.tmpl +2 -4
  104. crawlo/templates/project/items.py.tmpl +13 -17
  105. crawlo/templates/project/middlewares.py.tmpl +38 -38
  106. crawlo/templates/project/pipelines.py.tmpl +35 -36
  107. crawlo/templates/project/settings.py.tmpl +109 -111
  108. crawlo/templates/project/settings_distributed.py.tmpl +156 -159
  109. crawlo/templates/project/settings_gentle.py.tmpl +170 -176
  110. crawlo/templates/project/settings_high_performance.py.tmpl +171 -177
  111. crawlo/templates/project/settings_minimal.py.tmpl +98 -100
  112. crawlo/templates/project/settings_simple.py.tmpl +168 -174
  113. crawlo/templates/project/spiders/__init__.py.tmpl +9 -9
  114. crawlo/templates/run.py.tmpl +23 -23
  115. crawlo/templates/spider/spider.py.tmpl +32 -40
  116. crawlo/templates/spiders_init.py.tmpl +5 -10
  117. crawlo/tools/__init__.py +86 -189
  118. crawlo/tools/date_tools.py +289 -289
  119. crawlo/tools/distributed_coordinator.py +384 -384
  120. crawlo/tools/scenario_adapter.py +262 -262
  121. crawlo/tools/text_cleaner.py +232 -232
  122. crawlo/utils/__init__.py +50 -50
  123. crawlo/utils/batch_processor.py +276 -259
  124. crawlo/utils/config_manager.py +442 -0
  125. crawlo/utils/controlled_spider_mixin.py +439 -439
  126. crawlo/utils/db_helper.py +250 -250
  127. crawlo/utils/error_handler.py +410 -410
  128. crawlo/utils/fingerprint.py +121 -121
  129. crawlo/utils/func_tools.py +82 -82
  130. crawlo/utils/large_scale_helper.py +344 -344
  131. crawlo/utils/leak_detector.py +335 -0
  132. crawlo/utils/log.py +79 -79
  133. crawlo/utils/misc.py +81 -81
  134. crawlo/utils/mongo_connection_pool.py +157 -0
  135. crawlo/utils/mysql_connection_pool.py +197 -0
  136. crawlo/utils/performance_monitor.py +285 -285
  137. crawlo/utils/queue_helper.py +175 -175
  138. crawlo/utils/redis_checker.py +91 -0
  139. crawlo/utils/redis_connection_pool.py +578 -388
  140. crawlo/utils/redis_key_validator.py +198 -198
  141. crawlo/utils/request.py +278 -256
  142. crawlo/utils/request_serializer.py +225 -225
  143. crawlo/utils/resource_manager.py +337 -0
  144. crawlo/utils/selector_helper.py +137 -137
  145. crawlo/utils/singleton.py +70 -0
  146. crawlo/utils/spider_loader.py +201 -201
  147. crawlo/utils/text_helper.py +94 -94
  148. crawlo/utils/{url.py → url_utils.py} +39 -39
  149. crawlo-1.4.7.dist-info/METADATA +689 -0
  150. crawlo-1.4.7.dist-info/RECORD +347 -0
  151. examples/__init__.py +7 -7
  152. tests/__init__.py +7 -7
  153. tests/advanced_tools_example.py +217 -275
  154. tests/authenticated_proxy_example.py +110 -110
  155. tests/baidu_performance_test.py +108 -108
  156. tests/baidu_test.py +59 -59
  157. tests/bug_check_test.py +250 -250
  158. tests/cleaners_example.py +160 -160
  159. tests/comprehensive_framework_test.py +212 -212
  160. tests/comprehensive_test.py +81 -81
  161. tests/comprehensive_testing_summary.md +186 -186
  162. tests/config_validation_demo.py +142 -142
  163. tests/controlled_spider_example.py +205 -205
  164. tests/date_tools_example.py +180 -180
  165. tests/debug_configure.py +69 -69
  166. tests/debug_framework_logger.py +84 -84
  167. tests/debug_log_config.py +126 -126
  168. tests/debug_log_levels.py +63 -63
  169. tests/debug_pipelines.py +66 -66
  170. tests/detailed_log_test.py +233 -233
  171. tests/direct_selector_helper_test.py +96 -96
  172. tests/distributed_dedup_test.py +467 -0
  173. tests/distributed_test.py +66 -66
  174. tests/distributed_test_debug.py +76 -76
  175. tests/dynamic_loading_example.py +523 -523
  176. tests/dynamic_loading_test.py +104 -104
  177. tests/error_handling_example.py +171 -171
  178. tests/explain_mysql_update_behavior.py +76 -76
  179. tests/final_comprehensive_test.py +151 -151
  180. tests/final_log_test.py +260 -260
  181. tests/final_validation_test.py +182 -182
  182. tests/fix_log_test.py +142 -142
  183. tests/framework_performance_test.py +202 -202
  184. tests/log_buffering_test.py +111 -111
  185. tests/log_generation_timing_test.py +153 -153
  186. tests/monitor_redis_dedup.sh +72 -0
  187. tests/ofweek_scrapy/ofweek_scrapy/items.py +12 -12
  188. tests/ofweek_scrapy/ofweek_scrapy/middlewares.py +100 -100
  189. tests/ofweek_scrapy/ofweek_scrapy/pipelines.py +13 -13
  190. tests/ofweek_scrapy/ofweek_scrapy/settings.py +84 -84
  191. tests/ofweek_scrapy/scrapy.cfg +11 -11
  192. tests/optimized_performance_test.py +211 -211
  193. tests/performance_comparison.py +244 -244
  194. tests/queue_blocking_test.py +113 -113
  195. tests/queue_test.py +89 -89
  196. tests/redis_key_validation_demo.py +130 -130
  197. tests/request_params_example.py +150 -150
  198. tests/response_improvements_example.py +144 -144
  199. tests/scrapy_comparison/ofweek_scrapy.py +138 -138
  200. tests/scrapy_comparison/scrapy_test.py +133 -133
  201. tests/simple_cli_test.py +55 -0
  202. tests/simple_command_test.py +119 -119
  203. tests/simple_crawlo_test.py +126 -126
  204. tests/simple_follow_test.py +38 -38
  205. tests/simple_log_test2.py +137 -137
  206. tests/simple_optimization_test.py +128 -128
  207. tests/simple_queue_type_test.py +41 -41
  208. tests/simple_response_selector_test.py +94 -94
  209. tests/simple_selector_helper_test.py +154 -154
  210. tests/simple_selector_test.py +207 -207
  211. tests/simple_spider_test.py +49 -49
  212. tests/simple_url_test.py +73 -73
  213. tests/simulate_mysql_update_test.py +139 -139
  214. tests/spider_log_timing_test.py +177 -177
  215. tests/test_advanced_tools.py +148 -148
  216. tests/test_all_commands.py +230 -230
  217. tests/test_all_pipeline_fingerprints.py +133 -133
  218. tests/test_all_redis_key_configs.py +145 -145
  219. tests/test_asyncmy_usage.py +56 -56
  220. tests/test_batch_processor.py +178 -178
  221. tests/test_cleaners.py +54 -54
  222. tests/test_cli_arguments.py +119 -0
  223. tests/test_component_factory.py +174 -174
  224. tests/test_config_consistency.py +80 -80
  225. tests/test_config_merge.py +152 -152
  226. tests/test_config_validator.py +182 -182
  227. tests/test_controlled_spider_mixin.py +79 -79
  228. tests/test_crawler_process_import.py +38 -38
  229. tests/test_crawler_process_spider_modules.py +47 -47
  230. tests/test_crawlo_proxy_integration.py +114 -114
  231. tests/test_date_tools.py +123 -123
  232. tests/test_dedup_fix.py +220 -220
  233. tests/test_dedup_pipeline_consistency.py +124 -124
  234. tests/test_default_header_middleware.py +313 -313
  235. tests/test_distributed.py +65 -65
  236. tests/test_double_crawlo_fix.py +204 -204
  237. tests/test_double_crawlo_fix_simple.py +124 -124
  238. tests/test_download_delay_middleware.py +221 -221
  239. tests/test_downloader_proxy_compatibility.py +272 -272
  240. tests/test_edge_cases.py +305 -305
  241. tests/test_encoding_core.py +56 -56
  242. tests/test_encoding_detection.py +126 -126
  243. tests/test_enhanced_error_handler.py +270 -270
  244. tests/test_enhanced_error_handler_comprehensive.py +245 -245
  245. tests/test_error_handler_compatibility.py +112 -112
  246. tests/test_factories.py +252 -252
  247. tests/test_factory_compatibility.py +196 -196
  248. tests/test_final_validation.py +153 -153
  249. tests/test_fingerprint_consistency.py +135 -135
  250. tests/test_fingerprint_simple.py +51 -51
  251. tests/test_get_component_logger.py +83 -83
  252. tests/test_hash_performance.py +99 -99
  253. tests/test_integration.py +169 -169
  254. tests/test_item_dedup_redis_key.py +122 -122
  255. tests/test_large_scale_helper.py +235 -235
  256. tests/test_logging_enhancements.py +374 -374
  257. tests/test_logging_final.py +184 -184
  258. tests/test_logging_integration.py +312 -312
  259. tests/test_logging_system.py +282 -282
  260. tests/test_middleware_debug.py +141 -141
  261. tests/test_mode_consistency.py +51 -51
  262. tests/test_multi_directory.py +67 -67
  263. tests/test_multiple_spider_modules.py +80 -80
  264. tests/test_mysql_pipeline_config.py +164 -164
  265. tests/test_mysql_pipeline_error.py +98 -98
  266. tests/test_mysql_pipeline_init_log.py +82 -82
  267. tests/test_mysql_pipeline_integration.py +132 -132
  268. tests/test_mysql_pipeline_refactor.py +143 -143
  269. tests/test_mysql_pipeline_refactor_simple.py +85 -85
  270. tests/test_mysql_pipeline_robustness.py +195 -195
  271. tests/test_mysql_pipeline_types.py +88 -88
  272. tests/test_mysql_update_columns.py +93 -93
  273. tests/test_offsite_middleware.py +244 -244
  274. tests/test_offsite_middleware_simple.py +203 -203
  275. tests/test_optimized_selector_naming.py +100 -100
  276. tests/test_parsel.py +29 -29
  277. tests/test_performance.py +327 -327
  278. tests/test_performance_monitor.py +115 -115
  279. tests/test_pipeline_fingerprint_consistency.py +86 -86
  280. tests/test_priority_behavior.py +211 -211
  281. tests/test_priority_consistency.py +151 -151
  282. tests/test_priority_consistency_fixed.py +249 -249
  283. tests/test_proxy_health_check.py +32 -32
  284. tests/test_proxy_middleware.py +217 -217
  285. tests/test_proxy_middleware_enhanced.py +212 -212
  286. tests/test_proxy_middleware_integration.py +142 -142
  287. tests/test_proxy_middleware_refactored.py +207 -207
  288. tests/test_proxy_only.py +83 -83
  289. tests/test_proxy_providers.py +56 -56
  290. tests/test_proxy_stats.py +19 -19
  291. tests/test_proxy_strategies.py +59 -59
  292. tests/test_proxy_with_downloader.py +152 -152
  293. tests/test_queue_empty_check.py +41 -41
  294. tests/test_queue_manager_double_crawlo.py +173 -173
  295. tests/test_queue_manager_redis_key.py +179 -179
  296. tests/test_queue_naming.py +154 -154
  297. tests/test_queue_type.py +106 -106
  298. tests/test_queue_type_redis_config_consistency.py +130 -130
  299. tests/test_random_headers_default.py +322 -322
  300. tests/test_random_headers_necessity.py +308 -308
  301. tests/test_random_user_agent.py +72 -72
  302. tests/test_redis_config.py +28 -28
  303. tests/test_redis_connection_pool.py +294 -294
  304. tests/test_redis_key_naming.py +181 -181
  305. tests/test_redis_key_validator.py +123 -123
  306. tests/test_redis_queue.py +224 -224
  307. tests/test_redis_queue_name_fix.py +175 -175
  308. tests/test_redis_queue_type_fallback.py +129 -129
  309. tests/test_request_ignore_middleware.py +182 -182
  310. tests/test_request_params.py +111 -111
  311. tests/test_request_serialization.py +70 -70
  312. tests/test_response_code_middleware.py +349 -349
  313. tests/test_response_filter_middleware.py +427 -427
  314. tests/test_response_follow.py +104 -104
  315. tests/test_response_improvements.py +152 -152
  316. tests/test_response_selector_methods.py +92 -92
  317. tests/test_response_url_methods.py +70 -70
  318. tests/test_response_urljoin.py +86 -86
  319. tests/test_retry_middleware.py +333 -333
  320. tests/test_retry_middleware_realistic.py +273 -273
  321. tests/test_scheduler.py +252 -252
  322. tests/test_scheduler_config_update.py +133 -133
  323. tests/test_scrapy_style_encoding.py +112 -112
  324. tests/test_selector_helper.py +100 -100
  325. tests/test_selector_optimizations.py +146 -146
  326. tests/test_simple_response.py +61 -61
  327. tests/test_spider_loader.py +49 -49
  328. tests/test_spider_loader_comprehensive.py +69 -69
  329. tests/test_spider_modules.py +84 -84
  330. tests/test_spiders/test_spider.py +9 -9
  331. tests/test_telecom_spider_redis_key.py +205 -205
  332. tests/test_template_content.py +87 -87
  333. tests/test_template_redis_key.py +134 -134
  334. tests/test_tools.py +159 -159
  335. tests/test_user_agent_randomness.py +176 -176
  336. tests/test_user_agents.py +96 -96
  337. tests/untested_features_report.md +138 -138
  338. tests/verify_debug.py +51 -51
  339. tests/verify_distributed.py +117 -117
  340. tests/verify_log_fix.py +111 -111
  341. tests/verify_mysql_warnings.py +109 -109
  342. crawlo/logging/async_handler.py +0 -181
  343. crawlo/logging/monitor.py +0 -153
  344. crawlo/logging/sampler.py +0 -167
  345. crawlo/tools/authenticated_proxy.py +0 -241
  346. crawlo/tools/data_formatter.py +0 -226
  347. crawlo/tools/data_validator.py +0 -181
  348. crawlo/tools/encoding_converter.py +0 -127
  349. crawlo/tools/network_diagnostic.py +0 -365
  350. crawlo/tools/request_tools.py +0 -83
  351. crawlo/tools/retry_mechanism.py +0 -224
  352. crawlo/utils/env_config.py +0 -143
  353. crawlo/utils/large_scale_config.py +0 -287
  354. crawlo/utils/system.py +0 -11
  355. crawlo/utils/tools.py +0 -5
  356. crawlo-1.4.6.dist-info/METADATA +0 -329
  357. crawlo-1.4.6.dist-info/RECORD +0 -361
  358. tests/env_config_example.py +0 -134
  359. tests/ofweek_scrapy/ofweek_scrapy/spiders/ofweek_spider.py +0 -162
  360. tests/test_authenticated_proxy.py +0 -142
  361. tests/test_comprehensive.py +0 -147
  362. tests/test_dynamic_downloaders_proxy.py +0 -125
  363. tests/test_dynamic_proxy.py +0 -93
  364. tests/test_dynamic_proxy_config.py +0 -147
  365. tests/test_dynamic_proxy_real.py +0 -110
  366. tests/test_env_config.py +0 -122
  367. tests/test_framework_env_usage.py +0 -104
  368. tests/test_large_scale_config.py +0 -113
  369. tests/test_proxy_api.py +0 -265
  370. tests/test_real_scenario_proxy.py +0 -196
  371. tests/tools_example.py +0 -261
  372. {crawlo-1.4.6.dist-info → crawlo-1.4.7.dist-info}/WHEEL +0 -0
  373. {crawlo-1.4.6.dist-info → crawlo-1.4.7.dist-info}/entry_points.txt +0 -0
  374. {crawlo-1.4.6.dist-info → crawlo-1.4.7.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)