crawlo 1.1.0__py3-none-any.whl → 1.1.2__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 (120) hide show
  1. crawlo/__init__.py +34 -24
  2. crawlo/__version__.py +1 -1
  3. crawlo/cli.py +40 -40
  4. crawlo/commands/__init__.py +13 -13
  5. crawlo/commands/check.py +594 -155
  6. crawlo/commands/genspider.py +152 -111
  7. crawlo/commands/list.py +156 -119
  8. crawlo/commands/run.py +285 -170
  9. crawlo/commands/startproject.py +196 -101
  10. crawlo/commands/stats.py +188 -167
  11. crawlo/commands/utils.py +187 -0
  12. crawlo/config.py +280 -0
  13. crawlo/core/__init__.py +2 -2
  14. crawlo/core/engine.py +171 -158
  15. crawlo/core/enhanced_engine.py +190 -0
  16. crawlo/core/processor.py +40 -40
  17. crawlo/core/scheduler.py +162 -57
  18. crawlo/crawler.py +1028 -493
  19. crawlo/downloader/__init__.py +242 -78
  20. crawlo/downloader/aiohttp_downloader.py +212 -199
  21. crawlo/downloader/cffi_downloader.py +252 -277
  22. crawlo/downloader/httpx_downloader.py +257 -246
  23. crawlo/event.py +11 -11
  24. crawlo/exceptions.py +78 -78
  25. crawlo/extension/__init__.py +31 -31
  26. crawlo/extension/log_interval.py +49 -49
  27. crawlo/extension/log_stats.py +44 -44
  28. crawlo/extension/logging_extension.py +34 -34
  29. crawlo/filters/__init__.py +154 -37
  30. crawlo/filters/aioredis_filter.py +242 -150
  31. crawlo/filters/memory_filter.py +269 -202
  32. crawlo/items/__init__.py +23 -23
  33. crawlo/items/base.py +21 -21
  34. crawlo/items/fields.py +53 -53
  35. crawlo/items/items.py +104 -104
  36. crawlo/middleware/__init__.py +21 -21
  37. crawlo/middleware/default_header.py +32 -32
  38. crawlo/middleware/download_delay.py +28 -28
  39. crawlo/middleware/middleware_manager.py +135 -135
  40. crawlo/middleware/proxy.py +248 -245
  41. crawlo/middleware/request_ignore.py +30 -30
  42. crawlo/middleware/response_code.py +18 -18
  43. crawlo/middleware/response_filter.py +26 -26
  44. crawlo/middleware/retry.py +125 -90
  45. crawlo/mode_manager.py +201 -0
  46. crawlo/network/__init__.py +21 -7
  47. crawlo/network/request.py +311 -203
  48. crawlo/network/response.py +269 -166
  49. crawlo/pipelines/__init__.py +13 -13
  50. crawlo/pipelines/console_pipeline.py +39 -39
  51. crawlo/pipelines/csv_pipeline.py +317 -0
  52. crawlo/pipelines/json_pipeline.py +219 -0
  53. crawlo/pipelines/mongo_pipeline.py +116 -116
  54. crawlo/pipelines/mysql_pipeline.py +195 -195
  55. crawlo/pipelines/pipeline_manager.py +56 -56
  56. crawlo/project.py +153 -0
  57. crawlo/queue/pqueue.py +37 -0
  58. crawlo/queue/queue_manager.py +304 -0
  59. crawlo/queue/redis_priority_queue.py +192 -0
  60. crawlo/settings/__init__.py +7 -7
  61. crawlo/settings/default_settings.py +226 -169
  62. crawlo/settings/setting_manager.py +99 -99
  63. crawlo/spider/__init__.py +639 -129
  64. crawlo/stats_collector.py +59 -59
  65. crawlo/subscriber.py +106 -106
  66. crawlo/task_manager.py +30 -27
  67. crawlo/templates/crawlo.cfg.tmpl +10 -10
  68. crawlo/templates/project/__init__.py.tmpl +3 -3
  69. crawlo/templates/project/items.py.tmpl +17 -17
  70. crawlo/templates/project/middlewares.py.tmpl +87 -76
  71. crawlo/templates/project/pipelines.py.tmpl +336 -64
  72. crawlo/templates/project/run.py.tmpl +239 -0
  73. crawlo/templates/project/settings.py.tmpl +248 -54
  74. crawlo/templates/project/spiders/__init__.py.tmpl +5 -5
  75. crawlo/templates/spider/spider.py.tmpl +178 -32
  76. crawlo/utils/__init__.py +7 -7
  77. crawlo/utils/controlled_spider_mixin.py +336 -0
  78. crawlo/utils/date_tools.py +233 -233
  79. crawlo/utils/db_helper.py +343 -343
  80. crawlo/utils/func_tools.py +82 -82
  81. crawlo/utils/large_scale_config.py +287 -0
  82. crawlo/utils/large_scale_helper.py +344 -0
  83. crawlo/utils/log.py +128 -128
  84. crawlo/utils/queue_helper.py +176 -0
  85. crawlo/utils/request.py +267 -267
  86. crawlo/utils/request_serializer.py +220 -0
  87. crawlo/utils/spider_loader.py +62 -62
  88. crawlo/utils/system.py +11 -11
  89. crawlo/utils/tools.py +4 -4
  90. crawlo/utils/url.py +39 -39
  91. crawlo-1.1.2.dist-info/METADATA +567 -0
  92. crawlo-1.1.2.dist-info/RECORD +108 -0
  93. examples/__init__.py +7 -0
  94. tests/__init__.py +7 -7
  95. tests/test_final_validation.py +154 -0
  96. tests/test_proxy_health_check.py +32 -32
  97. tests/test_proxy_middleware_integration.py +136 -136
  98. tests/test_proxy_providers.py +56 -56
  99. tests/test_proxy_stats.py +19 -19
  100. tests/test_proxy_strategies.py +59 -59
  101. tests/test_redis_config.py +29 -0
  102. tests/test_redis_queue.py +225 -0
  103. tests/test_request_serialization.py +71 -0
  104. tests/test_scheduler.py +242 -0
  105. crawlo/pipelines/mysql_batch_pipline.py +0 -273
  106. crawlo/utils/concurrency_manager.py +0 -125
  107. crawlo/utils/pqueue.py +0 -174
  108. crawlo/utils/project.py +0 -197
  109. crawlo-1.1.0.dist-info/METADATA +0 -49
  110. crawlo-1.1.0.dist-info/RECORD +0 -97
  111. examples/gxb/items.py +0 -36
  112. examples/gxb/run.py +0 -16
  113. examples/gxb/settings.py +0 -72
  114. examples/gxb/spider/__init__.py +0 -2
  115. examples/gxb/spider/miit_spider.py +0 -180
  116. examples/gxb/spider/telecom_device.py +0 -129
  117. {examples/gxb → crawlo/queue}/__init__.py +0 -0
  118. {crawlo-1.1.0.dist-info → crawlo-1.1.2.dist-info}/WHEEL +0 -0
  119. {crawlo-1.1.0.dist-info → crawlo-1.1.2.dist-info}/entry_points.txt +0 -0
  120. {crawlo-1.1.0.dist-info → crawlo-1.1.2.dist-info}/top_level.txt +0 -0
crawlo/__init__.py CHANGED
@@ -1,25 +1,35 @@
1
- #!/usr/bin/python
2
- # -*- coding: UTF-8 -*-
3
- """
4
- Crawlo - 一个异步爬虫框架
5
- """
6
- from crawlo.spider import Spider
7
- from crawlo.items.items import Item
8
- from crawlo.network.request import Request
9
- from crawlo.network.response import Response
10
- from crawlo.downloader import DownloaderBase
11
- from crawlo.middleware import BaseMiddleware
12
-
13
- # 版本号
14
- from crawlo.__version__ import __version__
15
-
16
- # 可选:定义对外暴露的接口
17
- __all__ = [
18
- 'Spider',
19
- 'Item',
20
- 'Request',
21
- 'Response',
22
- 'DownloaderBase',
23
- 'BaseMiddleware',
24
- '__version__',
1
+ #!/usr/bin/python
2
+ # -*- coding: UTF-8 -*-
3
+ """
4
+ Crawlo - 一个异步爬虫框架
5
+ """
6
+ from crawlo.spider import Spider
7
+ from crawlo.items import Item, Field
8
+ from crawlo.network.request import Request
9
+ from crawlo.network.response import Response
10
+ from crawlo.downloader import DownloaderBase
11
+ from crawlo.middleware import BaseMiddleware
12
+
13
+
14
+ # 版本号:优先从元数据读取
15
+ try:
16
+ from importlib.metadata import version
17
+ __version__ = version("crawlo")
18
+ except Exception:
19
+ # 开发模式下可能未安装,回退到 __version__.py 或 dev
20
+ try:
21
+ from crawlo.__version__ import __version__
22
+ except ImportError:
23
+ __version__ = "dev"
24
+
25
+ # 定义对外 API
26
+ __all__ = [
27
+ 'Spider',
28
+ 'Item',
29
+ 'Field',
30
+ 'Request',
31
+ 'Response',
32
+ 'DownloaderBase',
33
+ 'BaseMiddleware',
34
+ '__version__',
25
35
  ]
crawlo/__version__.py CHANGED
@@ -1 +1 @@
1
- __version__ = "1.1.0"
1
+ __version__ = "1.1.2"
crawlo/cli.py CHANGED
@@ -1,41 +1,41 @@
1
- # crawlo/cli.py
2
- # !/usr/bin/python
3
- # -*- coding: UTF-8 -*-
4
- import sys
5
- import argparse
6
- from crawlo.commands import get_commands
7
-
8
-
9
- def main():
10
- # 获取所有可用命令
11
- commands = get_commands()
12
-
13
- parser = argparse.ArgumentParser(
14
- description="Crawlo: A lightweight web crawler framework.",
15
- usage="crawlo <command> [options]"
16
- )
17
- parser.add_argument('command', help='Available commands: ' + ', '.join(commands.keys()))
18
- # 注意:这里不添加具体参数,由子命令解析
19
-
20
- # 只解析命令
21
- args, unknown = parser.parse_known_args()
22
-
23
- if args.command not in commands:
24
- print(f"Unknown command: {args.command}")
25
- print(f"Available commands: {', '.join(commands.keys())}")
26
- sys.exit(1)
27
-
28
- # 动态导入并执行命令
29
- try:
30
- module = __import__(commands[args.command], fromlist=['main'])
31
- sys.exit(module.main(unknown))
32
- except ImportError as e:
33
- print(f"Failed to load command '{args.command}': {e}")
34
- sys.exit(1)
35
- except Exception as e:
36
- print(f"Command '{args.command}' failed: {e}")
37
- sys.exit(1)
38
-
39
-
40
- if __name__ == '__main__':
1
+ # crawlo/cli.py
2
+ # !/usr/bin/python
3
+ # -*- coding: UTF-8 -*-
4
+ import sys
5
+ import argparse
6
+ from crawlo.commands import get_commands
7
+
8
+
9
+ def main():
10
+ # 获取所有可用命令
11
+ commands = get_commands()
12
+
13
+ parser = argparse.ArgumentParser(
14
+ description="Crawlo: A lightweight web crawler framework.",
15
+ usage="crawlo <command> [options]"
16
+ )
17
+ parser.add_argument('command', help='Available commands: ' + ', '.join(commands.keys()))
18
+ # 注意:这里不添加具体参数,由子命令解析
19
+
20
+ # 只解析命令
21
+ args, unknown = parser.parse_known_args()
22
+
23
+ if args.command not in commands:
24
+ print(f"Unknown command: {args.command}")
25
+ print(f"Available commands: {', '.join(commands.keys())}")
26
+ sys.exit(1)
27
+
28
+ # 动态导入并执行命令
29
+ try:
30
+ module = __import__(commands[args.command], fromlist=['main'])
31
+ sys.exit(module.main(unknown))
32
+ except ImportError as e:
33
+ print(f"Failed to load command '{args.command}': {e}")
34
+ sys.exit(1)
35
+ except Exception as e:
36
+ print(f"Command '{args.command}' failed: {e}")
37
+ sys.exit(1)
38
+
39
+
40
+ if __name__ == '__main__':
41
41
  main()
@@ -1,14 +1,14 @@
1
- #!/usr/bin/python
2
- # -*- coding:UTF-8 -*-
3
-
4
- _commands = {
5
- 'startproject': 'crawlo.commands.startproject',
6
- 'genspider': 'crawlo.commands.genspider',
7
- 'run': 'crawlo.commands.run',
8
- 'check': 'crawlo.commands.check',
9
- 'list': 'crawlo.commands.list',
10
- 'stats': 'crawlo.commands.stats'
11
- }
12
-
13
- def get_commands():
1
+ #!/usr/bin/python
2
+ # -*- coding:UTF-8 -*-
3
+
4
+ _commands = {
5
+ 'startproject': 'crawlo.commands.startproject',
6
+ 'genspider': 'crawlo.commands.genspider',
7
+ 'run': 'crawlo.commands.run',
8
+ 'check': 'crawlo.commands.check',
9
+ 'list': 'crawlo.commands.list',
10
+ 'stats': 'crawlo.commands.stats'
11
+ }
12
+
13
+ def get_commands():
14
14
  return _commands