ProxyCheck.py 1.9 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061
  1. # -*- coding: utf-8 -*-
  2. """
  3. -------------------------------------------------
  4. File Name: ProxyCheck
  5. Description : 多线程验证useful_proxy
  6. Author : J_hao
  7. date: 2017/9/26
  8. -------------------------------------------------
  9. Change Activity:
  10. 2017/9/26: 多线程验证useful_proxy
  11. -------------------------------------------------
  12. """
  13. __author__ = 'J_hao'
  14. import sys
  15. import threading
  16. from threading import Thread
  17. sys.path.append('../')
  18. from Util.utilFunction import validUsefulProxy
  19. from Manager.ProxyManager import ProxyManager
  20. from Util.LogHandler import LogHandler
  21. FAIL_COUNT = 2 # 校验失败次数, 超过次数删除代理
  22. class ProxyCheck(ProxyManager, Thread):
  23. def __init__(self, queue, item_dict):
  24. ProxyManager.__init__(self)
  25. Thread.__init__(self)
  26. self.log = LogHandler('proxy_check')
  27. self.queue = queue
  28. self.item_dict = item_dict
  29. def run(self):
  30. self.db.changeTable(self.useful_proxy_queue)
  31. while self.queue.qsize():
  32. proxy = self.queue.get()
  33. count = self.item_dict[proxy]
  34. if validUsefulProxy(proxy):
  35. # 验证通过计数器减1
  36. if count and int(count) > 0:
  37. self.db.put(proxy, num=int(count) - 1)
  38. else:
  39. pass
  40. print('ProxyCheck: {} validation pass'.format(proxy))
  41. else:
  42. print('ProxyCheck: {} validation fail'.format(proxy))
  43. if count and int(count) > FAIL_COUNT:
  44. print('ProxyCheck: {} fail too many, delete!'.format(proxy))
  45. self.db.delete(proxy)
  46. else:
  47. self.db.put(proxy, num=int(count) + 1)
  48. self.queue.task_done()
  49. if __name__ == '__main__':
  50. # p = ProxyCheck()
  51. # p.run()
  52. pass