guanjun.py 2.8 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576
  1. # -*- coding: utf-8 -*-
  2. # import copy
  3. import datetime
  4. import json
  5. # import scrapy
  6. from scrapy.http import Request
  7. from .. items import Guanjun
  8. from scrapy_redis.spiders import RedisSpider
  9. class LqguanjunSpider(RedisSpider):
  10. name = 'guanjun'
  11. allowed_domains = ['hg3535z.com']
  12. to_day = datetime.datetime.now()
  13. redis_key = "guanjun: start_urls"
  14. custom_settings = {
  15. "ITEM_PIPELINES": {
  16. 'hg3535.pipeline.guanjun.Guanjunpipeline': 300,
  17. },
  18. # 'LOG_LEVEL': 'DEBUG',
  19. # 'LOG_FILE': "../hg3535/log/guanjun_{}_{}_{}.log".format(to_day.year, to_day.month, to_day.day)
  20. }
  21. def start_requests(self):
  22. for z in range(1, 5):
  23. for x in range(0, 4):
  24. url = 'https://odata.jiushan6688.com/odds6i/d/getodds/zh-cn/sid/' + str(z) + '/pt/1/ubt/or/pn/' + str(
  25. x) + '/sb/2/dc/null/pid/0'
  26. # redis_key = "guanjun: start_urls"
  27. yield Request(url=url, callback=self.parse, meta={'pt': 1, 'sid': z}, dont_filter=True)
  28. def parse(self, response):
  29. try:
  30. datas = json.loads(response.text)
  31. except:
  32. datas = ""
  33. try:
  34. results = datas['n-ot']['egs']
  35. except:
  36. results = ""
  37. try:
  38. ball = datas['n-ot']['s']['n']
  39. except:
  40. ball = ""
  41. if results:
  42. for result in results:
  43. # 联赛id
  44. league_id = result['c']['k']
  45. # 联赛名
  46. league_name = result['c']['n']
  47. new_results = result['es']
  48. for new_result in new_results:
  49. data_game = new_result['i'][4]
  50. time_game = new_result['i'][5]
  51. new_league_name = new_result['egn'].replace('\t', '')
  52. for i in new_result['n-o']:
  53. new_champion = i['mn']
  54. for y in i['o']:
  55. item = Guanjun()
  56. game_id = str(new_result['i'][16])
  57. tema_home = y[0]
  58. champion_team = float(y[2]) - 1
  59. item['league_name'] = league_name
  60. item['tema_home'] = tema_home
  61. item['league_id'] = league_id
  62. item['game_id'] = game_id
  63. item['data_game'] = data_game
  64. item['time_game'] = time_game
  65. item['new_league_name'] = new_league_name
  66. item['champion_team'] = champion_team
  67. item['new_champion'] = new_champion
  68. # item['pt'] = pt
  69. item['ball'] = ball
  70. yield item