zhufudaigeinimenxiqiyangyang_recommend.py 7.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164
  1. import json
  2. import os
  3. import random
  4. import sys
  5. import time
  6. import uuid
  7. import requests
  8. sys.path.append(os.getcwd())
  9. from common.video_item import VideoItem
  10. from common import PiaoQuanPipeline, AliyunLogger, tunnel_proxies, get_redirect_url
  11. from common.mq import MQ
  12. class Zfdgnmxqyyspcheduling(object):
  13. def __init__(self, platform, mode, rule_dict, user_list, env):
  14. self.platform = platform
  15. self.mode = mode
  16. self.rule_dict = rule_dict
  17. self.user_list = user_list
  18. self.env = env
  19. self.download_cnt = 0
  20. self.mq = MQ(topic_name="topic_crawler_etl_" + self.env)
  21. self.limit_flag = False
  22. def get_videoList(self):
  23. headers = {
  24. 'Host': 'api.xiahong.top',
  25. 'Content-Type': 'application/json',
  26. 'Accept-Language': 'zh-cn',
  27. 'Accept': '*/*',
  28. 'User-Agent': 'Mozilla/5.0 (iPhone; CPU iPhone OS 11_3 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Mobile/15E217 MicroMessenger/6.8.0(0x16080000) NetType/WIFI Language/en Branch/Br_trunk MiniProgramEnv/Mac',
  29. 'Referer': 'https://servicewechat.com/wxd0e272b032f12c4b/2/page-frame.html',
  30. 'token': 'eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpYXQiOjE3MDI1MjQyNzcsIm5iZiI6MTcwMjUyNDI3NywiZXhwIjoxNzAyNTMxNDc3LCJkYXRhIjp7InVzZXJfaWQiOjI0NDc3MjQ5N319.jFp2L-QKLudo0sOtwfyk8noIwn8e54a7OQUX3GSn_nQ',
  31. 'ik': 'b326b5062b2f0e69046810717534cb09'
  32. }
  33. page = 1
  34. while True:
  35. time.sleep(random.randint(1, 10))
  36. try:
  37. if self.limit_flag:
  38. message = "本轮已经抓取到足够的数据,自动退出\t{}".format(self.download_cnt)
  39. AliyunLogger.logging(
  40. code="2000",
  41. platform=self.platform,
  42. env=self.env,
  43. mode=self.mode,
  44. message=message
  45. )
  46. return
  47. else:
  48. url = "https://api.xiahong.top/index.php?s=mobile/Home/getHomeList&page={}&appid=wxd0e272b032f12c4b&version=1.9.1&env_version=release&scene=1053".format(
  49. page)
  50. response = requests.post(url, headers=headers, proxies=tunnel_proxies())
  51. video_list = response.json()['data']['list']
  52. page += 1
  53. if video_list:
  54. for index, video_obj in enumerate(video_list, 1):
  55. try:
  56. AliyunLogger.logging(
  57. code="1001",
  58. platform=self.platform,
  59. mode=self.mode,
  60. env=self.env,
  61. message="扫描到一条视频",
  62. data=video_obj
  63. )
  64. self.process_video_obj(video_obj)
  65. except Exception as e:
  66. AliyunLogger.logging(
  67. code="3000",
  68. platform=self.platform,
  69. mode=self.mode,
  70. env=self.env,
  71. message="抓取第{}条的时候出现问题, 报错信息是{}".format(index, e)
  72. )
  73. else:
  74. AliyunLogger.logging(
  75. code="2000",
  76. platform=self.platform,
  77. mode=self.mode,
  78. env=self.env,
  79. message="视频已经抓完",
  80. )
  81. return
  82. except Exception as e:
  83. AliyunLogger.logging(
  84. code="3000",
  85. platform=self.platform,
  86. mode=self.mode,
  87. env=self.env,
  88. message="抓取推荐页的时候出现错误, 报错信息是{}".format(e)
  89. )
  90. def process_video_obj(self, video_obj):
  91. trace_id = self.platform + str(uuid.uuid1())
  92. video_id = video_obj['id']
  93. url = "https://api.xiahong.top/index.php?s=mobile/Video/getVideoInfo&vid={}&appid=wxd0e272b032f12c4b&version=1.9.1&scene=1053".format(
  94. video_id)
  95. headers = {
  96. 'Host': 'api.xiahong.top',
  97. 'Content-Type': 'application/json',
  98. 'Accept-Language': 'zh-cn',
  99. 'Accept': '*/*',
  100. 'User-Agent': 'Mozilla/5.0 (iPhone; CPU iPhone OS 11_3 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Mobile/15E217 MicroMessenger/6.8.0(0x16080000) NetType/WIFI Language/en Branch/Br_trunk MiniProgramEnv/Mac',
  101. 'Referer': 'https://servicewechat.com/wxd0e272b032f12c4b/2/page-frame.html',
  102. 'token': 'eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpYXQiOjE3MDI1MjQyNzcsIm5iZiI6MTcwMjUyNDI3NywiZXhwIjoxNzAyNTMxNDc3LCJkYXRhIjp7InVzZXJfaWQiOjI0NDc3MjQ5N319.jFp2L-QKLudo0sOtwfyk8noIwn8e54a7OQUX3GSn_nQ',
  103. 'ik': 'b326b5062b2f0e69046810717534cb09'
  104. }
  105. response = requests.post(url, headers=headers, proxies=tunnel_proxies())
  106. video_info = response.json()["data"]["video_info"]
  107. video_url = video_info['video_url']
  108. video_url = get_redirect_url(video_url)
  109. our_user = random.choice(self.user_list)
  110. item = VideoItem()
  111. item.add_video_info("user_id", our_user["uid"])
  112. item.add_video_info("user_name", our_user["nick_name"])
  113. item.add_video_info("video_id", video_id)
  114. item.add_video_info("video_title", video_info["title"])
  115. item.add_video_info("video_url", video_url)
  116. item.add_video_info("cover_url", video_obj["images"])
  117. item.add_video_info("avatar_url", video_obj["images"])
  118. item.add_video_info("out_video_id", video_id)
  119. item.add_video_info("platform", self.platform)
  120. item.add_video_info("strategy", self.mode)
  121. item.add_video_info("session", "{}-{}".format(self.platform, int(time.time())))
  122. mq_obj = item.produce_item()
  123. pipeline = PiaoQuanPipeline(
  124. platform=self.platform,
  125. mode=self.mode,
  126. rule_dict=self.rule_dict,
  127. env=self.env,
  128. item=mq_obj,
  129. trace_id=trace_id,
  130. )
  131. if pipeline.process_item():
  132. self.download_cnt += 1
  133. self.mq.send_msg(mq_obj)
  134. AliyunLogger.logging(
  135. code="1002",
  136. platform=self.platform,
  137. mode=self.mode,
  138. env=self.env,
  139. message="成功发送至 ETL",
  140. data=mq_obj,
  141. trace_id=trace_id
  142. )
  143. if self.download_cnt >= int(
  144. self.rule_dict.get("videos_cnt", {}).get("min", 200)
  145. ):
  146. self.limit_flag = True
  147. if __name__ == "__main__":
  148. ZL = Zfdgnmxqyyspcheduling(
  149. platform="Jxzfwncdhy",
  150. mode="recommend",
  151. env="dev",
  152. rule_dict={},
  153. user_list=[{'nick_name': "Ivring", 'uid': "1997"}, {'nick_name': "paul", 'uid': "1998"}]
  154. )
  155. ZL.get_videoList()