jingdianfuqiwang_recommend_scheduling.py 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159
  1. import os
  2. import json
  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
  11. from common.mq import MQ
  12. class TFuQiWangRecommend(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_video_list(self):
  23. base_url = "https://jdfqw.wentingyou.cn/index.php/v111/index/index"
  24. headers = {
  25. 'chatkey': 'wx79ef316d416e5da1',
  26. 'user-agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/107.0.0.0 Safari/537.36 MicroMessenger/6.8.0(0x16080000) NetType/WIFI MiniProgramEnv/Mac MacWechat/WMPF MacWechat/3.8.5(0x13080510)XWEB/1100',
  27. 'content-type': 'application/x-www-form-urlencoded',
  28. 'xweb_xhr': '1',
  29. 'vision': '1.1.0',
  30. 'token': '',
  31. 'accept': '*/*',
  32. 'referer': 'https://servicewechat.com/wx79ef316d416e5da1/1/page-frame.html',
  33. 'accept-language': 'en-US,en;q=0.9'
  34. }
  35. page_index = 1
  36. while True:
  37. time.sleep(random.randint(1, 10))
  38. try:
  39. if self.limit_flag:
  40. AliyunLogger.logging(
  41. code="2000",
  42. platform=self.platform,
  43. mode=self.mode,
  44. env=self.env,
  45. message="本轮已经抓取到足够的数据,自动退出\t{}".format(self.download_cnt),
  46. )
  47. return
  48. else:
  49. obj = {
  50. "cid": "",
  51. "page": 1,
  52. "is_ads": 1
  53. }
  54. params = {
  55. "parameter": json.dumps(obj)
  56. }
  57. response = requests.get(
  58. url=base_url,
  59. headers=headers,
  60. params=params,
  61. proxies=tunnel_proxies(),
  62. )
  63. video_list = response.json()['data']['list']
  64. if video_list:
  65. for index, video_obj in enumerate(video_list, 1):
  66. try:
  67. if video_obj.get("title"):
  68. AliyunLogger.logging(
  69. code="1001",
  70. platform=self.platform,
  71. mode=self.mode,
  72. env=self.env,
  73. message="扫描到一条视频",
  74. data=video_obj,
  75. )
  76. self.process_video_obj(video_obj)
  77. except Exception as e:
  78. AliyunLogger.logging(
  79. code="3000",
  80. platform=self.platform,
  81. mode=self.mode,
  82. env=self.env,
  83. data=video_obj,
  84. message="抓取第{}条的时候出现问题, 报错信息是{}".format(index, e),
  85. )
  86. page_index += 1
  87. else:
  88. AliyunLogger.logging(
  89. code="2000",
  90. platform=self.platform,
  91. mode=self.mode,
  92. env=self.env,
  93. message="已经抓完了,自动退出"
  94. )
  95. return
  96. except Exception as e:
  97. AliyunLogger.logging(
  98. code="3000",
  99. platform=self.platform,
  100. mode=self.mode,
  101. env=self.env,
  102. message="抓取第{}页时候出现错误, 报错信息是{}".format(i + 1, e),
  103. )
  104. def process_video_obj(self, video_obj):
  105. trace_id = self.platform + str(uuid.uuid1())
  106. our_user = random.choice(self.user_list)
  107. item = VideoItem()
  108. item.add_video_info("user_id", our_user["uid"])
  109. item.add_video_info("user_name", our_user["nick_name"])
  110. item.add_video_info("video_id", video_obj["nid"])
  111. item.add_video_info("video_title", video_obj["title"])
  112. item.add_video_info("publish_time_stamp", int(video_obj['update_time']))
  113. item.add_video_info("video_url", video_obj["video_url"])
  114. item.add_video_info("cover_url", video_obj["video_cover"])
  115. item.add_video_info("out_video_id", video_obj["nid"])
  116. item.add_video_info("platform", self.platform)
  117. item.add_video_info("strategy", self.mode)
  118. item.add_video_info("session", "{}-{}".format(self.platform, int(time.time())))
  119. mq_obj = item.produce_item()
  120. pipeline = PiaoQuanPipeline(
  121. platform=self.platform,
  122. mode=self.mode,
  123. rule_dict=self.rule_dict,
  124. env=self.env,
  125. item=mq_obj,
  126. trace_id=trace_id,
  127. )
  128. if pipeline.process_item():
  129. self.download_cnt += 1
  130. self.mq.send_msg(mq_obj)
  131. AliyunLogger.logging(
  132. code="1002",
  133. platform=self.platform,
  134. mode=self.mode,
  135. env=self.env,
  136. message="成功发送至 ETL",
  137. data=mq_obj,
  138. )
  139. if self.download_cnt >= int(
  140. self.rule_dict.get("videos_cnt", {}).get("min", 200)
  141. ):
  142. self.limit_flag = True
  143. if __name__ == '__main__':
  144. S = TFuQiWangRecommend(
  145. platform="jingdianfuqiwang",
  146. mode="recommend",
  147. env="dev",
  148. rule_dict={},
  149. user_list=[{'nick_name': "Ivring", 'uid': "1997"}, {'nick_name': "paul", 'uid': "1998"}]
  150. )
  151. S.get_video_list()