jingdianfuqiwang_recommend_scheduling.py 6.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169
  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. from common.redis_db import get_data, store_data
  13. class TFuQiWangRecommend(object):
  14. def __init__(self, platform, mode, rule_dict, user_list, env):
  15. self.platform = platform
  16. self.mode = mode
  17. self.rule_dict = rule_dict
  18. self.user_list = user_list
  19. self.env = env
  20. self.download_cnt = 0
  21. self.mq = MQ(topic_name="topic_crawler_etl_" + self.env)
  22. self.limit_flag = False
  23. def get_video_list(self):
  24. base_url = "https://jdfqw.wentingyou.cn/index.php/v111/index/index"
  25. headers = {
  26. 'chatkey': 'wx79ef316d416e5da1',
  27. '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',
  28. 'content-type': 'application/x-www-form-urlencoded',
  29. 'xweb_xhr': '1',
  30. 'vision': '1.1.0',
  31. 'token': '',
  32. 'accept': '*/*',
  33. 'referer': 'https://servicewechat.com/wx79ef316d416e5da1/1/page-frame.html',
  34. 'accept-language': 'en-US,en;q=0.9'
  35. }
  36. page_index = 1
  37. while True:
  38. time.sleep(random.randint(1, 10))
  39. try:
  40. if self.limit_flag:
  41. AliyunLogger.logging(
  42. code="2000",
  43. platform=self.platform,
  44. mode=self.mode,
  45. env=self.env,
  46. message="本轮已经抓取到足够的数据,自动退出\t{}".format(self.download_cnt),
  47. )
  48. return
  49. else:
  50. obj = {
  51. "cid": "",
  52. "page": 1,
  53. "is_ads": 1
  54. }
  55. params = {
  56. "parameter": json.dumps(obj)
  57. }
  58. response = requests.get(
  59. url=base_url,
  60. headers=headers,
  61. params=params
  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(page_index + 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. value = get_data(self.platform, video_obj["nid"])
  121. if int(value) == 1:
  122. AliyunLogger.logging(
  123. code="2004",
  124. platform=self.platform,
  125. mode=self.mode,
  126. env=self.env,
  127. message="redis重复视频")
  128. pipeline = PiaoQuanPipeline(
  129. platform=self.platform,
  130. mode=self.mode,
  131. rule_dict=self.rule_dict,
  132. env=self.env,
  133. item=mq_obj,
  134. trace_id=trace_id,
  135. )
  136. if pipeline.process_item():
  137. self.download_cnt += 1
  138. self.mq.send_msg(mq_obj)
  139. store_data(self.platform, video_obj["nid"], False, 12)
  140. time.sleep(60 * random.randint(1, 5))
  141. AliyunLogger.logging(
  142. code="1002",
  143. platform=self.platform,
  144. mode=self.mode,
  145. env=self.env,
  146. message="成功发送至 ETL",
  147. data=mq_obj,
  148. )
  149. if self.download_cnt >= int(
  150. self.rule_dict.get("videos_cnt", {}).get("min", 200)
  151. ):
  152. self.limit_flag = True
  153. if __name__ == '__main__':
  154. S = TFuQiWangRecommend(
  155. platform="jingdianfuqiwang",
  156. mode="recommend",
  157. env="dev",
  158. rule_dict={},
  159. user_list=[{'nick_name': "Ivring", 'uid': "1997"}, {'nick_name': "paul", 'uid': "1998"}]
  160. )
  161. S.get_video_list()