download_play_sendtime.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232
  1. # -*- coding: utf-8 -*-
  2. # @Author: wangkun
  3. # @Time: 2022/6/17
  4. import json
  5. import os
  6. import sys
  7. import time
  8. import requests
  9. import urllib3
  10. sys.path.append(os.getcwd())
  11. from main.common import Common
  12. from main.get_feeds import get_feeds
  13. from main.publish import Publish
  14. from main.feishu_lib import Feishu
  15. proxies = {"http": None, "https": None}
  16. class DownloadPlaySendtime:
  17. # 抓取基础规则,写入recommend_feeds表
  18. @staticmethod
  19. def send_time_rule(send_time_width, send_time_height, send_time_duration, send_time_share_cnt):
  20. """
  21. 1.分辨率,宽或者高 >= 720 or == 0
  22. 2.时长 >= 60s
  23. 3.视频播放量 >= 0
  24. """
  25. if int(send_time_width) >= 720 or int(send_time_height) >= 720:
  26. if int(send_time_duration) >= 60:
  27. if int(send_time_share_cnt) > 0:
  28. return True
  29. else:
  30. return False
  31. else:
  32. return False
  33. else:
  34. return False
  35. @classmethod
  36. def download_play_sendtime(cls, env):
  37. """
  38. - 7日内播放大于2万
  39. - 或播放大于15万
  40. :param env: 测试环境:dev;正式环境:prod
  41. :return: 下载并上传视频
  42. """
  43. try:
  44. for i in range(1, len(Feishu.get_values_batch("SdCHOM"))+1):
  45. time.sleep(1)
  46. sendtime_session = Common.get_session()
  47. # Common.logger().info("获取视频info时,session:{}", sendtime_session)
  48. download_video_id = Feishu.get_values_batch("SdCHOM")[i][2]
  49. download_video_title = Feishu.get_values_batch("SdCHOM")[i][3]
  50. url = "https://search.weixin.qq.com/cgi-bin/recwxa/recwxagetonevideoinfo?"
  51. param = {
  52. "session": sendtime_session,
  53. "vid": download_video_id,
  54. "wxaVersion": "3.9.2",
  55. "channelid": "208201",
  56. "scene": "32",
  57. "subscene": "1089",
  58. "model": "iPhone 11<iPhone12,1>14.7.1",
  59. "clientVersion": "8.0.18",
  60. "sharesearchid": "447665862521758270",
  61. "sharesource": "-1"
  62. }
  63. urllib3.disable_warnings()
  64. r = requests.get(url=url, params=param, proxies=proxies, verify=False)
  65. response = json.loads(r.content.decode("utf8"))
  66. if "data" not in response:
  67. Common.logger().error("获取视频info时错误,删除该视频:{}", download_video_title)
  68. # 删除行或列,可选 ROWS、COLUMNS
  69. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  70. else:
  71. data = response["data"]
  72. v_duration = data["duration"]
  73. v_play_cnt_sendtime = data["played_cnt"]
  74. v_comment_cnt = data["comment_cnt"]
  75. v_liked_cnt = data["liked_cnt"]
  76. v_shared_cnt = data["shared_cnt"]
  77. v_width = data["width"]
  78. v_height = data["height"]
  79. v_resolution = str(v_width) + "*" + str(v_height)
  80. v_send_date = data["upload_time"]
  81. v_username = data["user_info"]["nickname"].strip().replace("\n", "")
  82. v_user_id = data["openid"]
  83. v_user_cover = data["user_info"]["headimg_url"]
  84. v_video_cover = data["cover_url"]
  85. if "items" not in data["play_info"]:
  86. if len(data["play_info"]) > 2:
  87. download_url_sendtime = data["play_info"][2]["play_url"]
  88. else:
  89. download_url_sendtime = data["play_info"][0]["play_url"]
  90. else:
  91. if len(data["play_info"]["items"]) > 2:
  92. download_url_sendtime = data["play_info"]["items"][2]["play_url"]
  93. else:
  94. download_url_sendtime = data["play_info"]["items"][0]["play_url"]
  95. Common.logger().info("正在判断第{}行,视频:{}", i, download_video_title)
  96. # 判断无效视频
  97. if download_video_id == "" \
  98. or download_video_id is None\
  99. and download_video_title == ""\
  100. or download_video_title is None\
  101. and v_duration == "" \
  102. and v_play_cnt_sendtime == ""\
  103. and v_send_date == ""\
  104. and v_user_cover == "" \
  105. and v_video_cover == ""\
  106. and download_url_sendtime == "":
  107. Common.logger().info("无效视频,删除该视频信息:{}", download_video_title)
  108. # 从云文档删除该视频信息:https://w42nne6hzg.feishu.cn/sheets/shtcngRPoDYAi24x52j2nDuHMih?sheet=SdCHOM
  109. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  110. return
  111. elif v_send_date < 1622476800:
  112. Common.logger().info(
  113. "发布时间小于2021年6月:{},{}", download_video_title, v_send_date)
  114. # 删除行或列,可选 ROWS、COLUMNS
  115. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  116. return
  117. # 抓取基础规则
  118. elif cls.send_time_rule(v_width, v_height, v_duration, v_play_cnt_sendtime) is False:
  119. Common.logger().info("不满足发布时间榜下载规则,删除该视频信息:{}", download_video_title)
  120. # 从云文档删除该视频信息:https://w42nne6hzg.feishu.cn/sheets/shtcngRPoDYAi24x52j2nDuHMih?sheet=SdCHOM
  121. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  122. return
  123. # 不满足规则:发布时间 > 7 天 and 播放量 < 150000
  124. elif int(time.time()) - int(v_send_date) > 604800 and int(v_play_cnt_sendtime) < 150000:
  125. # 从云文档删除该视频信息:https://w42nne6hzg.feishu.cn/sheets/shtcngRPoDYAi24x52j2nDuHMih?sheet=SdCHOM
  126. Common.logger().info("播放量:{} < 150000", int(v_play_cnt_sendtime))
  127. # 删除行或列,可选 ROWS、COLUMNS
  128. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  129. return
  130. # 不满足规则:发布时间 <= 7 天 and 播放量 < 20000
  131. elif int(time.time()) - int(v_send_date) <= 604800 and int(v_play_cnt_sendtime) < 20000:
  132. # 从云文档删除该视频信息:https://w42nne6hzg.feishu.cn/sheets/shtcngRPoDYAi24x52j2nDuHMih?sheet=SdCHOM
  133. Common.logger().info("视频7天播放量:{} < 20000", int(v_play_cnt_sendtime))
  134. # 删除行或列,可选 ROWS、COLUMNS
  135. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  136. return
  137. elif download_video_id in [j for m in Feishu.get_values_batch("20ce0c") for j in m]:
  138. Common.logger().info("视频已下载,删除该视频信息:{}", download_video_title)
  139. # 从云文档删除该视频信息:https://w42nne6hzg.feishu.cn/sheets/shtcngRPoDYAi24x52j2nDuHMih?sheet=SdCHOM
  140. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  141. return
  142. else:
  143. Common.logger().info("开始下载视频:{}", download_video_title)
  144. # 下载封面
  145. Common.download_method("cover", download_video_title, v_video_cover)
  146. # 下载视频
  147. Common.download_method("video", download_video_title, download_url_sendtime)
  148. # 保存视频信息到 "./files/{视频标题}/videoinfo.txt"
  149. with open(r"./videos/" + download_video_title +
  150. "/" + "info.txt", "a", encoding="utf8") as f_a2:
  151. f_a2.write(str(download_video_id) + "\n" +
  152. str(download_video_title) + "\n" +
  153. str(v_duration) + "\n" +
  154. str(v_play_cnt_sendtime) + "\n" +
  155. str(v_comment_cnt) + "\n" +
  156. str(v_liked_cnt) + "\n" +
  157. str(v_shared_cnt) + "\n" +
  158. str(v_resolution) + "\n" +
  159. str(v_send_date) + "\n" +
  160. str(v_username) + "\n" +
  161. str(v_user_cover) + "\n" +
  162. str(download_url_sendtime) + "\n" +
  163. str(v_video_cover) + "\n" +
  164. str(sendtime_session))
  165. Common.logger().info("==========视频信息已保存至info.txt==========")
  166. # 上传该视频
  167. Common.logger().info("开始上传视频:{}", download_video_title)
  168. Publish.upload_and_publish(env, "send_time")
  169. # 保存视频 ID 到云文档:
  170. # https://w42nne6hzg.feishu.cn/sheets/shtcngRPoDYAi24x52j2nDuHMih?sheet=20ce0c
  171. Common.logger().info("保存视频ID至云文档:{}", download_video_title)
  172. # 看一看+ ,视频ID工作表,插入首行
  173. Feishu.insert_columns("20ce0c", "ROWS", 1, 2)
  174. # 看一看+ ,视频ID工作表,首行写入数据
  175. upload_time = int(time.time())
  176. values = [[time.strftime("%Y/%m/%d %H:%M:%S", time.localtime(upload_time)),
  177. "推荐榜",
  178. str(download_video_id),
  179. str(download_video_title),
  180. int(v_play_cnt_sendtime),
  181. v_comment_cnt,
  182. v_liked_cnt,
  183. v_shared_cnt,
  184. v_duration,
  185. v_resolution,
  186. time.strftime("%Y/%m/%d %H:%M:%S", time.localtime(v_send_date)),
  187. str(v_username),
  188. str(v_user_id),
  189. str(v_user_cover),
  190. str(v_video_cover),
  191. str(download_url_sendtime)]]
  192. time.sleep(1)
  193. Feishu.update_values("20ce0c", "A2:Q2", values)
  194. # 从云文档删除该视频信息:https://w42nne6hzg.feishu.cn/sheets/shtcngRPoDYAi24x52j2nDuHMih?sheet=SdCHOM
  195. Common.logger().info("从云文档删除该视频信息:{}", download_video_title)
  196. # 删除行或列,可选 ROWS、COLUMNS
  197. Feishu.dimension_range("SdCHOM", "ROWS", i + 1, i + 1)
  198. return
  199. except Exception as e:
  200. Common.logger().error("获取视频info异常:{}", e)
  201. Feishu.dimension_range("SdCHOM", "ROWS", 2, 2)
  202. # 执行上传及下载
  203. @classmethod
  204. def run_download_play_sendtime(cls):
  205. try:
  206. while True:
  207. if len(Feishu.get_values_batch("SdCHOM")) == 1:
  208. break
  209. else:
  210. cls.download_play_sendtime("prod")
  211. except Exception as e:
  212. Common.logger().error("执行上传及下载异常:{}", e)
  213. if __name__ == "__main__":
  214. download_sendtime = DownloadPlaySendtime()
  215. get_feeds()
  216. download_sendtime.download_play_sendtime("prod")