videos_filter.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272
  1. import time
  2. import json
  3. import traceback
  4. from datetime import date, timedelta, datetime
  5. from utils import filter_video_status, send_msg_to_feishu
  6. from db_helper import RedisHelper
  7. from config import set_config
  8. from log import Log
  9. config_, env = set_config()
  10. log_ = Log()
  11. def filter_relevant_videos():
  12. """运营强插相关推荐视频过滤"""
  13. log_.info("relevant videos with op filter start...")
  14. # 读取需要过滤的头部视频id
  15. redis_helper = RedisHelper()
  16. head_videos = redis_helper.get_data_from_set(key_name=config_.RELEVANT_TOP_VIDEOS_KEY_NAME)
  17. if head_videos is None or len(head_videos) == 0:
  18. return
  19. # 过滤
  20. remove_head_vids = []
  21. for head_vid in head_videos:
  22. key_name = '{}{}'.format(config_.RELEVANT_VIDEOS_WITH_OP_KEY_NAME, head_vid)
  23. # 头部视频 对应的key不存在时,将head_vid移除对应redis
  24. if not redis_helper.key_exists(key_name=key_name):
  25. remove_head_vids.append(head_vid)
  26. log_.info('head_vid = {} relevant redis key not exist!'.format(head_vid))
  27. continue
  28. # 获取头部视频对应的相关视频
  29. relevant_videos = redis_helper.get_data_from_redis(key_name=key_name)
  30. # 该视频没有指定的相关性视频,将head_vid移除对应redis
  31. if relevant_videos is None:
  32. remove_head_vids.append(head_vid)
  33. log_.info('head_vid = {} not have relevant videos!'.format(head_vid))
  34. continue
  35. # 过滤
  36. relevant_videos = json.loads(relevant_videos)
  37. relevant_video_ids = [int(item['recommend_vid']) for item in relevant_videos]
  38. filtered_videos = filter_video_status(video_ids=relevant_video_ids)
  39. # 保留可推荐 且生效中 的视频
  40. relevant_videos_new = [
  41. item for item in relevant_videos
  42. if int(item['recommend_vid']) in filtered_videos and int(item['finish_time'] <= int(time.time()))
  43. ]
  44. # 过滤后没有符合的视频,将head_vid移除对应redis,删除对应的相关推荐的key
  45. if len(relevant_videos_new) == 0:
  46. remove_head_vids.append(head_vid)
  47. redis_helper.del_keys(key_name=key_name)
  48. log_.info('head_vid = {} filtered finished! new relevant videos count = {}'.format(
  49. head_vid, len(relevant_videos_new)))
  50. continue
  51. # 重新写入redis
  52. # 以最晚结束的视频的结束时间 - 当前时间 + 5s 作为key的过期时间
  53. finish_time_list = [item['finish_time'] for item in relevant_videos_new]
  54. expire_time = max(finish_time_list) - int(time.time()) + 5
  55. # 存入redis
  56. redis_helper.set_data_to_redis(key_name=key_name,
  57. value=json.dumps(relevant_videos_new),
  58. expire_time=expire_time)
  59. log_.info('head_vid = {} filtered finished! new relevant videos count = {}'.format(
  60. head_vid, len(relevant_videos_new)))
  61. # 将需要移除的头部视频id进行移除
  62. redis_helper.remove_value_from_set(key_name=config_.RELEVANT_TOP_VIDEOS_KEY_NAME, values=tuple(remove_head_vids))
  63. log_.info('head videos remove finished! remove_head_vids = {}'.format(remove_head_vids))
  64. log_.info("relevant videos with op filter end!")
  65. def filter_rov_pool(app_type=None):
  66. """ROV召回池视频过滤"""
  67. log_.info("rov recall pool filter start ...")
  68. # 拼接redis-key
  69. if app_type is None:
  70. key_name, _ = get_pool_redis_key(pool_type='rov')
  71. else:
  72. log_.info("appType = {}".format(app_type))
  73. key_name, _ = get_pool_redis_key(pool_type='rov', app_type=app_type)
  74. # 获取视频
  75. redis_helper = RedisHelper()
  76. data = redis_helper.get_data_zset_with_index(key_name=key_name, start=0, end=-1)
  77. if data is None:
  78. log_.info("data is None")
  79. log_.info("rov recall pool filter end!")
  80. return
  81. # 过滤
  82. video_ids = [int(video_id) for video_id in data]
  83. filtered_result = filter_video_status(video_ids=video_ids)
  84. # 求差集,获取需要过滤掉的视频,并从redis中移除
  85. filter_videos = set(video_ids) - set(filtered_result)
  86. log_.info("video_ids size = {}, filtered size = {}, filter sizer = {}".format(len(video_ids),
  87. len(filtered_result),
  88. len(filter_videos)))
  89. if len(filter_videos) == 0:
  90. log_.info("rov recall pool filter end!")
  91. return
  92. redis_helper.remove_value_from_zset(key_name=key_name, value=list(filter_videos))
  93. log_.info("rov recall pool filter end!")
  94. def filter_flow_pool():
  95. """流量池视频过滤"""
  96. log_.info("flow pool filter start ...")
  97. for _, app_type in config_.APP_TYPE.items():
  98. log_.info('app_type {} videos filter start...'.format(app_type))
  99. # 拼接redis-key
  100. key_name = get_pool_redis_key(pool_type='flow', app_type=app_type)
  101. # 获取视频
  102. redis_helper = RedisHelper()
  103. data = redis_helper.get_data_zset_with_index(key_name=key_name, start=0, end=-1)
  104. if data is None:
  105. log_.info("data is None")
  106. log_.info("app_type {} videos filter end!".format(app_type))
  107. continue
  108. # videoId与flowPool做mapping
  109. video_ids = []
  110. mapping = {}
  111. for video in data:
  112. video_id, flow_pool = video.split('-')
  113. video_id = int(video_id)
  114. if video_id not in video_ids:
  115. video_ids.append(video_id)
  116. mapping[video_id] = [flow_pool]
  117. else:
  118. mapping[video_id].append(flow_pool)
  119. # 过滤
  120. if len(video_ids) == 0:
  121. log_.info("data size = {}, video_ids size = {}, data = {}".format(len(data), len(video_ids), data))
  122. log_.info("app_type {} videos filter end!".format(app_type))
  123. continue
  124. filtered_result = filter_video_status(video_ids=video_ids)
  125. # 求差集,获取需要过滤掉的视频,并从redis中移除
  126. filter_videos = set(video_ids) - set(filtered_result)
  127. log_.info("data size = {}, video_ids size = {}, filtered size = {}, filter sizer = {}".format(
  128. len(data), len(video_ids), len(filtered_result), len(filter_videos)))
  129. # 移除
  130. if len(filter_videos) == 0:
  131. log_.info("app_type {} videos filter end!".format(app_type))
  132. continue
  133. remove_videos = ['{}-{}'.format(video_id, flow_pool)
  134. for video_id in filter_videos
  135. for flow_pool in mapping[video_id]]
  136. redis_helper.remove_value_from_zset(key_name=key_name, value=remove_videos)
  137. log_.info("app_type {} videos filter end!".format(app_type))
  138. log_.info("flow pool filter end!")
  139. def filter_bottom():
  140. """兜底视频过滤"""
  141. log_.info("bottom videos filter start ...")
  142. # 获取视频
  143. redis_helper = RedisHelper()
  144. data = redis_helper.get_data_zset_with_index(key_name=config_.BOTTOM_KEY_NAME, start=0, end=-1)
  145. if data is None:
  146. log_.info("data is None")
  147. log_.info("bottom videos filter end!")
  148. return
  149. # 过滤
  150. video_ids = [int(video_id) for video_id in data]
  151. filtered_result = filter_video_status(video_ids=video_ids)
  152. # 求差集,获取需要过滤掉的视频,并从redis中移除
  153. filter_videos = set(video_ids) - set(filtered_result)
  154. log_.info("video_ids size = {}, filtered size = {}, filter sizer = {}".format(len(video_ids),
  155. len(filtered_result),
  156. len(filter_videos)))
  157. if len(filter_videos) == 0:
  158. log_.info("bottom videos filter end!")
  159. return
  160. redis_helper.remove_value_from_zset(key_name=config_.BOTTOM_KEY_NAME, value=list(filter_videos))
  161. log_.info("bottom videos filter end!")
  162. def filter_rov_updated():
  163. """修改过ROV的视频过滤"""
  164. log_.info("update rov videos filter start ...")
  165. # 获取视频
  166. redis_helper = RedisHelper()
  167. data = redis_helper.get_data_zset_with_index(key_name=config_.UPDATE_ROV_KEY_NAME, start=0, end=-1)
  168. if data is None:
  169. log_.info("data is None")
  170. log_.info("update rov videos filter end!")
  171. return
  172. # 过滤
  173. video_ids = [int(video_id) for video_id in data]
  174. filtered_result = filter_video_status(video_ids=video_ids)
  175. # 求差集,获取需要过滤掉的视频,并从redis中移除
  176. filter_videos = set(video_ids) - set(filtered_result)
  177. log_.info("video_ids size = {}, filtered size = {}, filter sizer = {}".format(len(video_ids),
  178. len(filtered_result),
  179. len(filter_videos)))
  180. if len(filter_videos) == 0:
  181. log_.info("update rov videos filter end!")
  182. return
  183. redis_helper.remove_value_from_zset(key_name=config_.UPDATE_ROV_KEY_NAME, value=list(filter_videos))
  184. log_.info("update rov videos filter end!")
  185. def get_pool_redis_key(pool_type, app_type=None):
  186. """
  187. 拼接key
  188. :param pool_type: type-string {'rov': rov召回池, 'flow': 流量池}
  189. :param app_type: 产品标识
  190. :return: key_name
  191. """
  192. redis_helper = RedisHelper()
  193. if pool_type == 'rov':
  194. # appType = 6
  195. if app_type == config_.APP_TYPE['SHORT_VIDEO']:
  196. # 获取当前所在小时
  197. redis_date = datetime.now().hour
  198. # 判断热度列表是否更新,未更新则使用前一小时的热度列表
  199. key_name = '{}{}.{}'.format(config_.RECALL_KEY_NAME_PREFIX_APP_TYPE, app_type, redis_date)
  200. if redis_helper.key_exists(key_name):
  201. return key_name, redis_date
  202. else:
  203. if redis_date == 0:
  204. redis_date = 23
  205. else:
  206. redis_date = redis_date - 1
  207. key_name = '{}{}.{}'.format(config_.RECALL_KEY_NAME_PREFIX_APP_TYPE, app_type, redis_date)
  208. return key_name, redis_date
  209. # 其他
  210. else:
  211. # 判断热度列表是否更新,未更新则使用前一天的热度列表
  212. key_name = config_.RECALL_KEY_NAME_PREFIX + time.strftime('%Y%m%d')
  213. if redis_helper.key_exists(key_name):
  214. redis_date = date.today().strftime('%Y%m%d')
  215. else:
  216. redis_date = (date.today() - timedelta(days=1)).strftime('%Y%m%d')
  217. key_name = config_.RECALL_KEY_NAME_PREFIX + redis_date
  218. return key_name, redis_date
  219. elif pool_type == 'flow':
  220. # 流量池
  221. return config_.FLOWPOOL_KEY_NAME_PREFIX + str(app_type)
  222. else:
  223. log_.error('pool type error')
  224. return None, None
  225. def main():
  226. try:
  227. # ROV召回池视频过滤
  228. filter_rov_pool()
  229. # appType = 6,ROV召回池视频过滤
  230. filter_rov_pool(app_type=config_.APP_TYPE['SHORT_VIDEO'])
  231. # 流量池视频过滤
  232. filter_flow_pool()
  233. # 兜底视频过滤
  234. filter_bottom()
  235. # 修改过ROV的视频过滤
  236. filter_rov_updated()
  237. # 运营强插相关推荐视频过滤
  238. filter_relevant_videos()
  239. except Exception as e:
  240. log_.error(traceback.format_exc())
  241. send_msg_to_feishu('{} - 过滤失败 \n {}'.format(config_.ENV_TEXT, traceback.format_exc()))
  242. return
  243. if __name__ == '__main__':
  244. main()