video_recall.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243
  1. import time
  2. from datetime import date, timedelta, datetime
  3. from log import Log
  4. from db_helper import RedisHelper
  5. from config import set_config
  6. from utils import FilterVideos, get_videos_remain_view_count, get_videos_local_distribute_count
  7. log_ = Log()
  8. config_ = set_config()
  9. class PoolRecall(object):
  10. """召回"""
  11. def __init__(self, app_type, mid='', uid='', ab_code=''):
  12. """
  13. 初始化
  14. :param app_type: 产品标识 type-int
  15. :param mid: mid type-string
  16. :param uid: uid type-string
  17. :param ab_code: ab_code type-int
  18. """
  19. self.app_type = app_type
  20. self.mid = mid
  21. self.uid = uid
  22. self.ab_code = ab_code
  23. self.redis_helper = RedisHelper()
  24. def rov_pool_recall(self, size=10):
  25. """从ROV召回池中获取视频"""
  26. log_.info('====== rov pool recall')
  27. # 获取相关redis key, 用户上一次在rov召回池对应的位置
  28. rov_pool_key, last_rov_recall_key, idx = self.get_video_last_idx()
  29. if not rov_pool_key:
  30. log_.info('ROV召回池中无视频')
  31. return []
  32. rov_pool_recall_result = []
  33. # 每次获取的视频数
  34. get_size = size * 2
  35. # 记录获取频次
  36. freq = 0
  37. while len(rov_pool_recall_result) < size:
  38. freq += 1
  39. if freq > config_.MAX_FREQ_FROM_ROV_POOL:
  40. break
  41. # 获取数据
  42. st_get = time.time()
  43. data = self.redis_helper.get_data_zset_with_index(key_name=rov_pool_key,
  44. start=idx, end=idx + get_size - 1,
  45. with_scores=True)
  46. et_get = time.time()
  47. log_.info('get data from rov pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  48. freq, data, (et_get - st_get) * 1000))
  49. if not data:
  50. log_.info('ROV召回池中的视频已取完')
  51. break
  52. # 获取视频id,并转换类型为int,并存储为key-value{videoId: score}
  53. video_ids = []
  54. video_score = {}
  55. for value in data:
  56. video_ids.append(eval(value[0]))
  57. video_score[eval(value[0])] = value[1]
  58. # 过滤
  59. debug_tm_b = time.time()
  60. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  61. filtered_result = filter_.filter_videos()
  62. debug_tm_e = time.time()
  63. if filtered_result:
  64. # 添加视频源参数 pushFrom, abCode
  65. temp_result = [{'videoId': item, 'rovScore': video_score[item],
  66. 'pushFrom': 'recall_pool', 'abCode': self.ab_code}
  67. for item in filtered_result]
  68. rov_pool_recall_result.extend(temp_result)
  69. else:
  70. # 将此次获取的末位视频id同步刷新到Redis中,方便下次快速定位到召回位置,过期时间为1天
  71. self.redis_helper.set_data_to_redis(key_name=last_rov_recall_key, value=data[-1][0])
  72. idx += get_size
  73. return rov_pool_recall_result[:size]
  74. def flow_pool_recall(self, size=10):
  75. """从流量池中获取视频"""
  76. log_.info('====== flow pool recall')
  77. flow_pool_key = self.get_pool_redis_key('flow')
  78. flow_pool_recall_result = []
  79. flow_pool_recall_videos = []
  80. # 每次获取的视频数
  81. get_size = size * 3
  82. # 记录获取频次
  83. freq = 0
  84. idx = 0
  85. while len(flow_pool_recall_result) < size:
  86. freq += 1
  87. # 获取数据
  88. st_get = time.time()
  89. data = self.redis_helper.get_data_zset_with_index(key_name=flow_pool_key,
  90. start=idx, end=idx + get_size - 1,
  91. with_scores=True)
  92. et_get = time.time()
  93. log_.info('get data from flow pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  94. freq, data, (et_get - st_get) * 1000))
  95. if not data:
  96. log_.info('流量池中的视频已取完')
  97. break
  98. # 将video_id 与 flow_pool, score做mapping整理
  99. video_ids = []
  100. video_mapping = {}
  101. video_score = {}
  102. for value in data:
  103. video_id, flow_pool = value[0].split('-')
  104. video_id = eval(video_id)
  105. if video_id not in video_ids:
  106. video_ids.append(video_id)
  107. video_score[video_id] = value[1]
  108. if video_id not in video_mapping:
  109. video_mapping[video_id] = [flow_pool]
  110. else:
  111. video_mapping[video_id].append(flow_pool)
  112. # 过滤
  113. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  114. filtered_result = filter_.filter_videos()
  115. # 检查可分发数
  116. if filtered_result:
  117. st_check = time.time()
  118. check_result, error_flag = self.check_video_counts(video_ids=filtered_result,
  119. flow_pool_mapping=video_mapping)
  120. # 判断错误标记, True为错误
  121. if error_flag:
  122. # 结束流量池召回
  123. break
  124. for item in check_result:
  125. if item[0] not in flow_pool_recall_videos:
  126. # 取其中一个 flow_pool 作为召回结果
  127. # 添加视频源参数 pushFrom, abCode
  128. flow_pool_recall_result.append(
  129. {'videoId': item[0], 'flowPool': item[1], 'distributeCount': item[2],
  130. 'rovScore': video_score[item[0]], 'pushFrom': 'flow_pool', 'abCode': self.ab_code}
  131. )
  132. flow_pool_recall_videos.append(item[0])
  133. et_check = time.time()
  134. log_.info('check result: result = {}, execute time = {}ms'.format(
  135. check_result, (et_check - st_check) * 1000))
  136. idx += get_size
  137. return flow_pool_recall_result[:size]
  138. def check_video_counts(self, video_ids, flow_pool_mapping):
  139. """
  140. 检查视频剩余可分发数
  141. :param video_ids: 视频id type-list
  142. :param flow_pool_mapping: 视频id-流量池标记mapping, type-dict
  143. :return: check_result, error_flag
  144. """
  145. flow_pool_key = self.get_pool_redis_key('flow')
  146. videos = []
  147. for video_id in video_ids:
  148. for flow_pool in flow_pool_mapping[video_id]:
  149. videos.append({'videoId': video_id, 'flowPool': flow_pool})
  150. view_count_result, error_flag = get_videos_remain_view_count(app_type=self.app_type, videos=videos)
  151. log_.info('error_flag = {}, view_count_result = {}'.format(error_flag, view_count_result))
  152. # 判断返回的错误标记,True为错误
  153. if error_flag:
  154. # 从流量召回池移除视频videos
  155. for item in videos:
  156. value = '{}-{}'.format(item['videoId'], item['flowPool'])
  157. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  158. check_result = []
  159. for item in view_count_result:
  160. if item[2] > 0:
  161. # viewCount > 0
  162. # 判断本地分发数
  163. cur_count = get_videos_local_distribute_count(video_id=item[0], flow_pool=item[1])
  164. # 无记录 或 cur_count > 0
  165. if cur_count is None or cur_count > 0:
  166. check_result.append(item)
  167. # cur_count <= 0,从流量召回池移除
  168. else:
  169. value = '{}-{}'.format(item[0], item[1])
  170. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  171. else:
  172. # viewCount <= 0
  173. # 从流量召回池移除
  174. value = '{}-{}'.format(item[0], item[1])
  175. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  176. return check_result, error_flag
  177. def get_pool_redis_key(self, pool_type):
  178. """
  179. 拼接key
  180. :param pool_type: type-string {'rov': rov召回池, 'flow': 流量池}
  181. :return: key_name
  182. """
  183. if pool_type == 'rov':
  184. # appType = 6
  185. if self.app_type == config_.APP_TYPE['SHORT_VIDEO']:
  186. # 获取当前所在小时
  187. redis_date = datetime.now().hour
  188. # 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
  189. key_name = '{}{}.{}'.format(config_.RECALL_KEY_NAME_PREFIX_APP_TYPE, self.app_type, redis_date)
  190. if self.redis_helper.key_exists(key_name):
  191. return key_name, redis_date
  192. else:
  193. key_name = '{}{}.{}'.format(config_.RECALL_KEY_NAME_PREFIX_APP_TYPE, self.app_type, redis_date - 1)
  194. return key_name, redis_date - 1
  195. # 其他
  196. else:
  197. # 判断热度列表是否更新,未更新则使用前一天的热度列表
  198. key_name = config_.RECALL_KEY_NAME_PREFIX + time.strftime('%Y%m%d')
  199. if self.redis_helper.key_exists(key_name):
  200. redis_date = date.today().strftime('%Y%m%d')
  201. else:
  202. redis_date = (date.today() - timedelta(days=1)).strftime('%Y%m%d')
  203. key_name = config_.RECALL_KEY_NAME_PREFIX + redis_date
  204. # if not self.redis_helper.key_exists(key_name):
  205. # return None, None
  206. return key_name, redis_date
  207. elif pool_type == 'flow':
  208. return config_.FLOW_POOL_KEY_NAME_PREFIX + str(self.app_type)
  209. else:
  210. log_.error('pool type error')
  211. return None, None
  212. def get_video_last_idx(self):
  213. """获取用户上一次在rov召回池对应的位置"""
  214. rov_pool_key, redis_date = self.get_pool_redis_key('rov')
  215. if not rov_pool_key:
  216. return None, None, None
  217. last_rov_recall_key = config_.LAST_VIDEO_FROM_ROV_POOL_PREFIX + '{}.{}.{}'.format(
  218. self.app_type, self.mid, redis_date)
  219. value = self.redis_helper.get_data_from_redis(last_rov_recall_key)
  220. if value:
  221. idx = self.redis_helper.get_index_with_data(rov_pool_key, value)
  222. if not idx:
  223. idx = 0
  224. else:
  225. idx += 1
  226. else:
  227. idx = 0
  228. return rov_pool_key, last_rov_recall_key, idx