video_recall.py 7.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187
  1. import time
  2. from datetime import date, timedelta
  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
  7. log_ = Log()
  8. config_ = set_config()
  9. class PoolRecall(object):
  10. """召回"""
  11. def __init__(self, app_type, mid, uid):
  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. """
  18. self.app_type = app_type
  19. self.mid = mid
  20. self.uid = uid
  21. self.redis_helper = RedisHelper()
  22. def rov_pool_recall(self, size=10):
  23. """从ROV召回池中获取视频"""
  24. log_.info('====== rov pool recall')
  25. # 获取相关redis key, 用户上一次在rov召回池对应的位置
  26. rov_pool_key, last_rov_recall_key, idx = self.get_video_last_idx()
  27. if not rov_pool_key:
  28. log_.info('ROV召回池中无视频')
  29. return []
  30. rov_pool_recall_result = []
  31. # 每次获取的视频数
  32. get_size = size * 2
  33. # 记录获取频次
  34. freq = 0
  35. while len(rov_pool_recall_result) < size:
  36. freq += 1
  37. if freq > config_.MAX_FREQ_FROM_ROV_POOL:
  38. break
  39. # 获取数据
  40. st_get = time.time()
  41. data = self.redis_helper.get_data_zset_with_index(key_name=rov_pool_key,
  42. start=idx, end=idx + get_size - 1,
  43. with_scores=True)
  44. et_get = time.time()
  45. log_.info('get data from rov pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  46. freq, data, (et_get - st_get) * 1000))
  47. if not data:
  48. log_.info('ROV召回池中的视频已取完')
  49. break
  50. # 获取视频id,并转换类型为int,并存储为key-value{videoId: score}
  51. video_ids = []
  52. video_score = {}
  53. for value in data:
  54. video_ids.append(eval(value[0]))
  55. video_score[eval(value[0])] = value[1]
  56. # 过滤
  57. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  58. filtered_result = filter_.filter_videos()
  59. if filtered_result:
  60. temp_result = [{'videoId': item, 'rovScore': video_score[item]} for item in filtered_result]
  61. rov_pool_recall_result.extend(temp_result)
  62. else:
  63. # 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
  64. self.redis_helper.set_data_to_redis(key_name=last_rov_recall_key, value=data[-1])
  65. idx += get_size
  66. return rov_pool_recall_result[:size]
  67. def flow_pool_recall(self, size=10):
  68. """从流量池中获取视频"""
  69. log_.info('====== flow pool recall')
  70. flow_pool_key = config_.FLOW_POOL_KEY_NAME
  71. flow_pool_recall_result = []
  72. flow_pool_recall_videos = []
  73. # 每次获取的视频数
  74. get_size = size * 3
  75. # 记录获取频次
  76. freq = 0
  77. idx = 0
  78. while len(flow_pool_recall_result) < size:
  79. # 获取数据
  80. st_get = time.time()
  81. data = self.redis_helper.get_data_zset_with_index(key_name=flow_pool_key,
  82. start=idx, end=idx + get_size - 1,
  83. with_scores=True)
  84. et_get = time.time()
  85. log_.info('get data from flow pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  86. freq, data, (et_get - st_get) * 1000))
  87. if not data:
  88. log_.info('流量池中的视频已取完')
  89. break
  90. # 将video_id 与 flow_pool, score做mapping整理
  91. video_ids = []
  92. video_mapping = {}
  93. video_score = {}
  94. for value in data:
  95. video_id, flow_pool = value[0].split('-')
  96. video_id = eval(video_id)
  97. if video_id not in video_ids:
  98. video_ids.append(video_id)
  99. video_score[video_id] = value[1]
  100. if video_id not in video_mapping:
  101. video_mapping[video_id] = [flow_pool]
  102. else:
  103. video_mapping[video_id].append(flow_pool)
  104. # 过滤
  105. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  106. filtered_result = filter_.filter_videos()
  107. # 检查可分发数
  108. if filtered_result:
  109. st_check = time.time()
  110. check_result = self.check_video_counts(video_ids=filtered_result, flow_pool_mapping=video_mapping)
  111. for item in check_result:
  112. if item[0] not in flow_pool_recall_videos:
  113. # 取其中一个 flow_pool 作为召回结果
  114. flow_pool_recall_result.append(
  115. {'videoId': item[0], 'flowPool': item[1], 'rovScore': video_score[item[0]]}
  116. )
  117. flow_pool_recall_videos.append(item[0])
  118. et_check = time.time()
  119. log_.info('check result: result = {}, execute time = {}ms'.format(
  120. check_result, (et_check - st_check) * 1000))
  121. idx += get_size
  122. return flow_pool_recall_result[:size]
  123. def check_video_counts(self, video_ids, flow_pool_mapping):
  124. """
  125. 检查视频剩余可分发数
  126. :param video_ids: 视频id type-list
  127. :param flow_pool_mapping: 视频id-流量池标记mapping, type-dict
  128. :return:
  129. """
  130. videos = []
  131. for video_id in video_ids:
  132. for flow_pool in flow_pool_mapping[video_id]:
  133. videos.append({'videoId': video_id, 'flowPool': flow_pool})
  134. check_result = get_videos_remain_view_count(app_type=self.app_type, videos=videos)
  135. if not check_result:
  136. return None
  137. return check_result
  138. def get_pool_redis_key(self, pool_type):
  139. """
  140. 拼接key
  141. :param pool_type: type-string {'rov': rov召回池, 'flow': 流量池}
  142. :return: key_name
  143. """
  144. if pool_type == 'rov':
  145. # 判断热度列表是否更新,未更新则使用前一天的热度列表
  146. key_name = config_.RECALL_KEY_NAME_PREFIX + time.strftime('%Y%m%d')
  147. if self.redis_helper.key_exists(key_name):
  148. redis_date = date.today().strftime('%Y%m%d')
  149. else:
  150. redis_date = (date.today() - timedelta(days=1)).strftime('%Y%m%d')
  151. key_name = config_.RECALL_KEY_NAME_PREFIX + redis_date
  152. if not self.redis_helper.key_exists(key_name):
  153. return None, None
  154. return key_name, redis_date
  155. elif pool_type == 'flow':
  156. return config_.FLOW_POOL_KEY_NAME
  157. else:
  158. log_.error('pool type error')
  159. return None, None
  160. def get_video_last_idx(self):
  161. """获取用户上一次在rov召回池对应的位置"""
  162. rov_pool_key, redis_date = self.get_pool_redis_key('rov')
  163. if not rov_pool_key:
  164. return None, None, None
  165. last_rov_recall_key = config_.LAST_VIDEO_FROM_ROV_POOL_PREFIX + '{}.{}'.format(self.mid, redis_date)
  166. value = self.redis_helper.get_data_from_redis(last_rov_recall_key)
  167. if value:
  168. idx = self.redis_helper.get_index_with_data(rov_pool_key, value)
  169. if not idx:
  170. idx = 0
  171. else:
  172. idx += 1
  173. else:
  174. idx = 0
  175. return rov_pool_key, last_rov_recall_key, idx