video_recall.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253
  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. import gevent
  8. log_ = Log()
  9. config_ = set_config()
  10. class PoolRecall(object):
  11. """召回"""
  12. def __init__(self, app_type, mid='', uid='', ab_code=''):
  13. """
  14. 初始化
  15. :param app_type: 产品标识 type-int
  16. :param mid: mid type-string
  17. :param uid: uid type-string
  18. :param ab_code: ab_code type-int
  19. """
  20. self.app_type = app_type
  21. self.mid = mid
  22. self.uid = uid
  23. self.ab_code = ab_code
  24. self.redis_helper = RedisHelper()
  25. def rov_pool_recall(self, size=10):
  26. """从ROV召回池中获取视频"""
  27. log_.info('====== rov pool recall')
  28. # 获取相关redis key, 用户上一次在rov召回池对应的位置
  29. rov_pool_key, last_rov_recall_key, idx = self.get_video_last_idx()
  30. if not rov_pool_key:
  31. log_.info('ROV召回池中无视频')
  32. return []
  33. rov_pool_recall_result = []
  34. # 每次获取的视频数
  35. get_size = size * 2
  36. # 记录获取频次
  37. freq = 0
  38. while len(rov_pool_recall_result) < size:
  39. freq += 1
  40. if freq > config_.MAX_FREQ_FROM_ROV_POOL:
  41. break
  42. # 获取数据
  43. st_get = time.time()
  44. data = self.redis_helper.get_data_zset_with_index(key_name=rov_pool_key,
  45. start=idx, end=idx + get_size - 1,
  46. with_scores=True)
  47. et_get = time.time()
  48. log_.info('get data from rov pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  49. freq, data, (et_get - st_get) * 1000))
  50. if not data:
  51. log_.info('ROV召回池中的视频已取完')
  52. break
  53. # 获取视频id,并转换类型为int,并存储为key-value{videoId: score}
  54. video_ids = []
  55. video_score = {}
  56. for value in data:
  57. video_ids.append(eval(value[0]))
  58. video_score[eval(value[0])] = value[1]
  59. # 过滤
  60. debug_tm_b = time.time()
  61. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  62. ge = gevent.spawn(filter_.filter_videos)
  63. ge.join()
  64. filtered_result = ge.get()
  65. #filtered_result = filter_.filter_videos()
  66. debug_tm_e = time.time()
  67. if filtered_result:
  68. # 添加视频源参数 pushFrom, abCode
  69. temp_result = [{'videoId': item, 'rovScore': video_score[item],
  70. 'pushFrom': 'recall_pool', 'abCode': self.ab_code}
  71. for item in filtered_result]
  72. rov_pool_recall_result.extend(temp_result)
  73. else:
  74. # 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
  75. self.redis_helper.set_data_to_redis(key_name=last_rov_recall_key, value=data[-1][0])
  76. idx += get_size
  77. return rov_pool_recall_result[:size]
  78. def flow_pool_recall(self, size=10):
  79. """从流量池中获取视频"""
  80. log_.info('====== flow pool recall')
  81. flow_pool_key = self.get_pool_redis_key('flow')
  82. flow_pool_recall_result = []
  83. flow_pool_recall_videos = []
  84. # 每次获取的视频数
  85. get_size = size * 3
  86. # 记录获取频次
  87. freq = 0
  88. idx = 0
  89. while len(flow_pool_recall_result) < size:
  90. freq += 1
  91. # 获取数据
  92. st_get = time.time()
  93. data = self.redis_helper.get_data_zset_with_index(key_name=flow_pool_key,
  94. start=idx, end=idx + get_size - 1,
  95. with_scores=True)
  96. et_get = time.time()
  97. log_.info('get data from flow pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  98. freq, data, (et_get - st_get) * 1000))
  99. if not data:
  100. log_.info('流量池中的视频已取完')
  101. break
  102. # 将video_id 与 flow_pool, score做mapping整理
  103. video_ids = []
  104. video_mapping = {}
  105. video_score = {}
  106. for value in data:
  107. video_id, flow_pool = value[0].split('-')
  108. video_id = eval(video_id)
  109. if video_id not in video_ids:
  110. video_ids.append(video_id)
  111. video_score[video_id] = value[1]
  112. if video_id not in video_mapping:
  113. video_mapping[video_id] = [flow_pool]
  114. else:
  115. video_mapping[video_id].append(flow_pool)
  116. # 过滤
  117. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  118. ge = gevent.spawn(filter_.filter_videos)
  119. ge.join()
  120. filtered_result = ge.get()
  121. #filtered_result = filter_.filter_videos()
  122. # 检查可分发数
  123. if filtered_result:
  124. st_check = time.time()
  125. ge = gevent.spawn(self.check_video_counts, video_ids=filtered_result, flow_pool_mapping=video_mapping)
  126. ge.join()
  127. check_result = ge.get()
  128. #check_result, error_flag = self.check_video_counts(video_ids=filtered_result,
  129. flow_pool_mapping=video_mapping)
  130. # 判断错误标记, True为错误
  131. if error_flag:
  132. # 结束流量池召回
  133. break
  134. for item in check_result:
  135. if item[0] not in flow_pool_recall_videos:
  136. # 取其中一个 flow_pool 作为召回结果
  137. # 添加视频源参数 pushFrom, abCode
  138. flow_pool_recall_result.append(
  139. {'videoId': item[0], 'flowPool': item[1], 'distributeCount': item[2],
  140. 'rovScore': video_score[item[0]], 'pushFrom': 'flow_pool', 'abCode': self.ab_code}
  141. )
  142. flow_pool_recall_videos.append(item[0])
  143. et_check = time.time()
  144. log_.info('check result: result = {}, execute time = {}ms'.format(
  145. check_result, (et_check - st_check) * 1000))
  146. idx += get_size
  147. return flow_pool_recall_result[:size]
  148. def check_video_counts(self, video_ids, flow_pool_mapping):
  149. """
  150. 检查视频剩余可分发数
  151. :param video_ids: 视频id type-list
  152. :param flow_pool_mapping: 视频id-流量池标记mapping, type-dict
  153. :return: check_result, error_flag
  154. """
  155. flow_pool_key = self.get_pool_redis_key('flow')
  156. videos = []
  157. for video_id in video_ids:
  158. for flow_pool in flow_pool_mapping[video_id]:
  159. videos.append({'videoId': video_id, 'flowPool': flow_pool})
  160. view_count_result, error_flag = get_videos_remain_view_count(app_type=self.app_type, videos=videos)
  161. log_.info('error_flag = {}, view_count_result = {}'.format(error_flag, view_count_result))
  162. # 判断返回的错误标记,True为错误
  163. if error_flag:
  164. # 从流量召回池移除视频videos
  165. for item in videos:
  166. value = '{}-{}'.format(item['videoId'], item['flowPool'])
  167. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  168. check_result = []
  169. for item in view_count_result:
  170. if item[2] > 0:
  171. # viewCount > 0
  172. # 判断本地分发数
  173. cur_count = get_videos_local_distribute_count(video_id=item[0], flow_pool=item[1])
  174. # 无记录 或 cur_count > 0
  175. if cur_count is None or cur_count > 0:
  176. check_result.append(item)
  177. # cur_count <= 0,从流量召回池移除
  178. else:
  179. value = '{}-{}'.format(item[0], item[1])
  180. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  181. else:
  182. # viewCount <= 0
  183. # 从流量召回池移除
  184. value = '{}-{}'.format(item[0], item[1])
  185. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  186. return check_result, error_flag
  187. def get_pool_redis_key(self, pool_type):
  188. """
  189. 拼接key
  190. :param pool_type: type-string {'rov': rov召回池, 'flow': 流量池}
  191. :return: key_name
  192. """
  193. if pool_type == 'rov':
  194. # appType = 6
  195. if self.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, self.app_type, redis_date)
  200. if self.redis_helper.key_exists(key_name):
  201. return key_name, redis_date
  202. else:
  203. key_name = '{}{}.{}'.format(config_.RECALL_KEY_NAME_PREFIX_APP_TYPE, self.app_type, redis_date - 1)
  204. return key_name, redis_date - 1
  205. # 其他
  206. else:
  207. # 判断热度列表是否更新,未更新则使用前一天的热度列表
  208. key_name = config_.RECALL_KEY_NAME_PREFIX + time.strftime('%Y%m%d')
  209. if self.redis_helper.key_exists(key_name):
  210. redis_date = date.today().strftime('%Y%m%d')
  211. else:
  212. redis_date = (date.today() - timedelta(days=1)).strftime('%Y%m%d')
  213. key_name = config_.RECALL_KEY_NAME_PREFIX + redis_date
  214. # if not self.redis_helper.key_exists(key_name):
  215. # return None, None
  216. return key_name, redis_date
  217. elif pool_type == 'flow':
  218. return config_.FLOW_POOL_KEY_NAME_PREFIX + str(self.app_type)
  219. else:
  220. log_.error('pool type error')
  221. return None, None
  222. def get_video_last_idx(self):
  223. """获取用户上一次在rov召回池对应的位置"""
  224. rov_pool_key, redis_date = self.get_pool_redis_key('rov')
  225. if not rov_pool_key:
  226. return None, None, None
  227. last_rov_recall_key = config_.LAST_VIDEO_FROM_ROV_POOL_PREFIX + '{}.{}.{}'.format(
  228. self.app_type, self.mid, redis_date)
  229. value = self.redis_helper.get_data_from_redis(last_rov_recall_key)
  230. if value:
  231. idx = self.redis_helper.get_index_with_data(rov_pool_key, value)
  232. if not idx:
  233. idx = 0
  234. else:
  235. idx += 1
  236. else:
  237. idx = 0
  238. return rov_pool_key, last_rov_recall_key, idx