video_recall.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300
  1. import time
  2. import traceback
  3. from datetime import date, timedelta, datetime
  4. from log import Log
  5. from db_helper import RedisHelper
  6. from config import set_config
  7. from utils import FilterVideos, get_videos_remain_view_count, get_videos_local_distribute_count, send_msg_to_feishu
  8. import gevent
  9. log_ = Log()
  10. config_ = set_config()
  11. class PoolRecall(object):
  12. """召回"""
  13. def __init__(self, app_type, mid='', uid='', ab_code=''):
  14. """
  15. 初始化
  16. :param app_type: 产品标识 type-int
  17. :param mid: mid type-string
  18. :param uid: uid type-string
  19. :param ab_code: ab_code type-int
  20. """
  21. self.app_type = app_type
  22. self.mid = mid
  23. self.uid = uid
  24. self.ab_code = ab_code
  25. self.redis_helper = RedisHelper()
  26. def rov_pool_recall(self, size=10):
  27. """从ROV召回池中获取视频"""
  28. log_.info('====== rov pool recall')
  29. # 获取相关redis key, 用户上一次在rov召回池对应的位置
  30. rov_pool_key, last_rov_recall_key, idx = self.get_video_last_idx()
  31. if not rov_pool_key:
  32. log_.info('ROV召回池中无视频')
  33. return []
  34. rov_pool_recall_result = []
  35. # 每次获取的视频数
  36. get_size = size * 2
  37. # 记录获取频次
  38. freq = 0
  39. while len(rov_pool_recall_result) < size:
  40. freq += 1
  41. if freq > config_.MAX_FREQ_FROM_ROV_POOL:
  42. break
  43. # 获取数据
  44. st_get = time.time()
  45. data = self.redis_helper.get_data_zset_with_index(key_name=rov_pool_key,
  46. start=idx, end=idx + get_size - 1,
  47. with_scores=True)
  48. et_get = time.time()
  49. log_.info('get data from rov pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  50. freq, data, (et_get - st_get) * 1000))
  51. if not data:
  52. log_.info('ROV召回池中的视频已取完')
  53. break
  54. # 获取视频id,并转换类型为int,并存储为key-value{videoId: score}
  55. video_ids = []
  56. video_score = {}
  57. for value in data:
  58. video_ids.append(eval(value[0]))
  59. video_score[eval(value[0])] = value[1]
  60. # 过滤
  61. debug_tm_b = time.time()
  62. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  63. ge = gevent.spawn(filter_.filter_videos)
  64. ge.join()
  65. filtered_result = ge.get()
  66. #filtered_result = filter_.filter_videos()
  67. debug_tm_e = time.time()
  68. if filtered_result:
  69. # 添加视频源参数 pushFrom, abCode
  70. temp_result = [{'videoId': item, 'rovScore': video_score[item],
  71. 'pushFrom': 'recall_pool', 'abCode': self.ab_code}
  72. for item in filtered_result]
  73. rov_pool_recall_result.extend(temp_result)
  74. else:
  75. # 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
  76. self.redis_helper.set_data_to_redis(key_name=last_rov_recall_key, value=data[-1][0])
  77. idx += get_size
  78. return rov_pool_recall_result[:size]
  79. def flow_pool_recall(self, size=10):
  80. """从流量池中获取视频"""
  81. log_.info('====== flow pool recall')
  82. flow_pool_key = self.get_pool_redis_key('flow')
  83. flow_pool_recall_result = []
  84. flow_pool_recall_videos = []
  85. # 每次获取的视频数
  86. get_size = size * 3
  87. # 记录获取频次
  88. freq = 0
  89. idx = 0
  90. while len(flow_pool_recall_result) < size:
  91. freq += 1
  92. # 获取数据
  93. st_get = time.time()
  94. data = self.redis_helper.get_data_zset_with_index(key_name=flow_pool_key,
  95. start=idx, end=idx + get_size - 1,
  96. with_scores=True)
  97. et_get = time.time()
  98. log_.info('get data from flow pool redis: freq = {}, data = {}, execute time = {}ms'.format(
  99. freq, data, (et_get - st_get) * 1000))
  100. if not data:
  101. log_.info('流量池中的视频已取完')
  102. break
  103. # 将video_id 与 flow_pool, score做mapping整理
  104. video_ids = []
  105. video_mapping = {}
  106. video_score = {}
  107. for value in data:
  108. video_id, flow_pool = value[0].split('-')
  109. video_id = eval(video_id)
  110. if video_id not in video_ids:
  111. video_ids.append(video_id)
  112. video_score[video_id] = value[1]
  113. if video_id not in video_mapping:
  114. video_mapping[video_id] = [flow_pool]
  115. else:
  116. video_mapping[video_id].append(flow_pool)
  117. # 过滤
  118. filter_ = FilterVideos(app_type=self.app_type, mid=self.mid, uid=self.uid, video_ids=video_ids)
  119. ge = gevent.spawn(filter_.filter_videos)
  120. ge.join()
  121. filtered_result = ge.get()
  122. #filtered_result = filter_.filter_videos()
  123. # 检查可分发数
  124. if filtered_result:
  125. st_check = time.time()
  126. ge = gevent.spawn(self.check_video_counts, video_ids=filtered_result, flow_pool_mapping=video_mapping)
  127. ge.join()
  128. check_result, error_flag = ge.get()
  129. #check_result, error_flag = self.check_video_counts(video_ids=filtered_result, flow_pool_mapping=video_mapping)
  130. for item in check_result:
  131. if item[0] not in flow_pool_recall_videos:
  132. # 取其中一个 flow_pool 作为召回结果
  133. # 添加视频源参数 pushFrom, abCode
  134. flow_pool_recall_result.append(
  135. {'videoId': item[0], 'flowPool': item[1],
  136. 'rovScore': video_score[item[0]], 'pushFrom': 'flow_pool', 'abCode': self.ab_code}
  137. )
  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. # 判断错误标记, True为错误
  147. if error_flag:
  148. # 结束流量池召回
  149. break
  150. idx += get_size
  151. return flow_pool_recall_result[:size]
  152. def check_video_counts(self, video_ids, flow_pool_mapping):
  153. """
  154. 检查视频剩余可分发数
  155. :param video_ids: 视频id type-list
  156. :param flow_pool_mapping: 视频id-流量池标记mapping, type-dict
  157. :return: check_result, error_flag
  158. """
  159. flow_pool_key = self.get_pool_redis_key('flow')
  160. videos = []
  161. check_result = []
  162. for video_id in video_ids:
  163. for flow_pool in flow_pool_mapping[video_id]:
  164. # 判断是否有本地分发记录
  165. cur_count = get_videos_local_distribute_count(video_id=video_id, flow_pool=flow_pool)
  166. # 无记录
  167. if cur_count is None:
  168. videos.append({'videoId': video_id, 'flowPool': flow_pool})
  169. # 本地分发数 cur_count > 0
  170. elif cur_count > 0:
  171. check_result.append((video_id, flow_pool))
  172. # 本地分发数 cur_count <= 0,从流量召回池移除
  173. else:
  174. value = '{}-{}'.format(video_id, flow_pool)
  175. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  176. # 本次视频都有本地记录
  177. if len(video_ids) == 0:
  178. error_flag = False
  179. return check_result, error_flag
  180. # 本地无记录视频,检查实时分发数
  181. view_count_result, error_flag = get_videos_remain_view_count(app_type=self.app_type, videos=videos)
  182. log_.info('error_flag = {}, view_count_result = {}'.format(error_flag, view_count_result))
  183. # 判断返回的错误标记,True为错误
  184. if error_flag:
  185. return check_result, error_flag
  186. # 从流量召回池移除视频videos
  187. # for item in videos:
  188. # value = '{}-{}'.format(item['videoId'], item['flowPool'])
  189. # self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  190. redis_helper = RedisHelper()
  191. for item in view_count_result:
  192. try:
  193. # 接口超时,item[2]可能为None
  194. remain_count = int(item[2])
  195. except Exception as e:
  196. log_.error('remain_count type error...')
  197. log_.error(traceback.format_exc())
  198. continue
  199. if remain_count > 0:
  200. # viewCount > 0
  201. check_result.append(item)
  202. # 将分发数更新到本地记录
  203. key_name = '{}{}.{}'.format(config_.LOCAL_DISTRIBUTE_COUNT_PREFIX, item[0], item[1])
  204. redis_helper.incr_key(key_name=key_name, amount=remain_count, expire_time=5 * 60)
  205. # # 判断本地分发数
  206. # cur_count = get_videos_local_distribute_count(video_id=item[0], flow_pool=item[1])
  207. # # 无记录 或 cur_count > 0
  208. # if cur_count is None or cur_count > 0:
  209. # check_result.append(item)
  210. # # cur_count <= 0,从流量召回池移除
  211. # else:
  212. # value = '{}-{}'.format(item[0], item[1])
  213. # self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  214. else:
  215. # viewCount <= 0
  216. # 从流量召回池移除
  217. value = '{}-{}'.format(item[0], item[1])
  218. self.redis_helper.remove_value_from_zset(key_name=flow_pool_key, value=value)
  219. return check_result, error_flag
  220. def get_pool_redis_key(self, pool_type):
  221. """
  222. 拼接key
  223. :param pool_type: type-string {'rov': rov召回池, 'flow': 流量池}
  224. :return: key_name
  225. """
  226. if pool_type == 'rov':
  227. # appType = 6
  228. if self.app_type == config_.APP_TYPE['SHORT_VIDEO']:
  229. # 获取当前所在小时
  230. redis_date = datetime.now().hour
  231. # 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
  232. key_name = '{}{}.{}'.format(config_.RECALL_KEY_NAME_PREFIX_APP_TYPE, self.app_type, redis_date)
  233. if self.redis_helper.key_exists(key_name):
  234. return key_name, redis_date
  235. else:
  236. key_name = '{}{}.{}'.format(config_.RECALL_KEY_NAME_PREFIX_APP_TYPE, self.app_type, redis_date - 1)
  237. return key_name, redis_date - 1
  238. # 其他
  239. else:
  240. # 判断热度列表是否更新,未更新则使用前一天的热度列表
  241. key_name = config_.RECALL_KEY_NAME_PREFIX + time.strftime('%Y%m%d')
  242. if self.redis_helper.key_exists(key_name):
  243. redis_date = date.today().strftime('%Y%m%d')
  244. else:
  245. redis_date = (date.today() - timedelta(days=1)).strftime('%Y%m%d')
  246. key_name = config_.RECALL_KEY_NAME_PREFIX + redis_date
  247. # if not self.redis_helper.key_exists(key_name):
  248. # return None, None
  249. # 判断当前时间是否晚于rov召回池更新时间 + 1h,发送消息到飞书
  250. now_h = datetime.now().hour
  251. now_m = datetime.now().minute
  252. feishu_text = '今日ROV召回池数据未按时更新,请及时查看解决。'
  253. if now_h == config_.ROV_UPDATE_H + 1 and now_m > config_.ROV_UPDATE_MINUTE:
  254. send_msg_to_feishu(feishu_text)
  255. elif now_h > config_.ROV_UPDATE_H + 1:
  256. send_msg_to_feishu(feishu_text)
  257. return key_name, redis_date
  258. elif pool_type == 'flow':
  259. return config_.FLOW_POOL_KEY_NAME_PREFIX + str(self.app_type)
  260. else:
  261. log_.error('pool type error')
  262. return None, None
  263. def get_video_last_idx(self):
  264. """获取用户上一次在rov召回池对应的位置"""
  265. rov_pool_key, redis_date = self.get_pool_redis_key('rov')
  266. if not rov_pool_key:
  267. return None, None, None
  268. last_rov_recall_key = config_.LAST_VIDEO_FROM_ROV_POOL_PREFIX + '{}.{}.{}'.format(
  269. self.app_type, self.mid, redis_date)
  270. value = self.redis_helper.get_data_from_redis(last_rov_recall_key)
  271. if value:
  272. idx = self.redis_helper.get_index_with_data(rov_pool_key, value)
  273. if not idx:
  274. idx = 0
  275. else:
  276. idx += 1
  277. else:
  278. idx = 0
  279. return rov_pool_key, last_rov_recall_key, idx