zhangbo il y a 1 an
Parent
commit
bb97eff79e
2 fichiers modifiés avec 134 ajouts et 1 suppressions
  1. 133 0
      alg_recsys_recall_tags_videos.py
  2. 1 1
      config.py

+ 133 - 0
alg_recsys_recall_tags_videos.py

@@ -0,0 +1,133 @@
+# -*- coding: utf-8 -*-
+import traceback
+from config import set_config
+from log import Log
+from utils import execute_sql_from_odps
+from db_helper import RedisHelper
+import datetime
+from alg_recsys_recall_4h_region_trend import records_process_for_list
+config_, _ = set_config()
+log_ = Log()
+redis_helper = RedisHelper()
+
+PROJECT = "videoods"
+TABLE = "videoods.dim_video"
+REDIS_PREFIX = "alg_recsys_video_tags_"
+
+TAG_SET = set(
+    '元旦','腊八节','小年','小年','除夕','春节','情人节','元宵节','龙抬头','妇女节','劳动节','母亲节',
+    '儿童节','端午节','父亲节','建党节','七七事变','建军节','七夕节','中元节','中秋节','毛主席逝世',
+    '国庆节','重阳节','感恩节','公祭日','平安夜','圣诞节','毛主席诞辰','小寒','大寒','立春','雨水',
+    '惊蛰','春分','清明','谷雨','立夏','小满','芒种','夏至','小暑','大暑','立秋','处暑','白露','秋分',
+    '寒露','霜降','立冬','小雪','大雪','冬至','早上好','中午好','下午好','晚上好','晚安','祝福',
+    'P1高风险','P0高风险'
+)
+
+def get_video_tags():
+    """获取视频的tag"""
+    try:
+        sql = "SELECT  videoid \
+,tags \
+FROM    {} \
+LATERAL VIEW EXPLODE(SPLIT(tags,',')) exploded AS exploded_value \
+WHERE   tags IS NOT NULL \
+AND     ( \
+exploded_value = '元旦' \
+OR      exploded_value = '腊八节' \
+OR      exploded_value = '小年' \
+OR      exploded_value = '除夕' \
+OR      exploded_value = '春节' \
+OR      exploded_value = '情人节' \
+OR      exploded_value = '元宵节' \
+OR      exploded_value = '龙抬头' \
+OR      exploded_value = '妇女节' \
+OR      exploded_value = '劳动节' \
+OR      exploded_value = '母亲节' \
+OR      exploded_value = '儿童节' \
+OR      exploded_value = '端午节' \
+OR      exploded_value = '父亲节' \
+OR      exploded_value = '建党节' \
+OR      exploded_value = '七七事变' \
+OR      exploded_value = '建军节' \
+OR      exploded_value = '七夕节' \
+OR      exploded_value = '中元节' \
+OR      exploded_value = '中秋节' \
+OR      exploded_value = '毛主席逝世' \
+OR      exploded_value = '国庆节' \
+OR      exploded_value = '重阳节' \
+OR      exploded_value = '感恩节' \
+OR      exploded_value = '公祭日' \
+OR      exploded_value = '平安夜' \
+OR      exploded_value = '圣诞节' \
+OR      exploded_value = '毛主席诞辰' \
+OR      exploded_value = '小寒' \
+OR      exploded_value = '大寒' \
+OR      exploded_value = '立春' \
+OR      exploded_value = '雨水' \
+OR      exploded_value = '惊蛰' \
+OR      exploded_value = '春分' \
+OR      exploded_value = '清明' \
+OR      exploded_value = '谷雨' \
+OR      exploded_value = '立夏' \
+OR      exploded_value = '小满' \
+OR      exploded_value = '芒种' \
+OR      exploded_value = '夏至' \
+OR      exploded_value = '小暑' \
+OR      exploded_value = '大暑' \
+OR      exploded_value = '立秋' \
+OR      exploded_value = '处暑' \
+OR      exploded_value = '白露' \
+OR      exploded_value = '秋分' \
+OR      exploded_value = '寒露' \
+OR      exploded_value = '霜降' \
+OR      exploded_value = '立冬' \
+OR      exploded_value = '小雪' \
+OR      exploded_value = '大雪' \
+OR      exploded_value = '冬至' \
+OR      exploded_value = '早上好' \
+OR      exploded_value = '中午好' \
+OR      exploded_value = '下午好' \
+OR      exploded_value = '晚上好' \
+OR      exploded_value = '晚安' \
+OR      exploded_value = '祝福' \
+OR      exploded_value = 'P1高风险' \
+OR      exploded_value = 'P0高风险' \
+)".format(TABLE)
+        print("sql:"+sql)
+        records = execute_sql_from_odps(project=PROJECT, sql=sql)
+        video_tags_list = []
+        with records.open_reader() as reader:
+            for record in reader:
+                video_id = int(record['videoid'])
+                tags = ",".join([i for i in str(record['tags']).split(",") if i in TAG_SET])
+                d = {}
+                d["video_id"] = video_id
+                d["tags"] = tags
+                video_tags_list.append(d)
+                log_.info("{}:{}".format(video_id, tags))
+
+        records_process_for_list(video_tags_list, process_and_store, max_size=10, num_workers=5)
+
+    except Exception as e:
+        log_.error(str(e) + str(traceback.format_exc()))
+
+def process_and_store(row):
+    video_id = row["video_id"]
+    tags = row["tags"]
+    key = REDIS_PREFIX + str(video_id)
+    expire_time = 24 * 3600 * 2
+    redis_helper.set_data_to_redis(key, tags, expire_time)
+    log_.info("trend-avg写入数据key={},value={}".format(key, tags))
+
+def main():
+    log_.info("开始执行:" + datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
+    get_video_tags()
+    log_.info("完成执行:" + datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
+
+if __name__ == '__main__':
+    main()
+
+
+
+# cd /root/zhangbo/rov-offline
+# python alg_recsys_recall_shield_videos.py

+ 1 - 1
config.py

@@ -2805,7 +2805,7 @@ def set_config():
     # 获取环境变量 ROV_OFFLINE_ENV
     env = os.environ.get('ROV_OFFLINE_ENV')
     # print("ROV_OFFLINE_ENV:{}".format(str(env)))
-    # env = 'dev'
+    env = 'dev'
     if env is None:
         # log_.error('ENV ERROR: is None!')
         return