Browse Source

update data

linfan 1 year ago
parent
commit
9d47145ebd
3 changed files with 43 additions and 36 deletions
  1. 1 1
      config.py
  2. 38 31
      utils.py
  3. 4 4
      video_recall.py

+ 1 - 1
config.py

@@ -1286,7 +1286,7 @@ class ProductionConfig(BaseConfig):
 def set_config():
     # 获取环境变量 ROV_SERVER_ENV
     env = os.environ.get('ROV_SERVER_ENV')
-    # env = 'dev'
+    env = 'pro'
     if env is None:
         # log_.error('ENV ERROR: is None!')
         return

+ 38 - 31
utils.py

@@ -383,7 +383,7 @@ class FilterVideos(object):
         key_name = f"{config_.PREVIEW_KEY_PREFIX}{self.app_type}:{self.mid}"
         print("key_name:", key_name)
         pe_videos_list = redis_helper.get_data_from_set(key_name)
-        print("pe_videos_list:", pe_videos_list)
+        #print("pe_videos_list:", pe_videos_list)
         # log_.info('****app_type = {}, mid = {}, uid = {}, pe_videos_list = {}'.format(
         #     self.app_type, self.mid, self.uid, pe_videos_list))
         # log_.info('****app_type = {}, mid = {}, uid = {}, video_ids = {}'.format(
@@ -391,9 +391,9 @@ class FilterVideos(object):
         if not pe_videos_list:
             return video_ids
         pe_videos = [int(video) for video in pe_videos_list]
-        print("pe_videos:", pe_videos)
+        print("pe_videos:", len(pe_videos))
         filtered_videos = [video_id for video_id in video_ids if video_id not in pe_videos]
-        print(f"pre res: {filtered_videos}\nexecute_time: {(time.time() - pre_time) * 1000}")
+        #print(f"pre res: {filtered_videos}\nexecute_time: {(time.time() - pre_time) * 1000}")
         return filtered_videos
 
     # def filter_video_status(self, video_ids):
@@ -481,7 +481,7 @@ class FilterVideos(object):
         if 6 in types:
             types = list(types)
             types.remove(6)
-        print(types)
+        #print(types)
         request_data = {"appType": self.app_type,
                         "mid": self.mid,
                         "uid": self.uid,
@@ -489,7 +489,7 @@ class FilterVideos(object):
                         "videoIds": video_ids}
         # 调用http接口
         result = request_post(request_url=config_.VIDEO_FILTER_URL, request_data=request_data, timeout=(0.1, 1))
-        print(f"view res: {result}\nexecute_time: {(time.time() - st_time) * 1000}")
+        #print(f"view res: {result}\nexecute_time: {(time.time() - st_time) * 1000}")
         if result is None:
             # log_.info('过滤失败,types: {}'.format(types))
             return []
@@ -508,6 +508,7 @@ class FilterVideos(object):
         :param shield_key_name_list: 过滤视频 redis-key
         :return: filtered_videos  过滤后的列表  type-list
         """
+        print("filter_shield_video:", len(filter_shield_video))
         if len(video_ids) == 0:
             return video_ids
         # 根据Redis缓存中的数据过滤
@@ -522,6 +523,7 @@ class FilterVideos(object):
             #     continue
             # shield_videos = [int(video) for video in shield_videos_list]
             # video_ids = [int(video_id) for video_id in video_ids if int(video_id) not in shield_videos]
+        print("video_ids:", len(video_ids))
         return video_ids
 
     def new_filter_video(self):
@@ -546,7 +548,7 @@ class FilterVideos(object):
         #2. 视频已曝光过滤
         st_viewed = time.time()
         print("---filtered viewed---")
-        print("filtered_pre_result:",filtered_pre_result)
+        #print("filtered_pre_result:",filtered_pre_result)
         filtered_viewed_result = self.filter_video_viewed(video_ids=filtered_pre_result)
         if not filtered_viewed_result:
             return None
@@ -586,7 +588,7 @@ class FilterVideos(object):
             # 过滤 影视 + 宗教
             keys = [f"rov.filter.religion.{video_id}" for video_id in video_ids]
         else:
-            print(f"m_r res: {video_ids}\nexecute_time: {(time.time() - st_time) * 1000}")
+            #print(f"m_r res: {video_ids}\nexecute_time: {(time.time() - st_time) * 1000}")
             return video_ids
         redis_helper = RedisHelper(redis_info=config_.REDIS_INFO_FILTER)
         filter_videos = []
@@ -598,20 +600,21 @@ class FilterVideos(object):
             filter_videos.extend([int(data) for data in mget_res if data is not None])
         if len(filter_videos) > 0:
             filtered_videos = set(video_ids) - set(filter_videos)
-            print(f"m_r res: {list(filtered_videos)}\nexecute_time: {(time.time() - st_time) * 1000}")
+            #print(f"m_r res: {list(filtered_videos)}\nexecute_time: {(time.time() - st_time) * 1000}")
             return list(filtered_videos)
         else:
-            print(f"m_r res: {video_ids}\nexecute_time: {(time.time() - st_time) * 1000}")
+            #print(f"m_r res: {video_ids}\nexecute_time: {(time.time() - st_time) * 1000}")
             return video_ids
 
     def filter_videos_new(self, pool_type='rov', region_code=None, shield_config=None):
         """视频过滤"""
         # 预曝光过滤
         st_pre = time.time()
+        print("self.video_ids:", len(self.video_ids))
         filtered_pre_result = self.filter_video_previewed(self.video_ids)
         if not filtered_pre_result:
             return None
-
+        print("filtered_pre_result:", len(filtered_pre_result))
         # 视频已曝光过滤/白名单过滤
         st_viewed = time.time()
         t = [
@@ -619,36 +622,40 @@ class FilterVideos(object):
             gevent.spawn(self.filter_movie_religion_video, filtered_pre_result)]
         gevent.joinall(t)
         filtered_result_list = [i.get() for i in t]
+        print("filtered_result_list:",len(filtered_result_list))
         filtered_viewed_result = list(set(filtered_result_list[0]) & set(filtered_result_list[1]))
-        print(f"view&m_r res: {filtered_viewed_result}\nexecute_time: {(time.time() - st_viewed) * 1000}")
+        #print(f"view&m_r res: {filtered_viewed_result}\nexecute_time: {(time.time() - st_viewed) * 1000}")
+        print("filtered:",len(filtered_viewed_result))
         if not filtered_viewed_result:
+            print("12334")
             return None
         filtered_viewed_videos = [int(video_id) for video_id in filtered_viewed_result]
+        print("result:", len(filtered_viewed_videos))
         if pool_type == 'flow':
             # 流量池视频需过滤屏蔽视频
             if region_code is None or shield_config is None:
                 return filtered_viewed_videos
+        else:
+            shield_key_name_list = shield_config.get(region_code, None)
+            if shield_key_name_list is not None:
+                filtered_shield_video_ids = self.filter_shield_video(
+                   video_ids=filtered_viewed_videos, shield_key_name_list=shield_key_name_list
+                )
+                log_.info({
+                    'logTimestamp': int(time.time() * 1000),
+                    'pool_type': pool_type,
+                    'request_id': self.request_id,
+                    'app_type': self.app_type,
+                    'mid': self.mid,
+                    'uid': self.uid,
+                    'operation': 'shield_filter',
+                    'request_videos': filtered_viewed_videos,
+                    'shield_filter_result': filtered_shield_video_ids,
+                    'executeTime': (time.time() - st_viewed) * 1000
+                 })
+                return filtered_shield_video_ids
             else:
-                shield_key_name_list = shield_config.get(region_code, None)
-                if shield_key_name_list is not None:
-                    filtered_shield_video_ids = self.filter_shield_video(
-                        video_ids=filtered_viewed_videos, shield_key_name_list=shield_key_name_list
-                    )
-                    log_.info({
-                        'logTimestamp': int(time.time() * 1000),
-                        'pool_type': pool_type,
-                        'request_id': self.request_id,
-                        'app_type': self.app_type,
-                        'mid': self.mid,
-                        'uid': self.uid,
-                        'operation': 'shield_filter',
-                        'request_videos': filtered_viewed_videos,
-                        'shield_filter_result': filtered_shield_video_ids,
-                        'executeTime': (time.time() - st_viewed) * 1000
-                    })
-                    return filtered_shield_video_ids
-                else:
-                    return filtered_viewed_videos
+                return filtered_viewed_videos
 
 
 if __name__ == '__main__':

+ 4 - 4
video_recall.py

@@ -2141,7 +2141,7 @@ class PoolRecall(object):
                      'rovScore': per_item[1], 'pushFrom': config_.PUSH_FROM['rov_recall_region_h'],
                      'abCode': self.ab_code}
                 )
-        return recall_result[:500]
+        return recall_result[:100]
 
     # get region_day_recall
     def get_region_day_recall(self, size=4,region_code='-1'):
@@ -2161,7 +2161,7 @@ class PoolRecall(object):
                      'rovScore': per_item[1], 'pushFrom': config_.PUSH_FROM['rov_recall_region_24h'],
                      'abCode': self.ab_code}
                 )
-        return recall_result[:500]
+        return recall_result[:100]
 
 
     def get_selected_recall(self, size=4, region_code='-1'):
@@ -2182,7 +2182,7 @@ class PoolRecall(object):
                      'abCode': self.ab_code}
                 )
         #print("recall_result:", recall_result)
-        return recall_result[:500]
+        return recall_result[:100]
 
     def get_no_selected_recall(self, size=4, region_code='-1'):
         """未选择召回池召回视频"""
@@ -2201,7 +2201,7 @@ class PoolRecall(object):
                      'rovScore': per_item[1], 'pushFrom': config_.PUSH_FROM['rov_recall_24h_dup'],
                      'abCode': self.ab_code}
                 )
-        return recall_result[:500]
+        return recall_result[:100]
 
     def get_fast_flow_pool_recall(self, size=4):
         """快速流量池召回视频"""