update

task/daily_search_word_count_last_two_year.py

@@ -179,7 +179,8 @@ for t in range(1, task_days):
     sql_search_ctr = r"""
     SELECT query,
            all_search_uv as all_search_uv, --全部搜索uv
-           all_search_pv as all_search_pv --全部搜索pv
+           all_search_pv as all_search_pv, --全部搜索pv
+           case  when query in (select * from keywords_table) 0 else 1 end is_delete
     FROM (
              --搜索pvuv
              SELECT query
@@ -198,7 +199,7 @@ for t in range(1, task_days):
                         AND partition_date < '{end_date}'
                         AND ((action = 'do_search' AND params['input_type'] <> 'everyone_watch') or
                              action = 'search_result_click_search') 
-                        AND params['query'] in (select * from keywords_table)
+                        
 
                       UNION all
                       SELECT cl_id,
@@ -216,7 +217,7 @@ for t in range(1, task_days):
                         and params['tab'] = '精选'
                         and page_name = 'home'
                         AND params['query'] not in ('AI测颜值', 'AI测肤质') --这两个词不跳转搜索结果页
-                         AND params['query'] in (select * from keywords_table)
+             
 
                       union all
                       SELECT cl_id,
@@ -231,7 +232,7 @@ for t in range(1, task_days):
                         AND partition_date < '{end_date}'
                         AND action = 'on_click_card'
                         AND params['page_name'] = 'search_home'
-                        AND params['query'] in (select * from keywords_table)
+                       
 
                       union all
                       SELECT cl_id,
@@ -250,7 +251,7 @@ for t in range(1, task_days):
                         AND params['card_type'] = 'search_word'
                         AND params['card_name'] not in ('AI测颜值', 'AI测肤质') --这两个词不跳转搜索结果页
                         --AND page_name='home' android的page_name为空
-                         AND params['card_name'] in (select * from keywords_table)
+               
 
                       union all
                       SELECT cl_id,
@@ -267,7 +268,7 @@ for t in range(1, task_days):
                         AND params['page_name'] = 'welfare_home'
                         AND params['card_type'] = 'search_word'
                         AND params['in_page_pos'] = '大家都在搜'
-                         AND params['card_name'] in (select * from keywords_table)
+                        
 
                       union all
                       SELECT cl_id,
@@ -282,7 +283,6 @@ for t in range(1, task_days):
                         AND partition_date < '{end_date}'
                         AND action = 'on_click_card'
                         AND params['card_type'] = 'highlight_word'
-                        AND params['card_name'] in (select * from keywords_table)
                   ) click
              GROUP BY query
          ) 
@@ -304,8 +304,9 @@ for t in range(1, task_days):
         uv = name.all_search_uv
         is_tag = tag_dict.get(keywords, 0)
         servise_num = from_es_get_service_num(keywords)
+        is_delete = name.is_delete
         # pid = hashlib.md5((today_str + keywords).encode("utf8")).hexdigest()
-        tag_names_list_week.append((key_count, keywords, today_str, 0, servise_num, pv, is_tag))
+        tag_names_list_week.append((key_count, keywords, today_str, is_delete, servise_num, pv, is_tag))
 
     db = pymysql.connect(host='172.16.30.136', port=3306, user='doris', passwd='o5gbA27hXHHm',
                          db='doris_prod')