组合两个查询以获得总计数和过滤列表

Question

我有一个查询获取过去一年内登录的所有用户 ID。

查询 1:

SELECT user_id, user_name
FROM logins
WHERE login_date BETWEEN "2015-09-01" AND "2016-09-01"
GROUP BY user_id;

Execution time: 625ms

另一个与上面类似的查询，现在有额外的过滤器来列出带有 'jeff' 的用户名。

查询 2：

SELECT user_id, user_name
FROM logins
WHERE login_date BETWEEN "2015-09-01" AND "2016-09-01"
AND user_name LIKE '%jeff%'
GROUP BY user_id;

Execution time: 587ms

总执行时间为 1212 毫秒。

我只使用 Query1 来获取一年内登录用户的总数，使用 Query2 来获取 user_name 的用户列表，例如 'jeff'。

问题是，我需要执行这两个查询，而且需要更长的时间。

有没有办法将两个查询合并为一个并获取我需要的数据，即登录用户总数和 user_name（如 'jeff'）的登录用户列表？同时最小化执行时间？

示例数据：

user_id user_name   login_date
1   annie       2015-10-02
2   greg        2015-12-25
3   jeff        2015-11-05
4   aljeffar    2016-03-06
5   jeff1       2015-07-07
2   greg        2016-01-30
1   annie       2015-11-13
3   jeff        2016-02-14
2   greg        2016-09-30
3   jeff        2016-07-04

查询 1 结果：

user_id   user_name   
1 annie       
2 greg        
3 jeff        
4 aljeffar
Total result count: 4 -- this should be the fix value of total_loggedin_users column below

查询 2 结果：

  user_id user_name
  3   jeff    
  4   aljeffar

预期最终结果：

user_id   user_name   total_loggedin_users
3 jeff        4
4 aljeffar    4

Answer 1

SELECT 
       user_id, 
       count(*) as total, 
       sum(case when user_name LIKE '%jeff%' then 1 else 0 end) as jeff_count
FROM logins
WHERE login_date BETWEEN "2015-09-01" AND "2016-09-01"
GROUP BY user_id;

运行它只是一次，并添加一个条件总和来计算 jeffs。

Answer 2

试试这个。其他变体。

SELECT user_id, user_name, GROUP_CONCAT(user_name) AS list_of_jeffs, COUNT(*) AS total, SUM(IF(user_name LIKE "%jeff%",1,0)) as jeff
FROM `pt_terminal` 
WHERE login_date BETWEEN "2015-09-01" AND "2016-09-01" 
GROUP BY user_id;

Answer 3

如果您希望您的结果仅像 jeff 一样显示 user_name 并且还有总计数，您可以试试这个：

需要执行子选择以获得列中的总计数。更新查询：

SELECT  [USER_ID] ,
        [USER_NAME] ,
        CountAll
FROM    ( SELECT    [USER_ID] ,
                    [USER_NAME] ,
                    -- Window-Function (Only in SQL-Server) to get overall count
                    COUNT([USER_ID]) OVER ( ) AS CountAll
          FROM      dbo.logins
          WHERE     login_date BETWEEN '2015-09-01' AND '2016-09-01'
          --Group so we don't count multiple logins of one user 
          GROUP BY  [USER_ID] ,
                    [USER_NAME]
        ) subselect
WHERE   [USER_NAME] LIKE '%jeff%'

组合两个查询以获得总计数和过滤列表

Combining two queries to get total count and filtered list

sql

phoenix