給老婆的一篇文章

老婆公司有個這樣的需求：

查詢出王者榮耀的用戶迴流信息，當用戶連續兩天登陸，則判定爲2日迴流，如果間隔一天登陸，則判定爲3日迴流，如果間隔5天登陸，則判定爲7日迴流。用戶數據間隔時間短爲14天（固定）。

準備數據

database語法和mysql一致。

創建表：

create table user_login(
u_id int,
login_date timestamp default current_timestamp
);

插入數據：

DELIMITER //
create procedure loop_insert()
begin

declare days int;
declare usrs int;
declare mx int;
declare i int;

set days = 14;
set usrs = 30;
set mx = 500;
set i = 1;

repeat 
    insert into user_login(u_id,login_date) values (floor((RAND() * usrs)),subdate(sysdate(),(RAND() * (days+1))));
    set i = i + 1;
until i >= mx

end repeat;

end//

call loop_insert();

最開始我想到的是用group_concat,sql是這樣：

SELECT 
    u_id, group_concat(distinct DATE_FORMAT(login_date, '%Y%m%d') order by DATE_FORMAT(login_date, '%Y%m%d') desc separator '-') AS yyyymmdd
FROM
    user_login
GROUP BY u_id;

後來想想日期轉換成INTEGER相減是不準確的（比如跨月），而且這樣顯示並不能解決需求。

好吧，考慮行轉列。

行轉列 ##

行轉列，需要case when枚舉，好在日期只有14天，可以做到：

SELECT 
        u_id,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180413' THEN login_date
                ELSE ''
            END AS d20180413,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180414' THEN login_date
                ELSE ''
            END AS d20180414,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180415' THEN login_date
                ELSE ''
            END AS d20180415,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180416' THEN login_date
                ELSE ''
            END AS d20180416,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180417' THEN login_date
                ELSE ''
            END AS d20180417,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180418' THEN login_date
                ELSE ''
            END AS d20180418,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180419' THEN login_date
                ELSE ''
            END AS d20180419,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180420' THEN login_date
                ELSE ''
            END AS d20180420,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180421' THEN login_date
                ELSE ''
            END AS d20180421,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180422' THEN login_date
                ELSE ''
            END AS d20180422,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180423' THEN login_date
                ELSE ''
            END AS d20180423,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180424' THEN login_date
                ELSE ''
            END AS d20180424,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180425' THEN login_date
                ELSE ''
            END AS d20180425,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180426' THEN login_date
                ELSE ''
            END AS d20180426
    FROM
        user_login

查詢結果：

按用戶合併

按用戶合併日期，去重，用max可以保證單條數據，有數據置爲1無數據置爲0：

SELECT 
        u_id,
            CASE
                WHEN MAX(d20180413) = '' THEN '0'
                ELSE '1'
            END AS isZ20180413,
            CASE
                WHEN MAX(d20180414) = '' THEN '0'
                ELSE '1'
            END AS isZ20180414,
            CASE
                WHEN MAX(d20180415) = '' THEN '0'
                ELSE '1'
            END AS isZ20180415,
            CASE
                WHEN MAX(d20180416) = '' THEN '0'
                ELSE '1'
            END AS isZ20180416,
            CASE
                WHEN MAX(d20180417) = '' THEN '0'
                ELSE '1'
            END AS isZ20180417,
            CASE
                WHEN MAX(d20180418) = '' THEN '0'
                ELSE '1'
            END AS isZ20180418,
            CASE
                WHEN MAX(d20180419) = '' THEN '0'
                ELSE '1'
            END AS isZ20180419,
            CASE
                WHEN MAX(d20180420) = '' THEN '0'
                ELSE '1'
            END AS isZ20180420,
            CASE
                WHEN MAX(d20180421) = '' THEN '0'
                ELSE '1'
            END AS isZ20180421,
            CASE
                WHEN MAX(d20180422) = '' THEN '0'
                ELSE '1'
            END AS isZ20180422,
            CASE
                WHEN MAX(d20180423) = '' THEN '0'
                ELSE '1'
            END AS isZ20180423,
            CASE
                WHEN MAX(d20180424) = '' THEN '0'
                ELSE '1'
            END AS isZ20180424,
            CASE
                WHEN MAX(d20180425) = '' THEN '0'
                ELSE '1'
            END AS isZ20180425,
            CASE
                WHEN MAX(d20180426) = '' THEN '0'
                ELSE '1'
            END AS isZ20180426
    FROM
        (SELECT 
        u_id,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180413' THEN login_date
                ELSE ''
            END AS d20180413,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180414' THEN login_date
                ELSE ''
            END AS d20180414,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180415' THEN login_date
                ELSE ''
            END AS d20180415,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180416' THEN login_date
                ELSE ''
            END AS d20180416,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180417' THEN login_date
                ELSE ''
            END AS d20180417,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180418' THEN login_date
                ELSE ''
            END AS d20180418,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180419' THEN login_date
                ELSE ''
            END AS d20180419,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180420' THEN login_date
                ELSE ''
            END AS d20180420,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180421' THEN login_date
                ELSE ''
            END AS d20180421,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180422' THEN login_date
                ELSE ''
            END AS d20180422,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180423' THEN login_date
                ELSE ''
            END AS d20180423,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180424' THEN login_date
                ELSE ''
            END AS d20180424,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180425' THEN login_date
                ELSE ''
            END AS d20180425,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180426' THEN login_date
                ELSE ''
            END AS d20180426
    FROM
        user_login) t0
    GROUP BY u_id

查詢結果：

列轉行

使用group_concat:

SELECT 
        u_id,
            CONCAT(isZ20180413, isZ20180414, isZ20180415, isZ20180416, isZ20180417, isZ20180418, isZ20180419, isZ20180420, isZ20180421, isZ20180422, isZ20180423, isZ20180424, isZ20180425, isZ20180426) AS sumIsZ
    FROM
        (SELECT 
        u_id,
            CASE
                WHEN MAX(d20180413) = '' THEN '0'
                ELSE '1'
            END AS isZ20180413,
            CASE
                WHEN MAX(d20180414) = '' THEN '0'
                ELSE '1'
            END AS isZ20180414,
            CASE
                WHEN MAX(d20180415) = '' THEN '0'
                ELSE '1'
            END AS isZ20180415,
            CASE
                WHEN MAX(d20180416) = '' THEN '0'
                ELSE '1'
            END AS isZ20180416,
            CASE
                WHEN MAX(d20180417) = '' THEN '0'
                ELSE '1'
            END AS isZ20180417,
            CASE
                WHEN MAX(d20180418) = '' THEN '0'
                ELSE '1'
            END AS isZ20180418,
            CASE
                WHEN MAX(d20180419) = '' THEN '0'
                ELSE '1'
            END AS isZ20180419,
            CASE
                WHEN MAX(d20180420) = '' THEN '0'
                ELSE '1'
            END AS isZ20180420,
            CASE
                WHEN MAX(d20180421) = '' THEN '0'
                ELSE '1'
            END AS isZ20180421,
            CASE
                WHEN MAX(d20180422) = '' THEN '0'
                ELSE '1'
            END AS isZ20180422,
            CASE
                WHEN MAX(d20180423) = '' THEN '0'
                ELSE '1'
            END AS isZ20180423,
            CASE
                WHEN MAX(d20180424) = '' THEN '0'
                ELSE '1'
            END AS isZ20180424,
            CASE
                WHEN MAX(d20180425) = '' THEN '0'
                ELSE '1'
            END AS isZ20180425,
            CASE
                WHEN MAX(d20180426) = '' THEN '0'
                ELSE '1'
            END AS isZ20180426
    FROM
        (SELECT 
        u_id,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180413' THEN login_date
                ELSE ''
            END AS d20180413,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180414' THEN login_date
                ELSE ''
            END AS d20180414,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180415' THEN login_date
                ELSE ''
            END AS d20180415,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180416' THEN login_date
                ELSE ''
            END AS d20180416,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180417' THEN login_date
                ELSE ''
            END AS d20180417,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180418' THEN login_date
                ELSE ''
            END AS d20180418,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180419' THEN login_date
                ELSE ''
            END AS d20180419,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180420' THEN login_date
                ELSE ''
            END AS d20180420,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180421' THEN login_date
                ELSE ''
            END AS d20180421,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180422' THEN login_date
                ELSE ''
            END AS d20180422,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180423' THEN login_date
                ELSE ''
            END AS d20180423,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180424' THEN login_date
                ELSE ''
            END AS d20180424,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180425' THEN login_date
                ELSE ''
            END AS d20180425,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180426' THEN login_date
                ELSE ''
            END AS d20180426
    FROM
        user_login) t0
    GROUP BY u_id) t1

查詢結果：

統計

SELECT 
    u_id,
    sumIsZ,
    CASE
        WHEN LOCATE('11', sumIsZ) > 0 THEN 2
        ELSE 0
    END AS is2Back,/*2日迴流*/
    CASE
        WHEN LOCATE('101', sumIsZ) > 0 THEN 3
        ELSE 0
    END AS is3Back,/*3日迴流*/
    CASE
        WHEN LOCATE('1001', sumIsZ) > 0 THEN 4
        ELSE 0
    END AS is4Back,/*4日迴流*/
    CASE
        WHEN LOCATE('10001', sumIsZ) > 0 THEN 5
        ELSE 0
    END AS is5Back,/*5日迴流*/
    CASE
        WHEN LOCATE('100001', sumIsZ) > 0 THEN 6
        ELSE 0
    END AS is6Back,/*6日迴流*/
    CASE
        WHEN LOCATE('1000001', sumIsZ) > 0 THEN 7
        ELSE 0
    END AS is7Back/*7日迴流*/
FROM
    (SELECT 
        u_id,
            CONCAT(isZ20180413, isZ20180414, isZ20180415, isZ20180416, isZ20180417, isZ20180418, isZ20180419, isZ20180420, isZ20180421, isZ20180422, isZ20180423, isZ20180424, isZ20180425, isZ20180426) AS sumIsZ
    FROM
        (SELECT 
        u_id,
            CASE
                WHEN MAX(d20180413) = '' THEN '0'
                ELSE '1'
            END AS isZ20180413,
            CASE
                WHEN MAX(d20180414) = '' THEN '0'
                ELSE '1'
            END AS isZ20180414,
            CASE
                WHEN MAX(d20180415) = '' THEN '0'
                ELSE '1'
            END AS isZ20180415,
            CASE
                WHEN MAX(d20180416) = '' THEN '0'
                ELSE '1'
            END AS isZ20180416,
            CASE
                WHEN MAX(d20180417) = '' THEN '0'
                ELSE '1'
            END AS isZ20180417,
            CASE
                WHEN MAX(d20180418) = '' THEN '0'
                ELSE '1'
            END AS isZ20180418,
            CASE
                WHEN MAX(d20180419) = '' THEN '0'
                ELSE '1'
            END AS isZ20180419,
            CASE
                WHEN MAX(d20180420) = '' THEN '0'
                ELSE '1'
            END AS isZ20180420,
            CASE
                WHEN MAX(d20180421) = '' THEN '0'
                ELSE '1'
            END AS isZ20180421,
            CASE
                WHEN MAX(d20180422) = '' THEN '0'
                ELSE '1'
            END AS isZ20180422,
            CASE
                WHEN MAX(d20180423) = '' THEN '0'
                ELSE '1'
            END AS isZ20180423,
            CASE
                WHEN MAX(d20180424) = '' THEN '0'
                ELSE '1'
            END AS isZ20180424,
            CASE
                WHEN MAX(d20180425) = '' THEN '0'
                ELSE '1'
            END AS isZ20180425,
            CASE
                WHEN MAX(d20180426) = '' THEN '0'
                ELSE '1'
            END AS isZ20180426
    FROM
        (SELECT 
        u_id,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180413' THEN login_date
                ELSE ''
            END AS d20180413,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180414' THEN login_date
                ELSE ''
            END AS d20180414,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180415' THEN login_date
                ELSE ''
            END AS d20180415,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180416' THEN login_date
                ELSE ''
            END AS d20180416,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180417' THEN login_date
                ELSE ''
            END AS d20180417,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180418' THEN login_date
                ELSE ''
            END AS d20180418,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180419' THEN login_date
                ELSE ''
            END AS d20180419,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180420' THEN login_date
                ELSE ''
            END AS d20180420,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180421' THEN login_date
                ELSE ''
            END AS d20180421,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180422' THEN login_date
                ELSE ''
            END AS d20180422,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180423' THEN login_date
                ELSE ''
            END AS d20180423,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180424' THEN login_date
                ELSE ''
            END AS d20180424,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180425' THEN login_date
                ELSE ''
            END AS d20180425,
            CASE DATE_FORMAT(login_date, '%Y%m%d')
                WHEN '20180426' THEN login_date
                ELSE ''
            END AS d20180426
    FROM
        user_login) t0
    GROUP BY u_id) t1) t2
;

查詢結果：

結語

老婆，明天照着這個邏輯寫就可以了，麼麼噠。（2018年4月26日23點50分）