Question

请帮忙，这是我的表

   date     column1    column2          trx
2015-07-01  **side1    Internet**     777903315
2015-07-01  **side1    Internet**      41426210
2015-07-01  side1      Unlimited        2263500
2015-07-01  side1      Business          427000 
2015-07-01  side1      Extreme          3540900
2015-07-01  side1      Lifestyle       59360000
2015-07-01  side1      Socialita      240850500
2015-07-01  **side2    Unlimited**         6160
2015-07-01  **side2    Unlimited**    113502000

我想选择我的表格，结果如下：

   date     column1    column2          trx      type
2015-07-01  **side1    Internet**     777903315  pre
2015-07-01  **side1    Internet**      41426210  post
2015-07-01  side1      Unlimited        2263500  pre
2015-07-01  side1      Business          427000  pre
2015-07-01  side1      Extreme          3540900  pre
2015-07-01  side1      Lifestyle       59360000  pre
2015-07-01  side1      Socialita      240850500  pre
2015-07-01  **side2    Unlimited**         6160  post
2015-07-01  **side2    Unlimited**    113502000  pre

在column1和column2中具有相同值的行，在最小trx添加字段后面的行作为类型'post'和最大化'pre'

Answer 1

您可以使用以下查询：

SELECT m.*,       
       CASE trx 
          WHEN mintrx THEN 'post'
          WHEN maxtrx THEN 'pre'
          ELSE 'pre'
       END AS type
FROM mytable AS m
LEFT JOIN (
   SELECT column1, column2, MIN(trx) AS mintrx, MAX(trx) AS maxtrx
   FROM mytable
   GROUP BY column1, column2
   HAVING MIN(trx) <> MAX(trx) ) AS t
ON m.column1 = t.column1 AND m.column2 = t.column2

此查询执行LEFT JOIN，其派生表仅包含重复的column1，column2行。最小/最大trx匹配分别生成post / pre值，而pre是非匹配表行的默认值。

Demo here

Answer 2

SELECT t.date, t.c1, t.c2, t.trx
     , CASE WHEN t.trx = m.trx THEN 'pre' ELSE 'post' END AS `type`
  FROM so_q31531850 t
  LEFT OUTER JOIN ( SELECT DATE, c1, c2, MAX(trx) AS trx FROM so_q31531850 GROUP BY 1,2,3) m 
               ON m.trx = t.trx AND m.c1 = t.c1 AND m.c2 = t.c2 AND m.date = t.date

请注意，如果交易号相同，我建议的查询会将其视为post条目类型。

Answer 3

下面的查询（SQL）以上面给出的所需格式给出结果：

SELECT Date,column1,column2,trx
 ,CASE RANK() OVER(PARTITION BY Date,col1,col2 ORDER BY trx DESC)
    WHEN 1 THEN 'pre'
  ELSE 'post'
  END as type
FROM [TableName]

Answer 4

也许这段代码可以做得更好一点，我自己也不是一个sql人，但是小提琴很有趣。

这是一个示例表：

 CREATE TABLE EXAMPLE (
     id INT,
     c1 NVARCHAR(100), 
     c2 NVARCHAR(100)         
   );

INSERT INTO EXAMPLE (ID,c1,c2)
VALUES (1, '**side1', 'Internet**');

INSERT INTO EXAMPLE (ID,c1,c2)
VALUES (2, '**side2', 'Unlimited**');

INSERT INTO EXAMPLE (ID,c1,c2)
VALUES (3, 'side1', 'Unlimited');

INSERT INTO EXAMPLE (ID,c1,c2)
VALUES (4, 'side1', 'Lifestyle');

INSERT INTO EXAMPLE (ID,c1,c2)
VALUES (5, 'side1', 'Buisness');

INSERT INTO EXAMPLE (ID,c1,c2)
VALUES (6, '**side2', 'Unlimited**');

INSERT INTO EXAMPLE (ID,c1,c2)
VALUES (7, '**side1', 'Internet**');

查询：

select * from Example 
order by 
   case 
      when c1 like '**%' then substring(c1,3,Len(c1)) else c1   
   end,c1 ,c2 desc;

如何比较同一个表中的2行

4 个答案: