Question

create table t (col1 int,col2 varchar(10)) 
insert into t values(10,'val')
insert into t values(20,null)
insert into t values(10,'val3')
insert into t values(40,'val1')
insert into t values(50,null)
insert into t values(60,'val')

--1.
SELECT *,COUNT(COL2) OVER (ORDER BY COL1) FROM T

--2.
SELECT *,COUNT(COL1) OVER (ORDER BY COL2) FROM T

我无法理解上面提到的两个选择查询的输出..任何人都可以详细说明。

Answer 1

count列将包含col2中不为null的累积值，按col 1的值排序

SELECT *,COUNT(COL2) OVER (ORDER BY COL1) AS [count]
FROM T
ORDER BY COL1

结果：

col1    col2    count   --explanation
10      val3    2       --The first two values of col1 are the same, and the values of col2 are not null.
10      val     2       --The first two values of col1 are the same, and the values of col2 are not null.
20      NULL    2       --A new value of col1, but col2 is null so it's not counted.
40      val1    3       --A new value of col1, and col2 is not null.
50      NULL    3       --A new value of col1, but col2 is null so it's not counted.
60      val     4       --A new value of col1, and col2 is not null.

基本上相同，只需反转列：

SELECT *,COUNT(COL1) OVER (ORDER BY COL2) As col1Count
FROM T
ORDER BY  COL2

结果：

col1    col2    col1Count    --explanation
20      NULL    2            --The first two values of col2 are the same (both null), and the values of col1 are not null.
50      NULL    2            --The first two values of col2 are the same (both null), and the values of col1 are not null.
60      val     4            --The next two values of col2 are the same (both 'val'), and the values of col1 are not null.
10      val     4            --The next two values of col2 are the same (both 'val'), and the values of col1 are not null.
40      val1    5            --A new value of col2, and col1 is not null.
10      val3    6            --A new value of col2, and col1 is not null.

通过col2按顺序计数col1

1 个答案: