有表 tableA 如下:
-----------------------------------------------------
id, type , countNumber , countDate
1 A 10 2005-02-01 21:23:34
2 B 8 2005-02-01 21:23:34
3 C 5 2005-02-01 21:23:34
4 B 4 2005-02-01 11:23:34
5 C 5 2005-02-01 22:23:34
6 A 11 2005-02-02 12:23:34
7 B 9 2005-02-02 14:23:34
8 C 8 2005-02-02 17:23:34
9 A 15 2005-02-02 19:23:34
10 C 6 2005-02-02 04:23:34
11 A 7 2005-02-03 1:23:34
12 B 11 2005-02-03 2:23:34
13 C 12 2005-02-03 5:23:34
.......................
想要用一条语句生成如下的报表
日期 类型A 类型B 类型C
2005-02-01 10 12 10
2005-02-02 26 9 14
2005-02-03 7 11 12
我用的是如下的sql语句:
SELECT
a.DATEFMT
,(select sum(*) from tableA where type='A' and convert(char(10),countDate,20)= a.DATEFMT) typeA
,(select sum(*) from tableA where type='B' and convert(char(10),countDate,20)= a.DATEFMT) typeB
,(select sum(*) from tableA where type='C' and convert(char(10),countDate,20)= a.DATEFMT) typeC
FROM
(
SELECT convert(char(10),countDate,20) DATEFMT
FROM tableA
GROUP BY convert(char(10),countDate,20)
) a
但是这条语句在sqlserver 上执行起来特别的慢,要很久(20多秒),tabelA 中有9万条数据. 而我在mySql(表中有2万条数据)上执行就快很多!
后来在论坛上看到高手这样写
select convert(char(10),countDate,20),
typeA=sum(case type when A then CountNumber else 0 end),
typeB=sum(case type when B then CountNumber else 0 end),
typeC=sum(case type when C then CountNumber else 0 end)
From tableA
group by convert(char(10),countDate,20)
只用 1秒 不到! 唉! 厉害厉害!
本文地址:http://com.8s8s.com/it/it19147.htm