1 有时候我们会需要将多条数据根据一些特别的字段做一些合并。比如下面这个查询,正常会查询出3条数据,但是我们会希望根据create_by 分成两列显示
2 这时候需要用到string_agg函数,先通过group by分组,在进行合并,当然查询结果需要满足group by的限制;sql语句:
select create_by,string_agg(videoname,',') as videonames from w008_video_addr_info where id in (4248,538,546) group by create_by
查询结果:
3 复杂一些的应用场景(子查询):
下面的语句是我用来查询一个学生在什么时间看了哪些视频:
select sa.id, info.nickname, (select string_agg(v.videoname,',') from w008_school_assign_video sv join w008_video_addr_info v on sv.videoaddrinfo =v.id where sv.schoolassignment=sa.id and v.is_removed=0 and sv.is_removed=0 group by v.is_removed) as videos, (select string_agg(to_char(sv.create_date, 'MM-DD HH24:MI'),',') from w008_school_assign_video sv join w008_video_addr_info v on sv.videoaddrinfo =v.id where sv.schoolassignment=sa.id and v.is_removed=0 and sv.is_removed=0 group by v.is_removed) as viewtime from w008_school_assignment sa join w008_user_business_info info on sa.userlongid=info.id where sa.shchoolworkid=2514505674916356
结果:
当然,string_agg(field,'分隔符');分隔符可以填写其他任意的字符,方便后期处理即可;
补充:PostgreSql 聚合函数string_agg与array_agg,类似mysql中group_concat
string_agg,array_agg 这两个函数的功能大同小异,只不过合并数据的类型不同。
https://www.postgresql.org/docs/9.6/static/functions-aggregate.html
array_agg(expression)
把表达式变成一个数组 一般配合 array_to_string() 函数使用
string_agg(expression, delimiter)
直接把一个表达式变成字符串
案例:
create table(empno smallint, ename varchar(20), job varchar(20), mgr smallint, hiredate date, sal bigint, comm bigint, deptno smallint); insert into jinbo.employee(empno,ename,job, mgr, hiredate, sal, comm, deptno) values (7499, 'ALLEN', 'SALEMAN', 7698, '2014-11-12', 16000, 300, 30); insert into jinbo.employee(empno,ename,job, mgr, hiredate, sal, comm, deptno) values (7499, 'ALLEN', 'SALEMAN', 7698, '2014-11-12', 16000, 300, 30); insert into jinbo.employee(empno,ename,job, mgr, hiredate, sal, comm, deptno) values (7654, 'MARTIN', 'SALEMAN', 7698, '2016-09-12', 12000, 1400, 30); select * from jinbo.employee; empno | ename | job | mgr | hiredate | sal | comm | deptno -------+--------+---------+------+------------+-------+------+-------- 7499 | ALLEN | SALEMAN | 7698 | 2014-11-12 | 16000 | 300 | 30 7566 | JONES | MANAGER | 7839 | 2015-12-12 | 32000 | 0 | 20 7654 | MARTIN | SALEMAN | 7698 | 2016-09-12 | 12000 | 1400 | 30 (3 rows)
查询同一个部门下的员工且合并起来
方法1:
select deptno, string_agg(ename, ',') from jinbo.employee group by deptno; deptno | string_agg --------+-------------- 20 | JONES 30 | ALLEN,MARTIN
方法2:
select deptno, array_to_string(array_agg(ename),',') from jinbo.employee group by deptno; deptno | array_to_string --------+----------------- 20 | JONES 30 | ALLEN,MARTIN
在1条件的基础上,按ename 倒叙合并
select deptno, string_agg(ename, ',' order by ename desc) from jinbo.employee group by deptno; deptno | string_agg --------+-------------- 20 | JONES 30 | MARTIN,ALLEN
按数组格式输出使用 array_agg
select deptno, array_agg(ename) from jinbo.employee group by deptno; deptno | array_agg --------+---------------- 20 | {JONES} 30 | {ALLEN,MARTIN}
array_agg 去重元素,例如查询所有的部门
select array_agg(distinct deptno) from jinbo.employee; array_agg ----------- {20,30} (1 row) #不仅可以去重,还可以排序 select array_agg(distinct deptno order by deptno desc) from jinbo.employee; array_agg ----------- {30,20} (1 row)
以上为个人经验,希望能给大家一个参考,也希望大家多多支持好代码网。如有错误或未考虑完全的地方,望不吝赐教。