SQL语句复习--group by 的用法

今天自己在写需求的时候,运行自己写过的sql,总是报不是group by 分组函数,自己都搞了两个小时了,实在没有办法,最后请教林哥,记得上次请教林哥的也是一个关于group by函数的使用方法,并且,人家已经叮嘱自己下来之后上网上多看看group by函数的使用方法,看来自己真的应该长点记性了,不要再不会用了,因为这可是工作呀!就得要有认真的态度。

select lb.contno,
        decode(cont.conttype, '1', cont.appntno, '2', cont.insuredno),
        decode(cont.conttype, '1', cont.appntname, '2', cont.insuredname),
        to_date(lb.create_time),
        to_char(lb.sum_total, '9999999990.99'),
        to_char(lb.sum_price, '9999999990.99'),
        to_char(lb.sum_tax, '9999999990.99'),
        max(lb.sid),
        lb.managecom
   from LIS_BUSI_TRANSACTIONS lb, lccont cont
  where '1581260514000' = '1581260514000'
    and cont.contno = lb.contno
    and lb.invoiceflag in ('00')
    and lb.successflag = '1'
    and not exists
  (select 1 from lcgrpcont c where c.grpcontno = lb.contno)
    and lb.sum_total > 0
    and (cont.currency = '01' or cont.currency is null)
    and not exists (select 1
           from ljagetendorse a, LIS_BUSI_TRANSACTIONS b
          where a.actugetno = b.sourceid
            and a.getflag = '1'
            and b.ruleid in ('3', '6', '10', '12')
            and a.actugetno = lb.sourceid)
    and exists (SELECT 1 FROM ljapay d where d.otherno = cont.contno)
/*    and lb.contno = '1'
    and to_date(lb.create_time) >= '2019-02-10'
    and to_date(lb.create_time) <= '2020-02-14'
    and lb.managecom like '86%'
    and lb.managecom like '86%'*/
    group by lb.contno,
        decode(cont.conttype, '1', cont.appntno, '2', cont.insuredno),
        decode(cont.conttype, '1', cont.appntname, '2', cont.insuredname),
        to_date(lb.create_time),
        to_char(lb.sum_total, '9999999990.99'),
        to_char(lb.sum_price, '9999999990.99'),
        to_char(lb.sum_tax, '9999999990.99'),
        lb.sid,
        lb.managecom

  现在上面的函数是能够运行的,group by 函数就是要将查询结果中的每一列都要放到group by的后面。自己当时在写这个sql的时候,也不是怎么搞的,就是报错,可能是因为当时group by后面少放了字段的原因,看来遇到问题不要慌张,要冷静下来,仔细思考,结果一问林哥,人家也没有说啥,自己又尝试着改了一遍就好使了。

不过林哥说:你见哪个开发这样写sql了。

SQL语句复习--group by 的用法_第1张图片

 

 下面这一段,确实,自己都看着恶心。改造一下呗!

select lb.contno,
        decode(cont.conttype, '1', cont.appntno, '2', cont.insuredno),
        decode(cont.conttype, '1', cont.appntname, '2', cont.insuredname),
        to_date(lb.create_time),
        to_char(lb.sum_total, '9999999990.99'),
        to_char(lb.sum_price, '9999999990.99'),
        to_char(lb.sum_tax, '9999999990.99'),
        max(lb.sid),
        lb.managecom
   from LIS_BUSI_TRANSACTIONS lb, lccont cont
  where '1581260514000' = '1581260514000'
    and cont.contno = lb.contno
    and lb.invoiceflag in ('00')
    and lb.successflag = '1'
    and not exists
  (select 1 from lcgrpcont c where c.grpcontno = lb.contno)
    and lb.sum_total > 0
    and (cont.currency = '01' or cont.currency is null)
    and not exists (select 1
           from ljagetendorse a, LIS_BUSI_TRANSACTIONS b
          where a.actugetno = b.sourceid
            and a.getflag = '1'
            and b.ruleid in ('3', '6', '10', '12')
            and a.actugetno = lb.sourceid)
    and exists (SELECT 1 FROM ljapay d where d.otherno = cont.contno)
/*    and lb.contno = '1'
    and to_date(lb.create_time) >= '2019-02-10'
    and to_date(lb.create_time) <= '2020-02-14'
    and lb.managecom like '86%'
    and lb.managecom like '86%'*/
    group by lb.contno,
        cont.conttype, cont.appntno, cont.insuredno, cont.appntname,  cont.insuredname,
        lb.create_time,
        lb.sum_total,
        lb.sum_price, 
        lb.sum_tax,
        lb.sid,
        lb.managecom

  这样就好看多了,哈哈!

值得注意的是,像查询结果中使用到的decode函数,等等,一定要将函数中使用到的本表中的字段,一个不落的写在group by的后面才行。

假如我给要查询的结果列起别名,行不行呢?

select lb.contno a,
        decode(cont.conttype, '1', cont.appntno, '2', cont.insuredno) b,
        decode(cont.conttype, '1', cont.appntname, '2', cont.insuredname) c,
        to_date(lb.create_time) d,
        to_char(lb.sum_total, '9999999990.99') e,
        to_char(lb.sum_price, '9999999990.99') f,
        to_char(lb.sum_tax, '9999999990.99') g,
        max(lb.sid) h,
        lb.managecom j
   from LIS_BUSI_TRANSACTIONS lb, lccont cont
  where '1581260514000' = '1581260514000'
    and cont.contno = lb.contno
    and lb.invoiceflag in ('00')
    and lb.successflag = '1'
    and not exists
  (select 1 from lcgrpcont c where c.grpcontno = lb.contno)
    and lb.sum_total > 0
    and (cont.currency = '01' or cont.currency is null)
    and not exists (select 1
           from ljagetendorse a, LIS_BUSI_TRANSACTIONS b
          where a.actugetno = b.sourceid
            and a.getflag = '1'
            and b.ruleid in ('3', '6', '10', '12')
            and a.actugetno = lb.sourceid)
    and exists (SELECT 1 FROM ljapay d where d.otherno = cont.contno)
/*    and lb.contno = '1'
    and to_date(lb.create_time) >= '2019-02-10'
    and to_date(lb.create_time) <= '2020-02-14'
    and lb.managecom like '86%'
    and lb.managecom like '86%'*/
    group by a,b,c,d,e,f,g,h,j

  答案是不行的,会报J标识符无效。

哦!对了,这个是oracle数据库的。

 

你可能感兴趣的:(SQL语句复习--group by 的用法)