错误：子查询必须只返回一列

Question

我在尝试运行以下查询时收到错误 subquery must return only one column：

SELECT mat.mat as mat1, sum(stx.total ) as sumtotal1,
  (
    SELECT mat.mat  as mat, sum(stx.total)  as sumtotal
    FROM stx 
      LEFT JOIN mat ON stx.matid = mat.matid
      LEFT JOIN sale ON stx.saleid = sale.id
    WHERE stx.date BETWEEN '2013-05-01' AND '2013-08-31' 
      AND sale.userid LIKE 'A%'
    GROUP BY mat.mat
) AS MyField
FROM stx 
  LEFT JOIN mat ON stx.matid = mat.matid
  LEFT JOIN sale ON stx.saleid = sale.id
WHERE stx.date BETWEEN '2013-05-01' AND '2013-08-31'
  AND sale.userid LIKE 'B%'
GROUP BY mat.mat

是什么导致了这个错误？

Answer 1

代替子查询选择语句

SELECT mat.mat  as mat, sum(stx.total)  as sumtotal

试试这个说法

SELECT sum(stx.total)  as sumtotal

Answer 2

在FROM 列表中放置一个返回多列的子查询并从中进行选择。

开始时关联子查询是个坏主意。但是，您的查询甚至不相关，而是不相关（没有指向外部查询的链接）并且似乎返回多行。这会导致（可能非常昂贵且荒谬）交叉连接产生笛卡尔积，可能不是您的（秘密）意图。

看起来你真的很想要：

SELECT m1.mat AS mat1, m1.sumtotal AS sumtotal1
      ,m2.mat AS mat2, m2.sumtotal AS sumtotal2
FROM (
   SELECT mat.mat, sum(stx.total) AS sumtotal
   FROM   stx 
   LEFT   JOIN mat ON mat.matid = stx.matid
   LEFT   JOIN sale ON stx.saleid = sale.id
   WHERE  stx.date BETWEEN '2013-05-01' AND '2013-08-31'
   AND    sale.userid LIKE 'A%'
   GROUP  BY mat.mat
   ) m1
JOIN  (
   SELECT mat.mat, sum(stx.total) AS sumtotal
   FROM   stx 
   LEFT   JOIN mat ON mat.matid = stx.matid
   LEFT   JOIN sale ON sale.id = stx.saleid
   WHERE  stx.date BETWEEN '2013-05-01' AND '2013-08-31' 
   AND    sale.userid LIKE 'b%'
   GROUP  BY mat.mat
   ) m2 USING (mat);

LEFT JOIN 也毫无意义。 sale 上的那个被 WHERE 条件强制为 INNER JOIN。垫子上的那个似乎毫无意义，因为你 GROUP BY mat.mat - 除非你对 mat IS NULL 感兴趣？ （我怀疑。）

这个案例大概可以进一步简化为：

SELECT m.mat
      ,sum(CASE WHEN s.userid LIKE 'A%' THEN x.total END) AS total_a
      ,sum(CASE WHEN s.userid LIKE 'B%' THEN x.total END) AS total_b
FROM   sale s 
JOIN   stx  x ON x.saleid = s.id
JOIN   mat  m ON m.matid = x.matid
WHERE (s.userid LIKE 'A%' OR s.userid LIKE 'B%')
AND    x.date BETWEEN '2013-05-01' AND '2013-08-31'
GROUP  BY 1;

WHERE 条件可能可以进一步简化，具体取决于您的秘密数据类型和索引。 this related answer on dba.SE 中有大量关于该案例的信息。