MYSQL查询中GROUP_CONCAT的替代方案

Question

有人可以让我知道如何更改我的查询，以便它不使用GROUP_CONCAT 。 该查询适用于MYSQL和PHP，但我需要在Filemaker Pro运行相同的查询，但Filemaker不支持GROUP CONCAT。

SELECT sqlvalue
CAST(GROUP_CONCAT(IF(stat_number = 693, sqlvalue, NULL)) AS UNSIGNED) AS S693,
CAST(GROUP_CONCAT(IF(stat_number = 698, sqlvalue, NULL)) AS UNSIGNED) AS S698,
CAST(GROUP_CONCAT(IF(stat_number = 2010, sqlvalue, NULL)) AS UNSIGNED) AS S2010,
CAST(GROUP_CONCAT(IF(stat_number = 2020, sqlvalue, NULL)) AS UNSIGNED) AS S2020,
IFNULL(CAST(GROUP_CONCAT(IF(stat_number = 693, sqlvalue, NULL)) AS UNSIGNED),0) +
IFNULL(CAST(GROUP_CONCAT(IF(stat_number = 698, sqlvalue, NULL)) AS UNSIGNED),0) +
IFNULL(CAST(GROUP_CONCAT(IF(stat_number = 2010, sqlvalue, NULL)) AS UNSIGNED),0) +
IFNULL(CAST(GROUP_CONCAT(IF(stat_number = 2020, sqlvalue, NULL)) AS UNSIGNED),0) + AS STOTAL
FROM statsvalue
WHERE sqldate > ’01-07-2013′
ORDER BY sqldate ASC
GROUP By sqldate

field stat_number是给输入记录的代码， sqlvalue是数值， sqldate是日期。

上面的查询获取这些stat_number的值并将它们加在一起， sqldate field对总计进行sqldate field 。

Answer 1

也许我错过了一些东西，但在这种特殊情况下你不需要GROUP_CONCAT() （做条件聚合）。

因此恕我直言，您可以继续使用MAX()更改GROUP_CONCAT() ，这可能更有意义，至少对于读取查询。

我不知道Filemaker，但您也可以考虑改变

1. IF()到CASE END
2. IFNULL()到COALESCE()

除此之外，您的查询还有一些其他问题：

1. 即使MySQL允许你不应该在SELECT子句中使用sqlvalue列而没有将聚合函数应用于它作为group by的一部分。 它只是没有意义，其他主要的RDBMS（SQL Server，Oracle，Postgres）不允许它。 您最有可能打算使用sqldate 。
2. 即使你把sqlvalue放在SELECT也应该在列名之后使用逗号
3. 在AS STOTAL之前有一个额外的+权利
4. 使用日期字符串文字时，您可能需要使用正确的引号而不是智能引号
5. 您最好使用消除歧义的日期文字（而不是'01 -07-2013'，您应该使用'2013-07-01'或'2013-01-07'，具体取决于您的意思）
6. ORDER BY应该在GROUP BY

这就是说您的查询可能看起来像

SELECT sqldate,
       CAST(MAX(CASE WHEN stat_number = 693  THEN sqlvalue END) AS UNSIGNED) AS S693,
       CAST(MAX(CASE WHEN stat_number = 698  THEN sqlvalue END) AS UNSIGNED) AS S698,
       CAST(MAX(CASE WHEN stat_number = 2010 THEN sqlvalue END) AS UNSIGNED) AS S2010,
       CAST(MAX(CASE WHEN stat_number = 2020 THEN sqlvalue END) AS UNSIGNED) AS S2020,
       COALESCE(MAX(CASE WHEN stat_number = 693  THEN sqlvalue END) AS UNSIGNED), 0) +
       COALESCE(MAX(CASE WHEN stat_number = 698  THEN sqlvalue END) AS UNSIGNED), 0) +
       COALESCE(MAX(CASE WHEN stat_number = 2010 THEN sqlvalue END) AS UNSIGNED), 0) +
       COALESCE(MAX(CASE WHEN stat_number = 2020 THEN sqlvalue END) AS UNSIGNED), 0) AS STOTAL
  FROM statsvalue
 WHERE sqldate > '2013-01-07'
 GROUP BY sqldate
 ORDER BY sqldate

或使用这样的内联视图

SELECT sqldate, S693, S698, S2010, S2020,
       COALESCE(S693,  0) + 
       COALESCE(S698,  0) + 
       COALESCE(S2010, 0) + 
       COALESCE(S2020, 0) AS STOTAL
  FROM
(
  SELECT sqldate
         CAST(MAX(CASE WHEN stat_number = 693  THEN sqlvalue END) AS UNSIGNED) AS S693,
         CAST(MAX(CASE WHEN stat_number = 698  THEN sqlvalue END) AS UNSIGNED) AS S698,
         CAST(MAX(CASE WHEN stat_number = 2010 THEN sqlvalue END) AS UNSIGNED) AS S2010,
         CAST(MAX(CASE WHEN stat_number = 2020 THEN sqlvalue END) AS UNSIGNED) AS S2020
    FROM statsvalue
   WHERE sqldate > '2013-01-07'
   GROUP BY sqldate
) q
  ORDER BY sqldate