3

我有下表：

UID | ID  | Type
1   | 1   | product
1   | 2   | product
1   | 3   | service
1   | 4   | product
1   | 5   | product
2   | 6   | service
1   | 7   | order
2   | 8   | invoice
2   | 9   | product

我想结束：

UID | product | service | invoice | order
1   |  4      |  1      |  0      |  1
2   |  1      |  1      |  1      |  0

SQL 查询会是什么样子？或者至少，最足够的一个？

4

5 回答 5

7

如果你真的只需要这四种类型，那么你可以硬编码这些值，如下所示：

select UID,
    count(case when type='product' then 1 else null end) as product,
    count(case when type='service' then 1 else null end) as service,
    count(case when type='invoice' then 1 else null end) as invoice,
    count(case when type='order' then 1 else null end) as order
from MyTable
group by UID
order by UID
于 2009-09-26T20:42:12.553 回答
7

您要做的是一个透视操作，SQL 语法不直接支持它。但是，它并不太复杂，并且在概念上涉及 2 个步骤：

  1. 将数据“放大”为多列，原始数据集中每行一行。这通常使用 CASE WHEN ... ELSE ... END 或偶尔使用函数（如 oracle 中的 decode() ）来完成。我将在下面的示例中使用 CASE WHEN，因为它同样适用于大多数 RDBMS
  2. 使用 GROUP BY 和聚合函数（SUM、MIN、MAX 等）将许多行折叠到所需的输出行集中。

我正在使用这个数据集作为示例：

mysql> select * from foo;
+------+------+---------+
| uid  | id   | type    |
+------+------+---------+
|    1 |    1 | product | 
|    1 |    2 | product | 
|    1 |    3 | service | 
|    1 |    4 | product | 
|    1 |    5 | product | 
|    2 |    6 | service | 
|    1 |    7 | order   | 
|    2 |    8 | invoice | 
|    2 |    9 | product | 
+------+------+---------+

第 1 步是“炸毁”数据集：

select uid
     , case when type = 'product' then 1 else 0 end as is_product
     , case when type = 'service' then 1 else 0 end as is_service
     , case when type = 'invoice' then 1 else 0 end as is_invoice
     , case when type = 'order' then 1 else 0 end as is_order
  from foo;

这使：

+------+------------+------------+------------+----------+
| uid  | is_product | is_service | is_invoice | is_order |
+------+------------+------------+------------+----------+
|    1 |          1 |          0 |          0 |        0 | 
|    1 |          1 |          0 |          0 |        0 | 
|    1 |          0 |          1 |          0 |        0 | 
|    1 |          1 |          0 |          0 |        0 | 
|    1 |          1 |          0 |          0 |        0 | 
|    2 |          0 |          1 |          0 |        0 | 
|    1 |          0 |          0 |          0 |        1 | 
|    2 |          0 |          0 |          1 |        0 | 
|    2 |          1 |          0 |          0 |        0 | 
+------+------------+------------+------------+----------+

接下来，我们在每个日期的输出中折叠到一行，并对每个 is_* 列求和，使用或初始查询作为内联视图（也称为“子查询”）：

select uid
     , sum(is_product) as count_product
     , sum(is_service) as count_service
     , sum(is_invoice) as count_invoice
     , sum(is_order)   as count_order
  from (
         select uid
              , case when type = 'product' then 1 else 0 end as is_product
              , case when type = 'service' then 1 else 0 end as is_service
              , case when type = 'invoice' then 1 else 0 end as is_invoice
              , case when type = 'order' then 1 else 0 end as is_order
           from foo
       ) x
 group by uid;

（还请注意，您可以将这两个查询合并为一个，尽管为了清楚起见，我在这里分别显示它们；至少在 MySQL 中，这似乎导致执行计划更简单，这通常意味着执行速度更快——与往常一样，测试你在真实数据集上的 SQL 性能，不要相信我的话！）

这给了我们：

+------+---------------+---------------+---------------+-------------+
| uid  | count_product | count_service | count_invoice | count_order |
+------+---------------+---------------+---------------+-------------+
|    1 |             4 |             1 |             0 |           1 | 
|    2 |             1 |             1 |             1 |           0 | 
+------+---------------+---------------+---------------+-------------+

这是期望的结果。

于 2009-09-26T20:47:13.470 回答
2

您必须在 MySQL 中使用CASE 语句将行数据转换为列（反之亦然）：

  SELECT t.uid,
         SUM(CASE WHEN t.type = 'product' THEN COUNT(*) END) as PRODUCT,
         SUM(CASE WHEN t.type = 'service' THEN COUNT(*) END) as SERVICE,
         SUM(CASE WHEN t.type = 'invoice' THEN COUNT(*) END) as INVOICE,
         SUM(CASE WHEN t.type = 'order' THEN COUNT(*) END) as ORDER
    FROM TABLE t
GROUP BY t.uid, t.type
于 2009-09-26T20:44:09.913 回答
1

您正在寻找一种称为“数据透视表”的东西，这不是 MySQL 本身可以做的事情。

在这种情况下，您可能最好转换应用程序中的数据并使用 GROUP BY 和/或 DISTINCT 的某种组合来收集您正在寻找的数据。这个查询可能有效，我没有测试过：

SELECT Type, COUNT(ID), UID
  FROM tablename
 GROUP BY UID, Type
于 2009-09-26T20:38:01.673 回答
-3

从 mytable 中选择计数（产品、服务、发票、订单）

于 2009-09-26T20:34:22.553 回答