Question

我正在使用SQL Server 2012.我的SQL遵循以下结构。

SELECT A.attributeA
       ,A.attributeB
       ,Count(A.*) AS CountA  -- I know this is wrong. 
       ,Count(B.*) AS CountB
FROM
(
    SELECT ... FROM Foo1
) A
FULL OUTER JOIN 
(
    SELECT ... FROM Foo2
) B
ON A.attribute1 = B.attribute1
GROUP BY
A.attributeA
,A.attributeB

我想从子查询A和B中获取所有行的计数。我该怎么做？先感谢您。

Answer 1

假设目标是仅计算连接每一侧的非空记录，您可以指定您希望非空的列名（如注释中所述），通常与列中的列相同你的加入。例如，因为您加入了attribute1：

SELECT A.attributeA
       ,A.attributeB
       ,Count(A.attribute1) AS CountA 
       ,Count(B.attribute1) AS CountB
FROM
...

请注意，这不会告诉您2重叠的时间，如果这是您的目标的一部分。对于这种类型的计数，您可以使用SUM结合CASE：

SELECT A.attributeA
       ,A.attributeB
       ,Count(A.attribute1) AS CountA 
       ,Count(B.attribute1) AS CountB
       ,SUM(CASE WHEN A.attribute1 IS NOT NULL AND B.attribute1 IS NOT NULL
                 THEN 1 ELSE 0 
            END) as CountAAndBOverlap
FROM
...

Answer 2

如果您的目的只是获取两个子查询的计数，您可以执行类似这样的操作。我还为这两个子查询生成了常见的ID，以便我JOIN。最后，我使用了INNER JOIN代替FULL JOIN。

SELECT 
      CountSubqueryA AS CountA,
      CountSubqueryB AS CountB
FROM
(SELECT 1 AS ID,COUNT(*) AS CountSubqueryA FROM Foo1 ) AS A
INNER JOIN 
(SELECT 1 AS ID,COUNT(*) AS CountSubqueryB FROM Foo2 ) AS B
ON A.ID=B.ID

Answer 3

我看到了你的GROUP BY，想到也许你想要这个。它最多可以创建3个组，如下所示：

==========================
|   A   | NOT B |   24   |
--------------------------
| NOT A |   B   |   31   |
--------------------------
|   A   |   B   |   69   |
==========================

SELECT
    CASE WHEN A.attribute1 IS NOT NULL THEN 'A' ELSE 'NOT A' END,
    CASE WHEN B.attribute1 IS NOT NULL THEN 'B' ELSE 'NOT B' END
    COUNT(*)
FROM
(
    SELECT ... FROM Foo1
) A
FULL OUTER JOIN 
(
    SELECT ... FROM Foo2
) B
ON A.attribute1 = B.attribute1
GROUP BY
    CASE WHEN A.attribute1 IS NOT NULL THEN 'A' ELSE 'NOT A' END,
    CASE WHEN B.attribute1 IS NOT NULL THEN 'B' ELSE 'NOT B' END

在全外连接中计算子查询的数量

3 个答案: