Question

我有一个表，VehicleModelYear，包含列id，年份，品牌和型号。

以下两个查询按预期工作：

SELECT DISTINCT make, model
FROM VehicleModelYear

SELECT COUNT(DISTINCT make)
FROM VehicleModelYear

但是，此查询不起作用

SELECT COUNT(DISTINCT make, model)
FROM VehicleModelYear

很明显，答案是第一个查询返回的结果数，但只是想知道这个语法有什么问题或者为什么它不起作用。

Answer 1

COUNT()中的

SQL Server接受以下语法

COUNT(*)
COUNT(colName)
COUNT(DISTINCT colName)

您可以使用子查询返回可以计算的唯一make和model集。

SELECT  COUNT(*)
FROM
        (
            SELECT  DISTINCT make, model
            FROM    VehicleModelYear
        ) a

最后的“a”不是拼写错误。这是一个别名，没有它，SQL将给出错误ERROR 1248 (42000): Every derived table must have its own alias。

Answer 2

尝试将它们合并到一个字段中：

SELECT COUNT(DISTINCT make + ' ' + model)
FROM VehicleModelYear

Answer 3

语法是有效的SQL，但尚未在SQL-Server中实现。

尝试重写：

; WITH cte AS
  ( SELECT DISTINCT make, model
    FROM VehicleModelYear
  )
SELECT COUNT(*) AS total
FROM cte ;

或：

; WITH cte AS
  ( SELECT COUNT(DISTINCT model) AS cnt
    FROM VehicleModelYear
    GROUP BY make
  )
SELECT SUM(cnt) AS total
FROM cte ;

或：

; WITH cte AS
  ( SELECT NULL AS n
    FROM VehicleModelYear
    GROUP BY make, model
  )
SELECT COUNT(*) AS total
FROM cte ;

最后，修改了上面的第二个和第三个查询，没有子查询：

SELECT DISTINCT
    SUM(COUNT(DISTINCT model)) OVER () AS total
FROM VehicleModelYear
GROUP BY make ;

SELECT DISTINCT COUNT(*) OVER () AS total
FROM VehicleModelYear
GROUP BY make, model ;

@Alexander Fedorenko的后期添加：

SELECT TOP (1)
    SUM(COUNT(DISTINCT model)) OVER () AS total
FROM VehicleModelYear
GROUP BY make ;

SELECT TOP (1) COUNT(*) OVER () AS total
FROM VehicleModelYear
GROUP BY make, model ;

或：

; WITH cte AS
  ( SELECT DENSE_RANK() OVER(ORDER BY make, model) AS dr
    FROM VehicleModelYear
  )
SELECT MAX(dr) AS total
FROM cte ;

多列上的SELECT COUNT（DISTINCT ...）错误？

3 个答案: