将多个表中的值按公共列分组

Question

我在 MS SQL Server 2014 中有三个表。每个表都包含几个数值、一个描述和一个日期。为简洁起见，我们假设以下表格：

table "beverages"

day        beverage amount
---------- -------- ------
2018-12-01 water    2
2018-12-01 tea      1
2018-12-01 coffee   7
2018-12-02 water    4
2018-12-02 tea      2


table "meals"

day        meal   amount
---------- ------ ------
2018-12-01 burger 1
2018-12-01 bread  2
2018-12-02 steak  1


table "fruit"

day        fruit  amount
---------- ------ ------
2018-12-01 apple  4
2018-12-01 banana 1
2018-12-02 apple  2

然后我有另一个表只包含日期列表。

table "dates"

day
----------
2018-12-01
2018-12-02

我需要的是一个查询，它为dates 表中的每一行返回一行，每一行都有日期、饮料总量、膳食总量和水果总量为那一天。我不关心不同类型的饮料、膳食和水果，只关心总和。结果应该是：

expected result

day        beverages   meals       fruit
---------- ----------- ----------- -----------
2018-12-01 10          3           5
2018-12-02 6           1           2 

但是我收到了

received result

day        beverages   meals       fruit
---------- ----------- ----------- -----------
2018-12-01 40          18          30
2018-12-02 6           2           4

我已经知道问题出在哪里，只是不知道如何解决。更糟糕的是，我确信我曾经知道答案，但现在我什至无法找出正确的搜索词来让 Google 告诉我......

当我进行这样的查询时（我使用表变量进行测试）

SELECT 
     [d].[day]
    ,SUM([b].[amount]) AS [beverages]
    ,SUM([m].[amount]) AS [meals]
    ,SUM([f].[amount]) AS [fruit]
FROM @dates AS [d]
LEFT OUTER JOIN @beverages AS [b]
    ON [d].[day] = [b].[day]
LEFT OUTER JOIN @meals AS [m]
    ON [d].[day] = [m].[day]
LEFT OUTER JOIN @fruit AS [f]
    ON [d].[day] = [f].[day]
GROUP BY [d].[day]

它不止一次地对不同表中的每一行求和，因为它返回三个表的所有可能组合。删除 SUM() 和 GROUP BY 证明：

day        beverages   meals       fruit
---------- ----------- ----------- -----------
2018-12-01 2           1           4
2018-12-01 2           1           1
2018-12-01 2           2           4
2018-12-01 2           2           1
2018-12-01 1           1           4
2018-12-01 1           1           1
2018-12-01 1           2           4
2018-12-01 1           2           1
2018-12-01 7           1           4
2018-12-01 7           1           1
2018-12-01 7           2           4
2018-12-01 7           2           1
2018-12-02 4           1           2
2018-12-02 2           1           2

那么，我需要在查询中进行哪些更改，以使其对三个表中每个表的值求和，而不将其与其他表中的行数相乘？

Answer 1

在加入之前对表格进行分组，如下所示：

SELECT 
     [d].[day]
    ,[b].[amount] AS [beverages]
    ,[m].[amount] AS [meals]
    ,[f].[amount] AS [fruit]
FROM @dates AS [d]
LEFT OUTER JOIN (SELECT day, SUM(amount) as amount FROM @beverages GROUP BY day) AS [b]
    ON [d].[day] = [b].[day]
LEFT OUTER JOIN (SELECT day, SUM(amount) as amount FROM @meals GROUP BY day) AS [m]
    ON [d].[day] = [m].[day]
LEFT OUTER JOIN (SELECT day, SUM(amount) as amount FROM @fruit GROUP BY day) AS [f]
    ON [d].[day] = [f].[day]

Answer 2

改为 PIVOT 怎么样？

示例

Select *
 From (
        Select day,Item='beverage',amount from beverages
        Union All
        Select day,Item='meals'   ,amount from meals
        Union All
        Select day,Item='fruit'   ,amount from fruit
      ) src
Pivot ( sum(amount) for Item in ([beverages],[meals],[fruit]) ) pvt