具有复杂业务规则的 CROSS JOINS 或 UNION ALL？

问题描述

我之所以创建这个是因为我对答案很好奇。

我有一个基本上有 4 列的表格：GroupA、GroupB、Tag 和 value。

A组	B组	标签	价值
A	E	PrefixYear3.SubCategoryASubSubcategoryA	100
B	F	PrefixYear3.SubCategoryBSubSubcategoryA	A
C	G	PrefixYear4.SubCategoryCSubSubcategoryZ	300

我知道 Year3 是 2010 年，Year4 是 2011 年等等，我有一个特定标签的列表，而不是在表格中，以及它们匹配的类别（一个标签可以属于多个类别）。

我需要返回一个包含以下内容的表格：

子类别A	子类别B	子类别C	年	类别
1000	2000	2900	2010	A
...	...	...	...	...
1550	0	450	2020	C

我正在尝试改进的查询具有以下算法：

SELECT GROUPA,GROUPB,SUM(
CASE WHEN Tag in ('PrefixYear3.SubCategoryASubSubcategoryA') THEN TRY_CAST([Value] as float)END)
) as SubCategoryA,...,'2010' AS Year
'CategoryA' AS Category
GROUP BY GroupA,GroupB

UNION ALL

...

UNION ALL

 SELECT GROUPA,SUM(
CASE WHEN Tag in ('PrefixYear5.SubCategoryASubSubcategoryA') THEN 
TRY_CAST([Value] as float)END) as SubCategoryA,'2012' AS Year
'CategoryA' AS Category
GROUP BY GroupA,GroupB

我发现这样做的方法比那个解决方案要多得多（三倍）：

SELECT A.GROUPA,A.GROUPB,CASE C.Category
    WHEN 'A' THEN
        CASE B.Year WHEN 2010 THEN [2010 SubCategoryA CategoryA]
        ...
        CASE B.Year WHEN 2012 THEN [2012 SubCategoryA CategoryA]
    END AS [SubCategoryA]
...
CASE C.Category
    WHEN 'C' THEN
        CASE B.Year WHEN 2010 THEN [2010 SubCategoryC CategoryA]
        ...
        CASE B.Year WHEN 2012 THEN [2012 SubCategoryC CategoryA]
    END AS [SubCategoryC]
B.Year,C.Category FROM
(
SELECT GROUPA,sum(CASE WHEN Tag in ('PrefixYear3.SubCategoryASubSubcategoryA') THEN TRY_CAST([Value] as float)END) AS '2010 SubCategoryA CategoryA',...
            sum(CASE WHEN Tag in ('PrefixYear5.SubCategoryASubSubcategoryA') THEN TRY_CAST([Value] as float)END) AS '2012 SubCategoryA CategoryA'
FROM TableA GROUP BY GROUPA,GROUPB)
CROSS JOIN (SELECT      '2010' as [Year]
             UNION
             SELECT     '2011'
             ...) as B
CROSS JOIN (SELECT 'CategoryA' as [Category]
    UNION SELECT 'CategoryB'
...) AS C

答案应该是视图。这意味着没有临时表或创建其他表是有效的答案。

关于如何更好地做到这一点的任何建议表示赞赏。提前致谢！

解决方法

您的示例数据与显示的预期结果不匹配，因此需要进行一些猜测，特别是对于未包含在所提供数据中的 category，但我相信您可以通过以下方式获得所需数据：

CREATE TABLE mytable(
   GroupA  VARCHAR(1) NOT NULL,Group_B VARCHAR(1) NOT NULL,Tag     VARCHAR(39) NOT NULL,Value   VARCHAR(3) NOT NULL
);
INSERT INTO mytable(GroupA,Group_B,Tag,Value) VALUES ('A','E','PrefixYear3.SubCategoryASubSubcategoryA','100');
INSERT INTO mytable(GroupA,Value) VALUES ('B','F','PrefixYear3.SubCategoryBSubSubcategoryA','A');
INSERT INTO mytable(GroupA,Value) VALUES ('C','G','PrefixYear4.SubCategoryCSubSubcategoryZ','300');

然后这个查询：

select
        SubCategoryA,SubCategoryB,SubCategoryC,Year,GroupA as Category
from (
    select
           *,case when tag like '%.SubCategoryA%' then try_cast(value as float) end as SubCategoryA,case when tag like '%.SubCategoryB%' then try_cast(value as float) end as SubCategoryB,case when tag like '%.SubCategoryC%' then try_cast(value as float) end as SubCategoryC,case when tag like 'PrefixYear3%' then 2010
               when tag like 'PrefixYear4%' then 2011
               else NULL
           end as [Year]
    from mytable
    ) as sq

产生这个输出：

+--------------+--------------+--------------+------+----------+
| SubCategoryA | SubCategoryB | SubCategoryC | Year | Category |
+--------------+--------------+--------------+------+----------+
|          100 |              |              | 2010 | A        |
|              |              |              | 2010 | B        |
|              |              |          300 | 2011 | C        |
+--------------+--------------+--------------+------+----------+

在 dbfiddle here

注意：我使用了 try_cast()，因为如果它无法从 varchar 值返回浮点数，它将优雅地返回 NULL。