转换表格数据并重新格式化

Question

我有一个名为Table的表

id	Level1	Level2	Level3
1	US	CA	13000
2	FR	PA	30000
3	US	CA	24000
4	US	LA	10000
5	UK	LN	500
6	UK	LN	600
7	FR	PA	888
8	FR	DF	1000

我想将其转换为以下格式（第一次转换），该格式将 Level2 和列组按 Level1 ' 如下所示

select Level1, level2, sum(level3) as level3
from table
group by level2
order by level1

然后（第二次转换）将具有相同Level1 的公共行合并为如下所示

Answer 1

您可以考虑将数据分组并使用STRING_AGG，如下所示

转化 1

下面的代码按Level1 和Level2 分组，并在重命名为LEVEL3 之前对Level3 求和

SELECT
    Level1,
    Level2,
    SUM(Level3) as Level3
FROM
    my_table
GROUP BY
    Level1,
    Level2;

Level1	Level2	Level3
US	CA	37000
FR	PA	30888
US	LA	10000
UK	LN	1100
FR	DF	1000

---

转化 2

下面的查询使用上一个查询的结果作为子查询（如果需要，也可以用作 CTE）并使用STRING_AGG 函数连接基于LEVEL1 的分组结果。如果需要，您可以另外订购结果。

    
SELECT
    Level1,
    STRING_AGG(Level2,',') as Level2,
    STRING_AGG(Level3,',') as Level3
FROM (
    SELECT
        Level1,
        Level2,
        SUM(Level3) as Level3
    FROM
        my_table
    GROUP BY
        Level1,
        Level2
) t
GROUP BY
    Level1

Level1	Level2	Level3
FR	PA,DF	30888,1000
UK	LN	1100
US	CA,LA	37000,10000

---

Working example on db-fiddle here

编辑 1：

对于不支持 STRING_AGG 的 SQL Server 版本，可以使用以下方法使用 JOIN 和 CROSS APPLY 与 FOR XML PATH 来连接列：

WITH group_1 as (
    SELECT
        Level1,
        Level2,
        CONVERT(VARCHAR(10),SUM(Level3)) as Level3
    FROM
        my_table
    GROUP BY
        Level1,
        Level2
),
level2_grouped AS (
    SELECT
        Level1,
        LEFT(MergedValues,LEN(MergedValues)-1) AS Level2
    FROM group_1 extern
    CROSS APPLY (
        SELECT
            Level2+','
        FROM
            group_1 intern
        WHERE
            extern.Level1 = intern.Level1
        FOR XML PATH('')
    )merged(MergedValues)
),
level3_grouped AS (
    SELECT
        Level1,
        LEFT(MergedValues,LEN(MergedValues)-1) AS Level3
    FROM group_1 extern
    CROSS APPLY (
        SELECT
            Level3+','
        FROM
            group_1 intern
        WHERE
            extern.Level1 = intern.Level1
        FOR XML PATH('')
    )merged(MergedValues)
)
SELECT
    g1.Level1,
    l2.Level2,
    l3.Level3
FROM 
    group_1 g1
INNER JOIN 
    level2_grouped l2 ON g1.Level1 = l2.Level1
INNER JOIN 
    level3_grouped l3 ON g1.Level1 = l3.Level1
GROUP BY
    g1.Level1,
    l2.Level2,
    l3.Level3

或者更简单

WITH group_1 as (
    SELECT
        Level1,
        Level2,
        CONVERT(VARCHAR(10),SUM(Level3)) as Level3
    FROM
        my_table
    GROUP BY
        Level1,
        Level2
),
level_grouped AS (
    SELECT
        Level1,
        LEFT(Merged2Values,LEN(Merged2Values)-1) AS Level2,
        LEFT(Merged3Values,LEN(Merged3Values)-1) AS Level3
    FROM group_1 extern
    CROSS APPLY (
        SELECT
            Level2+','
        FROM
            group_1 intern
        WHERE
            extern.Level1 = intern.Level1
        FOR XML PATH('')
    )merged2(Merged2Values)
    CROSS APPLY (
        SELECT
            Level3+','
        FROM
            group_1 intern
        WHERE
            extern.Level1 = intern.Level1
        FOR XML PATH('')
    )merged3(Merged3Values)
)
SELECT
    Level1,
    Level2,
    Level3
FROM 
    level_grouped g1
GROUP BY
    Level1,
    Level2,
    Level3

Working db fiddle

让我知道这是否适合你