SQL Query显示树结构中的子项总数

时间:2016-05-23 15:51:57

标签: sql sql-server tsql common-table-expression

我正在尝试计算树结构的父节点的总计,并且由于某种原因,计算父节点的总值正在逃避我。

假设我有三张桌子

地区

GroupID     ParentID      Name
1           null          NorthAmerica
2           null          Asia
3           null          Europe
4           1             NorthEast
5           1             WestCoast
6           3             UK
7           3             Germany
8           2             Hong Kong
9           2             Japan

RegionMember

GroupID     EmpID   
4           10000   
4           10001   
5           10011   
6           20455   
6           10003   
7           34567   
9           43589   
9           54890   
8           84320   
8           84560   

EmployeeSales

EmployeeID     Name     Sales ($)
10000          Joe      $ 150,000.00 
10001          Mary     $ 200,000.00 
10011          John     $ 175,000.00 
20455          Fred     $ 100,000.00 
10003          Bill     $ 250,000.00 
34567          Abe      $ 142,000.00 
43589          Jack     $ 260,000.00 
54890          Amanda   $ 300,000.00 
84320          Jane     $ 15,000.00 
84560          Oscar    $ 175,000.00 

目标是查询树中的不同级别并查看这些区域的总计 例如,一个视图将显示具有销售总额的顶部区域:

NorthAmerica    525,000.00 *(The sum of NorthEast and WestCoast)*
Asia            750,000.00 *(The sum of Hong Kong and Japan)*
Europe          492,000.00 *(The sum of UK and Germany)*

另一种观点将显示区域总数(专注于单亲):

NorthAmerica    525,000.00 *(Total of the region members NorthEast and WestCoast)*
Northeast       350,000.00 *(Total of NorthEast Leaves Joe and Mary)*
WestCoast       175,000.00 *(Total of WestCoast Leaves John)*

当然,这些树在分支方面可以更深入,但我认为这个例子说明了我正在战斗的问题。

到目前为止,使用CTE我可以相当轻松地导航树结构,我可以得到最终分支(或分支的叶子)的总数,但我似乎无法通过总结来获得总结。

所以从上面的例子中,我可以得到以下输出:

NorthAmerica    NULL
NorthEast       350,000.00
WestCoast       175,000.00

我会提供现有代码,但实际表和连接数在我的实际表中有很大不同,可能只会混淆整体目标。但是,这个问题类似于我想要完成的问题,但它似乎并不完全适合填充:

CTE Sum of Child Levels

非常感谢任何帮助。

构建脚本关注:

create table Regions 
(
GroupID int, 
ParentID int,
Name Varchar(40)
)

create table RegionMember
(
GroupID int,
empid int
)

Create Table EmployeeSales
(
EmployeeID int,
Name Varchar(50),
Sales float,
)

Insert into Regions Values
(1, null, 'NorthAmerica'),
(2, null, 'Asia'),
(3, null, 'Europe'),
(4, 1, 'NorthEast'),
(5, 1, 'WestCoast'),
(6, 3, 'UK'),
(7, 3, 'Germany'),
(8, 2, 'Hong Kong'),
(9, 2, 'Japan');

Insert into RegionMember Values
(4, 10000),   
(4, 10001),   
(5, 10011),  
(6, 20455),   
(6, 10003),
(7, 34567),   
(9, 43589),   
(9, 54890),   
(8, 84320),   
(8, 84560);

Insert into EmployeeSales Values
(10000, 'Joe', 150000), 
(10001, 'Mary', 200000), 
(10011, 'John', 175000), 
(20455, 'Fred', 100000),
(10003, 'Bill', 250000),
(34567, 'Abe', 142000), 
(43589, 'Jack', 260000), 
(54890, 'Amanda', 300000), 
(84320, 'Jane', 15000), 
(84560, 'Oscar', 175000); 

还开始使用上面的SQL小提琴:http://sqlfiddle.com/#!6/4ee0c/1

1 个答案:

答案 0 :(得分:1)

我在示例数据中添加了几行,因为原始版本太简单了。这有三个层次。

Insert into Regions Values
(10, null, 'A1'),
(40, 10, 'B1'),
(50, 10, 'B2'),
(60, 10, 'B3'),
(70, 40, 'C1'),
(80, 40, 'C2');

Insert into RegionMember Values
(40, 104),
(50, 105),
(60, 106),
(70, 107),
(80, 108);

Insert into EmployeeSales Values
(104, '104', 104),
(105, '105', 105),
(106, '106', 106),
(107, '107', 107),
(108, '108', 108);

热门地区

此查询是直接递归CTE,它从最高级别(WHERE ParentID IS NULL)开始并总结其所有子级。这里的“技巧”是在我们遍历树时包含组的原始StartIDStartName,因此我们最后可以GROUP BY

WITH
CTE
AS
(
    SELECT
        Regions.GroupID AS StartID
        ,Regions.Name AS StartName
        ,Regions.GroupID
        ,Regions.ParentID
        ,Regions.Name
        ,1 AS Lvl
    FROM Regions
    WHERE ParentID IS NULL

    UNION ALL

    SELECT
        CTE.StartID
        ,CTE.StartName
        ,Regions.GroupID
        ,Regions.ParentID
        ,Regions.Name
        ,CTE.Lvl + 1 AS Lvl
    FROM
        Regions
        INNER JOIN CTE ON CTE.GroupID = Regions.ParentID
)
SELECT
    CTE.StartID
    ,CTE.StartName
    ,SUM(EmployeeSales.Sales) AS SumSales
FROM
    CTE
    INNER JOIN RegionMember ON RegionMember.GroupID = CTE.GroupID
    INNER JOIN EmployeeSales ON EmployeeSales.EmployeeID = RegionMember.empid
GROUP BY
    CTE.StartID
    ,CTE.StartName
ORDER BY
    CTE.StartID;

逐步运行查询以了解其工作原理。

<强>结果

+---------+--------------+----------+
| StartID |  StartName   | SumSales |
+---------+--------------+----------+
|       1 | NorthAmerica |   525000 |
|       2 | Asia         |   750000 |
|       3 | Europe       |   492000 |
|      10 | A1           |      530 |
+---------+--------------+----------+

区域总计和小计

第二个查询并不那么容易。第一部分CTE_Groups与上一个查询非常相似,但具有特定起始GroupID的过滤器。 CTE_Sums计算起始组及其每个子组的销售摘要。 CTE_Totals再次以递归方式查看CTE_Sums的结果,并根据需要重复子行,以获取每个组的总计,包括子项摘要。

再次,逐步运行查询CTE-by-CTE以了解其工作原理。 并非所有列都用于最终结果,但它们有助于了解中间步骤中发生的情况。

WITH
CTE_Groups
AS
(
    SELECT
        Regions.GroupID AS StartID
        ,Regions.Name AS StartName
        ,Regions.GroupID
        ,Regions.ParentID
        ,Regions.Name
        ,1 AS Lvl
    FROM Regions
    WHERE Regions.GroupID = 1 -- North America
    --WHERE Regions.GroupID = 10

    UNION ALL

    SELECT
        CTE_Groups.StartID
        ,CTE_Groups.StartName
        ,Regions.GroupID
        ,Regions.ParentID
        ,Regions.Name
        ,CTE_Groups.Lvl + 1 AS Lvl
    FROM
        Regions
        INNER JOIN CTE_Groups ON CTE_Groups.GroupID = Regions.ParentID
)
,CTE_Sums
AS
(
    SELECT
        CTE_Groups.GroupID
        ,CTE_Groups.ParentID
        ,CTE_Groups.Name
        ,SUM(EmployeeSales.Sales) AS SumSales
    FROM
        CTE_Groups
        LEFT JOIN RegionMember ON RegionMember.GroupID = CTE_Groups.GroupID
        LEFT JOIN EmployeeSales ON EmployeeSales.EmployeeID = RegionMember.empid
    GROUP BY
        CTE_Groups.GroupID
        ,CTE_Groups.ParentID
        ,CTE_Groups.Name
)
,CTE_Totals
AS
(
    SELECT
        CTE_Sums.GroupID AS StartID
        ,CTE_Sums.Name AS StartName
        ,CTE_Sums.GroupID
        ,CTE_Sums.ParentID
        ,CTE_Sums.Name
        ,CTE_Sums.SumSales
        ,1 AS Lvl
    FROM CTE_Sums

    UNION ALL

    SELECT
        CTE_Totals.StartID
        ,CTE_Totals.StartName
        ,CTE_Sums.GroupID
        ,CTE_Sums.ParentID
        ,CTE_Sums.Name
        ,CTE_Totals.SumSales
        ,CTE_Totals.Lvl + 1 AS Lvl
    FROM
        CTE_Sums
        INNER JOIN CTE_Totals ON CTE_Totals.ParentID = CTE_Sums.GroupID
)
SELECT
    GroupID
    ,Name
    ,SUM(SumSales) AS SumTotal
FROM CTE_Totals
GROUP BY
    GroupID
    ,Name
ORDER BY
    GroupID
    ,Name
;

GroupID = 1的结果

+---------+--------------+----------+
| GroupID |     Name     | SumTotal |
+---------+--------------+----------+
|       1 | NorthAmerica |   525000 |
|       4 | NorthEast    |   350000 |
|       5 | WestCoast    |   175000 |
+---------+--------------+----------+

GroupID = 10的结果

+---------+------+----------+
| GroupID | Name | SumTotal |
+---------+------+----------+
|      10 | A1   |      530 |
|      40 | B1   |      319 |
|      50 | B2   |      105 |
|      60 | B3   |      106 |
|      70 | C1   |      107 |
|      80 | C2   |      108 |
+---------+------+----------+