Question

任何人都可以帮我提出下面的查询？目标是转换实体，如预期输出所示。

Child   Parent  Name
0001    0001    HQ
0100    0001    HQ Accounting Dept
0200    0001    HQ Marketing Dept
0300    0001    HQ HR Dept
0101    0100    Branch North 111 
0102    0100    Branch North 112
0201    0200    Branch North 113
0301    0300    Branch North 114
8900    0300    Branch North 115
0387    8900    Sub Branch North 115

Expected output
----------------
Level1   Level2   Level3   Level4  Name
0001     0100     0101     N/A     Branch North 111
0001     0100     0102     N/A     Branch North 112
0001     0200     0201     N/A     Branch North 113
0001     0300     0301     N/A     Branch North 114
0001     0300     8900     0387    Sub Branch North 115

我试图查询它，但答案并不正确

with cte as
(select Child,Parent from cmbc_entity
 where Parent = '0001'),
 cte2 as
 (select A.Parent as Level1, B.Child as Level2, C.Child as Level3, C.Name
 from cmbc_entity B inner join cte A on A.Child = B.Parent inner join cmbc_entity C on B.Child   =     C.Parent
 where B.Child != '0001')
 select * from cte2

Answer 1

以这种方式拉动数据集的限制是你当然永远不能达到4级以上。动态列是尝试和实现的主要痛苦。

你可能会设置一个递归ID，但是你的列更像是Self，Parent，Grandparent，grandgrandparent，name，这可能会很快变得有些奇怪。

CREATE TABLE Department (
  DeptID INT,
  ParentID INT,
  Name VARCHAR(255));

INSERT INTO Department
VALUES
  (1,1,'HQ'),
  (100,1,'HQ Accounting Dept'),
  (200,1,'HQ Marketing Dept'),
  (300,1,'HQ HR Dept'),
  (101,100,'Branch North 111'),
  (102,100,'Branch North 112'),
  (201,200,'Branch North 113'),
  (301,300,'Branch North 114'),
  (8900,300,'Branch North 115'),
  (387,8900,'Sub Branch North 115');

;WITH cte_Level1 AS (
    SELECT
        CONVERT(VARCHAR(50), DeptID) [Level1],
        CONVERT(VARCHAR(50), 'N/A')  [Level2],
        CONVERT(VARCHAR(50), 'N/A')  [Level3],
        CONVERT(VARCHAR(50), 'N/A')  [Level4],
        Name
    FROM Department
    WHERE DeptID = ParentID
), cte_Level2 AS (
    SELECT * FROM cte_Level1 UNION ALL
    SELECT
        CONVERT(VARCHAR(50), A.Level1) [Level1],
        CONVERT(VARCHAR(50), D.DeptID) [Level2],
        CONVERT(VARCHAR(50), 'N/A')    [Level3],
        CONVERT(VARCHAR(50), 'N/A')    [Level4],
    D.Name
    FROM Department D
    INNER JOIN cte_Level1 A ON A.Level1 = D.ParentID
    WHERE D.DeptID <> D.ParentID
), cte_Level3 AS (
    SELECT * FROM cte_Level2 UNION ALL
    SELECT
        CONVERT(VARCHAR(50), A.Level1) [Level1],
        CONVERT(VARCHAR(50), A.Level2) [Level2],
        CONVERT(VARCHAR(50), D.DeptID) [Level3],
        CONVERT(VARCHAR(50), 'N/A')    [Level4],
        D.Name
    FROM Department D
    INNER JOIN cte_Level2 A ON A.Level2 = D.ParentID
        AND A.Level2 <> 'N/A'
    WHERE D.DeptID <> D.ParentID
), cte_Level4 AS (
    SELECT * FROM cte_Level3 UNION ALL
    SELECT
        CONVERT(VARCHAR(50), A.Level1) [Level1],
        CONVERT(VARCHAR(50), A.Level2) [Level2],
        CONVERT(VARCHAR(50), A.Level3) [Level3],
        CONVERT(VARCHAR(50), D.DeptID)    [Level4],
        D.Name
    FROM Department D
    INNER JOIN cte_Level3 A ON A.Level3 = D.ParentID
        AND A.Level3 <> 'N/A'
    WHERE D.DeptID <> D.ParentID
)

SELECT * FROM cte_Level4

Answer 2

如何使用这种优雅的方法来展平您的层次结构，使用递归CTE。这是一个类似问题的解决方案 Stack Overflow: Flattening Hierarchy series using SQL Server CTE

Answer 3

用其他样本数据测试我的脚本。尽管我自己并不满意，因为我讨厌使用不同的。我将尝试修改和更新，原因如下： -

为什么我要使用不同的？
我使用b.level1的原因在我的最终查询中不为空。

同时我对其他事情非常肯定。就像我的脚本很短。因为levelid而很容易转换为动态sql。

您确定输出结果吗？

Declare @t table (Child varchar(10),  Parent  varchar(10),  Name varchar(50))

insert into @t
select '0001','0001','HQ' union all
select'0100','0001','HQ Accounting Dept' union all
select'0200','0001','HQ Marketing Dept' union all
select'0300','0001','HQ HR Dept' union all
select'0101','0100','Branch North 111' union all
select'0102','0100','Branch North 112'union all
select'0201','0200','Branch North 113'union all
select'0301','0300','Branch North 114'union all
select'8900','0300','Branch North 115'union all
select '0387','8900','Sub Branch North 115'
--select * from @t
;with CTE as
(
select child,parent,name,case when parent=child then 1 else 0 end level1 from @t a 
)
,cte2 as
(
select * from cte where level1=1
union all
select a.Child,a.Parent,a.Name,b.level1+1  from @t a inner join cte2 b on a.parent=b.child and a.child<>b.child
)
select distinct
isnull(b.child,'N/A') Level1
,isnull(c.child,'N/A')  Level2
,isnull(d.child,'N/A') Level3
,isnull(e.child,'N/A') Level4
,case when e.child is null then d.name
      when d.child is null then c.name
      when c.child is null then b.name
      when b.child is null then a.name else e.Name end Name
from cte2 a
left join cte2 b on  a.parent=b.child  and b.level1=1
left join cte2 c on  b.child=c.parent  and c.level1=2
left join cte2 d on  c.Child=d.Parent  and d.level1=3
left join cte2 e on  d.Child=e.Parent  and e.level1=4
where b.level1 is not null

层次结构关系SQL查询

3 个答案: