[英]Flattening a SQL hierarchical data structure in SQL without cursors
假設我有下表
SectionID ParentIDSection DatumID
1 NULL 1
2 1 2
3 1 3
4 2 4
現在,讓我假裝我想選擇SectionID = 1下的所有DatumID,即使它是它的后代孩子的一部分,所以我會得到
SectionID DatumID
1 1
1 2
1 3
1 4
是否有可能在沒有用游標遞歸地迭代第一個表的情況下?
這讓我困惑了一分鍾,因為它與我對遞歸的看法有相反的方向,但我想出了這個非常簡單的解決方案:
;WITH Rollups AS (
SELECT SectionId, ParentIdSection, DatumId
FROM SectionDataTable
UNION ALL
SELECT parent.SectionId, parent.ParentIdSection, child.DatumId
FROM SectionDataTable parent
INNER JOIN Rollups child ON child.ParentIdSection = parent.SectionId
)
SELECT *
FROM Rollups
WHERE SectionID = 1
(替換為您想要的部分ID)
declare @demo table
(
SectionId int not null primary key clustered
, ParentId int null --foreign key references @demo(SectionId)
, DatumId int
);
insert @demo
select 1, null, 1
union select 2, 1, 2
union select 3, 1, 3
union select 4, 2, 4
union select 5, null, 5
union select 6, 5, 6
;
with demoHierarchy
as
(
select SectionId
, SectionId ParentId
, SectionId OriginalAncestorId
, DatumId
from @demo
where ParentId is null --remove this line if you want OriginalAncestorId to be AnyAncestorId (if that doesn't make sense, try it and see)
union all
select d.SectionId
, d.ParentId
, h.OriginalAncestorId
, d.DatumId
from @demo d
inner join demoHierarchy h
on d.ParentId = h.SectionId
where d.ParentId is not null --implied by the join, but here to make it explicit
)
select OriginalAncestorId SectionId
, DatumId
from demoHierarchy
where OriginalAncestorId = 1 --not needed given your sample data only contains this data, but required when you have multiple root ancestors
有關分層查詢的更多信息,請訪問: http : //blog.sqlauthority.com/2012/04/24/sql-server-introduction-to-hierarchical-query-using-a-recursive-cte-a-primer/
您可以在MS SQL 2008中使用層次結構ID。另一個問題是關於SQL Server 2008層次結構數據類型性能的相對性能? 。
以下是處理遞歸層次結構的另一個參考: http : //www.bimonkey.com/2009/09/handling-recursive-hierarchies-in-sql-server/
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.