Sql server 列中ID列表上的SQL联接

我的数据库中有一个名为MyFolder的多层对象。MyFolder可以是另一个无限级别MyFolder的子级。该表定义如下：

CREATE TABLE dbo.MyFolders
(
    MyFolderId INT IDENTITY(1,1) NOT NULL,
    ParentMyFolderId INT NULL,
    Name NVARCHAR(50) NOT NULL,
    Depth INT NOT NULL,
    Ancestry NVARCHAR(max) NOT NULL,
    CONSTRAINT PK_MyFolders PRIMARY KEY CLUSTERED (MyFolderId ASC),
    CONSTRAINT FK_MyFolders_MyFolders FOREIGN KEY(ParentMyFolderId) REFERENCES dbo.MyFolders (MyFolderId)
)

它有如下数据：

MyFolderId ParentMyFolderId Name           Depth Ancestry
1          NULL             Folder1        0     /
2          1                Folder1A       1     /1/
3          1                Folder1B       1     /1/
4          1                Folder1C       1     /1/
5          4                Folder1C1      2     /1/4/
6          4                Folder1C2      2     /1/4/
7          6                Folder1C2a     3     /1/4/6/
8          6                Folder1C2b     3     /1/4/6/

这对于我的系统中所需的一切都非常有效。但是，如果要检索如下查询，则会变得很棘手：

MyFolderId Name
1          Folder1
2          Folder1/Folder1A
3          Folder1/Folder1B
4          Folder1/Folder1C
5          Folder1/Folder1C/Folder1C1
6          Folder1/Folder1C/Folder1C2
7          Folder1/Folder1C/Folder1C2/Folder1C2a
8          Folder1/Folder1C/Folder1C2/Folder1C2b

---

有没有办法加入祖先场以获得祖先的名字？还是使用ParentMyFolderId列的另一种方法？我确实有一个名为SplitString（value，delimiter）的表值拆分字符串函数。

这可以使用递归查询来完成，只需将当前文件夹名称附加到以前的文件夹中即可

查询：

;WITH Source (MyFolderId, ParentMyFolderId, Name, Depth, Ancestry)
AS (
SELECT 1, NULL, 'Folder1', 0, '/'
UNION ALL
SELECT 2, 1, 'Folder1A', 1, '/1/'
UNION ALL
SELECT 3, 1, 'Folder1B', 1, '/1/'
UNION ALL
SELECT 4, 1, 'Folder1C', 1, '/1/'
UNION ALL
SELECT 5, 4, 'Folder1C1', 2, '/1/4/'
UNION ALL
SELECT 6, 4, 'Folder1C2', 2, '/1/4/'
UNION ALL
SELECT 7, 6, 'Folder1C2a', 3, '/1/4/6/'
UNION ALL
SELECT 8, 6, 'Folder1C2b', 3, '/1/4/6/'
),
cte AS
(
SELECT S.MyFolderID, S.ParentMyFolderId, CAST(S.Name AS VARCHAR(MAX)) AS Name
FROM Source AS S
WHERE ParentMyFolderId IS NULL
UNION ALL
SELECT S.MyFolderID, S.ParentMyFolderId, C.Name + '/' + S.Name
FROM Source AS S
INNER JOIN cte AS C
    ON C.MyFolderId = S.ParentMyFolderId
)
SELECT *
FROM cte

---

下面是注释中提到的递归CTE：

WITH TreeStructure(MyFolderId, Name) AS 
(
    SELECT MyFolderId, CONVERT(varchar(500), Name)
    FROM MyFolders WHERE ParentMyFolderId IS NULL
    UNION ALL
    SELECT sd.MyFolderId, CONVERT(varchar(500), t.Name + '/' + sd.Name)
    FROM MyFolders sd
    JOIN TreeStructure t ON sd.ParentMyFolderId = t.MyFolderId
    WHERE sd.ParentMyFolderId IS NOT NULL
)
SELECT * FROM TreeStructure

结果:

MyFolderId名称
----------- ----------------------------------------
1文件夹1
2折叠1/折叠1a
3文件夹1/文件夹1b
4折叠1/折叠1c
5文件夹1/文件夹1C/文件夹1C1
6文件夹1/文件夹1C/文件夹1C2
7文件夹1/文件夹1C/文件夹1C2/文件夹1C2A
8文件夹1/文件夹1C/文件夹1C2/文件夹1C2B

递归cte应该能够非常轻松地完成您想要的操作。只是好奇——您是否考虑过将

祖先

列切换为

hierarchyid

类型？这将为您提供一些引用完整性，这样您就不必担心无效的

祖先
值（例如
/1/4/17
）。