查询具有多个父节点的闭包表中所有节点的所有路径

Question

我有一个闭包表，其中可能有多个父级，关联的数据表如下所示：

CREATE TABLE [data] (
    id int IDENTITY(1,1) PRIMARY KEY,
    name VARCHAR(100)
)

CREATE TABLE closure (
    id int IDENTITY(1,1) PRIMARY KEY,
    [src_id] int,
    [dst_id] int,
    [depth] int,
    FOREIGN KEY ([src_id]) REFERENCES [data](id),
    FOREIGN KEY ([dst_id]) REFERENCES [data](id)
)

现在我创建了一些看起来像这样的测试数据，DDL 看起来像这样：

数据表

INSERT [data] ([name]) VALUES (N'data1')
INSERT [data] ([name]) VALUES (N'data2')
INSERT [data] ([name]) VALUES (N'data3')
INSERT [data] ([name]) VALUES (N'data4')
INSERT [data] ([name]) VALUES (N'data5')
INSERT [data] ([name]) VALUES (N'data6')

闭包表

INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 1, 0)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (2, 2, 0)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (3, 3, 0)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (4, 4, 0)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (5, 5, 0)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (6, 6, 0)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 2, 1)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (2, 5, 1)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 5, 2)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (5, 6, 1)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (2, 6, 2)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 6, 3)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 3, 1)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (3, 4, 1)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 4, 2)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (4, 5, 1)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (3, 5, 2)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 5, 3)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (4, 6, 2)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (3, 6, 3)
INSERT [closure] ([src_id], [dst_id], [depth]) VALUES (1, 6, 4)

现在我想要的是一个查询，它输出每个数据集的每个路径，直到该路径不再有src_id， 我想用这个语句查询：

;WITH hierarchy_cte AS (
    SELECT [data].id, [data].name, CAST ([data].name AS VARCHAR(100)) AS path , 0 as pathDepth  
    FROM closure
    INNER JOIN [data]
    ON [data].id = closure.src_id
    WHERE closure.depth = 0
    UNION ALL
    SELECT closure.src_id, hierarchy_cte.name, CAST(([srcdata].name + ' / ' + hierarchy_cte.path) AS VARCHAR(100)) , hierarchy_cte.pathDepth + 1
    FROM closure    
    INNER JOIN [data] srcdata
    ON srcdata.id = closure.src_id 
    INNER JOIN [data]
    ON [data].id = closure.dst_id
    inner join [hierarchy_cte]
    ON [data].id = hierarchy_cte.id
    AND closure.depth = 1
) 
SELECT * FROM hierarchy_cte

结果如下所示：

+----+-------+---------------------------------------+-----------+
| id | name  |                 path                  | pathDepth |
+----+-------+---------------------------------------+-----------+
|  1 | data1 | data1                                 |         0 |
|  2 | data2 | data2                                 |         0 |
|  3 | data3 | data3                                 |         0 |
|  4 | data4 | data4                                 |         0 |
|  5 | data5 | data5                                 |         0 |
|  6 | data6 | data6                                 |         0 |
|  5 | data6 | data5 / data6                         |         1 |
|  2 | data6 | data2 / data5 / data6                 |         2 |
|  4 | data6 | data4 / data5 / data6                 |         2 |
|  3 | data6 | data3 / data4 / data5 / data6         |         3 |
|  1 | data6 | data1 / data3 / data4 / data5 / data6 |         4 |
|  1 | data6 | data1 / data2 / data5 / data6         |         3 |
|  2 | data5 | data2 / data5                         |         1 |
|  4 | data5 | data4 / data5                         |         1 |
|  3 | data5 | data3 / data4 / data5                 |         2 |
|  1 | data5 | data1 / data3 / data4 / data5         |         3 |
|  1 | data5 | data1 / data2 / data5                 |         2 |
|  3 | data4 | data3 / data4                         |         1 |
|  1 | data4 | data1 / data3 / data4                 |         2 |
|  1 | data3 | data1 / data3                         |         1 |
|  1 | data2 | data1 / data2                         |         1 |
+----+-------+---------------------------------------+-----------+

我认为我的陈述将沿着所有可能的路径进行，并在没有更多 src_id 时结束。我只是卡住了。 我想要的结果是：

+----+-------+---------------------------------------+-----------+
| id | name  |                 path                  | pathDepth |
+----+-------+---------------------------------------+-----------+
|  1 | data1 | data1                                 |         0 |
|  2 | data2 | data1 / data2                         |         1 |
|  3 | data3 | data1 / data3                         |         1 |
|  4 | data4 | data1 / data3 / data4                 |         2 |
|  5 | data5 | data1 / data3 / data4 / data5         |         3 |
|  5 | data5 | data1 / data2 / data5                 |         2 |
|  6 | data6 | data1 / data3 / data4 / data5 / data6 |         4 |
|  6 | data6 | data1 / data2 / data5 / data6         |         3 |
+----+-------+---------------------------------------+-----------+

这个查询似乎并不难，但我就是想不通。

Answer 1

你需要加入目的地，我在 CTE 中添加了closure.dst_id

然后我从 hierarchy_cte 中选择所有内容，并将 dst_id 与 src_id 连接

FROM hierarchy_cte
    inner join closure 
    on closure.src_id=hierarchy_cte.dst_id
    AND closure.depth = 1

最后，我们只需要为 CTE 的第一步选择根。让我们将根定义为不是闭包关系目的地的任何节点，

select * from closure c1
where depth=0
and not exists(select 1 from closure c2 where c2.depth>=1 and c1.src_id=c2.dst_id)

或许

select * from closure c1
where depth=0
and not exists(select 1 from closure c2 where c2.src_id<>c2.dst_id and c1.src_id=c2.dst_id)

使用这个我们得到这个最终查询：

;WITH hierarchy_cte AS (
    SELECT [data].id, [data].name, CAST ([data].name AS VARCHAR(100)) AS path , closure.depth as pathDepth, closure.dst_id
    FROM closure
    INNER JOIN [data]
    ON [data].id = closure.src_id
    WHERE closure.depth = 0
    and not exists(select 1 from closure c2 where c2.depth>=1 and closure.src_id=c2.dst_id)
    UNION ALL
    SELECT hierarchy_cte.id,data.name,CAST ( [hierarchy_cte].path+'/'+ [data].name AS VARCHAR(100)) AS path,pathDepth+1,closure.dst_id
    from [hierarchy_cte] 
    inner join closure 
    on closure.src_id=hierarchy_cte.dst_id
    and closure.depth=1
    INNER JOIN [data] srcdata
    ON srcdata.id = closure.src_id 
    INNER JOIN [data]
    ON [data].id = closure.dst_id
) 
SELECT id,name,path,pathDepth FROM hierarchy_cte
order by path