使用特定逻辑选择不重复的记录

Question

我有一张表 A、B 和 C，其中包含很多列 (30+)。所有的主要列是Id，RefNumber

我还有表 LinkedEntity 可以匹配来自不同表的记录（A、B 或 C）

我需要从表 A 中选择所有记录并显示来自 B 和 C 的链接记录

一个

Id	RefNumber	OtherColumns
101	A101	...
102	A102	...

B

Id	RefNumber	OtherColumns
201	B101	...
202	B102	...

C

Id	RefNumber	OtherColumns
301	C101	...
302	C102	...

关联实体

Id	EntityId	LinkedEntityId
1	101	202
2	102	301
3	102	201
4	102	202

预期结果：

Id	RefNumber	LinkedB	LinkedBRefNumb	LinkedC	LinkedCRefNumb
101	A101	202	B102	NULL	NULL
102	A102	201,202	B101,B102	301	C101

第一个想法写类似的东西

SELECT A.Id, A.RefNumber, L1.Id, L1.RefNumber, L2.Id, L2.RefNumber
FROM A
LEFT JOIN (SELECT B.Id, B.RefNumber, le.EntityId, le.LinkedEntityId FROM B JOIN LinkedEntity le ON le.EntityId = B.Id OR le.LinkedEntityId = B.Id) L1
ON A.Id = L1.EntityId OR A.Id = L1.LinkedEntityId 
LEFT JOIN (SELECT C.Id, C.RefNumber, le.EntityId, le.LinkedEntityId FROM C JOIN LinkedEntity le ON le.EntityId = C.Id OR le.LinkedEntityId = C.Id) L2
ON A.Id = L2.EntityId OR A.Id = L2.LinkedEntityId

但此查询返回 A 表的重复记录。 有什么方法可以删除重复项并加入linkedEntities 的值？ （也许使用STRING_AGG）？

Answer 1

这是一种使用连接的简单方法，我在 LinkedEntity 中使用一些额外的重复条目进行了测试（您当前的设计允许这样做，您可以使用 EntityId 和 LinkedEntityId 的复合键来解决此问题，并从该表中删除 Id）。

对于旧版本的 sql server，您确实需要使用 STRING_AGG 或 one of these other approaches 对记录进行分组。此外，如果您有兴趣对分组的连接数据进行排序，docs 提供了如何排序数据的其他示例。

这使用IN 运算符和左连接来检索链接的实体，并使用 distinct 关键字来删除重复项。

查询 1

SELECT
    AId as Id,
    ARefNumber as RefNumber,
    STRING_AGG(BId,',') as LinkedB,
    STRING_AGG(BRefNumber,',') as LinkedBRefNumb,
    STRING_AGG(CId,',') as LinkedC,
    STRING_AGG(CRefNumber,',') as LinkedCRefNumb
FROM (
    SELECT DISTINCT
        A.Id as AId,
        A.RefNumber as ARefNumber,
        B.Id as BId,
        B.RefNumber as BRefNumber,
        C.Id as CId,
        C.RefNumber as CRefNumber
    FROM 
        A
    LEFT JOIN
        LinkedEntity le ON A.Id IN (le.EntityId,le.LinkedEntityId)
    LEFT JOIN
        B ON B.Id IN  (le.EntityId,le.LinkedEntityId) 
    LEFT JOIN
        C ON C.Id IN  (le.EntityId,le.LinkedEntityId) 
) t
GROUP BY
    AId,
    ARefNumber

Id	RefNumber	LinkedB	LinkedBRefNumb	LinkedC	LinkedCRefNumb
101	A101	202	B102	null	null
102	A102	201,202	B101,B102	301	C101

View working demo db fiddle

让我知道这是否适合你。

Answer 2

您是否尝试过外键（据我所知，我不确定该表是通过 id 还是 refnumber 链接的）：

ALTER TABLE A
   ADD CONSTRAINT Foreign1 FOREIGN KEY (Id)
      REFERENCES B (Id)
      ON DELETE RESTRICT
      ON DELETE RESTRICT

从 b 到 c 以及从 Linkedentity 到 a、b、c 以此类推。这里是ms官方页面https://docs.microsoft.com/it-it/sql/relational-databases/tables/create-foreign-key-relationships?view=sql-server-ver15

那么你所要做的应该是：

SELECT * FROM A JOIN B JOIN C JOIN LinkedEntity

Answer 3

SELECT
a_id,
a_RefNumber,
GROUP_AGGR(b_id) AS b_aggr_id,
GROUP_AGGR(b_RefNumber) AS b_aggr_ref,
GROUP_AGGR(c_id) as c_aggr,
GROUP_AGGR(c_RefNumber) AS c_aggr_ref
FROM (

    SELECT
    A.id AS a_id,
    A.RefNumber AS a_RefNumber,
    B.id AS b_id,
    B.RefNumber AS b_RefNumber,
    NULL AS c_id,
    NULL AS c_RefNumber
    FROM A
    LEFT JOIN LinkedEntity X ON X.EntityId=A.id
    LEFT JOIN B ON X.LinkedEntityId=B.id

    UNION ALL

    SELECT
    A.id AS a_id,
    A.RefNumber AS a_RefNumber,
    NULL AS b_id,
    NULL AS b_RefNumber,
    C.id AS c_id,
    C.RefNumber AS c_RefNumber
    FROM A
    LEFT JOIN LinkedEntity X ON X.EntityId=A.id
    LEFT JOIN C ON X.LinkedEntityId=C.id

) AS matrix
GROUP BY a_id, a_RefNumber
ORDER BY 1