替换T-SQL中节点数组中一个节点的值

Question

如果我在表中有一个列作为 xml 具有以下节点结构：

<ArrayOfPickListObjectBaseModel xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema">
  <PickListObjectBaseModel>
    <MasterObjectId>3964405</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>405716</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>5872525</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
</ArrayOfPickListObjectBaseModel>

我可以成功检索到具有特定 MasterObjectId 的所有记录

SELECT  p.MyPrimaryKey, 
        m.value('(MasterObjectId)[1]', 'INT')  as MasterObjectID,
        p.MasterObjectIDs
    FROM PickList p
    CROSS APPLY MasterObjectIDs.nodes('//PickListObjectBaseModel') AS t1(m)
    WHERE m.value('(MasterObjectId)[1]', 'INT') = 3964405

但是，如果需要，我将如何替换所有匹配记录中的那个数字？ 所以在上面的例子中，将3964405替换为999

这是不好的语法，但我认为它类似于：

    DECLARE @oldId INT = 13131180;
    DECLARE @newId INT = 99999;

    UPDATE PickList
        SET MasterObjectIDs.modify('replace value of 
    (//PickListObjectBaseModel/MasterObjectId/text()=sql:variable("@oldId"))[1] with sql:variable("@newId")');

Answer 1

这里是基于值替换文本节点的完整解决方案 FROM 部分将加速它，因为它减少了受影响的行数，否则它将对所有行执行操作

    DECLARE @oldId INT = 13131180;
    DECLARE @newId INT = 99999;

    UPDATE PickList
        SET MasterObjectIDs.modify('replace value of 
        (//PickListObjectBaseModel/MasterObjectId[text() = sql:variable("@oldId")]/text())[1]
            with sql:variable("@newId")')
        FROM PickList p
                CROSS APPLY MasterObjectIDs.nodes('//PickListObjectBaseModel') AS t1(m)
                WHERE m.value('(MasterObjectId)[1]', 'INT') = @oldId

Answer 2

太好了，您找到了解决方案（投票赞成），只是一些提示：

一个三行的虚拟表：

DECLARE @DummyTable TABLE(ID INT IDENTITY, Comment VARCHAR(100), MasterObjectIDs XML);
INSERT INTO @DummyTable VALUES
('Your example', N'<ArrayOfPickListObjectBaseModel xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema">
  <PickListObjectBaseModel>
    <MasterObjectId>3964405</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>405716</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>5872525</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
</ArrayOfPickListObjectBaseModel>'
)
,('More than one target', N'<ArrayOfPickListObjectBaseModel xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema">
  <PickListObjectBaseModel>
    <MasterObjectId>3964405</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>405716</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>3964405</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
</ArrayOfPickListObjectBaseModel>'
),
('no target', N'<ArrayOfPickListObjectBaseModel xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema">
  <PickListObjectBaseModel>
    <MasterObjectId>1111</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>405716</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
  <PickListObjectBaseModel>
    <MasterObjectId>5872525</MasterObjectId>
    <IsSelected>false</IsSelected>
  </PickListObjectBaseModel>
</ArrayOfPickListObjectBaseModel>'); 

--变量

DECLARE @oldId INT = 3964405;
DECLARE @newId INT = 99999;

--我使用xml.exist()，比您过滤目标行的方法更快：

UPDATE @DummyTable
        SET MasterObjectIDs.modify('replace value of 
        (//PickListObjectBaseModel/MasterObjectId[text() = sql:variable("@oldId")]/text())[1]
            with sql:variable("@newId")')
FROM @DummyTable p
WHERE p.MasterObjectIDs.exist('/ArrayOfPickListObjectBaseModel/PickListObjectBaseModel/MasterObjectId[text()=sql:variable("@oldId")]') = 1;

--结果

SELECT @@ROWCOUNT;         --good the filter hits only 2 rows
SELECT * FROM @DummyTable; --bad, the XML more than one occurances is not edited everywhere...

如果您要查找的 id 可能在一个 XML 中出现多次，您的方法将不适用于所有这些，仅适用于第一个...

您可以循环运行您的语句，直到exist() 返回0，或者您可以分解 XML 并从头开始重建它，如下所示：

WITH UpdateableCTE AS
(
    SELECT p.MasterObjectIds AS Original
          ,B.NewXML
    FROM @DummyTable AS p
    CROSS APPLY
    (
        SELECT
        (
            SELECT CASE WHEN plo.value(N'(MasterObjectId/text())[1]','int')=@oldId THEN @newId ELSE plo.value(N'(MasterObjectId/text())[1]','int') END AS MasterObjectId
                  ,plo.value(N'(IsSelected/text())[1]','nvarchar(max)') AS IsSelected
            FROM MasterObjectIDs.nodes(N'/ArrayOfPickListObjectBaseModel/PickListObjectBaseModel') AS A(plo)
            FOR XML PATH(N'PickListObjectBaseMode'),ROOT(N'ArrayOfPickListObjectBaseModel'),TYPE
        )
    ) AS B(NewXML)
    WHERE p.MasterObjectIDs.exist('/ArrayOfPickListObjectBaseModel/PickListObjectBaseModel/MasterObjectId[text()=sql:variable("@oldId")]') = 1
)
UPDATE UpdateableCTE SET Original=NewXML;

--只有2行受到影响

SELECT @@ROWCOUNT;         --good the filter hits only 2 rows
SELECT * FROM @DummyTable; --all occurances are switched...