在 Cross Apply 上使用嵌套节点查询 XML

Question

给定一个结构如下的 XML：

<ROOT_NODE>
    <FOLDER_LIST>
        <FOLDER>
            <CODE_FOLDER>1</CODE_FOLDER>
            <DESCRIPTION>This is a folder</DESCRIPTION>
            <DATA_LIST>
                <DATA>
                    <CODE_DATA>100</CODE_DATA>
                    <OPTIONS>
                        <OPTION>
                            <CODE_OPTION>200</CODE_OPTION>
                            <PRINT_TEXT>This is a test</PRINT_TEXT>
                        </OPTION>
                        <OPTION>
                            <CODE_OPTION>200</CODE_OPTION>
                            <PRINT_TEXT>This is a test</PRINT_TEXT>
                        </OPTION>
                    </OPTIONS>
                </DATA>
            </DATA_LIST>
        </FOLDER>
    </FOLDER_LIST>
</ROOT_NODE>

首先，我使用

将第一级 (FOLDER) 的值放入名为 @tmpFolders 的临时表中

FROM @xml.nodes('ROOT_NODE/FOLDER_LIST/FOLDER') as folder(id)

然后我在@tmpFolders 上声明了一个光标

DECLARE cur CURSOR FOR
SELECT CODE_FOLDER, DESCRIPTION FROM @tmpFolders 
OPEN cur 
FETCH NEXT FROM cur INTO @codeFolder, @description
WHILE (@@FETCH_STATUS = 0)        

在游标内，我使用 CROSS APPLY 将第二级 (DATA) 的值插入到另一个名为 @tmpData 的临时表中

INSERT INTO @tmpData(CODE_DATA)
SELECT data.id.value('CODE_DATA[1]','INT'))
FROM @xml.nodes('ROOT_NODE/FOLDER_LIST/FOLDER') as folder(Id)
    CROSS APPLY folder.Id.nodes('DATA_LIST/DATA') as data(Id)

到目前为止，一切正常。 现在我需要从第三级（OPTION）获取值并将它们插入另一个名为@tmpOptions 的临时表中 我尝试添加另一个 CROSS APPLY 但没有成功

INSERT INTO @tmpOptions(CODE_OPTION, PRINT_TEXT)
SELECT data.id.value('CODE_DATA[1]','INT')),
       option.id.value('CODE_OPTION[1]','INT'))
       option.id.value('PRINT_TEXT[1]','VARCHAR(50)'))
FROM @xml.nodes('ROOT_NODE/FOLDER_LIST/FOLDER') as folder(Id)
    CROSS APPLY folder.Id.nodes('DATA_LIST/DATA') as data(Id)
    CROSS APPLY data.Id.nodes('OPTIONS/OPTION') as option(Id)

我没有收到任何错误，所以我不确定自己做错了什么。

Answer 1

您发布的代码不正确...

我没有收到任何错误，所以我不确定自己做错了什么。

有很多右括号，缺少逗号，并且您使用的是保留字，应该像[option] 一样引用。这必须抛出错误...

这样试试

SELECT [data].id.value('CODE_DATA[1]','INT'),
       [option].id.value('CODE_OPTION[1]','INT'),
       [option].id.value('PRINT_TEXT[1]','VARCHAR(50)')
FROM @xml.nodes('ROOT_NODE/FOLDER_LIST/FOLDER') as folder(Id)
    CROSS APPLY folder.Id.nodes('DATA_LIST/DATA') as [data](Id)
    CROSS APPLY [data].Id.nodes('OPTIONS/OPTION') as [option](Id)

但是...

您的代码是 - 可能！ - 如果有多个<FOLDER> 或多个<DATA>，则不会按照您的预期进行。在您的CURSOR 中，您可以读取所有元素，而无需对给定父级进行任何过滤...

无论如何，这不是你应该这样做的方式。尽可能避免CURSOR！

你的最终目标是什么？如果要在相关表中传输此结构。是否故意使两者的选项代码相同（200）？可能是复制粘贴错误...如果所有内部代码都是唯一的，则很简单：

SELECT Fld.value(N'(CODE_FOLDER/text())[1]',N'int') AS Folder_Code
      ,Fld.value(N'(DESCRIPTION/text())[1]',N'nvarchar(max)') AS Folder_Description
      ,Dt.value(N'(CODE_DATA/text())[1]',N'int') AS Data_Code
      ,Opt.value(N'(CODE_OPTION/text())[1]',N'int') AS Option_Code
      ,Opt.value(N'(PRINT_TEXT/text())[1]',N'nvarchar(max)') AS Option_Text
      --Generate running IDs, you might add an existing max id if you have to insert into filled tables
      ,DENSE_RANK() OVER(ORDER BY Fld.value(N'(CODE_FOLDER/text())[1]',N'int')) AS FolderId
      ,DENSE_RANK() OVER(ORDER BY Fld.value(N'(CODE_FOLDER/text())[1]',N'int')
                                 ,Dt.value(N'(CODE_DATA/text())[1]',N'int')) AS DataId
      ,DENSE_RANK() OVER(ORDER BY Fld.value(N'(CODE_FOLDER/text())[1]',N'int')
                                 ,Dt.value(N'(CODE_DATA/text())[1]',N'int')
                                 ,Opt.value(N'(CODE_OPTION/text())[1]',N'int')) AS OptionId
FROM @xml.nodes(N'/ROOT_NODE/FOLDER_LIST/FOLDER') AS A(Fld)
OUTER APPLY Fld.nodes(N'DATA_LIST/DATA') AS B(Dt)
OUTER APPLY Dt.nodes(N'OPTIONS/OPTION') AS C(Opt);

如果内码不唯一，可以这样走：

WITH Folders AS
(
    SELECT  ROW_NUMBER() OVER(ORDER BY (SELECT NULL)) AS FolderId
           ,Fld.value(N'(CODE_FOLDER/text())[1]',N'int') AS Folder_Code
           ,Fld.value(N'(DESCRIPTION/text())[1]',N'nvarchar(max)') AS Folder_Description
           ,Fld.query(N'DATA_LIST/DATA') AS Node_data
    FROM @xml.nodes(N'/ROOT_NODE/FOLDER_LIST/FOLDER') AS A(Fld)
)
,FoldersWithDatas AS
(
    SELECT Folders.FolderId
          ,Folders.Folder_Code
          ,Folders.Folder_Description
          ,ROW_NUMBER() OVER(ORDER BY (SELECT NULL)) AS DataId
          ,Dt.value(N'(CODE_DATA/text())[1]',N'int') AS Data_Code
          ,Dt.query(N'OPTIONS/OPTION') AS Node_data
    FROM Folders
    OUTER APPLY Folders.Node_data.nodes(N'DATA') AS A(Dt)
)
SELECT   FoldersWithDatas.FolderId
        ,FoldersWithDatas.Folder_Code
        ,FoldersWithDatas.Folder_Description
        ,FoldersWithDatas.DataId
        ,FoldersWithDatas.Data_Code
        ,ROW_NUMBER() OVER(ORDER BY (SELECT NULL)) AS OptionId
        ,Dt.value(N'(CODE_OPTION/text())[1]',N'int') AS Option_Code
FROM FoldersWithDatas
OUTER APPLY FoldersWithDatas.Node_data.nodes(N'OPTION') AS A(Dt);

这适用于任意数量的文件夹、嵌套数据和嵌套选项...

将其写入临时表并使用SELECT DISTINCT 将每组数据连同适当的外键一起插入其表中。