如何在 XML 列上进行 PIVOT？

Question

是否可以使用 SQL Server 2014 标准版对 XML 列名的值进行 PIVOT？最终我想接受这个：

<table>
  <id>{3d2699c4-3159-4e8b-b48c-c2c4c9b5bd77}</id>
  <rows>
    <row>
      <columns>
        <column name="DESC" value="DACS" type="System.String" />
        <column name="ec_amount" value="5000" type="System.Decimal" />
        <column name="ec_exrate" value="1" type="System.Decimal" />
        <column name="ec_total" value="5000.00" type="System.Decimal" />
        <column name="ItemNo" value="PVT-C30" type="System.String" />
        <column name="UOM" value="EA" type="System.String" />
        <column name="DefaultKey" value="1" type="System.Int32" /><----THIS IS THE COLUMN ON WHICH I WOULD LIKE TO PIVOT
      </columns>
    </row>
    <row>
      <columns>
        <column name="DESC" value="DACS" type="System.String" />
        <column name="ec_amount" value="1500" type="System.Decimal" />
        <column name="ec_exrate" value="5" type="System.Decimal" />
        <column name="ec_total" value="7500.00" type="System.Decimal" />
        <column name="ItemNo" value="PVT-C30" type="System.String" />
        <column name="UOM" value="EA" type="System.String" />
        <column name="DefaultKey" value="2" type="System.Int32" />
      </columns>
    </row>
  </rows>
  <key>DefaultKey</key>
  <total>12500.00</total>
  <data />
  <parameters />
</table>

....并创建这些结果：

DefaultKey  DESC    ec_amount   ec_exrate   ec_total    ItemNo      UOM
1           DACS    5000        1           5000        PVT-C30     EA
2           DACS    1500        5           7500        PVT-C30     EA

我搜索了 Stack Overflow 网站，在我遇到的所有内容中，这两个帖子都很接近，但它们并没有完全让我到达那里：

SQL Pivot using an XML column

How do I Pivot on an XML column's attributes in T-SQL

如果这个问题已经得到解决，我很抱歉，我只是太早放弃了。

Answer 1

无需旋转...试试这个

DECLARE @xml XML=N'<table>
  <id>{3d2699c4-3159-4e8b-b48c-c2c4c9b5bd77}</id>
  <rows>
    <row>
      <columns>
        <column name="DESC" value="DACS" type="System.String" />
        <column name="ec_amount" value="5000" type="System.Decimal" />
        <column name="ec_exrate" value="1" type="System.Decimal" />
        <column name="ec_total" value="5000.00" type="System.Decimal" />
        <column name="ItemNo" value="PVT-C30" type="System.String" />
        <column name="UOM" value="EA" type="System.String" />
        <column name="DefaultKey" value="1" type="System.Int32" />
      </columns>
    </row>
    <row>
      <columns>
        <column name="DESC" value="DACS" type="System.String" />
        <column name="ec_amount" value="1500" type="System.Decimal" />
        <column name="ec_exrate" value="5" type="System.Decimal" />
        <column name="ec_total" value="7500.00" type="System.Decimal" />
        <column name="ItemNo" value="PVT-C30" type="System.String" />
        <column name="UOM" value="EA" type="System.String" />
        <column name="DefaultKey" value="2" type="System.Int32" />
      </columns>
    </row>
  </rows>
  <key>DefaultKey</key>
  <total>12500.00</total>
  <data />
  <parameters />
</table>';

SELECT r.value(N'(columns/column[@name="DefaultKey"]/@value)[1]',N'int') AS DefaultKey
      ,r.value(N'(columns/column[@name="DESC"]/@value)[1]',N'nvarchar(max)') AS [DESC]
      ,r.value(N'(columns/column[@name="ec_amount"]/@value)[1]',N'decimal(10,4)') AS ec_amount
      --similar with your other elements
FROM @xml.nodes(N'/table/rows/row') AS A(r)

Answer 2

您基本上需要为 XML 的每个“行”分配一个行号（或找出区分每个“行”的其他方法）。您可以解决此问题的一种方法是从每个“行”中提取默认键（通过使用窗口函数和分区）：

SELECT *
FROM
(
    SELECT Name = C2.X.value('@name', 'varchar(max)'),
           Val = C2.X.value('@value', 'varchar(max)'),
           DefaultKey = MAX(CASE WHEN C2.X.value('@name', 'varchar(max)') = 'DefaultKey' THEN C2.X.value('@value', 'varchar(max)') END) OVER(PARTITION BY C1.X)
    FROM myTable
    CROSS APPLY myXMLColumn.nodes('table/rows/row/columns') AS C1(X)
    CROSS APPLY C1.X.nodes('column') AS C2(X)
) AS T
PIVOT (MAX(Val) FOR Name IN ([DESC], [ec_amount], [ec_exrate], [ec_total], [ItemNo], [UOM])) AS P;

如果 DefaultKey 不是唯一的，您可以使用行号或排名窗口函数来代替每个类似的结果：

SELECT *
FROM
(
    SELECT Name = C2.X.value('@name', 'varchar(max)'),
           Val = C2.X.value('@value', 'varchar(max)'),
           RowInXML = DENSE_RANK() OVER (ORDER BY C1.X)
    FROM myTable
    CROSS APPLY myXMLColumn.nodes('table/rows/row/columns') AS C1(X)
    CROSS APPLY C1.X.nodes('column') AS C2(X)
) AS T
PIVOT (MAX(Val) FOR Name IN ([DefaultKey], [DESC], [ec_amount], [ec_exrate], [ec_total], [ItemNo], [UOM])) AS P;