如何將複雜的XML解析爲SQL服務器中的多個插入和更新？

Question

我找到了解析xml插入的例子。然而這些例子非常簡單。 他們通常只是這樣的：

<person> 
    <name>Martin</name> 
</person> 
<person> 
    <name>John</name> 
</person> 

但我有類似這樣的XML - 如果我需要插入到其他表的子元素。

<root> 
    <family> 
     <name>Smith</name> 
     <address>Some road 1</address> 
     <persons> 
      <person> 
       <name>Tina</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
      <person> 
       <name>Martin</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
     </persons> 
    </family> 
    <family> 
     <name>Lane</name> 
     <address>Some road 1</address> 
     <persons> 
      <person> 
       <name>Kevin</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
      <person> 
       <name>Julia</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
     </persons> 
    </family> 
</root> 

我需要通過這個XML迭代和第一行插入表「家」 後，我返回ID爲家庭和在表中使用它作爲下一個INSERT外鍵一個人「人」與愛好相同。我想你應該已經明白了。在「家庭」之後，我需要做一些更新陳述，然後再轉到下一個家庭。

難道有人指着我正確的方向嗎？將不勝感激。

Answer 1

不幸的是，SQL Server不支持多表插入，所以你需要做的單刀片這樣的：

insert into family 
    select f.node.value('name[1]', 'varchar(32)') as name 
    from @xml.nodes('/root/family') f(node) 

insert into person 
    select family.ID as familyID, p.node.value('name[1]', 'varchar(32)') as name 
    from @xml.nodes('/root/family') f(node) 
    cross apply f.node.nodes('persons/person') p(node) 
    inner join family on f.node.value('name[1]', 'varchar(32)') = family.name 

Answer 2

這裏的主要問題是，你可以有相同的名字，許多家庭。我的解決方案考慮到這個「方面」（注意：我使用了修改的XML進行測試）。在這個例子中，你可以看到兩個同名的家庭（史密斯），但差異。人。這些家庭有差異。 RowNum的（和FamilyID的）。最棘手的部分是在最後CROSS APPLY提取每一個家庭的名字出現的所有/person元素：

CROSS APPLY @x.nodes('/root/family[name=sql:column("f.Name")][sql:column("f.RowNum")]/persons/person') AS a(b)

注意：您可以開發這個解決方案中提取，另外，每一個愛好。

DECLARE @x XML = N' 
<root> 
    <family> 
     <name>Smith</name> 
     <address>Some road 1</address> 
     <persons> 
      <person> 
       <name>Tina</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
      <person> 
       <name>Martin</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
     </persons> 
    </family> 
    <family> 
     <name>Lane</name> 
     <address>Some road 1</address> 
     <persons> 
      <person> 
       <name>Kevin</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
      <person> 
       <name>Julia</name> 
       <hobbies> 
        <hobby>Some hobby 1</hobby> 
        <hobby>Some hobby 2</hobby> 
       </hobbies> 
      </person> 
     </persons> 
    </family> 
    <family> 
     <name>Smith</name> 
     <address>Some another road 11</address> 
     <persons> 
      <person> 
       <name>Coco</name> 
      </person> 
      <person> 
       <name>Jambo</name> 
      </person> 
     </persons> 
    </family> 
</root>'; 

DECLARE @Family TABLE (
    FamilyID INT IDENTITY(1,1) UNIQUE, 
    Name NVARCHAR(50) NOT NULL, 
    RowNum INT NOT NULL, 
    PRIMARY KEY (Name, RowNum) 
); 
INSERT @Family (Name, RowNum) 
SELECT src.Name, 
     ROW_NUMBER() OVER(PARTITION BY src.Name ORDER BY @@SPID) AS RowNum 
FROM (
    SELECT a.b.value('(name)[1]', 'NVARCHAR(50)') AS Name 
    FROM @x.nodes('/root/family') AS a(b) 
) src; 

SELECT f.* 
FROM @Family AS f 

DECLARE @Person TABLE (
    PersonID INT IDENTITY(1,1) UNIQUE, 
    FamilyID INT NOT NULL, -- Kind of FK 
    Name NVARCHAR(50) NOT NULL, 
    RowNum INT NOT NULL, 
    PRIMARY KEY (Name, RowNum) 
) 
INSERT @Person (FamilyID, Name, RowNum) 
SELECT src.FamilyID, 
     src.Name, 
     ROW_NUMBER() OVER(PARTITION BY src.FamilyID, src.Name ORDER BY @@SPID) AS RowNum 
FROM (
    SELECT f.FamilyID, 
      a.b.value('(name)[1]', 'NVARCHAR(50)') AS Name 
    FROM @family f 
    CROSS APPLY @x.nodes('/root/family[name=sql:column("f.Name")][sql:column("f.RowNum")]/persons/person') AS a(b) 
) src; 

SELECT p.* 
FROM @Person AS p; 

結果：

FamilyID Name RowNum 
-------- ----- ------ 
1  Lane 1 
2  Smith 1 
3  Smith 2 

PersonID FamilyID Name RowNum 
-------- -------- ------ ------ 
5  3  Coco 1 
6  3  Jambo 1 
1  1  Julia 1 
2  1  Kevin 1 
3  2  Martin 1 
4  2  Tina 1