USE Test
--Create 2 tables as an example
CREATE TABLE ExampleTable
(
[ID] int PRIMARY KEY
,[Name] nvarchar(256)
)
CREATE TABLE ExampleTable2
(
[ID] int PRIMARY KEY
,[Name] nvarchar(256)
)
----way1
SELECT *
FROM sys.objects [table]
WHERE
[Name] LIKE 'ExampleTable%'
FOR XML AUTO, ROOT('tables')
--way2
SELECT[name] AS [@name], object_id AS [@object_id],
* FROM sys.objects
WHERE [Name] LIKE 'ExampleTable%'
FOR XML PATH('table'), ROOT('tables')
--SELECT * FROM sys.objects WHERE [Name] LIKE 'ExampleTable%'
--Way3
-------------------------------------
--Nested Results using FOR XML PATH--
-------------------------------------
-------------------------------------
--Nested Results using FOR XML PATH--
-------------------------------------
SELECT
t.[name] AS [@name]
,t.object_id AS [@object_id]
,t.[create_date] AS [date/@cdate]
,t.[modify_date] AS [date/mdate]
,CAST('<mydata>some random xml</mydata>' AS xml) [*]
,(
SELECT
c.[name] AS [@name]
,c.[column_id] AS [@column_id]
,c.[is_nullable] AS [@allows_nulls]
,ty.[name] AS [@datatype]
FROM sys.columns c
INNER JOIN sys.types ty ON c.[user_type_id] = ty.[user_type_id]
WHERE t.[object_id] = c.[object_id]
FOR XML PATH('columns'), TYPE
)
FROM sys.objects t
WHERE [Name] LIKE 'ExampleTable%'
FOR XML PATH('table'), ROOT('tables')
SELECT cdate, * FROM syscolumns WHERE id=1077578877
SELECT * FROM sysobjects WHERE id=1077578877
SELECT * FROM sys.objects WHERE object_id=1077578877
--------------------
--Set Xml Variable--
--------------------
/*
In 2000, it wasn't possible to set any variable to the result of a query
*/
DECLARE @xml xml
SELECT @xml =
(
SELECT TOP 3
[Name]
FROM sys.objects
FOR XML AUTO
)
SELECT @xml
/*
The query method allows reformatting of a document into other formats
It could be considered a little like XSLT
*/
DECLARE @doc xml
SET @doc =
'<people>
<person>Martin</person>
<person>Simon</person>
<person>Mark</person>
</people>'
SELECT @doc.query('
for $p in /people/person/text()
return
<root>
<name>
{$p}
</name>
</root>
') AS queryresults
/*
The value method allows a single value to be extracted from an xml document
*/
DECLARE @doc xml, @count int,@name1 nvarchar(20)
DECLARE @maxName AS nvarchar(20)
SET @doc =
'<people>
<person>Martin</person>
<person>Simon</person>
<person>Mark</person>
</people>'
SELECT @count = @doc.value('count(/people/person)', 'int')
SELECT @name1 = @doc.value('(/people/person/text())[2]','nvarchar(20)')
SELECT @count AS [count]
SELECT @name1 AS Name1
----------------------
--Using sql:variable--
----------------------
/*
It is not possible to construct a custom xquery string
However, it is possible to pass in a sql variables as parameters
to the XQuery statement.
NOTE: SQL must be sure that a single value will be returned,
hence the final [1], even though the statement
/people/person[sql:variable("@pos")]/text()
would only ever return 1 value
*/
DECLARE @doc xml, @pos int
SET @doc =
'<people>
<person>Martin</person>
<person>Simon</person>
<person>Mark</person>
</people>'
SELECT @pos = 2
SELECT @doc.value('(/people/person[sql:variable("@pos")]/text())[1]', 'nvarchar(256)')AS SecondName
SELECT @doc.value('(//people/person/text())[2]','nvarchar(20)')
/*
.nodes can be used to shred xml into a table
In 2000 we had to use OPENXML
*/
DECLARE @doc xml
SET @doc =
'<people>
<person>Martin</person>
<person>Simon</person>
<person>Mark</person>
</people>'
SELECT
t.c.query('text()') AS [Name]
FROM
@doc.nodes('people/person') AS t(c);
DECLARE @doc xml
SET @doc =
'<people>
<person id=''1''>
<name>benny</name>
<name>aa</name>
<phone>49313</phone>
<address/>
</person>
<person id=''2''>
<name>skype</name>
<phone>138</phone>
<address>park</address>
</person>
</people>'
SELECT
T.c.value('@id','char(1)') AS Id
,T.c.value('name[2]','nvarchar(20)')
,T.c.value('phone[1]','nvarchar(10)')
FROM @doc.nodes('/people/person') T(c)
WHERE T.c.value('@id','int') = 1
AND T.c.value('phone[1]','nvarchar(10)')='49313'
/*
ID Name phone address
1 benny 19313 ''
2 skype 138 park
*/
SELECT
PeopleTable.PersonColumn.query('/name/text()') AS [Name]
FROM @doc.nodes('/people/person') AS PeopleTable(PersonColumn)
WHERE PeopleTable.PersonColumn.value('@id','int') = 1
--其實XML相對路徑的問題，請確認是Namespace不確定還是XML節點路徑都不確定(如果是這樣，那麼設計的XML穩定性有待增強)。
--下面的Demo，是在解析多個Namespace的XML節點，但是路徑至少可以確定的：
DECLARE @XML xml
SET @XML=N'
<Publish xmlns="http://soa.newegg.com/SOA/USA/InfrastructureService/V10/PubSubService">
<UserData xmlns="">
<SessionID>0xABde12345</SessionID>
<OrderNumber>1000</OrderNumber>
<Node>
<SessionID>0xABde12345</SessionID>
<OrderNumber>102365</OrderNumber>
</Node>
</UserData>
</Publish>'
--這個是相對路徑的解析方式,這種解析方式很容易出現解析到的數據不是我們想要的,比如,我們先想要的Node節點下的SessionID和OrderNumber
--;WITH XMLNAMESPACES (DEFAULT 'http://soa.newegg.com/SOA/USA/InfrastructureService/V10/PubSubService')
SELECT @XML.value('(//SessionID/text())[1]','CHAR(15)')
,@XML.value('(//OrderNumber/text())[1]','int')
--雖然XML的Namespace不確定，但是，我們可以確定XML的路徑，這樣子避免解析到其他節點的數據。
SELECT @XML.value('(/*:Publish/*:UserData/*:Node/*:SessionID/text())[1]','CHAR(15)')
,@XML.value('(/*:Publish/*:UserData/*:Node/*:OrderNumber/text())[1]','int')
