how to get the node names as well as node values from xml in sqlserver

Question

Post reply

how to get the node names as well as node values from xml in sqlserver

suvo.kundu

SSC Eights!

Points: 953
More actions
September 6, 2012 at 1:31 pm

#100852

i have an XML like:-
<XML>
<Provider>
<providerID>1</providerID>
<Address>address1</Address>
</Provider>
<Provider>
<providerID>2</providerID>
<Address>address2</Address>
</Provider>
here i know the node names .So i could get the all the node values in tabular from.But in ideal situation i will be getting the xml where i will dont know the node names as well as node values.please help me to solve the problem
alter PROCEDURE spXML
@id varchar (50)
as
Declare @xml xml
Set @xml =
'<XML>
<Provider>
<providerID>1</providerID>
<Address>address1</Address>
</Provider>
<Provider>
<providerID>2</providerID>
<Address>address2</Address>
</Provider>
<Provider>
<providerID>3</providerID>
<Address>address3</Address>
<city>kol</city>
</Provider>
</XML>'
Select
X.T.value('(providerID)[1]', 'int')as providerID,
X.T.value('(Address)[1]', 'varchar(500)')as Address,
X.T.value('(city)[1]', 'varchar(500)')as city
From @xml.nodes('/XML/Provider') as X(T)

Viewing 9 posts - 1 through 8 (of 8 total)

You must be logged in to reply to this topic. Login to reply

suvo.kundu SSC Eights! Points: 953 More actions · Answer 1

I could get the node names as column names

SELECT

distinct NodeName = C.value('local-name(.)', 'varchar(50)')

-- , NodeValue = C.value('(.)[1]', 'varchar(50)')

FROM @xml.nodes('/XML/Provider/*') AS T(C)

OUTPUT

1address1NULL

2address2NULL

3address3kol

Address

city

providerID

Arthur Olcot SSCertifiable Points: 6639 More actions · Answer 2

I sometimes use this function if I find myself in a similar situation where I need to shred an xml structure to a tabular form but do not know the xml schema.

http://beyondrelational.com/modules/2/blogs/28/posts/10495/xquery-lab-58-select-from-xml.aspx

suvo.kundu SSC Eights! Points: 953 More actions · Answer 3

The output of the query is:-

Declare @xml xml

Set @xml =

'<XML>

<Address>address1</Address>

</Provider>

<Address>address2</Address>

</Provider>

<Address>address3</Address>

</Provider>

</XML>'

SELECT

distinct

NodeName = C.value('local-name(.)', 'varchar(50)')

, NodeValue = C.value('(.)[1]', 'varchar(50)')

FROM @xml.nodes('/XML/Provider/*') AS T(C)

output

Addressaddress1

Addressaddress2

Addressaddress3

citykol

providerID1

providerID2

providerID3

if any body can help me to change the above out put to this one it can still serve the purpose:-

ProvideId Address City

1 address1 NULL

2 address2 NULL

3 address3 kol

please help

Arthur Olcot SSCertifiable Points: 6639 More actions · Answer 4

Try this revised query:

Declare @xml xml

Set @xml =

'<XML>

<Provider>

<providerID>1</providerID>

<Address>address1</Address>

</Provider>

<Provider>

<providerID>2</providerID>

<Address>address2</Address>

</Provider>

<Provider>

<providerID>3</providerID>

<Address>address3</Address>

<city>kol</city>

</Provider>

</XML>'

SELECT t.c.value('providerID[1]', 'int') AS 'ProviderID'

, t.c.value('Address[1]', 'varchar(100)') AS 'Address'

, t.c.value('city[1]', 'varchar(100)') AS 'City'

FROM @xml.nodes('/XML/Provider') AS T(C)

This outputs it in the format that you have specified.

suvo.kundu SSC Eights! Points: 953 More actions · Answer 5

see this is basically for statically getting the tabular structure where i know the node names.

but for the generalized work around whats the procedure can it be possible?

I will be getting a random xml based on a particular id

suvo.kundu SSC Eights! Points: 953 More actions · Answer 6

if i can manage to get the the starting and the root node as reference...i can make it very generalized like:-

Declare @xml xml

Set @xml =

'<XML>

<Address>address1</Address>

</Provider>

<Address>address2</Address>

</Provider>

<Address>address3</Address>

</Provider>

</XML>'

SELECT

dense_rank() OVER (ORDER BY C.value('local-name(.)', 'varchar(50)')) AS 'SN',

NodeName = C.value('local-name(.)', 'varchar(50)')

, NodeValue = C.value('(.)[1]', 'varchar(50)')

FROM @xml.nodes('/XML/Provider/*') AS T(C)

its out put is

1Addressaddress1

1Addressaddress2

1Addressaddress3

2city kol

3providerID 1

3providerID 3

3providerID 2

now i have to just customize it to

1address1NULL

2address2NULL

3address3 kol

please help me to achieve this thing..atleast if not fully generalised

suvo.kundu SSC Eights! Points: 953 More actions · Answer 7

i tried to achieve this by creating a table variable and altering its structure with the node name by traversing a loop. and again inserting the node values to the table variable..

i am finding some problem...in this

is this the right way to do so..plz advice

i have deined an udf like

set ANSI_NULLS ON

set QUOTED_IDENTIFIER ON

GO

ALTER function [dbo].[udf_GetXMLNodeValue_Names](@xml xml)

returns table

as

--Declare @NodeName varchar(100)

return

(

--Declare @xml xml

--Set @xml =

--'<XML>

--<Provider>

-- <providerID>1</providerID>

--

-- <Address>address1</Address>

--

--</Provider>

--

--<Provider>

-- <providerID>2</providerID>

--

-- <Address>address2</Address>

--

--</Provider>

--<Provider>

-- <providerID>3</providerID>

--

-- <Address>address3</Address>

-- <city>kol</city>

--

--</Provider>

--

--</XML>'

SELECT

dense_rank() OVER (ORDER BY C.value('local-name(.)', 'varchar(50)')) AS 'SN',

NodeName = C.value('local-name(.)', 'varchar(50)')

, NodeValue = C.value('(.)[1]', 'varchar(50)')

FROM @xml.nodes('/XML/Provider/*') AS T(C)

)

my actual stored proc of getting the tabular structure is:-

set ANSI_NULLS ON

set QUOTED_IDENTIFIER ON

go

alter PROCEDURE [dbo].[spFetchNodeNames_ValuesFromXML]

AS

--begin try

--begin tran

Declare @xml xml

Set @xml =

'<XML>

<Address>address1</Address>

</Provider>

<Address>address2</Address>

</Provider>

<Address>address3</Address>

</Provider>

</XML>'

declare @colName varchar(100);

declare @Selectsql varchar(100);

declare @sql varchar(100);

declare @count int;

declare @i int;

set @i=1;

set @sql='';

select @count=count(distinct NodeName) from dbo.udf_GetXMLNodeValue_Names (@xml)

--DECLARE @finaltable TABLE

--(

-- SN int identity

--

--)

IF EXISTS (SELECT * FROM sys.tables WHERE name LIKE '#temp%')

begin

DROP TABLE #temp

print ('hi')

end

SET @sql = 'CREATE TABLE temptab (SN int )'

EXEC (@sql)

WHILE (@i <=@count)

BEGIN

select @colName= NodeName from dbo.udf_GetXMLNodeValue_Names (@xml) where SN=@i;

select(@colName)

SET @sql = 'ALTER TABLE temptab ADD ' + @colName + ' VARCHAR(100)'

EXEC (@sql)

--INSERT INTO temptab VALUES ('COl 1')

SET @i = @i + 1

END

SET @sql = 'ALTER TABLE temptab DROP COLUMN SN'

EXEC (@sql)

select * from temptab

DROP TABLE temptab

select * from temptab

--commit

--end try

--begin catch

--rollback

--end catch

Arthur Olcot SSCertifiable Points: 6639 More actions · Answer 8

Did you have a look at my previous reply regarding the function developed by jacob sebastian? That will shred an xml structure to a flat table similar to what you are trying to do

http://beyondrelational.com/modules/2/blogs/28/posts/10495/xquery-lab-58-select-from-xml.aspx

If you truly do not know the xml schema then this function will probably be your best course of action. However, you say that you will know the schema based upon a certain ID? You never mentioned the number of potential schemas, but if fairly low (prob less than about 50), personally I would be inclined to go down the road of creating a view for each schema rather than trying to create something generic that "fits all". In my experience, if you try to create something generic like that, you invariably at some point end up have to cater for a particular schema that doesn't quite fit in which either makes the code quite heavyweight or you end up with lots of conditional statements in there trying to support it. It'll be easier to maintain long term (devs won't be scared picking up the code) and would think better performing as well. However, if you are sure that the schemas are not going to change/get added to then maybe something generic will work better. As always.. it depends. But start off by looking at that link above and trying out the function.