Reputation: 3242
Essentially I am trying to transform the XML I have into a flat data structure using SQL.
My XML is in the following format (I've altered the XML into a subset to make it simpler for my example):
<Actions>
<AddComponent>
<Action>
<DataItem>
<GroupId>1</GroupId>
<Data>
<Id>100</Id>
<Value>Value A</Value>
<Children>
<Data>
<Id>200</Id>
<Value>Value B</Value>
<Children>
<Data>
<Id>300</Id>
<Value>Value C1</Value>
</Data>
<Data>
<Id>301</Id>
<Value>Value C2</Value>
<Children />
</Data>
</Children>
</Data>
</Children>
</Data>
</DataItem>
<DataItem>
<GroupId>2</GroupId>
<Data>
<Id>101</Id>
<Value>Value A</Value>
<Children>
<Data>
<Id>200</Id>
<Value>Value B</Value>
<Children>
<Data>
<Id>302</Id>
<Value>Value C3</Value>
</Data>
</Children>
</Data>
</Children>
</Data>
</DataItem>
</Action>
</AddComponent>
</Actions>
The output I am looking for is the following:
+---------+-----+----------+----------+
| GroupId | Id | Value | ParentId |
+---------+-----+----------+----------+
| 1 | 100 | Value A | NULL |
| 1 | 200 | Value B | 100 |
| 1 | 300 | Value C1 | 200 |
| 1 | 301 | Value C2 | 200 |
| 2 | 101 | Value A | NULL |
| 2 | 200 | Value B | 101 |
| 2 | 302 | Value C3 | 200 |
+---------+-----+----------+----------+
I am not sure on the best method of recursively going through the 'Children' elements. As the number of children could be ad infinitum.
Upvotes: 2
Views: 223
Reputation: 138960
select DI.X.value('(GroupId/text())[1]', 'int') as GroupId,
D.X.value('(Id/text())[1]', 'int') as Id,
D.X.value('(Value/text())[1]', 'nvarchar(50)') as Value,
D.X.value('(../../Id/text())[1]', 'int') as ParentID
from @XML.nodes('/Actions/AddComponent/Action/DataItem') as DI(X)
cross apply DI.X.nodes('.//Data') as D(X)
.//Data
will give you all the Data
nodes recursively.
Note: Using the parent axis (../../Id/text())[1]
to get the ParentID
could be a performance killer for you. Try it out with your data to see if it performance is acceptable.
Update:
Doing the parent axis in a nodes()
call looks like it will give you a much better query plan.
select DI.X.value('(GroupId/text())[1]', 'int') as GroupId,
D.X.value('(Id/text())[1]', 'int') as Id,
D.X.value('(Value/text())[1]', 'nvarchar(50)') as Value,
P.X.value('text()[1]', 'int') as ParentID
from @XML.nodes('/Actions/AddComponent/Action/DataItem') as DI(X)
cross apply DI.X.nodes('.//Data') as D(X)
outer apply D.X.nodes('../../Id') as P(X)
Upvotes: 4
Reputation: 3202
Check this query if it can help :
DECLARE @XML XML = '<Actions><AddComponent><Action><DataItem><GroupId>1</GroupId><Data><Id>100</Id><Value>Value A</Value><Children><Data><Id>200</Id><Value>Value B</Value><Children><Data><Id>300</Id><Value>Value C1</Value></Data><Data><Id>301</Id><Value>Value C2</Value><Children /></Data></Children></Data></Children></Data></DataItem><DataItem><GroupId>2</GroupId><Data><Id>101</Id><Value>Value A</Value><Children><Data><Id>200</Id><Value>Value B</Value><Children><Data><Id>302</Id><Value>Value C3</Value></Data></Children></Data></Children></Data></DataItem></Action></AddComponent></Actions>';
;WITH cte
AS (SELECT c.value('(GroupId)[1]', 'varchar(30)') groupid,
c.value('(Data/Id)[1]', 'varchar(30)') id,
c.value('(Data/Value)[1]', 'varchar(30)') value,
c.query('Data/Children') AS childdata,
Cast(NULL AS VARCHAR(30)) AS parentId
FROM (SELECT @xml) temp(x)
CROSS apply x.nodes('//Actions/AddComponent/Action/DataItem') tabl(c)
UNION ALL
SELECT groupid,
c.value('(Id)[1]', 'varchar(30)') id,
c.value('(Value)[1]', 'varchar(30)') value,
c.query('Children') AS childdata,
Cast(cte.id AS VARCHAR(30)) AS parentId
FROM cte
CROSS apply childdata.nodes('/Children/Data') tabl(c))
SELECT groupid,
id,
value,
parentId
FROM cte
ORDER BY groupid,id
As you can see in my CTE, I first extracted all the parent nodes data and their child XML then recursively get all child data from child XML.
Upvotes: 1