asked on

Comingling relational and XML data

I have a table that consists of a few keys and an XML fragment. I need to decompose that xml fragment and load it into a new table, with one row for each XML node. The XML node needs to be broken into 2 columns. That's the easy part. For each row of the XML data in the new table, I also need to populate the column with the keys from the original table. I have been unable to find a way to do this without a very slow cursor going through each row in the source table and inserting each node along with its keys into the destination table.

Example of a row in the source table:

IsntanceID PatientID EncounterID XMLData
12867 46 63481 (XML Fragment stored in this column shown below)

XML Fragment:
<finding>
<item id="314" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
<item id="353" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
<item id="341" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
<item id="650" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
<item id="1613" sectionname="ChiefComplaintSection" />
</finding>
<finding>
<item id="663" sectionname="ChiefComplaintSection" />
</finding>

So once transformed, this would end up with 6 rows in the destination table, looking like the following:

IsntanceID PatientID EncounterID id sectionname
12867 46 63481 314 HistoryofPresentIllnessSection
12867 46 63481 353 HistoryofPresentIllnessSection
12867 46 63481 341 HistoryofPresentIllnessSection
12867 46 63481 650 HistoryofPresentIllnessSection
12867 46 63481 1613 ChiefComplaintSection
12867 46 63481 663 ChiefComplaintSection

chapmandew

this should do it:

select t.intstanceid,
patientid , encounterid , id ,
tc.c.value('@sectionname','varchar(50)')
from SourceTable t
cross apply t.XMLData.nodes('/finding/item') tc(c)

gustavlasko

ASKER

I gave that a shot (edited id to be tc.c.value('@sectionname','varchar(50)') as id is in the xml):

select t.intstanceid,
patientid , encounterid ,
tc.c.value('@id','varchar(50)') as ID,
tc.c.value('@sectionname','varchar(50)') as SectionName
from SourceTable t
cross apply t.XMLData.nodes('/finding/item') tc(c)

This doesn't yield any results, though. I'm going to continue to look into using apply, as this seems like my best bet.

chapmandew

It is...and it works for me. Here is the sample data I used:

create table #t(intstanceid int, patientid int, encounterid int, id int, sectionname xml)
 
insert into #t
select 12867,46,63481,314, '<finding>
  <item id="314" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
  <item id="353" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
  <item id="341" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
  <item id="650" sectionname="HistoryofPresentIllnessSection" />
</finding>
<finding>
  <item id="1613" sectionname="ChiefComplaintSection" />
</finding>
<finding>
  <item id="663" sectionname="ChiefComplaintSection" />
</finding>'
 
select t.intstanceid, 
patientid , encounterid , id , 
tc.c.value('@sectionname','varchar(50)')
from #t t
cross apply t.sectionname.nodes('/finding/item') tc(c)

Open in new window

gustavlasko

ASKER

It looks like that solution requires an ID column that will match up with the ID attribute in the XML... my source table doesn't have that available. I need to get both the id and sectionname from the XML

ASKER CERTIFIED SOLUTION

chapmandew

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

gustavlasko

ASKER

Thank you. This returns the results I am looking for.