How to iterate through every potential XML subelement using SQL Server - sql

Question

I have a large XML file of over 45K contacts and I need to iterate through their subelement transactions into a SQL table. I've looked at several solutions to this, using value(), node(), etc..., but no examples appear to have an XML structure close to mine:
<Contacts>
<Contact>
<ContactID>1234</ContactID>
<ContactName>’John Doe’</ContactName>
<DOB>09031978</DOB>
<Address>’123 Main Street’</Address>
<Transactions>
<Transaction>
<TransactionID>4490</TransactionID>
<ProductName>’Recliner’</ProductName>
<Cost>123.00</Cost>
<PurchaseDate>07042020
</Transaction>
<Transaction>
<TransactionID>5678</TransactionID>
<ProductName>’Lamp’</ProductName>
<Cost>45.00</Cost>
<PurchaseDate>07042020
<Transaction>
</Transactions>
</Contact>
<Contact>
<ContactID>4567</ContactID>
<ContactName>’Jane Doe’</ContactName>
<DOB>05191984</DOB>
<Address>’567 Fake Street’</Address>
<Transactions>
<Transaction>
<TransactionID>4378</TransactionID>
<ProductName>’Coffee Table’</ProductName>
<Cost>225.00</Cost>
<PurchaseDate>07042018
</Transaction>
</Transactions>
</Contact>
</Contacts>
I need these data in a result like below:
ContactID
TransactionID
ProductName
Cost
PurchaseDate
1234
4490
Recliner
123.00
4 July 2020
1234
5678
Lamp
45.00
4 July 2020
4567
4378
Coffee Table
225.00
4 July 2018
I've tried a query using the following script:
EXEC sp_xml_preparedocument #idoc OUTPUT, #doc
-- Execute a SELECT stmt using OPENXML rowset provider.
SELECT *
FROM OPENXML (#idoc, '/Contacts/Contact/Transactions/Transaction',2)
WITH (ContactID int '../ContactID',
TransactionID int 'TransactionID',
ProductName nvarchar(50) 'ProductName',
Cost float 'Cost',
PurchaseDate date 'PurchaseDate')
But this will return either a null for ContactID; or return only one transaction for each ContactID. But I need it to iterate and get as many transactions as exist for a contact.
Any insights would be most welcome!

Answer 1

Try to avoid sp_xml_preparedocument because it uses large amounts of memory that can't be used by SQL Server until you remember to free it up by invoking sp_xml_removedocument.
What you're asking for can be easily achieved using nodes() with a cross apply, e.g. (after fixing your XML sample):
declare #doc xml = N'<Contacts>
<Contact>
<ContactID>1234</ContactID>
<ContactName>’John Doe’</ContactName>
<DOB>09031978</DOB>
<Address>’123 Main Street’</Address>
<Transactions>
<Transaction>
<TransactionID>4490</TransactionID>
<ProductName>’Recliner’</ProductName>
<Cost>123.00</Cost>
<PurchaseDate>07042020</PurchaseDate>
</Transaction>
<Transaction>
<TransactionID>5678</TransactionID>
<ProductName>’Lamp’</ProductName>
<Cost>45.00</Cost>
<PurchaseDate>07042020</PurchaseDate>
</Transaction>
</Transactions>
</Contact>
<Contact>
<ContactID>4567</ContactID>
<ContactName>’Jane Doe’</ContactName>
<DOB>05191984</DOB>
<Address>’567 Fake Street’</Address>
<Transactions>
<Transaction>
<TransactionID>4378</TransactionID>
<ProductName>’Coffee Table’</ProductName>
<Cost>225.00</Cost>
<PurchaseDate>07042018</PurchaseDate>
</Transaction>
</Transactions>
</Contact>
</Contacts>';
select
Cont.value('ContactID[1]', 'int') as ContactID
,Trans.value('TransactionID[1]', 'int') as TransactionID
,Trans.value('ProductName[1]', 'nvarchar(50)') as ProductName
,Trans.value('Cost[1]', 'float') as Cost
,convert(date, concat(substring(purDate,1,2), '/', substring(purDate,3,2), '/', substring(purDate,5,4)), 101) as PurchaseDate
from #doc.nodes('//Contact') nodes1(Cont)
cross apply nodes1.Cont.nodes('Transactions/Transaction') nodes2(Trans)
outer apply (
select purDate = Trans.value('PurchaseDate[1]', 'nvarchar(8)')
) temp;
Which yields:
ContactID
TransactionID
ProductName
Cost
PurchaseDate
1234
4490
’Recliner’
123
2020-07-04
1234
5678
’Lamp’
45
2020-07-04
4567
4378
’Coffee Table’
225
2018-07-04