Delete XML nodes in SQL server based on conditions - sql-server

Question

I have xml that looks like
<MeasureSet>
<MeasureSetType val_type="name">Variant</MeasureSetType>
<Measure>
<MeasureType val_type="name">single nucleotide variant</MeasureType>
<AttributeSet>
<MeasureAttributeType val_type="name">HGVS</MeasureAttributeType>
<Attribute>NM_000054.4:c.838dupT</Attribute>
</AttributeSet>
<Citation>
<CitationType val_type="name">general</CitationType>
<ID Source="PubMed">10820168</ID>
</Citation>
<Citation>
<CitationType val_type="name">general</CitationType>
<ID Source="PubMed" />
</Citation>
<Citation>
<CitationType val_type="name">general</CitationType>
<ID Source="PubMed">9773787</ID>
</Citation>
<Citation>
<CitationType val_type="name">general</CitationType>
<ID Source="PubMed">18726898</ID>
</Citation>
<Citation>
<CitationType val_type="name">general</CitationType>
<ID Source="PubMed" />
</Citation>
</Measure>
</MeasureSet>
I want to delete the Citation element nodes that have Citation/ID/#Source = "PubMed" and Citation/ID is empty
This deletes only the ID element, but the correct ones
SET #myBlob.modify('delete //Citation/ID[#Source = ''PubMed''
and string-length(string(number(.))) = 0]')
and this deletes all Citation elements
SET #myBlob.modify('delete //Citation[ID/#Source = ''PubMed''
and string-length(string(number(.))) = 0]')
Seems there should be an easy solution I'm not hitting on. Any suggestions? Thanks

Answer 1

Your second one is checking the length of the wrong element - it's looking at (.), which is here the Citation element (because it's in a predicate on the Citation element). Use instead
SET #myBlob.modify('
delete //Citation[ID/#Source = ''PubMed''
and string-length(string(number(ID[1]))) = 0]')
where I have changed . to ID[1]. The [1] is needed because "'number()' requires a singleton (or empty sequence)".