Convert nested XML via XQuery to flat xml - sql

Question

I've a nested piece of XML i want to insert in to SQL.
Import XML:
<RECORD>
<RECID>118810</RECID>
<FIELD TYPE="C">
<NAME>proj_code</NAME>
<VALUE>118810</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>sub_nr</NAME>
<VALUE>99900</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>proj_desc</NAME>
<VALUE>Nagekomen kosten Oktober 2018</VALUE>
</FIELD>
<FIELD TYPE="N">
<NAME>pro_stat</NAME>
<VALUE>9</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>comment</NAME>
<VALUE></VALUE>
</FIELD>
</RECORD>
I want to convert via XQuery it into:
<RECORD>
<RECID>118810</RECID>
<proj_code>118810</proj_code>
<sub_nr>99900</sub_nr>
<proj_desc>Nagekomen kosten Oktober 2018</proj_desc>
<pro_stat>9</pro_stat>
<comment></comment>
</RECORD>
so i can import it into SQL.
Any thoughts?

Answer 1

As Martin Honnen pointed out, MS SQL Server XQuery doesn't support computed dynamic element names, just literals. Unfortunately, including even the latest SQL Server 2019. Here is an ugly solution.
SQL
DECLARE #xml XML = N'<root>
<RECORD>
<RECID>118810</RECID>
<FIELD TYPE="C">
<NAME>proj_code</NAME>
<VALUE>118810</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>sub_nr</NAME>
<VALUE>99900</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>proj_desc</NAME>
<VALUE>Nagekomen kosten Oktober 2018</VALUE>
</FIELD>
<FIELD TYPE="N">
<NAME>pro_stat</NAME>
<VALUE>9</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>comment</NAME>
<VALUE></VALUE>
</FIELD>
</RECORD>
<RECORD>
<RECID>118811</RECID>
<FIELD TYPE="C">
<NAME>proj_code</NAME>
<VALUE>118811</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>sub_nr</NAME>
<VALUE>99901</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>proj_desc</NAME>
<VALUE>Nagekomen kosten November 2019</VALUE>
</FIELD>
<FIELD TYPE="N">
<NAME>pro_stat</NAME>
<VALUE>19</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>comment</NAME>
<VALUE>wow</VALUE>
</FIELD>
</RECORD>
</root>';
DECLARE #tbl TABLE (ID INT IDENTITY(1,1) PRIMARY KEY, RECID VARCHAR(10), [col_name] VARCHAR(20), [col_value] VARCHAR(100));
INSERT INTO #tbl
SELECT c.value('(../RECID/text())[1]', 'VARCHAR(100)') AS [RECID]
, c.value('(NAME/text())[1]', 'VARCHAR(30)') AS [name]
, c.value('(VALUE/text())[1]', 'VARCHAR(100)') AS [value]
FROM #xml.nodes('root/RECORD/FIELD') AS t(c);
DECLARE #RowCount INT = (SELECT MAX(ID) FROM #tbl)
, #recID varchar(10) = (SELECT TOP(1) RECID FROM #tbl WHERE ID = 1)
, #xml_data VARCHAR(MAX) = '<root><RECORD>';
WHILE #RowCount > 0 BEGIN
SELECT #xml_data += IIF(#recID != RECID, '</RECORD><RECORD>', '') +
--'<' + [col_Name] + '>' + COALESCE([col_value],'') + '</' + [col_Name] + '>'
CONCAT('<',[col_Name],'>',[col_value],'</',[col_Name],'>')
, #recID = RECID
FROM #tbl
ORDER BY ID DESC OFFSET #RowCount - 1 ROWS FETCH NEXT 1 ROWS ONLY;
SET #RowCount -= 1;
END;
SET #xml_data += '</RECORD></root>';
SELECT CAST(#xml_data AS XML);

Answer 2

Simply map each RECORD to a new one where you map all FIELDs to elements with the name from NAME and the value from VALUE:
RECORD !
<RECORD>
{
RECID,
FIELD ! element { NAME } { data(VALUE) }
}
</RECORD>
https://xqueryfiddle.liberty-development.net/nbUY4kB
For XQuery 1 you would need to use for return instead of the map operator !:
for $record in //RECORD
return
<RECORD>
{
$record/RECID,
for $field in $record/FIELD
return element { $field/NAME } { data($field/VALUE) }
}
</RECORD>
https://xqueryfiddle.liberty-development.net/nbUY4kB/1

Answer 3

I tried this:
DECLARE #xml xml;
SET #xml =N'
<RECORD>
<RECID>118810</RECID>
<FIELD TYPE="C">
<NAME>proj_code</NAME>
<VALUE>118810</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>sub_nr</NAME>
<VALUE>99900</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>proj_desc</NAME>
<VALUE>Nagekomen kosten Oktober 2018</VALUE>
</FIELD>
<FIELD TYPE="N">
<NAME>pro_stat</NAME>
<VALUE>9</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>comment</NAME>
<VALUE></VALUE>
</FIELD>
</RECORD>
<RECORD>
<RECID>118811</RECID>
<FIELD TYPE="C">
<NAME>proj_code</NAME>
<VALUE>118811</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>sub_nr</NAME>
<VALUE>99900</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>proj_desc</NAME>
<VALUE>Nagekomen kosten November 2018</VALUE>
</FIELD>
<FIELD TYPE="N">
<NAME>pro_stat</NAME>
<VALUE>9</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>comment</NAME>
<VALUE></VALUE>
</FIELD>
</RECORD>
';
SELECT #xml.query('
for $record in //RECORD
return
<RECORD>
{
$record/RECID,
for $field in $record/FIELD
return element { $field/NAME } { data($field/VALUE) }
}
</RECORD>
' ) as result
But i get an error on the return element part:
XQuery [query()]: Only constant expressions are supported for the name expression of computed element and attribute constructors.

Answer 4

This is the final version with select for inserting it into SQL:
DECLARE #xml XML = N'
<AVXML>
<SIGNONMSGRS>
<DTSERVER>2019-09-10T15:54:32</DTSERVER>
<APPID>ACCOUNTVIEW</APPID>
<APPVER>0908-</APPVER>
</SIGNONMSGRS>
<EBUSMSGSRS>
<EBUSQRYRS xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<RECORD>
<RECID>118810</RECID>
<FIELD TYPE="C">
<NAME>proj_code</NAME>
<VALUE>118810</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>sub_nr</NAME>
<VALUE>99900</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>proj_desc</NAME>
<VALUE>Nagekomen kosten Oktober 2018</VALUE>
</FIELD>
<FIELD TYPE="N">
<NAME>pro_stat</NAME>
<VALUE>9</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>comment</NAME>
<VALUE></VALUE>
</FIELD>
</RECORD>
<RECORD>
<RECID>118811</RECID>
<FIELD TYPE="C">
<NAME>proj_code</NAME>
<VALUE>118811</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>sub_nr</NAME>
<VALUE>99901</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>proj_desc</NAME>
<VALUE>Nagekomen kosten November 2019</VALUE>
</FIELD>
<FIELD TYPE="N">
<NAME>pro_stat</NAME>
<VALUE>19</VALUE>
</FIELD>
<FIELD TYPE="C">
<NAME>comment</NAME>
<VALUE>wow</VALUE>
</FIELD>
</RECORD>
</EBUSQRYRS>
</EBUSMSGSRS>
</AVXML>
';
DECLARE #tbl TABLE (ID INT IDENTITY(1,1) PRIMARY KEY, RECID VARCHAR(10), [col_name] VARCHAR(20), [col_value] VARCHAR(100));
INSERT INTO #tbl
SELECT c.value('(../RECID/text())[1]', 'VARCHAR(100)') AS [RECID]
, c.value('(NAME/text())[1]', 'VARCHAR(30)') AS [name]
, c.value('(VALUE/text())[1]', 'VARCHAR(500)') AS [value]
FROM #xml.nodes('//RECORD/FIELD') AS t(c);
DECLARE #RowCount INT = (SELECT MAX(ID) FROM #tbl)
, #recID varchar(10) = (SELECT TOP(1) RECID FROM #tbl WHERE ID = 1)
, #xml_data VARCHAR(MAX) = '<root><RECORD>';
WHILE #RowCount > 0 BEGIN
SELECT #xml_data +=
IIF(#recID != RECID,
'<RECID>' + #recID + '</RECID>' +
'</RECORD><RECORD>',
'') +
CONCAT('<',[col_Name],'>',[col_value],'</',[col_Name],'>')
, #recID = RECID
FROM #tbl
ORDER BY ID DESC OFFSET #RowCount - 1 ROWS FETCH NEXT 1 ROWS ONLY;
SET #RowCount -= 1;
END;
SET #xml_data += '<RECID>' + #recID + '</RECID>' + '</RECORD></root>';
DECLARE #handler int;
exec sys.sp_xml_preparedocument #handler OUTPUT, #xml_data;
--print #xml_data
select *
from OPENXML(#handler,'/root/RECORD',11)
WITH
(
id nvarchar(50) 'RECID',
proj_code nvarchar(50) 'proj_code',
sub_nr nvarchar(50) 'sub_nr',
proj_desc nvarchar(1000) 'proj_desc',
pro_stat nvarchar(50) 'pro_stat',
comment nvarchar(50) 'comment'
)
exec sys.sp_xml_removedocument #handler