This is the code I am currently using:
SET serveroutput ON
CREATE OR REPLACE
PROCEDURE test_proc(i_xml varchar2)
IS
l_name VARCHAR2(20);
l_age NUMBER;
l_xml xmltype;
BEGIN
l_xml := xmltype(i_xml);
FOR x IN
(SELECT VALUE(p) col_val
FROM TABLE(XMLSEQUENCE(EXTRACT(l_xml, '/ROWSET/ROW'))) p
)
LOOP
IF x.col_val.existSNode('/ROW/name/text()') > 0 THEN
l_name:= x.col_val.EXTRACT('/ROW/name/text()').getstringVal();
END IF;
IF x.col_val.existSNode('/ROW/age/text()') > 0 THEN
l_age := x.col_val.EXTRACT('/ROW/age/text()').getstringVal();
END IF;
end loop;
end;
/
BEGIN
test_proc('<ROWSET>
<ROW>
<name>aa</name>
<age>20</age>
</ROW>
<ROW>
<name>bbb</name>
<age>25</age>
</ROW>
</ROWSET>');
END;
/
The above code uses xml to extract and save the existing node values to particular local variables. It is been used in the case for multiple sets of data and is working fine. I just wanted to know whether can I able to use the same without "for x loop", because I will only have one data in the i_xml from now onwards and I will only have either
name
or age
tags .
The following code should be used to save into l_name or l_age without the "loop" method like I used above:
<ROWSET>
<ROW>
<name>aa</name>
</ROW>
</ROWSET>
or
<ROWSET>
<ROW>
<age>18</age>
</ROW>
</ROWSET>
/ And I've tried using the following:
SELECT
CASE
WHEN VALUE(p).existsNode('/ROW/name/text()') = 1
THEN p.EXTRACT('/ROW/name/text()').getstringVal()
WHEN VALUE(P).existsNode('/ROW/age/text()') = 1
THEN p.EXTRACT('/ROW/age/text()').getstringVal()
END
INTO l_new
FROM TABLE(xmlsequence(EXTRACT(l_xml, '/ROWSET/ROW'))) p;
/ Any better way is appreciated.. Thanks
If you're really sure you'll only have one ROW
then you can do:
begin
l_xml := xmltype(i_xml);
if l_xml.existsnode('/ROWSET/ROW/name') > 0 then
l_name := l_xml.extract('/ROWSET/ROW/name/text()').getstringval();
end if;
if l_xml.existsnode('/ROWSET/ROW/age') > 0 then
l_age := l_xml.extract('/ROWSET/ROW/age/text()').getnumberval();
end if;
end;
That will work if you have name
or age
, or both, or neither (where 'work' means doesn't error, at least). If you did have more than one row it would concatenate the results, so with your original data, l_name
would be aabbb
, and l_age
would be 2025
. Which might not be what you expect.