Creating a PLSQL script to parse structured XML with repeated tags

Question

Hi, I'm trying to parse an xml feed but the way it's structured makes it difficult to pull out the values. (see example) Would anyone be able to recommend a solution or some ideas as to how to get around this?
SAMPLE XML:<env:Envelope xmlns:env='http://schemas.xmlsoap.org/soap/envelope/'>
     <env:Header>
     </env:Header>
     <env:Body>
          <ns3:commandResponse xmlns:ns2="http://test.com/2011/Generic/schema" xmlns:ns3="http://test.com/2011/Generic">
               <return>
               <ns2:return>success</ns2:return>
               <ns2:Command>issues</ns2:Command>
                    <ns2:WorkItems>
                         <ns2:Id>216141</ns2:Id>
                         <ns2:ModelType>Issue</ns2:ModelType>
                         <ns2:DisplayId>216141</ns2:DisplayId>
                         <ns2:Field>
                              <ns2:Name>Type</ns2:Name>
                              <ns2:Value>
                                   <ns2:Item>
                                        <ns2:Id>Dev Task</ns2:Id>
                                        <ns2:ModelType>Type</ns2:ModelType>
                                        <ns2:DisplayId>Dev Task</ns2:DisplayId>
                                   </ns2:Item>
                              </ns2:Value>
                         </ns2:Field>
                         <ns2:Field>
                              <ns2:Name>ID</ns2:Name>
                              <ns2:Value>
                                   <ns2:int>216141</ns2:int>
                              </ns2:Value>
                         </ns2:Field>
                         <ns2:Field>
                              <ns2:Name>Reason</ns2:Name>
                              <ns2:Value>
                                   <ns2:string>Integrating</ns2:string>
                              </ns2:Value>
                         </ns2:Field>
                         <ns2:Field>
                              <ns2:Name>Dev Task Component</ns2:Name>
                              <ns2:Value>
                                   <ns2:string>Java Tools</ns2:string>
                              </ns2:Value>
                         </ns2:Field>
                         <ns2:Field>
                              <ns2:Name>Created Date</ns2:Name>
                              <ns2:Value>
                                   <ns2:datetime>2009-08-10T15:52:39.000-04:00</ns2:datetime>
                              </ns2:Value>
                         </ns2:Field>
                         <ns2:Field>
                              <ns2:Name>Date Closed</ns2:Name>
                              <ns2:Value/>
                         </ns2:Field>
                         <ns2:Field>
                              <ns2:Name>Modified Date</ns2:Name>
                              <ns2:Value>
                                   <ns2:datetime>2011-03-04T12:57:05.000-05:00</ns2:datetime>
                              </ns2:Value>
                         </ns2:Field>
                    </ns2:WorkItems>
               </return>
          </ns3:commandResponse>
     </env:Body>
</env:Envelope>This is just a sample with just one WorkItem, but there would be much more, N number of items with 9 fields per item. (Not all of the fields were put in the sample, and some can have null values)
I only need to pull the content from /ns2:WorkItems/ns2:Field/ns2:Value/ns2:Item/ns2:Id for the first field and the /ns2:value/* tag of all the other fields. Then put this in a table where each row is a workitem and the fields are the columns (create table workitems (Type,ID,Reason,Dev Task Component,Created Date, Date Closed, Modified Date) --all the fields should be varchar2 except the dates)
What I've been trying so far seems rather brute force by running a nested loop to go through every item and field and then an IF case for each field 1,2,...9 which would insert the value into a table.
At the moment I'm using something like below to pull a single value
path1 = '//ns2:WorkItems[1]/ns2:Field[1]/ns2:Value[1]/ns2:Item[1]/ns2:Id[1]';
nameserve = 'xmlns:ns2="http://test.com/2011/Generic/schema"';
extractvalue(xmltype(src_clob),path1,nameserve);I'm not entirely sure if I would be able to substitute the [1]'s with [' || nitem || '] where nitem is loop number to do something like:
for nitem in 1..itemcount
loop
    FOR nfield in 1..9
    loop
        if nfield=1 then
            path1 := '//ns2:WorkItems[' || nitem || ']/ns2:Field[' || nfield || ']/ns2:Value[1]/ns2:Item[1]/ns2:Id';
            fieldvalue := extractvalue(xmltype(src_clob),path1,nameserve);';
        else
            path2 := '//ns2:WorkItems[' || nitem || ']/ns2:Field[' || nfield || ']/ns2:Value[1]/*[1]';
            fieldvalue := extractvalue(xmltype(src_clob),path2,nameserve);';
        end if;
    end loop;
end loop;The problem with the above script is how do I insert this fieldvalue into different columns on a table without using an IF case for each field.
I was wondering if there is simpler way to put each field into a different column and loop through every workitem. I looked into dynamically naming variables but I don't think plsql supports that.
Any help/advice is appreciated,
Thanks!
Edited by: 843508 on Mar 10, 2011 1:56 PM
Edited by: 843508 on Mar 10, 2011 1:57 PM
Edited by: 843508 on Mar 10, 2011 2:01 PM

Answer 1

If it were me, I wouldn't use PL/SQL to try and process XML, but would use SQL's XMLTABLE functionality e.g.
SQL> ed
Wrote file afiedt.buf
  1  WITH t as (select XMLTYPE('
  2  <RECSET xmlns:aa="http://www.w3.org">
  3    <aa:REC>
  4      <aa:COUNTRY>1</aa:COUNTRY>
  5      <aa:POINT>1800</aa:POINT>
  6      <aa:USER_INFO>
  7        <aa:USER_ID>1</aa:USER_ID>
  8        <aa:TARGET>28</aa:TARGET>
  9        <aa:STATE>6</aa:STATE>
10        <aa:TASK>12</aa:TASK>
11      </aa:USER_INFO>
12      <aa:USER_INFO>
13        <aa:USER_ID>5</aa:USER_ID>
14        <aa:TARGET>19</aa:TARGET>
15        <aa:STATE>1</aa:STATE>
16        <aa:TASK>90</aa:TASK>
17      </aa:USER_INFO>
18    </aa:REC>
19    <aa:REC>
20      <aa:COUNTRY>2</aa:COUNTRY>
21      <aa:POINT>2400</aa:POINT>
22      <aa:USER_INFO>
23        <aa:USER_ID>3</aa:USER_ID>
24        <aa:TARGET>14</aa:TARGET>
25        <aa:STATE>7</aa:STATE>
26        <aa:TASK>5</aa:TASK>
27      </aa:USER_INFO>
28    </aa:REC>
29  </RECSET>') as xml from dual)
30  -- END OF TEST DATA
31  select x.country, x.point, y.user_id, y.target, y.state, y.task
32  from t
33      ,XMLTABLE(XMLNAMESPACES('http://www.w3.org' as "aa"),
34                '/RECSET/aa:REC'
35                PASSING t.xml
36                COLUMNS country NUMBER PATH '/aa:REC/aa:COUNTRY'
37                       ,point   NUMBER PATH '/aa:REC/aa:POINT'
38                       ,user_info XMLTYPE PATH '/aa:REC/*'
39               ) x
40      ,XMLTABLE(XMLNAMESPACES('http://www.w3.org' as "aa"),
41                '/aa:USER_INFO'
42                PASSING x.user_info
43                COLUMNS user_id NUMBER PATH '/aa:USER_INFO/aa:USER_ID'
44                       ,target  NUMBER PATH '/aa:USER_INFO/aa:TARGET'
45                       ,state   NUMBER PATH '/aa:USER_INFO/aa:STATE'
46                       ,task    NUMBER PATH '/aa:USER_INFO/aa:TASK'
47*              ) y
SQL> /
   COUNTRY      POINT    USER_ID     TARGET      STATE       TASK
         1       1800          1         28          6         12
         1       1800          5         19          1         90
         2       2400          3         14          7          5p.s. XML questions are better suited in the XML DB forum:
XML DB FAQ