Can dbms_lob.instr have end position?

Question

I am extracting data from xml using dbms_lob utility and the pattern in XML is as below.
<Atag>
<Btag>
<Ctag>value</Ctag>
</Btag>
<Btag>
<Ctag>value</Ctag>
<Dtag>value</Dtag>
</Btag>
<Btag>
<Ctag>value</Ctag>
<Dtag>value</Dtag>
</Btag>
</Atag>
I want to extract all tag values inside <Btag> which will be nohting but Ctag and Dtag values. There can be any occurence of <Btag> in this xml. When I use like this,
val:=0;
<LOOP>
val1:=dbms_lob.instr(xml,'<Btag>',val);
val2:=dbms_lob.instr(xml,'</Btag>,val1);
val3:=case(val1) when 0 then 0 else dbms_lob.instr(xml,'<Ctag>,val1) end;
val4:=case(val3) when 0 then 0 else dbms_lob.instr(xml,'</Ctag>,val3) end;
val5:=case(val1) when 0 then 0 else dbms_lob.instr(xml,'<Dtag>,val1) end;
val6:=case(val5) when 0 then 0 else dbms_lob.instr(xml,'</Dtag>,val5) end;
In this case as <Dtag> doesnt exist inside the first occurence of <Btag>, so it gets the value from the next one which is not what I require. I want that to be set to null/zero so for which I want to search only between the first occurence of <Btag> and </Btag>.
I know it is confusing but can some one of you tell me how to go and find the values of <Ctag> and <Dtag> between each occurence of <Btag>. If they are not present I want the value to be set to zero/null.
Thanks in advance.

Answer 1

test@ora>
test@ora> @ver1
BANNER
Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - Prod
1 row selected.
test@ora>
test@ora>
test@ora> --
test@ora> with t as (
  2    select '<Atag>
  3  <Btag>
  4  <Ctag>Tootsie</Ctag>
  5  </Btag>
  6  <Btag>
  7  <Ctag>Psycho</Ctag>
  8  <Dtag>Casablanca</Dtag>
  9  </Btag>
10  <Btag>
11  <Ctag>Goodfellas</Ctag>
12  <Dtag>Chinatown</Dtag>
13  </Btag>
14  </Atag>' as x from dual)
15  --
16  select
17    x,
18    regexp_replace(modx,'(.*?)<Btag>(.*?)</Btag>.*$','\2',1,1) as txt1,
19    regexp_replace(modx,'(.*?)(<Btag>(.*?)</Btag>){1}<Btag>(.*?)</Btag>.*$','\4',1,1) as txt2,
20    regexp_replace(modx,'(.*?)(<Btag>(.*?)</Btag>){2}<Btag>(.*?)</Btag>.*$','\4',1,1) as txt3
21  from (
22    select regexp_replace(x,'(>).(<)','\1\2',1,0,'n') as modx, x from t
23  );
X                              TXT1                 TXT2                                        TXT3
<Atag>                         <Ctag>Tootsie</Ctag> <Ctag>Psycho</Ctag><Dtag>Casablanca</Dtag>  <Ctag>Goodfellas</Ctag><Dtag>Chinatown</Dtag>
<Btag>
<Ctag>Tootsie</Ctag>
</Btag>
<Btag>
<Ctag>Psycho</Ctag>
<Dtag>Casablanca</Dtag>
</Btag>
<Btag>
<Ctag>Goodfellas</Ctag>
<Dtag>Chinatown</Dtag>
</Btag>
</Atag>
1 row selected.
test@ora>
test@ora>
test@ora>@Keith - Regular expressions are from 10g upwards.
isotope