Wildcards in CTXCAT grammar vs. CONTEXT grammar

Question

I've got a question about using wildcards in the CONTEXT grammar, when working with a catalog (CTXCAT index). For the application in question, the simpler CTXCAT grammar is almost exactly what I need, but there are a few circumstances in which I want to be able to take advantage of the flexibility of the CONTEXT grammar. I am able to do that using a query template, but when the query in question includes wildcards, I sometimes see behavior that I do not understand. Here's a query that works perfectly when expressed in the CTXCAT grammar:
SELECT count(*)
FROM ItemKeywordSearch iks
WHERE CATSEARCH(iks.ItemKeywords, '403(b)*', null) > 0;
When I try to restate that same query using the CONTEXT grammar like this:
SELECT count(*)
FROM ItemKeywordSearch iks
WHERE CATSEARCH(iks.ItemKeywords, '<query><textquery grammar="CONTEXT">403(b)%</textquery></query>', null) > 0;
I get a "DRG-51030: wildcard query expansion resulted in too many terms" error. I'm guessing the issue is related to some interaction between the parentheses characters and the wildcard; I don't see the same error if I remove the parentheses from the query.
Can anyone help me understand the cause of this issue, and suggest a way I can resolve or work around it?
Thanks,
Tim

Answer 1

I don't know what preference attributes you have set. If you want to search for parentheses, then you should set them as printjoins within a lexer so that they are indexed and you will also need to escape them by placing a \ before each one, so that they are not treated as special characters. You can also set the wildcard_maxterms to the maximum value for your verision in a wordlist. The two queries that use the different grammar may be viewing the search string differently. If one ignores the parentheses tries to search for b% or everything that starts with b then that would expand to a lot of terms very quickly and would cause the error. You can query the dr$token column of your dr$...$i domain index table to see how your data has been tokenized during indexing. Please see the demonstration below.
SCOTT@orcl_11g> CREATE TABLE ItemKeywordSearch
  2    (ItemKeywords  VARCHAR2 (20))
  3  /
Table created.
SCOTT@orcl_11g> INSERT ALL
  2  INTO ItemKeywordSearch VALUES ('403(b)')
  3  INTO ItemKeywordSearch VALUES ('403(b)A')
  4  INTO ItemKeywordSearch VALUES ('403(b)B')
  5  INTO ItemKeywordSearch VALUES ('403(b)C')
  6  INTO ItemKeywordSearch VALUES ('other')
  7  SELECT * FROM DUAL
  8  /
5 rows created.
SCOTT@orcl_11g> BEGIN
  2    CTX_DDL.CREATE_PREFERENCE ('your_lexer', 'BASIC_LEXER');
  3    CTX_DDL.SET_ATTRIBUTE ('your_lexer', 'PRINTJOINS', '()');
  4    --
  5    CTX_DDL.CREATE_PREFERENCE ('your_wordlist', 'BASIC_WORDLIST');
  6    CTX_DDL.SET_ATTRIBUTE ('your_wordlist', 'WILDCARD_MAXTERMS', '50000');
  7  END;
  8  /
PL/SQL procedure successfully completed.
SCOTT@orcl_11g> CREATE INDEX iks_idx
  2  ON ItemKeywordSearch (ItemKeywords)
  3  INDEXTYPE IS CTXSYS.CTXCAT
  4  PARAMETERS
  5    ('LEXER        your_lexer
  6        WORDLIST  your_wordlist')
  7  /
Index created.
SCOTT@orcl_11g> SELECT dr$token FROM dr$iks_idx$i
  2  /
DR$TOKEN
403(B)
403(B)A
403(B)B
403(B)C
OTHER
SCOTT@orcl_11g> SELECT count(*)
  2  FROM   ItemKeywordSearch iks
  3  WHERE  CATSEARCH
  4             (iks.ItemKeywords,
  5              '403\(b\)*',
  6              null) > 0
  7  /
  COUNT(*)
         4
SCOTT@orcl_11g> SELECT count(*)
  2  FROM   ItemKeywordSearch iks
  3  WHERE  CATSEARCH
  4             (iks.ItemKeywords,
  5              '<query>
  6              <textquery grammar="CONTEXT">403\(b\)%
  7              </textquery>
  8            </query>',
  9            null) > 0
10  /
  COUNT(*)
         4
SCOTT@orcl_11g>