>That looks feasable. Anyone want to take on re-writing the query parser?
I've made real progress with rewriting the query parser; not quite there,
but I'm pretty confident I'm getting close and its coming faster than I
expected. And, as an added bonus, AND and OR will work now (they didn't
before.) Also, the new parser won't be confused with things like pluses
and minuses in the query text. I also added a main() to QueryParser where
you can give it a query and it will spit out the query.toString() value so
you can test it, in addition to the JUnit test case for it.
The way it will work is this: the Analyzer only gets a crack at the actual
clause text, not the entire input text. So if the field name is a stop
word, no problem. If the analyzer expands or shortens the clause text, it
will be adjusted properly, including if the analyzer shortens it to
zero. I also plan to add the term boost feature, since that's easy.
If anyone wants to send me their nastiest queries, I could use all the test
cases I can get.
The changes are serious enough (basically, a total rewrite of the
QueryParser) that some peer review and testing is in order before
committing it. When I get finished, I will post my new version to the list
before committing it (thankfully, it is only one module) and I would
appreciate the help of anyone willing to try it out, peer review it, etc.
--
Brian Goetz
Quiotix Corporation
[EMAIL PROTECTED] Tel: 650-843-1300 Fax: 650-324-8032
http://www.quiotix.com
_______________________________________________
Lucene-dev mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/lucene-dev
