Skip to main content.
home | support | download

Back to List Archive

Re: Special Characters and Search

From: Bill Meier <bill(at)not-real.insulators.com>
Date: Tue May 15 2001 - 12:58:45 GMT
At 08:02 PM 05/10/01, you wrote:
>My searches fail if the keyword includes special characters, including 
>comma, period, forward slash, and hyphen. As a result, I miss hits like 
>3/30/01, $4,000, $40.10, and hell-bent.
>
>I can find these hits if I search with the special character omitted: 3 30 
>01, $4 000, $40 10, and hell bent.
>
>Is this another issue with query.pl, or must I re-configure swish-e itself 
>in some way? Thanks!

Swish only indexes and allows you to search for "words" -- the examples you 
site are not "words" to swish.

Note that a search for "3 30 01" just "happens" to come close to matching 
3/30/01, but what is it matching is ANY document with the numbers 3, 30, 
and 01 in them...

You can change the way swish looks for words. You can add / and - to the 
characters it considers words. But, you have to be aware of the 
consequences of this... if "/" is defined as a word character to swish, 
then "hot/cold" will be indexed as one word, and searching for "hot" (or 
"cold") will not match that word.

Read the documentation in the config file about WORDCHARS.

And, you may have to tweak query.pl to not filter out some of these 
characters too. You can certainly add $ and , as word characters, but you 
also need to set ENDCHARS IGNOREFIRSTCHAR IGNORELASTCHAR as otherwise comma 
at the end of a word will be indexed too... !!!

Bill
Received on Tue May 15 13:00:03 2001