Reputation: 3209
Not sure how exactly the regexp option is to be used.
What I want: That words like ['wick','wicks'] and ['television',''televisions'] give the same search results.
This is what I tried:
regexp_filter = (?i)\b([^\s]+?[^es])s\b => \1
After using this television matches televisions, but televisions doesn't match anything! Which is quite strange..
I also tried
regexp_filter = (?i)\b([^\s]+?[^es])s\b => \1s
which was like not having this filter at all..(television matched only television, televisions matched only televisions)
Sphinx Version 2.2.8
Installed Re2
Debian Squeeze
Upvotes: 0
Views: 208
Reputation: 3209
Edit: Not quite. The regexp filter
regexp_filter = (?i)\b([^\s]+?[^es])s\b => \1s \1
makes 'words' -> match 'words' and 'word' . However 'word' still doesn't match 'words' :(
I figured it out! again..
What I had to do was use both \1s and \1 , so 'word' matches both 'word' and 'words'
regexp_filter = (?i)\b([^\s]+?[^es])s\b => \1s \1
Got this info from here http://gotoanswer.stanford.edu/?q=Configure+Sphinx+to+handle+space+as+possible+words+separator
Upvotes: 1