489380 - New searches on hyphenated words treats as two separate words

Status: VERIFIED FIXED

(support.mozilla.org :: General, defect)

Description

[[:Cww]]

Searches on "marque-page" (Bookmarks in French) searches for page and marque as independent words, returning pretty much everything (because page = webpage).

Hyphenated phrases should be treated as one word in searches.

Comment 1

[Paul Craciunoiu [:paulc]]

There's this forum post that mentions possible solutions:
http://www.sphinxsearch.com/forum/view.html?id=2876
But I'm not yet sure where those variables (such as charset_table) appear, need to look through. I'll be busy till the evening but if no one figures out where sphinx is configured, I'll look into it later.

Comment 2

[Paul Craciunoiu [:paulc]]

r24770 / r24771
It seems that there's no need to modify the indexer. Hyphens were being stripped in tiki-nnewsearch.php.

Also modified the bolding a bit, it didn't seem to work for hyphenated words.

Comment 3

[krupa raj[:krupa]]

Searching for "marque-page" doesnt return any search results.

https://support-stage.mozilla.org/tiki-newsearch.php?locale=fr&q=marque-pages+&where=all&sa=&filter_lang=1&l=fr&en_too=1&lastmodif=0&type=0&author=

Reopening...

Comment 4

[Paul Craciunoiu [:paulc]]

I think we're waiting for reindexing (cache cleared, etc). Nothing really returns anything anymore, and I didn't change other stuff :)

Comment 5

[Paul Craciunoiu [:paulc]]

This works:
https://support-stage.mozilla.org/tiki-newsearch.php?locale=fr&q=marque-page+&where=all&sa=&filter_lang=1&l=fr&en_too=1&lastmodif=0&type=0&author
(for marque-page)
The fact that the plural doesn't work is not related to this bug.

See:
https://support-stage.mozilla.org/tiki-newsearch.php?locale=en&q=bookmarks+&where=all&sa=&filter_lang=1&l=fr&en_too=1&lastmodif=0&type=0&author
(bookmarks)
vs
https://support-stage.mozilla.org/tiki-newsearch.php?locale=en&q=bookmark+&where=all&sa=&filter_lang=1&l=fr&en_too=1&lastmodif=0&type=0&author

The results differ.

Comment 6

[Stephen Donner [:stephend] Not actively reading bugmail]

Verified FIXED on https://support-stage.mozilla.org/tiki-newsearch.php?locale=fr&q=marque-pages+&where=all&sa=&filter_lang=1&l=fr&en_too=1&lastmodif=0&type=0&author (Krupa and I looked at this).