I noticed that MySQL full text search has this nice optimisation to create better matches for the file namespace.......
if ( $ns === NS_FILE ) {
$t = preg_replace( "/ (png|gif|jpg|jpeg|ogg)$/", "", $t );
}Not exactly up to date :)
Description
Description
Details
Details
| Subject | Repo | Branch | Lines +/- | |
|---|---|---|---|---|
| Don't hardcode the file extension in searchindex normalization | mediawiki/core | master | +2 -10 |
Event Timeline
Comment Actions
Change #1040591 had a related patch set uploaded (by TheDJ; author: TheDJ):
[mediawiki/core@master] Don't hardcode the file extension in searchindex normalization
Comment Actions
@dcausse @Gehel As far as I can see, updateTitle is not implemented by CirrusSearch right, and thus a noop per the parent SearchEngine class ? If so, then i can safely modify this.
I do note that each SearchUpdate Job goes through multiple regexes on the title before it hands it to the noop, which is a tad wasteful
Comment Actions
Indeed CirrusSearch uses other events to trigger updates, reason is that it has to support the various ways the parsed content can be changed (mainly via LinksUpdate). So yes modifying this in core should not impact CirrusSearch.
I do note that each SearchUpdate Job goes through multiple regexes on the title before it hands it to the noop, which is a tad wasteful
I think that in the case of CirrusSearch this is optimized by returning false on SearchEngine::support( 'search-update' ) at https://gerrit.wikimedia.org/g/mediawiki/core/+/8e8093d627ec92fc7a1d9226dcab8135f09c25be/includes/deferred/SearchUpdate.php#82