Page MenuHomePhabricator

Searching for a subpage title using the search bar at the top of the page (prefix) doesn't propose it for autocomplete
Closed, DeclinedPublic


If a subpage like Blah/Yada exists, typing Yada into the search bar at the top of a page doesn't offer any auto-complete suggestions. I think it should treat the page title as if it were "Yada" for search purposes, since to a human, the page title really is Yada. The fact that it happens to be a sub-page is irrelevant.

Doing a search for "prefix: Yada" also returns no matches, which seems clearly wrong from a user perspective.

I'll file a separate similar task for the new completion suggester beta feature.

Event Timeline

ksmith raised the priority of this task from to Needs Triage.
ksmith updated the task description. (Show Details)
ksmith added a project: CirrusSearch.
ksmith added a subscriber: ksmith.
Restricted Application added subscribers: StudiesWorld, Aklapper. · View Herald Transcript
debt triaged this task as Medium priority.Aug 30 2016, 10:19 PM
debt moved this task from needs triage to Current work on the Discovery-Search board.
debt edited projects, added Discovery-Search (Current work); removed Discovery-Search.

Clarifying note: I created this task back when the completion suggester was in beta. At that point, this issue would have been the more important one. Now that the completion suggester is enabled for normal searches by default, T123015 is presumably more valuable than this one.

I think Erik said this would still kick in with some special types of searches (which we're not even sure how many people use).

Will wait for @dcausse to reply, but my impression is based on how prefix search currently works this would be a rather large undertaking.

I'm not sure yet how to handle that with prefix.
I think I can add subfields to title and redirect like subphrase_prefix
I'll then have to implement a custom lucene filter to reimplement the bruteforce and naive subphrase/subphrase extractor done in php.
I'm not sure yet how to exclude translate subpages that are easily ignored on php side. When writing a lucene analyzer/filter you don't have access to the context of the document thus I don't have access to the language field to detect the translate suffix.

On cirrus side I'll have to refactor the prefix search query out of the searcher class to allow variations of the fields used.
Then I'm not sure how to expose this new functionality to the user...

Let's review the demo here and see how things are looking based on ticket T123015. I'll close this ticket out as declined.