Page MenuHomePhabricator

[INVALID] number of words below search snippets
Closed, ResolvedPublic

Description

In example URL:

maggiore
Italiano Aggettivo maggiore m sing (pl: maggiori) (musica) relativo al
byte ([INVALID] parole) - 16:15, 5 set 2013

airone bianco maggiore
Un airone bianco maggiore (Casmerodius albus) Italiano Locuzione
byte ([INVALID] parole) - 20:27, 8 lug 2013

malfattore
Italiano Sostantivo malfattore m sing (pl: malfattori) chi commette reati Sillabazione mal | fat |
650 byte (650 parole) - 19:26, 24 ago 2013


Version: master
Severity: minor
URL: https://it.wiktionary.org/w/index.php?title=Speciale%3ARicerca&profile=default&search=ma*or&fulltext=Search

Details

Reference
bz56020

Event Timeline

bzimport raised the priority of this task from to High.Nov 22 2014, 2:25 AM
bzimport added a project: CirrusSearch.
bzimport set Reference to bz56020.
bzimport added a subscriber: Unknown Object (MLST).

Probably introduced by 2f801f15c07f.

I'm not familiar with the CirrusSearch code, though it would seem that for some CirrusSearchResults, both $result->text_words and $result->text_bytes are some non-integer, non-null value.

(In reply to comment #1)

I'm not familiar with the CirrusSearch code, though it would seem that for
some
CirrusSearchResults, both $result->text_words and $result->text_bytes are
some
non-integer, non-null value.

Actually, null (or an array or object containing null) would be the only thing CirrusSearchResult::getWordCount() could be returning, because nothing else would cause "[INVALID]" to be shown...

Could $result->textLen be null for some reason?

I'm with Kevin, it had to be some mistake I made with 2f801f15c07f. I'm not sure we'll ever know because we had a mini disaster and ended up rebuilding the search indexes before I could reproduce the problem. Now that the indexes are rebuilt, everything is working. The thing is, that wasn't supposed to be required. It was supposed to default to that textLen field we'd been writing to Elasticsearch since we started the project.

Its working so I'm resolving the bug, but I'm not happy with this because I don't know _what_ I didn't wrong and don't think I'm going to be able to figure it out.

I've been able to reproduce this locally so I should be able to figure out what caused this.

Change 92357 had a related patch set uploaded by Manybubbles:
Prevent search index updates of nonexistant pages

https://gerrit.wikimedia.org/r/92357

Change 92359 had a related patch set uploaded by Manybubbles:
Tests catching some invalid word count causes

https://gerrit.wikimedia.org/r/92359

Change 92357 merged by jenkins-bot:
Prevent search index updates of nonexistant pages

https://gerrit.wikimedia.org/r/92357

Change 92359 merged by jenkins-bot:
Tests catching some invalid word count causes

https://gerrit.wikimedia.org/r/92359