Jarry1250 has uploaded a new change for review.
https://gerrit.wikimedia.org/r/156885
Change subject: Phase out use of MessageGroupStats in favour of more brutal
methods
......................................................................
Phase out use of MessageGroupStats in favour of more brutal methods
MessageGroupStats uses a powerful but slow approach, iterating over
every possible language, and then compiling the results in the database.
This means it performs well if message groups are reused and developed,
but poor performance if there are lots of new message groups to crawl.
At the moment, such power seems unneccessary: getOnWikiLanguages is
called infrequently, and does not require the filtering options
provided by MessageGroupStats. Thus the first-time performance hit
is the more major concern.
It may be in future that the power/performance balance shifts, in which
case this change can be easily reverted.
Change-Id: I46253980a236d9f0b22aa3eee44540ef89fe11a1
---
M SVGMessageGroup.php
1 file changed, 22 insertions(+), 10 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/mediawiki/extensions/TranslateSvg
refs/changes/85/156885/1
diff --git a/SVGMessageGroup.php b/SVGMessageGroup.php
index 2562baa..9c2e3a0 100644
--- a/SVGMessageGroup.php
+++ b/SVGMessageGroup.php
@@ -157,6 +157,10 @@
if ( $language === 'fallback' ) {
$language = $this->getSourceLanguage();
}
+ if ( !Language::isValidBuiltInCode( $language )
) {
+ // Some weird attack vector?
+ continue;
+ }
$translation =
TranslateSvgUtils::arrayToTranslation( $innerArray );
$fullKey = $this->source . '/' . $key . '/' .
$language;
$title = Title::makeTitleSafe(
$this->getNamespace(), $fullKey );
@@ -210,19 +214,26 @@
/**
* Returns a list of languages the file has been translated into *on
wiki*
- * i.e. some of those may not have been saved back to the file yet.
+ * i.e. some of those may not have been saved back to the file yet.
Only the language code is
+ * (even minimally) verified. A pretty brutal method all things
considered, but effective and quick --
+ * the downsides being that we assume a certain file structure and
there's minimal caching.
*/
public function getOnWikiLanguages() {
- $stats = MessageGroupStats::forGroup( $this->getId() );
$languages = array();
- foreach ( $stats as $language => $data ) {
- $translatedCount = $data[MessageGroupStats::TRANSLATED];
- $fuzzyCount = $data[MessageGroupStats::FUZZY];
- if ( $translatedCount > 0 || $fuzzyCount > 0 ) {
- $languages[] = $language;
+ $subpages = Title::makeTitleSafe( $this->getNamespace(),
$this->source )->getSubpages();
+
+ foreach ( $subpages as $subpage ) {
+ /** @var Title $subpage */
+
+ // These are subpages of the form
File:Foo.svg/tspan2991/de, i.e. actually sub-subpages,
+ // but $subpage->getSubpageText() will nevertheless
just get the last bit (i.e. the language code).
+ $langCode = $subpage->getSubpageText();
+ if( Language::isSupportedLanguage( $langCode ) ) {
+ $languages[] = $langCode;
}
}
- return $languages;
+
+ return array_unique( $languages );
}
/*
@@ -236,10 +247,11 @@
return $this->onWikiTranslations;
}
- $onWikiTranslations = array();
+ // SVGs are unlikely to be translated into more than a dozen,
so let's check first to see which
+ // languages have any translations at all in order to narrow
our search.
$languages = $this->getOnWikiLanguages();
- // Translations generated onwiki
+ $onWikiTranslations = array();
foreach ( $languages as $language ) {
$collection = $this->initCollection( $language );
$collection->loadTranslations();
--
To view, visit https://gerrit.wikimedia.org/r/156885
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: I46253980a236d9f0b22aa3eee44540ef89fe11a1
Gerrit-PatchSet: 1
Gerrit-Project: mediawiki/extensions/TranslateSvg
Gerrit-Branch: master
Gerrit-Owner: Jarry1250 <jarry1...@gmail.com>
_______________________________________________
MediaWiki-commits mailing list
MediaWiki-commits@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
