wikimedia/mediawiki-extensions-SpamBlacklist
1
0
Fork 0
mirror of https://gerrit.wikimedia.org/r/mediawiki/extensions/SpamBlacklist synced 2024-12-18 01:30:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

Abandon EventLogging instrumentation

Browse source 
The instrumentation attempted here has been implemented more correctly
by https://phabricator.wikimedia.org/T214706.  This change removes
instrumentation from this extension.  It should not affect much,
because emitting events was already disabled.

Bug: T162365
Change-Id: I8f02c62d41260a62b37a7738d69c039d2ce7562f
This commit is contained in:
Dan Andreescu 2019-02-21 14:47:12 -05:00
parent 3f6da90ccf
commit a72c0f339a
3 changed files with 2 additions and 151 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
extension.json
View file

@ -84,9 +84,7 @@
],
"UploadVerifyUpload": [
"SpamBlacklistHooks::onUploadVerifyUpload"
],
"ArticleDelete": "SpamBlacklistHooks::onArticleDelete",
"ArticleDeleteComplete": "SpamBlacklistHooks::onArticleDeleteComplete"
]
},
"config": {
"BlacklistSettings": {
@ -95,8 +93,7 @@
},
"_merge_strategy": "array_plus_2d"
},
"LogSpamBlacklistHits": false,
"SpamBlacklistEventLogging": false
"LogSpamBlacklistHits": false
},
"AuthManagerAutoConfig": {
"preauth": {

102
includes/SpamBlacklist.php
View file

@ -11,12 +11,6 @@ class SpamBlacklist extends BaseBlacklist {
const STASH_TTL = 180;
const STASH_AGE_DYING = 150;
/**
* Changes to external links, for logging purposes
* @var array[]
*/
private $urlChangeLog = [];
/**
* Returns the code for the blacklist implementation
*
@ -56,12 +50,6 @@ class SpamBlacklist extends BaseBlacklist {
$statsd = MediaWikiServices::getInstance()->getStatsdDataFactory();
$cache = ObjectCache::getLocalClusterInstance();
// If there are no new links, and we are logging,
// mark all of the current links as being removed.
if ( !$links && $this->isLoggingEnabled() ) {
$this->logUrlChanges( $this->getCurrentLinks( $title ), [], [] );
}
if ( !$links ) {
return false;
}
@ -110,10 +98,6 @@ class SpamBlacklist extends BaseBlacklist {
wfDebugLog( 'SpamBlacklist', "New URLs: " . implode( ', ', $newLinks ) );
wfDebugLog( 'SpamBlacklist', "Added URLs: " . implode( ', ', $addedLinks ) );
if ( !$preventLog ) {
$this->logUrlChanges( $oldLinks, $newLinks, $addedLinks );
}
$links = implode( "\n", $addedLinks );
# Strip whitelisted URLs from the match
@ -176,92 +160,6 @@ class SpamBlacklist extends BaseBlacklist {
return $retVal;
}
public function isLoggingEnabled() {
global $wgSpamBlacklistEventLogging;
return $wgSpamBlacklistEventLogging &&
ExtensionRegistry::getInstance()->isLoaded( 'EventLogging' );
}
/**
* Diff added/removed urls and generate events for them
*
* @param string[] $oldLinks
* @param string[] $newLinks
* @param string[] $addedLinks
*/
public function logUrlChanges( $oldLinks, $newLinks, $addedLinks ) {
if ( !$this->isLoggingEnabled() ) {
return;
}
$removedLinks = array_diff( $oldLinks, $newLinks );
foreach ( $addedLinks as $url ) {
$this->logUrlChange( $url, 'insert' );
}
foreach ( $removedLinks as $url ) {
$this->logUrlChange( $url, 'remove' );
}
}
/**
* Actually push the url change events post-save
*
* @param User $user
* @param Title $title
* @param int $revId
*/
public function doLogging( User $user, Title $title, $revId ) {
if ( !$this->isLoggingEnabled() ) {
return;
}
$baseInfo = [
'revId' => $revId,
'pageId' => $title->getArticleID(),
'pageNamespace' => $title->getNamespace(),
'userId' => $user->getId(),
'userText' => $user->getName(),
];
$changes = $this->urlChangeLog;
// Empty the changes queue in case this function gets called more than once
$this->urlChangeLog = [];
DeferredUpdates::addCallableUpdate( function () use ( $changes, $baseInfo ) {
foreach ( $changes as $change ) {
EventLogging::logEvent(
'ExternalLinksChange',
15716074,
$baseInfo + $change
);
}
} );
}
/**
* Queue log data about change for a url addition or removal
*
* @param string $url
* @param string $action 'insert' or 'remove'
*/
private function logUrlChange( $url, $action ) {
$parsed = wfParseUrl( $url );
if ( !isset( $parsed['host'] ) ) {
wfDebugLog( 'SpamBlacklist', "Unable to parse $url" );
return;
}
$info = [
'action' => $action,
'protocol' => $parsed['scheme'],
'domain' => $parsed['host'],
'path' => $parsed['path'] ?? '',
'query' => $parsed['query'] ?? '',
'fragment' => $parsed['fragment'] ?? '',
];
$this->urlChangeLog[] = $info;
}
/**
* Look up the links currently in the article, so we can
* ignore them on a second run.

44
includes/SpamBlacklistHooks.php
View file

@ -171,11 +171,6 @@ class SpamBlacklistHooks {
Status $status,
$baseRevId
) {
if ( $revision ) {
BaseBlacklist::getSpamBlacklist()
->doLogging( $user, $wikiPage->getTitle(), $revision->getId() );
}
if ( !BaseBlacklist::isLocalSource( $wikiPage->getTitle() ) ) {
return true;
}
@ -242,43 +237,4 @@ class SpamBlacklistHooks {
return true;
}
/**
* @param WikiPage &$article
* @param User &$user
* @param string &$reason
* @param string &$error
*/
public static function onArticleDelete( WikiPage &$article, User &$user, &$reason, &$error ) {
$spam = BaseBlacklist::getSpamBlacklist();
if ( !$spam->isLoggingEnabled() ) {
return;
}
// Log the changes, but we only commit them once the deletion has happened.
// We do that since the external links table could get cleared before the
// ArticleDeleteComplete hook runs
$spam->logUrlChanges( $spam->getCurrentLinks( $article->getTitle() ), [], [] );
}
/**
* @param WikiPage &$page
* @param User &$user
* @param string $reason
* @param int $id
* @param Content|null $content
* @param LogEntry $logEntry
* @suppress PhanParamReqAfterOpt
*/
public static function onArticleDeleteComplete(
&$page,
User &$user,
$reason,
$id,
Content $content = null,
LogEntry $logEntry
) {
$spam = BaseBlacklist::getSpamBlacklist();
$spam->doLogging( $user, $page->getTitle(), $page->getLatest() );
}
}