Make LinkFilter handle Content objects
authordaniel <daniel.kinzler@wikimedia.de>
Tue, 21 Aug 2012 15:54:52 +0000 (17:54 +0200)
committerdaniel <daniel.kinzler@wikimedia.de>
Tue, 21 Aug 2012 15:54:52 +0000 (17:54 +0200)
Change-Id: Iced1c83642143faa79a5461ae6278267e0595aa9

includes/LinkFilter.php
maintenance/cleanupSpam.php

index 214f495..b19f6c4 100644 (file)
 class LinkFilter {
 
        /**
-        * Check whether $text contains a link to $filterEntry
+        * Check whether $content contains a link to $filterEntry
         *
-        * @param $text String: text to check
+        * @param $content Content: content to check
         * @param $filterEntry String: domainparts, see makeRegex() for more details
         * @return Integer: 0 if no match or 1 if there's at least one match
         */
-       static function matchEntry( $text, $filterEntry ) {
+       static function matchEntry( Content $content, $filterEntry ) {
+               if ( !( $content instanceof TextContent ) ) {
+                       //TODO: handle other types of content too.
+                       //      Maybe create ContentHandler::matchFilter( LinkFilter ).
+                       //      Think about a common base class for LinkFilter and MagicWord.
+                       return 0;
+               }
+
+               $text = $content->getNativeData();
+
                $regex = LinkFilter::makeRegex( $filterEntry );
                return preg_match( $regex, $text );
        }
index 03a36f5..bbe0609 100644 (file)
@@ -103,8 +103,8 @@ class CleanupSpam extends Maintenance {
                $rev = Revision::newFromTitle( $title );
                $currentRevId = $rev->getId();
 
-               //FIXME: LinkFilter needs to handle Content objects! Or rather, ContentHandler needs to provide the appropriate LinkFilter.
-               while ( $rev && ( $rev->isDeleted( Revision::DELETED_TEXT ) || LinkFilter::matchEntry( $rev->getText() , $domain ) ) ) {
+               while ( $rev && ( $rev->isDeleted( Revision::DELETED_TEXT )
+                                               || LinkFilter::matchEntry( $rev->getContent( Revision::RAW ), $domain ) ) ) {
                        $rev = $rev->getPrevious();
                }
 
@@ -118,8 +118,10 @@ class CleanupSpam extends Maintenance {
                        $page = WikiPage::factory( $title );
                        if ( $rev ) {
                                // Revert to this revision
+                               $content = $rev->getContent( Revision::RAW );
+
                                $this->output( "reverting\n" );
-                               $page->doEdit( $rev->getText(), wfMessage( 'spam_reverting', $domain )->inContentLanguage()->text(),
+                               $page->doEditContent( $content, wfMessage( 'spam_reverting', $domain )->inContentLanguage()->text(),
                                        EDIT_UPDATE, $rev->getId() );
                        } elseif ( $this->hasOption( 'delete' ) ) {
                                // Didn't find a non-spammy revision, blank the page
@@ -127,8 +129,11 @@ class CleanupSpam extends Maintenance {
                                $page->doDeleteArticle( wfMessage( 'spam_deleting', $domain )->inContentLanguage()->text() );
                        } else {
                                // Didn't find a non-spammy revision, blank the page
+                               $handler = ContentHandler::getForTitle( $title );
+                               $content = $handler->makeEmptyContent();
+
                                $this->output( "blanking\n" );
-                               $page->doEdit( '', wfMessage( 'spam_blanking', $domain )->inContentLanguage()->text() );
+                               $page->doEditContent( $content, wfMessage( 'spam_blanking', $domain )->inContentLanguage()->text() );
                        }
                        $dbw->commit( __METHOD__ );
                }