Extension:Urllogger

<?php
/**
Script to log added URLs to a text file, based loosly on Brion's early captcha work
Aerik Sylvan, January 2006
 */
 
if ( defined( 'MEDIAWIKI' ) ) {
 
	$wgHooks['ArticleSave'][] = "logurls";
}
 
 
	// ----------------------------------
 
	/**
	 * @param EditPage $editPage
	 * @param string $newtext
	 * @param string $section
	 * @return bool true if the captcha should run
	 */
	function logurls($article, $user, $newtext) {
	$logfile = "extensions/urllog.txt";
	print_r($ary);
		$oldtext = $article->getContent( TRUE );
		$oldLinks = findLinks( $oldtext );
		$newLinks = findLinks( $newtext );
		//$unknownLinks = array_filter( $newLinks, array( &$this, 'filterLink' ) 
 
		//$addedLinks = array_diff( $unknownLinks, $oldLinks );
		$addedLinks = array_diff( $newLinks, $oldLinks );
		print_r($addedLinks);
		$numLinks = count( $addedLinks );
 
		if( $numLinks > 0 ) {
			global $wgUser;
			$name = $user->getName();
			foreach($addedLinks as $link){
				$logentry .= $link."\t".date("U")."\t".$name."\n";
			}
			$handle = fopen ($logfile, "a");
            fwrite($handle,$logentry);
            fclose($handle);
		}
 
		return true;
	}
 
	function findLinks( $text ) {
		$regex = '/((?:' . HTTP_PROTOCOLS . ')' . EXT_LINK_URL_CLASS . '+)/';
		//print "***".$regex."***\n";
 
		if( preg_match_all( $regex, $text, $matches, PREG_PATTERN_ORDER ) ) {
			return $matches[0];
		} else {
			return array();
		}
	}
 
	function loadText( $editPage, $section ) {
		$rev = Revision::newFromTitle( $editPage->mTitle );
		if( is_null( $rev ) ) {
			return "";
		} else {
			$text = $rev->getText();
			if( $section != '' ) {
				return Article::getSection( $text, $section );
			} else {
				return $text;
			}
		}
	}