r39467 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r39466‎ \| r39467 \| r39468 >
Date:	10:13, 16 August 2008
Author:	krimpet
Status:	old
Tags:
Comment:	Fix for problems with r39414; LinkHolderArray::replaceInterwiki() was badly broken
Modified paths:	/trunk/phase3/RELEASE-NOTES (modified) (history) /trunk/phase3/includes/AutoLoader.php (modified) (history) /trunk/phase3/includes/LinkCache.php (modified) (history) /trunk/phase3/includes/MessageCache.php (modified) (history) /trunk/phase3/includes/StringUtils.php (modified) (history) /trunk/phase3/includes/Title.php (modified) (history) /trunk/phase3/includes/parser/LinkHolderArray.php (added) (history) /trunk/phase3/includes/parser/LinkHolderArray.php (added) (history) /trunk/phase3/includes/parser/Parser.php (modified) (history)

Diff [purge]

Index: trunk/phase3/includes/parser/LinkHolderArray.php
—	—	@@ -0,0 +1,406 @@
	2	+<?php
	3	+
	4	+class LinkHolderArray {
	5	+ var $batchSize = 1000;
	6	+
	7	+ var $internals = array(), $interwikis = array();
	8	+ var $size = 0;
	9	+ var $parent;
	10	+
	11	+ function __construct( $parent ) {
	12	+ $this->parent = $parent;
	13	+ }
	14	+
	15	+ /**
	16	+ * Merge another LinkHolderArray into this one
	17	+ */
	18	+ function merge( $other ) {
	19	+ foreach ( $other->internals as $ns => $entries ) {
	20	+ $this->size += count( $entries );
	21	+ if ( !isset( $this->internals[$ns] ) ) {
	22	+ $this->internals[$ns] = $entries;
	23	+ } else {
	24	+ $this->internals[$ns] += $entries;
	25	+ }
	26	+ }
	27	+ $this->interwikis += $other->interwikis;
	28	+ }
	29	+
	30	+ /**
	31	+ * Returns true if the memory requirements of this object are getting large
	32	+ */
	33	+ function isBig() {
	34	+ return $this->size > $this->batchSize;
	35	+ }
	36	+
	37	+ /**
	38	+ * Clear all stored link holders.
	39	+ * Make sure you don't have any text left using these link holders, before you call this
	40	+ */
	41	+ function clear() {
	42	+ $this->internals = array();
	43	+ $this->interwikis = array();
	44	+ $this->size = 0;
	45	+ }
	46	+
	47	+ /**
	48	+ * Make a link placeholder. The text returned can be later resolved to a real link with
	49	+ * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
	50	+ * parsing of interwiki links, and secondly to allow all existence checks and
	51	+ * article length checks (for stub links) to be bundled into a single query.
	52	+ *
	53	+ */
	54	+ function makeHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
	55	+ wfProfileIn( __METHOD__ );
	56	+ if ( ! is_object($nt) ) {
	57	+ # Fail gracefully
	58	+ $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
	59	+ } else {
	60	+ # Separate the link trail from the rest of the link
	61	+ list( $inside, $trail ) = Linker::splitTrail( $trail );
	62	+
	63	+ $entry = array(
	64	+ 'title' => $nt,
	65	+ 'text' => $prefix.$text.$inside,
	66	+ 'pdbk' => $nt->getPrefixedDBkey(),
	67	+ );
	68	+ if ( $query !== '' ) {
	69	+ $entry['query'] = $query;
	70	+ }
	71	+
	72	+ if ( $nt->isExternal() ) {
	73	+ // Use a globally unique ID to keep the objects mergable
	74	+ $key = $this->parent->nextLinkID();
	75	+ $this->interwikis[$key] = $entry;
	76	+ $retVal = "<!--IWLINK $key-->{$trail}";
	77	+ } else {
	78	+ $key = $this->parent->nextLinkID();
	79	+ $ns = $nt->getNamespace();
	80	+ $this->internals[$ns][$key] = $entry;
	81	+ $retVal = "<!--LINK $ns:$key-->{$trail}";
	82	+ }
	83	+ $this->size++;
	84	+ }
	85	+ wfProfileOut( __METHOD__ );
	86	+ return $retVal;
	87	+ }
	88	+
	89	+ /**
	90	+ * Replace <!--LINK--> link placeholders with actual links, in the buffer
	91	+ * Placeholders created in Skin::makeLinkObj()
	92	+ * Returns an array of link CSS classes, indexed by PDBK.
	93	+ */
	94	+ function replace( &$text ) {
	95	+ wfProfileIn( __METHOD__ );
	96	+
	97	+ $colours = $this->replaceInternal( $text );
	98	+ $this->replaceInterwiki( $text );
	99	+
	100	+ wfProfileOut( __METHOD__ );
	101	+ return $colours;
	102	+ }
	103	+
	104	+ /**
	105	+ * Replace internal links
	106	+ */
	107	+ protected function replaceInternal( &$text ) {
	108	+ if ( !$this->internals ) {
	109	+ return;
	110	+ }
	111	+
	112	+ wfProfileIn( __METHOD__ );
	113	+ global $wgUser, $wgContLang;
	114	+
	115	+ $pdbks = array();
	116	+ $colours = array();
	117	+ $linkcolour_ids = array();
	118	+ $sk = $this->parent->getOptions()->getSkin();
	119	+ $linkCache = LinkCache::singleton();
	120	+ $output = $this->parent->getOutput();
	121	+
	122	+ wfProfileIn( __METHOD__.'-check' );
	123	+ $dbr = wfGetDB( DB_SLAVE );
	124	+ $page = $dbr->tableName( 'page' );
	125	+ $threshold = $wgUser->getOption('stubthreshold');
	126	+
	127	+ # Sort by namespace
	128	+ ksort( $this->internals );
	129	+
	130	+ # Generate query
	131	+ $query = false;
	132	+ $current = null;
	133	+ foreach ( $this->internals as $ns => $entries ) {
	134	+ foreach ( $entries as $index => $entry ) {
	135	+ $key = "$ns:$index";
	136	+ $title = $entry['title'];
	137	+ $pdbk = $entry['pdbk'];
	138	+
	139	+ # Skip invalid entries.
	140	+ # Result will be ugly, but prevents crash.
	141	+ if ( is_null( $title ) ) {
	142	+ continue;
	143	+ }
	144	+
	145	+ # Check if it's a static known link, e.g. interwiki
	146	+ if ( $title->isAlwaysKnown() ) {
	147	+ $colours[$pdbk] = '';
	148	+ } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
	149	+ $colours[$pdbk] = '';
	150	+ $output->addLink( $title, $id );
	151	+ } elseif ( $linkCache->isBadLink( $pdbk ) ) {
	152	+ $colours[$pdbk] = 'new';
	153	+ } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
	154	+ $colours[$pdbk] = 'new';
	155	+ } else {
	156	+ # Not in the link cache, add it to the query
	157	+ if ( !isset( $current ) ) {
	158	+ $current = $ns;
	159	+ $query = "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
	160	+ $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
	161	+ } elseif ( $current != $ns ) {
	162	+ $current = $ns;
	163	+ $query .= ")) OR (page_namespace=$ns AND page_title IN(";
	164	+ } else {
	165	+ $query .= ', ';
	166	+ }
	167	+
	168	+ $query .= $dbr->addQuotes( $title->getDBkey() );
	169	+ }
	170	+ }
	171	+ }
	172	+ if ( $query ) {
	173	+ $query .= '))';
	174	+
	175	+ $res = $dbr->query( $query, __METHOD__ );
	176	+
	177	+ # Fetch data and form into an associative array
	178	+ # non-existent = broken
	179	+ while ( $s = $dbr->fetchObject($res) ) {
	180	+ $title = Title::makeTitle( $s->page_namespace, $s->page_title );
	181	+ $pdbk = $title->getPrefixedDBkey();
	182	+ $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
	183	+ $output->addLink( $title, $s->page_id );
	184	+ $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
	185	+ //add id to the extension todolist
	186	+ $linkcolour_ids[$s->page_id] = $pdbk;
	187	+ }
	188	+ unset( $res );
	189	+ //pass an array of page_ids to an extension
	190	+ wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
	191	+ }
	192	+ wfProfileOut( __METHOD__.'-check' );
	193	+
	194	+ # Do a second query for different language variants of links and categories
	195	+ if($wgContLang->hasVariants()){
	196	+ $linkBatch = new LinkBatch();
	197	+ $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
	198	+ $categoryMap = array(); // maps $category_variant => $category (dbkeys)
	199	+ $varCategories = array(); // category replacements oldDBkey => newDBkey
	200	+
	201	+ $categories = $output->getCategoryLinks();
	202	+
	203	+ // Add variants of links to link batch
	204	+ foreach ( $this->internals as $ns => $entries ) {
	205	+ foreach ( $entries as $index => $entry ) {
	206	+ $key = "$ns:$index";
	207	+ $pdbk = $entry['pdbk'];
	208	+ $title = $entry['title'];
	209	+ $titleText = $title->getText();
	210	+
	211	+ // generate all variants of the link title text
	212	+ $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
	213	+
	214	+ // if link was not found (in first query), add all variants to query
	215	+ if ( !isset($colours[$pdbk]) ){
	216	+ foreach($allTextVariants as $textVariant){
	217	+ if($textVariant != $titleText){
	218	+ $variantTitle = Title::makeTitle( $ns, $textVariant );
	219	+ if(is_null($variantTitle)) continue;
	220	+ $linkBatch->addObj( $variantTitle );
	221	+ $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
	222	+ }
	223	+ }
	224	+ }
	225	+ }
	226	+ }
	227	+
	228	+ // process categories, check if a category exists in some variant
	229	+ foreach( $categories as $category ){
	230	+ $variants = $wgContLang->convertLinkToAllVariants($category);
	231	+ foreach($variants as $variant){
	232	+ if($variant != $category){
	233	+ $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
	234	+ if(is_null($variantTitle)) continue;
	235	+ $linkBatch->addObj( $variantTitle );
	236	+ $categoryMap[$variant] = $category;
	237	+ }
	238	+ }
	239	+ }
	240	+
	241	+
	242	+ if(!$linkBatch->isEmpty()){
	243	+ // construct query
	244	+ $titleClause = $linkBatch->constructSet('page', $dbr);
	245	+
	246	+ $variantQuery = "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
	247	+
	248	+ $variantQuery .= " FROM $page WHERE $titleClause";
	249	+
	250	+ $varRes = $dbr->query( $variantQuery, __METHOD__ );
	251	+
	252	+ // for each found variants, figure out link holders and replace
	253	+ while ( $s = $dbr->fetchObject($varRes) ) {
	254	+
	255	+ $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
	256	+ $varPdbk = $variantTitle->getPrefixedDBkey();
	257	+ $vardbk = $variantTitle->getDBkey();
	258	+
	259	+ $holderKeys = array();
	260	+ if(isset($variantMap[$varPdbk])){
	261	+ $holderKeys = $variantMap[$varPdbk];
	262	+ $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
	263	+ $output->addLink( $variantTitle, $s->page_id );
	264	+ }
	265	+
	266	+ // loop over link holders
	267	+ foreach($holderKeys as $key){
	268	+ list( $ns, $index ) = explode( ':', $key, 2 );
	269	+ $entry =& $this->internals[$ns][$index];
	270	+ $pdbk = $entry['pdbk'];
	271	+
	272	+ if(!isset($colours[$pdbk])){
	273	+ // found link in some of the variants, replace the link holder data
	274	+ $entry['title'] = $variantTitle;
	275	+ $entry['pdbk'] = $varPdbk;
	276	+
	277	+ // set pdbk and colour
	278	+ $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
	279	+ $linkcolour_ids[$s->page_id] = $pdbk;
	280	+ }
	281	+ wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
	282	+ }
	283	+
	284	+ // check if the object is a variant of a category
	285	+ if(isset($categoryMap[$vardbk])){
	286	+ $oldkey = $categoryMap[$vardbk];
	287	+ if($oldkey != $vardbk)
	288	+ $varCategories[$oldkey]=$vardbk;
	289	+ }
	290	+ }
	291	+
	292	+ // rebuild the categories in original order (if there are replacements)
	293	+ if(count($varCategories)>0){
	294	+ $newCats = array();
	295	+ $originalCats = $output->getCategories();
	296	+ foreach($originalCats as $cat => $sortkey){
	297	+ // make the replacement
	298	+ if( array_key_exists($cat,$varCategories) )
	299	+ $newCats[$varCategories[$cat]] = $sortkey;
	300	+ else $newCats[$cat] = $sortkey;
	301	+ }
	302	+ $this->mOutput->parent->setCategoryLinks($newCats);
	303	+ }
	304	+ }
	305	+ }
	306	+
	307	+ # Construct search and replace arrays
	308	+ wfProfileIn( __METHOD__.'-construct' );
	309	+ $replacePairs = array();
	310	+ foreach ( $this->internals as $ns => $entries ) {
	311	+ foreach ( $entries as $index => $entry ) {
	312	+ $pdbk = $entry['pdbk'];
	313	+ $title = $entry['title'];
	314	+ $query = isset( $entry['query'] ) ? $entry['query'] : '';
	315	+ $key = "$ns:$index";
	316	+ $searchkey = "<!--LINK $key-->";
	317	+ if ( !isset( $colours[$pdbk] ) \|\| $colours[$pdbk] == 'new' ) {
	318	+ $linkCache->addBadLinkObj( $title );
	319	+ $colours[$pdbk] = 'new';
	320	+ $output->addLink( $title, 0 );
	321	+ $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
	322	+ $entry['text'],
	323	+ $query );
	324	+ } else {
	325	+ $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
	326	+ $entry['text'],
	327	+ $query );
	328	+ }
	329	+ }
	330	+ }
	331	+ $replacer = new HashtableReplacer( $replacePairs, 1 );
	332	+ wfProfileOut( __METHOD__.'-construct' );
	333	+
	334	+ # Do the thing
	335	+ wfProfileIn( __METHOD__.'-replace' );
	336	+ $text = preg_replace_callback(
	337	+ '/(<!--LINK .*?-->)/',
	338	+ $replacer->cb(),
	339	+ $text);
	340	+
	341	+ wfProfileOut( __METHOD__.'-replace' );
	342	+ wfProfileOut( __METHOD__ );
	343	+ }
	344	+
	345	+ /**
	346	+ * Replace interwiki links
	347	+ */
	348	+ protected function replaceInterwiki( &$text ) {
	349	+ if ( empty( $this->interwikis ) ) {
	350	+ return;
	351	+ }
	352	+
	353	+ wfProfileIn( __METHOD__ );
	354	+ # Make interwiki link HTML
	355	+ $sk = $this->parent->getOptions()->getSkin();
	356	+ $replacePairs = array();
	357	+ foreach( $this->interwikis as $key => $link ) {
	358	+ $replacePairs[$key] = $sk->link( $link['title'], $link['text'] );
	359	+ }
	360	+ $replacer = new HashtableReplacer( $replacePairs, 1 );
	361	+
	362	+ $text = preg_replace_callback(
	363	+ '/<!--IWLINK (.*?)-->/',
	364	+ $replacer->cb(),
	365	+ $text );
	366	+ wfProfileOut( __METHOD__ );
	367	+ }
	368	+
	369	+ /**
	370	+ * Replace <!--LINK--> link placeholders with plain text of links
	371	+ * (not HTML-formatted).
	372	+ * @param string $text
	373	+ * @return string
	374	+ */
	375	+ function replaceText( $text ) {
	376	+ wfProfileIn( __METHOD__ );
	377	+
	378	+ $text = preg_replace_callback(
	379	+ '/<!--(LINK\|IWLINK) (.*?)-->/',
	380	+ array( &$this, 'replaceTextCallback' ),
	381	+ $text );
	382	+
	383	+ wfProfileOut( __METHOD__ );
	384	+ return $text;
	385	+ }
	386	+
	387	+ /**
	388	+ * @param array $matches
	389	+ * @return string
	390	+ * @private
	391	+ */
	392	+ function replaceTextCallback( $matches ) {
	393	+ $type = $matches[1];
	394	+ $key = $matches[2];
	395	+ if( $type == 'LINK' ) {
	396	+ list( $ns, $index ) = explode( ':', $key, 2 );
	397	+ if( isset( $this->internals[$ns][$index]['text'] ) ) {
	398	+ return $this->internals[$ns][$index]['text'];
	399	+ }
	400	+ } elseif( $type == 'IWLINK' ) {
	401	+ if( isset( $this->interwikis[$key]['text'] ) ) {
	402	+ return $this->interwikis[$key]['text'];
	403	+ }
	404	+ }
	405	+ return $matches[0];
	406	+ }
	407	+}
Property changes on: trunk/phase3/includes/parser/LinkHolderArray.php
___________________________________________________________________
Added: svn:eol-style
1	408	+ native
Index: trunk/phase3/includes/parser/Parser.php
—	—	@@ -98,7 +98,7 @@
99	99	# Cleared with clearState():
100	100	var $mOutput, $mAutonumber, $mDTopen, $mStripState;
101	101	var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
102		~~- var $mInterwikiLinkHolders, $mLinkHolders;~~
	102	+ var $mLinkHolders, $mLinkID;
103	103	var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
104	104	var $mTplExpandCache; // empty-frame expansion cache
105	105	var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
—	—	@@ -179,17 +179,8 @@
180	180	$this->mStripState = new StripState;
181	181	$this->mArgStack = false;
182	182	$this->mInPre = false;
183		~~- $this->mInterwikiLinkHolders = array(~~
184		~~- 'texts' => array(),~~
185		~~- 'titles' => array()~~
186		~~- );~~
187		~~- $this->mLinkHolders = array(~~
188		~~- 'namespaces' => array(),~~
189		~~- 'dbkeys' => array(),~~
190		~~- 'queries' => array(),~~
191		~~- 'texts' => array(),~~
192		~~- 'titles' => array()~~
193		~~- );~~
	183	+ $this->mLinkHolders = new LinkHolderArray( $this );
	184	+ $this->mLinkID = 0;
194	185	$this->mRevisionTimestamp = $this->mRevisionId = null;
195	186
196	187	/**
—	—	@@ -204,7 +195,7 @@
205	196	*/
206	197	#$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
207	198	# Changed to \x7f to allow XML double-parsing -- TS
208		~~- $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();~~
	199	+ $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
209	200
210	201
211	202	# Clear these on every parse, bug 4549
—	—	@@ -294,7 +285,7 @@
295	286	*/
296	287
297	288	global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
298		~~- $fname = 'Parser::parse-' . wfGetCaller();~~
	289	+ $fname = __METHOD__.'-' . wfGetCaller();
299	290	wfProfileIn( __METHOD__ );
300	291	wfProfileIn( $fname );
301	292
—	—	@@ -328,7 +319,6 @@
329	320	);
330	321	$text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
331	322
332		~~- # only once and last~~
333	323	$text = $this->doBlockLevels( $text, $linestart );
334	324
335	325	$this->replaceLinkHolders( $text );
—	—	@@ -348,7 +338,7 @@
349	339	$uniq_prefix = $this->mUniqPrefix;
350	340	$matches = array();
351	341	$elements = array_keys( $this->mTransparentTagHooks );
352		~~- $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );~~
	342	+ $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
353	343
354	344	foreach( $matches as $marker => $data ) {
355	345	list( $element, $content, $params, $tag ) = $data;
—	—	@@ -366,7 +356,7 @@
367	357	$text = Sanitizer::normalizeCharReferences( $text );
368	358
369	359	if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
370		~~- $text = Parser::tidy($text);~~
	360	+ $text = self::tidy($text);
371	361	} else {
372	362	# attempt to sanitize at least some nesting problems
373	363	# (bug #2702 and quite a few others)
—	—	@@ -471,6 +461,8 @@
472	462	function &getTitle() { return $this->mTitle; }
473	463	function getOptions() { return $this->mOptions; }
474	464	function getRevisionId() { return $this->mRevisionId; }
	465	+ function getOutput() { return $this->mOutput; }
	466	+ function nextLinkID() { return $this->mLinkID++; }
475	467
476	468	function getFunctionLang() {
477	469	global $wgLang, $wgContLang;
—	—	@@ -658,9 +650,9 @@
659	651	' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
660	652	'<head><title>test</title></head><body>'.$text.'</body></html>';
661	653	if( $wgTidyInternal ) {
662		~~- $correctedtext = Parser::internalTidy( $wrappedtext );~~
	654	+ $correctedtext = self::internalTidy( $wrappedtext );
663	655	} else {
664		~~- $correctedtext = Parser::externalTidy( $wrappedtext );~~
	656	+ $correctedtext = self::externalTidy( $wrappedtext );
665	657	}
666	658	if( is_null( $correctedtext ) ) {
667	659	wfDebug( "Tidy error detected!\n" );
—	—	@@ -677,8 +669,7 @@
678	670	*/
679	671	function externalTidy( $text ) {
680	672	global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
681		~~- $fname = 'Parser::externalTidy';~~
682		~~- wfProfileIn( $fname );~~
	673	+ wfProfileIn( __METHOD__ );
683	674
684	675	$cleansource = '';
685	676	$opts = ' -utf8';
—	—	@@ -707,7 +698,7 @@
708	699	}
709	700	}
710	701
711		~~- wfProfileOut( $fname );~~
	702	+ wfProfileOut( __METHOD__ );
712	703
713	704	if( $cleansource == '' && $text != '') {
714	705	// Some kind of error happened, so we couldn't get the corrected text.
—	—	@@ -729,8 +720,7 @@
730	721	*/
731	722	function internalTidy( $text ) {
732	723	global $wgTidyConf, $IP, $wgDebugTidy;
733		~~- $fname = 'Parser::internalTidy';~~
734		~~- wfProfileIn( $fname );~~
	724	+ wfProfileIn( __METHOD__ );
735	725
736	726	$tidy = new tidy;
737	727	$tidy->parseString( $text, $wgTidyConf, 'utf8' );
—	—	@@ -748,7 +738,7 @@
749	739	"\n-->";
750	740	}
751	741
752		~~- wfProfileOut( $fname );~~
	742	+ wfProfileOut( __METHOD__ );
753	743	return $cleansource;
754	744	}
755	745
—	—	@@ -758,34 +748,35 @@
759	749	* @private
760	750	*/
761	751	function doTableStuff ( $text ) {
762		~~- $fname = 'Parser::doTableStuff';~~
763		~~- wfProfileIn( $fname );~~
	752	+ wfProfileIn( __METHOD__ );
764	753
765		~~- $lines = explode ( "\n" , $text );~~
	754	+ $lines = StringUtils::explode( "\n", $text );
	755	+ $out = '';
766	756	$td_history = array (); // Is currently a td tag open?
767	757	$last_tag_history = array (); // Save history of last lag activated (td, th or caption)
768	758	$tr_history = array (); // Is currently a tr tag open?
769	759	$tr_attributes = array (); // history of tr attributes
770	760	$has_opened_tr = array(); // Did this table open a <tr> element?
771	761	$indent_level = 0; // indent level of the table
772		~~- foreach ( $lines as $key => $line )~~
773		~~- {~~
774		~~- $line = trim ( $line );~~
775	762
	763	+ foreach ( $lines as $outLine ) {
	764	+ $line = trim( $outLine );
	765	+
776	766	if( $line == '' ) { // empty line, go to next line
	767	+ $out .= "\n";
777	768	continue;
778	769	}
779		~~- $first_character = $line{0};~~
	770	+ $first_character = $line[0];
780	771	$matches = array();
781	772
782		~~- if ( preg_match( '/^(:)\{\\|(.)$/' , $line , $matches ) ) {~~
	773	+ if ( preg_match( '/^(:)\{\\|(.)$/', $line , $matches ) ) {
783	774	// First check if we are starting a new table
784	775	$indent_level = strlen( $matches[1] );
785	776
786	777	$attributes = $this->mStripState->unstripBoth( $matches[2] );
787	778	$attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
788	779
789		~~- $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";~~
	780	+ $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
790	781	array_push ( $td_history , false );
791	782	array_push ( $last_tag_history , '' );
792	783	array_push ( $tr_history , false );
—	—	@@ -793,6 +784,7 @@
794	785	array_push ( $has_opened_tr , false );
795	786	} else if ( count ( $td_history ) == 0 ) {
796	787	// Don't do any of the following
	788	+ $out .= $outLine."\n";
797	789	continue;
798	790	} else if ( substr ( $line , 0 , 2 ) == '\|}' ) {
799	791	// We are ending a table
—	—	@@ -811,7 +803,7 @@
812	804	$line = "</{$last_tag}>{$line}";
813	805	}
814	806	array_pop ( $tr_attributes );
815		~~- $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );~~
	807	+ $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
816	808	} else if ( substr ( $line , 0 , 2 ) == '\|-' ) {
817	809	// Now we have a table row
818	810	$line = preg_replace( '#^\\|-+#', '', $line );
—	—	@@ -835,7 +827,7 @@
836	828	$line = "</{$last_tag}>{$line}";
837	829	}
838	830
839		~~- $lines[$key] = $line;~~
	831	+ $outLine = $line;
840	832	array_push ( $tr_history , false );
841	833	array_push ( $td_history , false );
842	834	array_push ( $last_tag_history , '' );
—	—	@@ -859,7 +851,7 @@
860	852	// attribute values containing literal "\|\|".
861	853	$cells = StringUtils::explodeMarkup( '\|\|' , $line );
862	854
863		~~- $lines[$key] = '';~~
	855	+ $outLine = '';
864	856
865	857	// Loop through each table cell
866	858	foreach ( $cells as $cell )
—	—	@@ -910,38 +902,42 @@
911	903	$cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
912	904	}
913	905
914		~~- $lines[$key] .= $cell;~~
	906	+ $outLine .= $cell;
915	907	array_push ( $td_history , true );
916	908	}
917	909	}
	910	+ $out .= $outLine . "\n";
918	911	}
919	912
920	913	// Closing open td, tr && table
921	914	while ( count ( $td_history ) > 0 )
922	915	{
923	916	if ( array_pop ( $td_history ) ) {
924		~~- $lines[] = '</td>' ;~~
	917	+ $out .= "</td>\n";
925	918	}
926	919	if ( array_pop ( $tr_history ) ) {
927		~~- $lines[] = '</tr>' ;~~
	920	+ $out .= "</tr>\n";
928	921	}
929	922	if ( !array_pop ( $has_opened_tr ) ) {
930		~~- $lines[] = "<tr><td></td></tr>" ;~~
	923	+ $out .= "<tr><td></td></tr>\n" ;
931	924	}
932	925
933		~~- $lines[] = '</table>' ;~~
	926	+ $out .= "</table>\n";
934	927	}
935	928
936		~~- $output = implode ( "\n" , $lines ) ;~~
	929	+ // Remove trailing line-ending (b/c)
	930	+ if ( substr( $out, -1 ) == "\n" ) {
	931	+ $out = substr( $out, 0, -1 );
	932	+ }
937	933
938	934	// special case: don't return empty table
939		~~- if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {~~
940		~~- $output = '';~~
	935	+ if( $out == "<table>\n<tr><td></td></tr>\n</table>" ) {
	936	+ $out = '';
941	937	}
942	938
943		~~- wfProfileOut( $fname );~~
	939	+ wfProfileOut( __METHOD__ );
944	940
945		~~- return $output;~~
	941	+ return $out;
946	942	}
947	943
948	944	/**
—	—	@@ -952,12 +948,11 @@
953	949	*/
954	950	function internalParse( $text ) {
955	951	$isMain = true;
956		~~- $fname = 'Parser::internalParse';~~
957		~~- wfProfileIn( $fname );~~
	952	+ wfProfileIn( __METHOD__ );
958	953
959	954	# Hook to suspend the parser in this state
960	955	if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
961		~~- wfProfileOut( $fname );~~
	956	+ wfProfileOut( __METHOD__ );
962	957	return $text ;
963	958	}
964	959
—	—	@@ -990,7 +985,7 @@
991	986	$text = $this->doMagicLinks( $text );
992	987	$text = $this->formatHeadings( $text, $isMain );
993	988
994		~~- wfProfileOut( $fname );~~
	989	+ wfProfileOut( __METHOD__ );
995	990	return $text;
996	991	}
997	992
—	—	@@ -1060,14 +1055,13 @@
1061	1056	* @private
1062	1057	*/
1063	1058	function doHeadings( $text ) {
1064		~~- $fname = 'Parser::doHeadings';~~
1065		~~- wfProfileIn( $fname );~~
	1059	+ wfProfileIn( __METHOD__ );
1066	1060	for ( $i = 6; $i >= 1; --$i ) {
1067	1061	$h = str_repeat( '=', $i );
1068	1062	$text = preg_replace( "/^$h(.+)$h\\s*$/m",
1069	1063	"<h$i>\\1</h$i>", $text );
1070	1064	}
1071		~~- wfProfileOut( $fname );~~
	1065	+ wfProfileOut( __METHOD__ );
1072	1066	return $text;
1073	1067	}
1074	1068
—	—	@@ -1077,15 +1071,14 @@
1078	1072	* @return string the altered text
1079	1073	*/
1080	1074	function doAllQuotes( $text ) {
1081		~~- $fname = 'Parser::doAllQuotes';~~
1082		~~- wfProfileIn( $fname );~~
	1075	+ wfProfileIn( __METHOD__ );
1083	1076	$outtext = '';
1084		~~- $lines = explode( "\n", $text );~~
	1077	+ $lines = StringUtils::explode( "\n", $text );
1085	1078	foreach ( $lines as $line ) {
1086		~~- $outtext .= $this->doQuotes ( $line ) . "\n";~~
	1079	+ $outtext .= $this->doQuotes( $line ) . "\n";
1087	1080	}
1088	1081	$outtext = substr($outtext, 0,-1);
1089		~~- wfProfileOut( $fname );~~
	1082	+ wfProfileOut( __METHOD__ );
1090	1083	return $outtext;
1091	1084	}
1092	1085
—	—	@@ -1264,8 +1257,7 @@
1265	1258	*/
1266	1259	function replaceExternalLinks( $text ) {
1267	1260	global $wgContLang;
1268		~~- $fname = 'Parser::replaceExternalLinks';~~
1269		~~- wfProfileIn( $fname );~~
	1261	+ wfProfileIn( __METHOD__ );
1270	1262
1271	1263	$sk = $this->mOptions->getSkin();
1272	1264
—	—	@@ -1335,11 +1327,11 @@
1336	1328	# Register link in the output object.
1337	1329	# Replace unnecessary URL escape codes with the referenced character
1338	1330	# This prevents spammers from hiding links from the filters
1339		~~- $pasteurized = Parser::replaceUnusualEscapes( $url );~~
	1331	+ $pasteurized = self::replaceUnusualEscapes( $url );
1340	1332	$this->mOutput->addExternalLink( $pasteurized );
1341	1333	}
1342	1334
1343		~~- wfProfileOut( $fname );~~
	1335	+ wfProfileOut( __METHOD__ );
1344	1336	return $s;
1345	1337	}
1346	1338
—	—	@@ -1349,8 +1341,7 @@
1350	1342	*/
1351	1343	function replaceFreeExternalLinks( $text ) {
1352	1344	global $wgContLang;
1353		~~- $fname = 'Parser::replaceFreeExternalLinks';~~
1354		~~- wfProfileIn( $fname );~~
	1345	+ wfProfileIn( __METHOD__ );
1355	1346
1356	1347	$bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1357	1348	$s = array_shift( $bits );
—	—	@@ -1412,7 +1403,7 @@
1413	1404	$text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1414	1405	# Register it in the output object...
1415	1406	# Replace unnecessary URL escape codes with their equivalent characters
1416		~~- $pasteurized = Parser::replaceUnusualEscapes( $url );~~
	1407	+ $pasteurized = self::replaceUnusualEscapes( $url );
1417	1408	$this->mOutput->addExternalLink( $pasteurized );
1418	1409	}
1419	1410	$s .= $text . $trail;
—	—	@@ -1420,7 +1411,7 @@
1421	1412	$s .= $protocol . $remainder;
1422	1413	}
1423	1414	}
1424		~~- wfProfileOut( $fname );~~
	1415	+ wfProfileOut( __METHOD__ );
1425	1416	return $s;
1426	1417	}
1427	1418
—	—	@@ -1436,7 +1427,7 @@
1437	1428	*/
1438	1429	static function replaceUnusualEscapes( $url ) {
1439	1430	return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1440		~~- array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );~~
	1431	+ array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1441	1432	}
1442	1433
1443	1434	/**
—	—	@@ -1480,35 +1471,48 @@
1481	1472
1482	1473	/**
1483	1474	* Process [[ ]] wikilinks
	1475	+ * @return processed text
1484	1476	*
1485	1477	* @private
1486	1478	*/
1487	1479	function replaceInternalLinks( $s ) {
	1480	+ $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
	1481	+ return $s;
	1482	+ }
	1483	+
	1484	+ /**
	1485	+ * Process [[ ]] wikilinks
	1486	+ * @return LinkHolderArray
	1487	+ *
	1488	+ * @private
	1489	+ */
	1490	+ function replaceInternalLinks2( &$s ) {
1488	1491	global $wgContLang;
1489		~~- static $fname = 'Parser::replaceInternalLinks' ;~~
1490	1492
1491		~~- wfProfileIn( $fname );~~
	1493	+ wfProfileIn( __METHOD__ );
1492	1494
1493		~~- wfProfileIn( $fname.'-setup' );~~
1494		~~- static $tc = FALSE;~~
	1495	+ wfProfileIn( __METHOD__.'-setup' );
	1496	+ static $tc = FALSE, $e1, $e1_img;
1495	1497	# the % is needed to support urlencoded titles as well
1496		~~- if ( !$tc ) { $tc = Title::legalChars() . '#%'; }~~
	1498	+ if ( !$tc ) {
	1499	+ $tc = Title::legalChars() . '#%';
	1500	+ # Match a link having the form [[namespace:link\|alternate]]trail
	1501	+ $e1 = "/^([{$tc}]+)(?:\\\|(.+?))?]](.*)\$/sD";
	1502	+ # Match cases where there is no "]]", which might still be images
	1503	+ $e1_img = "/^([{$tc}]+)\\\|(.*)\$/sD";
	1504	+ }
1497	1505
1498	1506	$sk = $this->mOptions->getSkin();
	1507	+ $holders = new LinkHolderArray( $this );
1499	1508
1500	1509	#split the entire text string on occurences of [[
1501		~~- $a = explode( '[[', ' ' . $s );~~
	1510	+ $a = StringUtils::explode( '[[', ' ' . $s );
1502	1511	#get the first element (all text up to first [[), and remove the space we added
1503		~~- $s = array_shift( $a );~~
	1512	+ $s = $a->current();
	1513	+ $a->next();
	1514	+ $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1504	1515	$s = substr( $s, 1 );
1505	1516
1506		~~- # Match a link having the form [[namespace:link\|alternate]]trail~~
1507		~~- static $e1 = FALSE;~~
1508		~~- if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\\|(.+?))?]](.*)\$/sD"; }~~
1509		~~- # Match cases where there is no "]]", which might still be images~~
1510		~~- static $e1_img = FALSE;~~
1511		~~- if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\\|(.*)\$/sD"; }~~
1512		-
1513	1517	$useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1514	1518	$e2 = null;
1515	1519	if ( $useLinkPrefixExtension ) {
—	—	@@ -1518,8 +1522,8 @@
1519	1523	}
1520	1524
1521	1525	if( is_null( $this->mTitle ) ) {
1522		~~- wfProfileOut( $fname );~~
1523		~~- wfProfileOut( $fname.'-setup' );~~
	1526	+ wfProfileOut( __METHOD__ );
	1527	+ wfProfileOut( __METHOD__.'-setup' );
1524	1528	throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1525	1529	}
1526	1530	$nottalk = !$this->mTitle->isTalkPage();
—	—	@@ -1541,13 +1545,20 @@
1542	1546	$selflink = array($this->mTitle->getPrefixedText());
1543	1547	}
1544	1548	$useSubpages = $this->areSubpagesAllowed();
1545		~~- wfProfileOut( $fname.'-setup' );~~
	1549	+ wfProfileOut( __METHOD__.'-setup' );
1546	1550
1547	1551	# Loop for each link
1548		~~- for ($k = 0; isset( $a[$k] ); $k++) {~~
1549		~~- $line = $a[$k];~~
	1552	+ for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
	1553	+ # Check for excessive memory usage
	1554	+ if ( $holders->isBig() ) {
	1555	+ # Too big
	1556	+ # Do the existence check, replace the link holders and clear the array
	1557	+ $holders->replace( $s );
	1558	+ $holders->clear();
	1559	+ }
	1560	+
1550	1561	if ( $useLinkPrefixExtension ) {
1551		~~- wfProfileIn( $fname.'-prefixhandling' );~~
	1562	+ wfProfileIn( __METHOD__.'-prefixhandling' );
1552	1563	if ( preg_match( $e2, $s, $m ) ) {
1553	1564	$prefix = $m[2];
1554	1565	$s = $m[1];
—	—	@@ -1559,12 +1570,12 @@
1560	1571	$prefix = $first_prefix;
1561	1572	$first_prefix = false;
1562	1573	}
1563		~~- wfProfileOut( $fname.'-prefixhandling' );~~
	1574	+ wfProfileOut( __METHOD__.'-prefixhandling' );
1564	1575	}
1565	1576
1566	1577	$might_be_img = false;
1567	1578
1568		~~- wfProfileIn( "$fname-e1" );~~
	1579	+ wfProfileIn( __METHOD__."-e1" );
1569	1580	if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1570	1581	$text = $m[2];
1571	1582	# If we get a ] at the beginning of $m[3] that means we have a link that's something like:
—	—	@@ -1598,18 +1609,18 @@
1599	1610	$trail = "";
1600	1611	} else { # Invalid form; output directly
1601	1612	$s .= $prefix . '[[' . $line ;
1602		~~- wfProfileOut( "$fname-e1" );~~
	1613	+ wfProfileOut( __METHOD__."-e1" );
1603	1614	continue;
1604	1615	}
1605		~~- wfProfileOut( "$fname-e1" );~~
1606		~~- wfProfileIn( "$fname-misc" );~~
	1616	+ wfProfileOut( __METHOD__."-e1" );
	1617	+ wfProfileIn( __METHOD__."-misc" );
1607	1618
1608	1619	# Don't allow internal links to pages containing
1609	1620	# PROTO: where PROTO is a valid URL protocol; these
1610	1621	# should be external links.
1611	1622	if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1612	1623	$s .= $prefix . '[[' . $line ;
1613		~~- wfProfileOut( "$fname-misc" );~~
	1624	+ wfProfileOut( __METHOD__."-misc" );
1614	1625	continue;
1615	1626	}
1616	1627
—	—	@@ -1626,27 +1637,30 @@
1627	1638	$link = substr($link, 1);
1628	1639	}
1629	1640
1630		~~- wfProfileOut( "$fname-misc" );~~
1631		~~- wfProfileIn( "$fname-title" );~~
	1641	+ wfProfileOut( __METHOD__."-misc" );
	1642	+ wfProfileIn( __METHOD__."-title" );
1632	1643	$nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1633	1644	if( !$nt ) {
1634	1645	$s .= $prefix . '[[' . $line;
1635		~~- wfProfileOut( "$fname-title" );~~
	1646	+ wfProfileOut( __METHOD__."-title" );
1636	1647	continue;
1637	1648	}
1638	1649
1639	1650	$ns = $nt->getNamespace();
1640	1651	$iw = $nt->getInterWiki();
1641		~~- wfProfileOut( "$fname-title" );~~
	1652	+ wfProfileOut( __METHOD__."-title" );
1642	1653
1643	1654	if ($might_be_img) { # if this is actually an invalid link
1644		~~- wfProfileIn( "$fname-might_be_img" );~~
	1655	+ wfProfileIn( __METHOD__."-might_be_img" );
1645	1656	if ($ns == NS_IMAGE && $noforce) { #but might be an image
1646	1657	$found = false;
1647		~~- while (isset ($a[$k+1]) ) {~~
	1658	+ while ( true ) {
1648	1659	#look at the next 'line' to see if we can close it there
1649		~~- $spliced = array_splice( $a, $k + 1, 1 );~~
1650		~~- $next_line = array_shift( $spliced );~~
	1660	+ $a->next();
	1661	+ $next_line = $a->current();
	1662	+ if ( $next_line === false \|\| $next_line === null ) {
	1663	+ break;
	1664	+ }
1651	1665	$m = explode( ']]', $next_line, 3 );
1652	1666	if ( count( $m ) == 3 ) {
1653	1667	# the first ]] closes the inner link, the second the image
—	—	@@ -1666,19 +1680,19 @@
1667	1681	if ( !$found ) {
1668	1682	# we couldn't find the end of this imageLink, so output it raw
1669	1683	#but don't ignore what might be perfectly normal links in the text we've examined
1670		~~- $text = $this->replaceInternalLinks($text);~~
	1684	+ $holders->merge( $this->replaceInternalLinks2( $text ) );
1671	1685	$s .= "{$prefix}[[$link\|$text";
1672	1686	# note: no $trail, because without an end, there is no trail
1673		~~- wfProfileOut( "$fname-might_be_img" );~~
	1687	+ wfProfileOut( __METHOD__."-might_be_img" );
1674	1688	continue;
1675	1689	}
1676	1690	} else { #it's not an image, so output it raw
1677	1691	$s .= "{$prefix}[[$link\|$text";
1678	1692	# note: no $trail, because without an end, there is no trail
1679		~~- wfProfileOut( "$fname-might_be_img" );~~
	1693	+ wfProfileOut( __METHOD__."-might_be_img" );
1680	1694	continue;
1681	1695	}
1682		~~- wfProfileOut( "$fname-might_be_img" );~~
	1696	+ wfProfileOut( __METHOD__."-might_be_img" );
1683	1697	}
1684	1698
1685	1699	$wasblank = ( '' == $text );
—	—	@@ -1688,41 +1702,38 @@
1689	1703	if( $noforce ) {
1690	1704
1691	1705	# Interwikis
1692		~~- wfProfileIn( "$fname-interwiki" );~~
	1706	+ wfProfileIn( __METHOD__."-interwiki" );
1693	1707	if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1694	1708	$this->mOutput->addLanguageLink( $nt->getFullText() );
1695	1709	$s = rtrim($s . $prefix);
1696	1710	$s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1697		~~- wfProfileOut( "$fname-interwiki" );~~
	1711	+ wfProfileOut( __METHOD__."-interwiki" );
1698	1712	continue;
1699	1713	}
1700		~~- wfProfileOut( "$fname-interwiki" );~~
	1714	+ wfProfileOut( __METHOD__."-interwiki" );
1701	1715
1702	1716	if ( $ns == NS_IMAGE ) {
1703		~~- wfProfileIn( "$fname-image" );~~
	1717	+ wfProfileIn( __METHOD__."-image" );
1704	1718	if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1705	1719	# recursively parse links inside the image caption
1706	1720	# actually, this will parse them in any other parameters, too,
1707	1721	# but it might be hard to fix that, and it doesn't matter ATM
1708	1722	$text = $this->replaceExternalLinks($text);
1709		~~- $text = $this->replaceInternalLinks($text);~~
	1723	+ $holders->merge( $this->replaceInternalLinks2( $text ) );
1710	1724
1711	1725	# cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1712		~~- $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;~~
1713		~~- $this->mOutput->addImage( $nt->getDBkey() );~~
	1726	+ $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1714	1727
1715		~~- wfProfileOut( "$fname-image" );~~
	1728	+ wfProfileOut( __METHOD__."-image" );
1716	1729	continue;
1717		~~- } else {~~
1718		~~- # We still need to record the image's presence on the page~~
1719		~~- $this->mOutput->addImage( $nt->getDBkey() );~~
1720	1730	}
1721		~~- wfProfileOut( "$fname-image" );~~
	1731	+ $this->mOutput->addImage( $nt->getDBkey() );
	1732	+ wfProfileOut( __METHOD__."-image" );
1722	1733
1723	1734	}
1724	1735
1725	1736	if ( $ns == NS_CATEGORY ) {
1726		~~- wfProfileIn( "$fname-category" );~~
	1737	+ wfProfileIn( __METHOD__."-category" );
1727	1738	$s = rtrim($s . "\n"); # bug 87
1728	1739
1729	1740	if ( $wasblank ) {
—	—	@@ -1741,7 +1752,7 @@
1742	1753	*/
1743	1754	$s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1744	1755
1745		~~- wfProfileOut( "$fname-category" );~~
	1756	+ wfProfileOut( __METHOD__."-category" );
1746	1757	continue;
1747	1758	}
1748	1759	}
—	—	@@ -1772,7 +1783,7 @@
1773	1784	if( SpecialPage::exists( $nt->getDBkey() ) ) {
1774	1785	$s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1775	1786	} else {
1776		~~- $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );~~
	1787	+ $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1777	1788	}
1778	1789	continue;
1779	1790	} elseif( $ns == NS_IMAGE ) {
—	—	@@ -1786,10 +1797,10 @@
1787	1798	continue;
1788	1799	}
1789	1800	}
1790		~~- $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );~~
	1801	+ $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1791	1802	}
1792		~~- wfProfileOut( $fname );~~
1793		~~- return $s;~~
	1803	+ wfProfileOut( __METHOD__ );
	1804	+ return $holders;
1794	1805	}
1795	1806
1796	1807	/**
—	—	@@ -1798,32 +1809,10 @@
1799	1810	* parsing of interwiki links, and secondly to allow all existence checks and
1800	1811	* article length checks (for stub links) to be bundled into a single query.
1801	1812	*
	1813	+ * @deprecated
1802	1814	*/
1803	1815	function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1804		~~- wfProfileIn( __METHOD__ );~~
1805		~~- if ( ! is_object($nt) ) {~~
1806		~~- # Fail gracefully~~
1807		~~- $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";~~
1808		~~- } else {~~
1809		~~- # Separate the link trail from the rest of the link~~
1810		~~- list( $inside, $trail ) = Linker::splitTrail( $trail );~~
1811		-
1812		~~- if ( $nt->isExternal() ) {~~
1813		~~- $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );~~
1814		~~- $this->mInterwikiLinkHolders['titles'][] = $nt;~~
1815		~~- $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";~~
1816		~~- } else {~~
1817		~~- $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );~~
1818		~~- $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();~~
1819		~~- $this->mLinkHolders['queries'][] = $query;~~
1820		~~- $this->mLinkHolders['texts'][] = $prefix.$text.$inside;~~
1821		~~- $this->mLinkHolders['titles'][] = $nt;~~
1822		-
1823		~~- $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";~~
1824		~~- }~~
1825		~~- }~~
1826		~~- wfProfileOut( __METHOD__ );~~
1827		~~- return $retVal;~~
	1816	+ return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1828	1817	}
1829	1818
1830	1819	/**
—	—	@@ -1889,8 +1878,7 @@
1890	1879	# ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1891	1880	# ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1892	1881
1893		~~- $fname = 'Parser::maybeDoSubpageLink';~~
1894		~~- wfProfileIn( $fname );~~
	1882	+ wfProfileIn( __METHOD__ );
1895	1883	$ret = $target; # default return value is no change
1896	1884
1897	1885	# Some namespaces don't allow subpages,
—	—	@@ -1949,7 +1937,7 @@
1950	1938	}
1951	1939	}
1952	1940
1953		~~- wfProfileOut( $fname );~~
	1941	+ wfProfileOut( __METHOD__ );
1954	1942	return $ret;
1955	1943	}
1956	1944
—	—	@@ -2036,50 +2024,53 @@
2037	2025	* @return string the lists rendered as HTML
2038	2026	*/
2039	2027	function doBlockLevels( $text, $linestart ) {
2040		~~- $fname = 'Parser::doBlockLevels';~~
2041		~~- wfProfileIn( $fname );~~
	2028	+ wfProfileIn( __METHOD__ );
2042	2029
2043	2030	# Parsing through the text line by line. The main thing
2044	2031	# happening here is handling of block-level elements p, pre,
2045	2032	# and making lists from lines starting with * # : etc.
2046	2033	#
2047		~~- $textLines = explode( "\n", $text );~~
	2034	+ $textLines = StringUtils::explode( "\n", $text );
2048	2035
2049	2036	$lastPrefix = $output = '';
2050	2037	$this->mDTopen = $inBlockElem = false;
2051	2038	$prefixLength = 0;
2052	2039	$paragraphStack = false;
2053	2040
2054		~~- if ( !$linestart ) {~~
2055		~~- $output .= array_shift( $textLines );~~
2056		~~- }~~
2057	2041	foreach ( $textLines as $oLine ) {
	2042	+ # Fix up $linestart
	2043	+ if ( !$linestart ) {
	2044	+ $output .= $oLine;
	2045	+ $linestart = true;
	2046	+ continue;
	2047	+ }
	2048	+
2058	2049	$lastPrefixLength = strlen( $lastPrefix );
2059	2050	$preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2060	2051	$preOpenMatch = preg_match('/<pre/i', $oLine );
2061	2052	if ( !$this->mInPre ) {
2062	2053	# Multiple prefixes may abut each other for nested lists.
2063	2054	$prefixLength = strspn( $oLine, '*#:;' );
2064		~~- $pref = substr( $oLine, 0, $prefixLength );~~
	2055	+ $prefix = substr( $oLine, 0, $prefixLength );
2065	2056
2066	2057	# eh?
2067		~~- $pref2 = str_replace( ';', ':', $pref );~~
	2058	+ $prefix2 = str_replace( ';', ':', $prefix );
2068	2059	$t = substr( $oLine, $prefixLength );
2069		~~- $this->mInPre = !empty($preOpenMatch);~~
	2060	+ $this->mInPre = (bool)$preOpenMatch;
2070	2061	} else {
2071	2062	# Don't interpret any other prefixes in preformatted text
2072	2063	$prefixLength = 0;
2073		~~- $pref = $pref2 = '';~~
	2064	+ $prefix = $prefix2 = '';
2074	2065	$t = $oLine;
2075	2066	}
2076	2067
2077	2068	# List generation
2078		~~- if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {~~
	2069	+ if( $prefixLength && $lastPrefix === $prefix2 ) {
2079	2070	# Same as the last item, so no need to deal with nesting or opening stuff
2080		~~- $output .= $this->nextItem( substr( $pref, -1 ) );~~
	2071	+ $output .= $this->nextItem( substr( $prefix, -1 ) );
2081	2072	$paragraphStack = false;
2082	2073
2083		~~- if ( substr( $pref, -1 ) == ';') {~~
	2074	+ if ( substr( $prefix, -1 ) == ';') {
2084	2075	# The one nasty exception: definition lists work like this:
2085	2076	# ; title : definition text
2086	2077	# So we check for : in the remainder text to split up the
—	—	@@ -2092,18 +2083,18 @@
2093	2084	}
2094	2085	} elseif( $prefixLength \|\| $lastPrefixLength ) {
2095	2086	# Either open or close a level...
2096		~~- $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );~~
	2087	+ $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2097	2088	$paragraphStack = false;
2098	2089
2099	2090	while( $commonPrefixLength < $lastPrefixLength ) {
2100		~~- $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );~~
	2091	+ $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2101	2092	--$lastPrefixLength;
2102	2093	}
2103	2094	if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2104		~~- $output .= $this->nextItem( $pref{$commonPrefixLength-1} );~~
	2095	+ $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2105	2096	}
2106	2097	while ( $prefixLength > $commonPrefixLength ) {
2107		~~- $char = substr( $pref, $commonPrefixLength, 1 );~~
	2098	+ $char = substr( $prefix, $commonPrefixLength, 1 );
2108	2099	$output .= $this->openList( $char );
2109	2100
2110	2101	if ( ';' == $char ) {
—	—	@@ -2115,10 +2106,10 @@
2116	2107	}
2117	2108	++$commonPrefixLength;
2118	2109	}
2119		~~- $lastPrefix = $pref2;~~
	2110	+ $lastPrefix = $prefix2;
2120	2111	}
2121	2112	if( 0 == $prefixLength ) {
2122		~~- wfProfileIn( "$fname-paragraph" );~~
	2113	+ wfProfileIn( __METHOD__."-paragraph" );
2123	2114	# No prefix (not in list)--go to paragraph mode
2124	2115	// XXX: use a stack for nestable elements like span, table and div
2125	2116	$openmatch = preg_match('/(?:<table\|<blockquote\|<h1\|<h2\|<h3\|<h4\|<h5\|<h6\|<pre\|<tr\|<p\|<ul\|<ol\|<li\|<\\/tr\|<\\/td\|<\\/th)/iS', $t );
—	—	@@ -2174,7 +2165,7 @@
2175	2166	}
2176	2167	}
2177	2168	}
2178		~~- wfProfileOut( "$fname-paragraph" );~~
	2169	+ wfProfileOut( __METHOD__."-paragraph" );
2179	2170	}
2180	2171	// somewhere above we forget to get out of pre block (bug 785)
2181	2172	if($preCloseMatch && $this->mInPre) {
—	—	@@ -2185,7 +2176,7 @@
2186	2177	}
2187	2178	}
2188	2179	while ( $prefixLength ) {
2189		~~- $output .= $this->closeList( $pref2{$prefixLength-1} );~~
	2180	+ $output .= $this->closeList( $prefix2[$prefixLength-1] );
2190	2181	--$prefixLength;
2191	2182	}
2192	2183	if ( '' != $this->mLastSection ) {
—	—	@@ -2193,7 +2184,7 @@
2194	2185	$this->mLastSection = '';
2195	2186	}
2196	2187
2197		~~- wfProfileOut( $fname );~~
	2188	+ wfProfileOut( __METHOD__ );
2198	2189	return $output;
2199	2190	}
2200	2191
—	—	@@ -2206,13 +2197,12 @@
2207	2198	* return string the position of the ':', or false if none found
2208	2199	*/
2209	2200	function findColonNoLinks($str, &$before, &$after) {
2210		~~- $fname = 'Parser::findColonNoLinks';~~
2211		~~- wfProfileIn( $fname );~~
	2201	+ wfProfileIn( __METHOD__ );
2212	2202
2213	2203	$pos = strpos( $str, ':' );
2214	2204	if( $pos === false ) {
2215	2205	// Nothing to find!
2216		~~- wfProfileOut( $fname );~~
	2206	+ wfProfileOut( __METHOD__ );
2217	2207	return false;
2218	2208	}
2219	2209
—	—	@@ -2221,7 +2211,7 @@
2222	2212	// Easy; no tag nesting to worry about
2223	2213	$before = substr( $str, 0, $pos );
2224	2214	$after = substr( $str, $pos+1 );
2225		~~- wfProfileOut( $fname );~~
	2215	+ wfProfileOut( __METHOD__ );
2226	2216	return $pos;
2227	2217	}
2228	2218
—	—	@@ -2245,7 +2235,7 @@
2246	2236	// We found it!
2247	2237	$before = substr( $str, 0, $i );
2248	2238	$after = substr( $str, $i + 1 );
2249		~~- wfProfileOut( $fname );~~
	2239	+ wfProfileOut( __METHOD__ );
2250	2240	return $i;
2251	2241	}
2252	2242	// Embedded in a tag; don't break it.
—	—	@@ -2255,7 +2245,7 @@
2256	2246	$colon = strpos( $str, ':', $i );
2257	2247	if( $colon === false ) {
2258	2248	// Nothing else interesting
2259		~~- wfProfileOut( $fname );~~
	2249	+ wfProfileOut( __METHOD__ );
2260	2250	return false;
2261	2251	}
2262	2252	$lt = strpos( $str, '<', $i );
—	—	@@ -2264,7 +2254,7 @@
2265	2255	// We found it!
2266	2256	$before = substr( $str, 0, $colon );
2267	2257	$after = substr( $str, $colon + 1 );
2268		~~- wfProfileOut( $fname );~~
	2258	+ wfProfileOut( __METHOD__ );
2269	2259	return $i;
2270	2260	}
2271	2261	}
—	—	@@ -2314,8 +2304,8 @@
2315	2305	if( $c == ">" ) {
2316	2306	$stack--;
2317	2307	if( $stack < 0 ) {
2318		~~- wfDebug( "Invalid input in $fname; too many close tags\n" );~~
2319		~~- wfProfileOut( $fname );~~
	2308	+ wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
	2309	+ wfProfileOut( __METHOD__ );
2320	2310	return false;
2321	2311	}
2322	2312	$state = self::COLON_STATE_TEXT;
—	—	@@ -2350,14 +2340,14 @@
2351	2341	}
2352	2342	break;
2353	2343	default:
2354		~~- throw new MWException( "State machine error in $fname" );~~
	2344	+ throw new MWException( "State machine error in " . __METHOD__ );
2355	2345	}
2356	2346	}
2357	2347	if( $stack > 0 ) {
2358		~~- wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );~~
	2348	+ wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2359	2349	return false;
2360	2350	}
2361		~~- wfProfileOut( $fname );~~
	2351	+ wfProfileOut( __METHOD__ );
2362	2352	return false;
2363	2353	}
2364	2354
—	—	@@ -2587,12 +2577,11 @@
2588	2578	* @private
2589	2579	*/
2590	2580	function initialiseVariables() {
2591		~~- $fname = 'Parser::initialiseVariables';~~
2592		~~- wfProfileIn( $fname );~~
	2581	+ wfProfileIn( __METHOD__ );
2593	2582	$variableIDs = MagicWord::getVariableIDs();
2594	2583
2595	2584	$this->mVariables = new MagicWordArray( $variableIDs );
2596		~~- wfProfileOut( $fname );~~
	2585	+ wfProfileOut( __METHOD__ );
2597	2586	}
2598	2587
2599	2588	/**
—	—	@@ -2661,8 +2650,7 @@
2662	2651	return $text;
2663	2652	}
2664	2653
2665		~~- $fname = __METHOD__;~~
2666		~~- wfProfileIn( $fname );~~
	2654	+ wfProfileIn( __METHOD__ );
2667	2655
2668	2656	if ( $frame === false ) {
2669	2657	$frame = $this->getPreprocessor()->newFrame();
—	—	@@ -2675,7 +2663,7 @@
2676	2664	$flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2677	2665	$text = $frame->expand( $dom, $flags );
2678	2666
2679		~~- wfProfileOut( $fname );~~
	2667	+ wfProfileOut( __METHOD__ );
2680	2668	return $text;
2681	2669	}
2682	2670
—	—	@@ -2738,8 +2726,7 @@
2739	2727	*/
2740	2728	function braceSubstitution( $piece, $frame ) {
2741	2729	global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2742		~~- $fname = __METHOD__;~~
2743		~~- wfProfileIn( $fname );~~
	2730	+ wfProfileIn( __METHOD__ );
2744	2731	wfProfileIn( __METHOD__.'-setup' );
2745	2732
2746	2733	# Flags
—	—	@@ -2926,7 +2913,7 @@
2927	2914	}
2928	2915	} else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2929	2916	$found = false; //access denied
2930		~~- wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );~~
	2917	+ wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2931	2918	} else {
2932	2919	list( $text, $title ) = $this->getTemplateDom( $title );
2933	2920	if ( $text !== false ) {
—	—	@@ -2960,7 +2947,7 @@
2961	2948	# Recover the source wikitext and return it
2962	2949	if ( !$found ) {
2963	2950	$text = $frame->virtualBracketedImplode( '{{', '\|', '}}', $titleWithSpaces, $args );
2964		~~- wfProfileOut( $fname );~~
	2951	+ wfProfileOut( __METHOD__ );
2965	2952	return array( 'object' => $text );
2966	2953	}
2967	2954
—	—	@@ -3019,7 +3006,7 @@
3020	3007	$ret = array( 'text' => $text );
3021	3008	}
3022	3009
3023		~~- wfProfileOut( $fname );~~
	3010	+ wfProfileOut( __METHOD__ );
3024	3011	return $ret;
3025	3012	}
3026	3013
—	—	@@ -3562,12 +3549,7 @@
3563	3550	# <!--LINK number-->
3564	3551	# turns into
3565	3552	# link text with suffix
3566		~~- $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',~~
3567		~~- "\$this->mLinkHolders['texts'][\$1]",~~
3568		~~- $safeHeadline );~~
3569		~~- $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',~~
3570		~~- "\$this->mInterwikiLinkHolders['texts'][\$1]",~~
3571		~~- $safeHeadline );~~
	3553	+ $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3572	3554
3573	3555	# Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3574	3556	$tocline = preg_replace(
—	—	@@ -3795,7 +3777,7 @@
3796	3778	} else {
3797	3779	# Failed to validate; fall back to the default
3798	3780	$nickname = $username;
3799		~~- wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );~~
	3781	+ wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3800	3782	}
3801	3783	}
3802	3784
—	—	@@ -3901,19 +3883,17 @@
3902	3884	global $wgTitle;
3903	3885	static $executing = false;
3904	3886
3905		~~- $fname = "Parser::transformMsg";~~
3906		-
3907	3887	# Guard against infinite recursion
3908	3888	if ( $executing ) {
3909	3889	return $text;
3910	3890	}
3911	3891	$executing = true;
3912	3892
3913		~~- wfProfileIn($fname);~~
	3893	+ wfProfileIn(__METHOD__);
3914	3894	$text = $this->preprocess( $text, $wgTitle, $options );
3915	3895
3916	3896	$executing = false;
3917		~~- wfProfileOut($fname);~~
	3897	+ wfProfileOut(__METHOD__);
3918	3898	return $text;
3919	3899	}
3920	3900
—	—	@@ -4010,7 +3990,7 @@
4011	3991	# Add to function cache
4012	3992	$mw = MagicWord::get( $id );
4013	3993	if( !$mw )
4014		~~- throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );~~
	3994	+ throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4015	3995
4016	3996	$synonyms = $mw->getSynonyms();
4017	3997	$sensitive = intval( $mw->isCaseSensitive() );
—	—	@@ -4046,266 +4026,9 @@
4047	4027	* Replace <!--LINK--> link placeholders with actual links, in the buffer
4048	4028	* Placeholders created in Skin::makeLinkObj()
4049	4029	* Returns an array of link CSS classes, indexed by PDBK.
4050		~~- * $options is a bit field, RLH_FOR_UPDATE to select for update~~
4051	4030	*/
4052	4031	function replaceLinkHolders( &$text, $options = 0 ) {
4053		~~- global $wgUser;~~
4054		~~- global $wgContLang;~~
4055		-
4056		~~- $fname = 'Parser::replaceLinkHolders';~~
4057		~~- wfProfileIn( $fname );~~
4058		-
4059		~~- $pdbks = array();~~
4060		~~- $colours = array();~~
4061		~~- $linkcolour_ids = array();~~
4062		~~- $sk = $this->mOptions->getSkin();~~
4063		~~- $linkCache = LinkCache::singleton();~~
4064		-
4065		~~- if ( !empty( $this->mLinkHolders['namespaces'] ) ) {~~
4066		~~- wfProfileIn( $fname.'-check' );~~
4067		~~- $dbr = wfGetDB( DB_SLAVE );~~
4068		~~- $page = $dbr->tableName( 'page' );~~
4069		~~- $threshold = $wgUser->getOption('stubthreshold');~~
4070		-
4071		~~- # Sort by namespace~~
4072		~~- asort( $this->mLinkHolders['namespaces'] );~~
4073		-
4074		~~- # Generate query~~
4075		~~- $query = false;~~
4076		~~- $current = null;~~
4077		~~- foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {~~
4078		~~- # Make title object~~
4079		~~- $title = $this->mLinkHolders['titles'][$key];~~
4080		-
4081		~~- # Skip invalid entries.~~
4082		~~- # Result will be ugly, but prevents crash.~~
4083		~~- if ( is_null( $title ) ) {~~
4084		~~- continue;~~
4085		~~- }~~
4086		~~- $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();~~
4087		-
4088		~~- # Check if it's a static known link, e.g. interwiki~~
4089		~~- if ( $title->isAlwaysKnown() ) {~~
4090		~~- $colours[$pdbk] = '';~~
4091		~~- } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {~~
4092		~~- $colours[$pdbk] = '';~~
4093		~~- $this->mOutput->addLink( $title, $id );~~
4094		~~- } elseif ( $linkCache->isBadLink( $pdbk ) ) {~~
4095		~~- $colours[$pdbk] = 'new';~~
4096		~~- } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {~~
4097		~~- $colours[$pdbk] = 'new';~~
4098		~~- } else {~~
4099		~~- # Not in the link cache, add it to the query~~
4100		~~- if ( !isset( $current ) ) {~~
4101		~~- $current = $ns;~~
4102		~~- $query = "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";~~
4103		~~- $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";~~
4104		~~- } elseif ( $current != $ns ) {~~
4105		~~- $current = $ns;~~
4106		~~- $query .= ")) OR (page_namespace=$ns AND page_title IN(";~~
4107		~~- } else {~~
4108		~~- $query .= ', ';~~
4109		~~- }~~
4110		-
4111		~~- $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );~~
4112		~~- }~~
4113		~~- }~~
4114		~~- if ( $query ) {~~
4115		~~- $query .= '))';~~
4116		~~- if ( $options & RLH_FOR_UPDATE ) {~~
4117		~~- $query .= ' FOR UPDATE';~~
4118		~~- }~~
4119		-
4120		~~- $res = $dbr->query( $query, $fname );~~
4121		-
4122		~~- # Fetch data and form into an associative array~~
4123		~~- # non-existent = broken~~
4124		~~- while ( $s = $dbr->fetchObject($res) ) {~~
4125		~~- $title = Title::makeTitle( $s->page_namespace, $s->page_title );~~
4126		~~- $pdbk = $title->getPrefixedDBkey();~~
4127		~~- $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );~~
4128		~~- $this->mOutput->addLink( $title, $s->page_id );~~
4129		~~- $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );~~
4130		~~- //add id to the extension todolist~~
4131		~~- $linkcolour_ids[$s->page_id] = $pdbk;~~
4132		~~- }~~
4133		~~- //pass an array of page_ids to an extension~~
4134		~~- wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );~~
4135		~~- }~~
4136		~~- wfProfileOut( $fname.'-check' );~~
4137		-
4138		~~- # Do a second query for different language variants of links and categories~~
4139		~~- if($wgContLang->hasVariants()){~~
4140		~~- $linkBatch = new LinkBatch();~~
4141		~~- $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)~~
4142		~~- $categoryMap = array(); // maps $category_variant => $category (dbkeys)~~
4143		~~- $varCategories = array(); // category replacements oldDBkey => newDBkey~~
4144		-
4145		~~- $categories = $this->mOutput->getCategoryLinks();~~
4146		-
4147		~~- // Add variants of links to link batch~~
4148		~~- foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {~~
4149		~~- $title = $this->mLinkHolders['titles'][$key];~~
4150		~~- if ( is_null( $title ) )~~
4151		~~- continue;~~
4152		-
4153		~~- $pdbk = $title->getPrefixedDBkey();~~
4154		~~- $titleText = $title->getText();~~
4155		-
4156		~~- // generate all variants of the link title text~~
4157		~~- $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);~~
4158		-
4159		~~- // if link was not found (in first query), add all variants to query~~
4160		~~- if ( !isset($colours[$pdbk]) ){~~
4161		~~- foreach($allTextVariants as $textVariant){~~
4162		~~- if($textVariant != $titleText){~~
4163		~~- $variantTitle = Title::makeTitle( $ns, $textVariant );~~
4164		~~- if(is_null($variantTitle)) continue;~~
4165		~~- $linkBatch->addObj( $variantTitle );~~
4166		~~- $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;~~
4167		~~- }~~
4168		~~- }~~
4169		~~- }~~
4170		~~- }~~
4171		-
4172		~~- // process categories, check if a category exists in some variant~~
4173		~~- foreach( $categories as $category ){~~
4174		~~- $variants = $wgContLang->convertLinkToAllVariants($category);~~
4175		~~- foreach($variants as $variant){~~
4176		~~- if($variant != $category){~~
4177		~~- $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );~~
4178		~~- if(is_null($variantTitle)) continue;~~
4179		~~- $linkBatch->addObj( $variantTitle );~~
4180		~~- $categoryMap[$variant] = $category;~~
4181		~~- }~~
4182		~~- }~~
4183		~~- }~~
4184		-
4185		-
4186		~~- if(!$linkBatch->isEmpty()){~~
4187		~~- // construct query~~
4188		~~- $titleClause = $linkBatch->constructSet('page', $dbr);~~
4189		-
4190		~~- $variantQuery = "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";~~
4191		-
4192		~~- $variantQuery .= " FROM $page WHERE $titleClause";~~
4193		~~- if ( $options & RLH_FOR_UPDATE ) {~~
4194		~~- $variantQuery .= ' FOR UPDATE';~~
4195		~~- }~~
4196		-
4197		~~- $varRes = $dbr->query( $variantQuery, $fname );~~
4198		-
4199		~~- // for each found variants, figure out link holders and replace~~
4200		~~- while ( $s = $dbr->fetchObject($varRes) ) {~~
4201		-
4202		~~- $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );~~
4203		~~- $varPdbk = $variantTitle->getPrefixedDBkey();~~
4204		~~- $vardbk = $variantTitle->getDBkey();~~
4205		-
4206		~~- $holderKeys = array();~~
4207		~~- if(isset($variantMap[$varPdbk])){~~
4208		~~- $holderKeys = $variantMap[$varPdbk];~~
4209		~~- $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );~~
4210		~~- $this->mOutput->addLink( $variantTitle, $s->page_id );~~
4211		~~- }~~
4212		-
4213		~~- // loop over link holders~~
4214		~~- foreach($holderKeys as $key){~~
4215		~~- $title = $this->mLinkHolders['titles'][$key];~~
4216		~~- if ( is_null( $title ) ) continue;~~
4217		-
4218		~~- $pdbk = $title->getPrefixedDBkey();~~
4219		-
4220		~~- if(!isset($colours[$pdbk])){~~
4221		~~- // found link in some of the variants, replace the link holder data~~
4222		~~- $this->mLinkHolders['titles'][$key] = $variantTitle;~~
4223		~~- $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();~~
4224		-
4225		~~- // set pdbk and colour~~
4226		~~- $pdbks[$key] = $varPdbk;~~
4227		~~- $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );~~
4228		~~- $linkcolour_ids[$s->page_id] = $pdbk;~~
4229		~~- }~~
4230		~~- wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );~~
4231		~~- }~~
4232		-
4233		~~- // check if the object is a variant of a category~~
4234		~~- if(isset($categoryMap[$vardbk])){~~
4235		~~- $oldkey = $categoryMap[$vardbk];~~
4236		~~- if($oldkey != $vardbk)~~
4237		~~- $varCategories[$oldkey]=$vardbk;~~
4238		~~- }~~
4239		~~- }~~
4240		-
4241		~~- // rebuild the categories in original order (if there are replacements)~~
4242		~~- if(count($varCategories)>0){~~
4243		~~- $newCats = array();~~
4244		~~- $originalCats = $this->mOutput->getCategories();~~
4245		~~- foreach($originalCats as $cat => $sortkey){~~
4246		~~- // make the replacement~~
4247		~~- if( array_key_exists($cat,$varCategories) )~~
4248		~~- $newCats[$varCategories[$cat]] = $sortkey;~~
4249		~~- else $newCats[$cat] = $sortkey;~~
4250		~~- }~~
4251		~~- $this->mOutput->setCategoryLinks($newCats);~~
4252		~~- }~~
4253		~~- }~~
4254		~~- }~~
4255		-
4256		~~- # Construct search and replace arrays~~
4257		~~- wfProfileIn( $fname.'-construct' );~~
4258		~~- $replacePairs = array();~~
4259		~~- foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {~~
4260		~~- $pdbk = $pdbks[$key];~~
4261		~~- $searchkey = "<!--LINK $key-->";~~
4262		~~- $title = $this->mLinkHolders['titles'][$key];~~
4263		~~- if ( !isset( $colours[$pdbk] ) \|\| $colours[$pdbk] == 'new' ) {~~
4264		~~- $linkCache->addBadLinkObj( $title );~~
4265		~~- $colours[$pdbk] = 'new';~~
4266		~~- $this->mOutput->addLink( $title, 0 );~~
4267		~~- $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,~~
4268		~~- $this->mLinkHolders['texts'][$key],~~
4269		~~- $this->mLinkHolders['queries'][$key] );~~
4270		~~- } else {~~
4271		~~- $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],~~
4272		~~- $this->mLinkHolders['texts'][$key],~~
4273		~~- $this->mLinkHolders['queries'][$key] );~~
4274		~~- }~~
4275		~~- }~~
4276		~~- $replacer = new HashtableReplacer( $replacePairs, 1 );~~
4277		~~- wfProfileOut( $fname.'-construct' );~~
4278		-
4279		~~- # Do the thing~~
4280		~~- wfProfileIn( $fname.'-replace' );~~
4281		~~- $text = preg_replace_callback(~~
4282		~~- '/(<!--LINK .*?-->)/',~~
4283		~~- $replacer->cb(),~~
4284		~~- $text);~~
4285		-
4286		~~- wfProfileOut( $fname.'-replace' );~~
4287		~~- }~~
4288		-
4289		~~- # Now process interwiki link holders~~
4290		~~- # This is quite a bit simpler than internal links~~
4291		~~- if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {~~
4292		~~- wfProfileIn( $fname.'-interwiki' );~~
4293		~~- # Make interwiki link HTML~~
4294		~~- $replacePairs = array();~~
4295		~~- foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {~~
4296		~~- $title = $this->mInterwikiLinkHolders['titles'][$key];~~
4297		~~- $replacePairs[$key] = $sk->link( $title, $link );~~
4298		~~- }~~
4299		~~- $replacer = new HashtableReplacer( $replacePairs, 1 );~~
4300		-
4301		~~- $text = preg_replace_callback(~~
4302		~~- '/<!--IWLINK (.*?)-->/',~~
4303		~~- $replacer->cb(),~~
4304		~~- $text );~~
4305		~~- wfProfileOut( $fname.'-interwiki' );~~
4306		~~- }~~
4307		-
4308		~~- wfProfileOut( $fname );~~
4309		~~- return $colours;~~
	4032	+ return $this->mLinkHolders->replace( $text );
4310	4033	}
4311	4034
4312	4035	/**
—	—	@@ -4315,39 +4038,10 @@
4316	4039	* @return string
4317	4040	*/
4318	4041	function replaceLinkHoldersText( $text ) {
4319		~~- $fname = 'Parser::replaceLinkHoldersText';~~
4320		~~- wfProfileIn( $fname );~~
4321		-
4322		~~- $text = preg_replace_callback(~~
4323		~~- '/<!--(LINK\|IWLINK) (.*?)-->/',~~
4324		~~- array( &$this, 'replaceLinkHoldersTextCallback' ),~~
4325		~~- $text );~~
4326		-
4327		~~- wfProfileOut( $fname );~~
4328		~~- return $text;~~
	4042	+ return $this->mLinkHolders->replaceText( $text );
4329	4043	}
4330	4044
4331	4045	/**
4332		~~- * @param array $matches~~
4333		~~- * @return string~~
4334		~~- * @private~~
4335		~~- */~~
4336		~~- function replaceLinkHoldersTextCallback( $matches ) {~~
4337		~~- $type = $matches[1];~~
4338		~~- $key = $matches[2];~~
4339		~~- if( $type == 'LINK' ) {~~
4340		~~- if( isset( $this->mLinkHolders['texts'][$key] ) ) {~~
4341		~~- return $this->mLinkHolders['texts'][$key];~~
4342		~~- }~~
4343		~~- } elseif( $type == 'IWLINK' ) {~~
4344		~~- if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {~~
4345		~~- return $this->mInterwikiLinkHolders['texts'][$key];~~
4346		~~- }~~
4347		~~- }~~
4348		~~- return $matches[0];~~
4349		~~- }~~
4350		-
4351		- /**
4352	4046	* Tag hook handler for 'pre'.
4353	4047	*/
4354	4048	function renderPreTag( $text, $attribs ) {
—	—	@@ -4398,7 +4092,7 @@
4399	4093
4400	4094	wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4401	4095
4402		~~- $lines = explode( "\n", $text );~~
	4096	+ $lines = StringUtils::explode( "\n", $text );
4403	4097	foreach ( $lines as $line ) {
4404	4098	# match lines like these:
4405	4099	# Image:someimage.jpg\|This is some image
—	—	@@ -4411,7 +4105,7 @@
4412	4106
4413	4107	if ( strpos( $matches[0], '%' ) !== false )
4414	4108	$matches[1] = urldecode( $matches[1] );
4415		~~- $tp = Title::newFromText( $matches[1] );~~
	4109	+ $tp = Title::newFromText( $matches[1], NS_IMAGE );
4416	4110	$nt =& $tp;
4417	4111	if( is_null( $nt ) ) {
4418	4112	# Bogus title. Ignore these so we don't bomb out later.
—	—	@@ -4477,8 +4171,11 @@
4478	4172
4479	4173	/**
4480	4174	* Parse image options text and use it to make an image
	4175	+ * @param Title $title
	4176	+ * @param string $options
	4177	+ * @param LinkHolderArray $holders
4481	4178	*/
4482		~~- function makeImage( $title, $options ) {~~
	4179	+ function makeImage( $title, $options, $holders = false ) {
4483	4180	# Check if the options text is of the form "options\|alt text"
4484	4181	# Options are:
4485	4182	# * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
—	—	@@ -4501,7 +4198,7 @@
4502	4199	# * bottom
4503	4200	# * text-bottom
4504	4201
4505		~~- $parts = array_map( 'trim', explode( '\|', $options) );~~
	4202	+ $parts = StringUtils::explode( "\|", $options );
4506	4203	$sk = $this->mOptions->getSkin();
4507	4204
4508	4205	# Give extensions a chance to select the file revision for us
—	—	@@ -4588,7 +4285,13 @@
4589	4286	}
4590	4287
4591	4288	# Strip bad stuff out of the alt text
4592		~~- $alt = $this->replaceLinkHoldersText( $caption );~~
	4289	+ # We can't just use replaceLinkHoldersText() here, because if this function
	4290	+ # is called from replaceInternalLinks2(), mLinkHolders won't be up to date.
	4291	+ if ( $holders ) {
	4292	+ $alt = $holders->replaceText( $caption );
	4293	+ } else {
	4294	+ $alt = $this->replaceLinkHoldersText( $caption );
	4295	+ }
4593	4296
4594	4297	# make sure there are no placeholders in thumbnail attributes
4595	4298	# that are later expanded to html- so expand them now and
Index: trunk/phase3/includes/MessageCache.php
—	—	@@ -44,7 +44,6 @@
45	45
46	46	/**
47	47	* ParserOptions is lazy initialised.
48		~~- * Access should probably be protected.~~
49	48	*/
50	49	function getParserOptions() {
51	50	if ( !$this->mParserOptions ) {
Index: trunk/phase3/includes/AutoLoader.php
—	—	@@ -66,6 +66,7 @@
67	67	'EnotifNotifyJob' => 'includes/EnotifNotifyJob.php',
68	68	'ErrorPageError' => 'includes/Exception.php',
69	69	'Exif' => 'includes/Exif.php',
	70	+ 'ExplodeIterator' => 'includes/StringUtils.php',
70	71	'ExternalEdit' => 'includes/ExternalEdit.php',
71	72	'ExternalStoreDB' => 'includes/ExternalStoreDB.php',
72	73	'ExternalStoreHttp' => 'includes/ExternalStoreHttp.php',
—	—	@@ -356,6 +357,7 @@
357	358	# includes/parser
358	359	'CoreParserFunctions' => 'includes/parser/CoreParserFunctions.php',
359	360	'DateFormatter' => 'includes/parser/DateFormatter.php',
	361	+ 'LinkHolderArray' => 'includes/parser/LinkHolderArray.php',
360	362	'OnlyIncludeReplacer' => 'includes/parser/Parser.php',
361	363	'PPDAccum_Hash' => 'includes/parser/Preprocessor_Hash.php',
362	364	'PPDPart' => 'includes/parser/Preprocessor_DOM.php',
Index: trunk/phase3/includes/Title.php
—	—	@@ -10,12 +10,6 @@
11	11
12	12	define ( 'GAID_FOR_UPDATE', 1 );
13	13
14		-/**
15		~~- * Title::newFromText maintains a cache to avoid expensive re-normalization of~~
16		~~- * commonly used titles. On a batch operation this can become a memory leak~~
17		~~- * if not bounded. After hitting this many titles reset the cache.~~
18		~~- */~~
19		~~-define( 'MW_TITLECACHE_MAX', 1000 );~~
20	14
21	15	/**
22	16	* Constants for pr_cascade bitfield
—	—	@@ -36,6 +30,14 @@
37	31	//@}
38	32
39	33	/**
	34	+ * Title::newFromText maintains a cache to avoid expensive re-normalization of
	35	+ * commonly used titles. On a batch operation this can become a memory leak
	36	+ * if not bounded. After hitting this many titles reset the cache.
	37	+ */
	38	+ const CACHE_MAX = 1000;
	39	+
	40	+
	41	+ /**
40	42	* @name Private member variables
41	43	* Please use the accessor functions instead.
42	44	* @private
—	—	@@ -131,7 +133,7 @@
132	134	static $cachedcount = 0 ;
133	135	if( $t->secureAndSplit() ) {
134	136	if( $defaultNamespace == NS_MAIN ) {
135		~~- if( $cachedcount >= MW_TITLECACHE_MAX ) {~~
	137	+ if( $cachedcount >= self::CACHE_MAX ) {
136	138	# Avoid memory leaks on mass operations...
137	139	Title::$titleCache = array();
138	140	$cachedcount=0;
Index: trunk/phase3/includes/LinkCache.php
—	—	@@ -9,7 +9,6 @@
10	10	// becomes incompatible with the new version.
11	11	/* private */ var $mClassVer = 4;
12	12
13		~~- /* private */ var $mPageLinks;~~
14	13	/* private */ var $mGoodLinks, $mBadLinks;
15	14	/* private */ var $mForUpdate;
16	15
—	—	@@ -26,7 +25,6 @@
27	26
28	27	function __construct() {
29	28	$this->mForUpdate = false;
30		~~- $this->mPageLinks = array();~~
31	29	$this->mGoodLinks = array();
32	30	$this->mGoodLinkFields = array();
33	31	$this->mBadLinks = array();
—	—	@@ -78,14 +76,12 @@
79	77	$dbkey = $title->getPrefixedDbKey();
80	78	$this->mGoodLinks[$dbkey] = $id;
81	79	$this->mGoodLinkFields[$dbkey] = array( 'length' => $len, 'redirect' => $redir );
82		~~- $this->mPageLinks[$dbkey] = $title;~~
83	80	}
84	81
85	82	public function addBadLinkObj( $title ) {
86	83	$dbkey = $title->getPrefixedDbKey();
87	84	if ( ! $this->isBadLink( $dbkey ) ) {
88	85	$this->mBadLinks[$dbkey] = 1;
89		~~- $this->mPageLinks[$dbkey] = $title;~~
90	86	}
91	87	}
92	88
—	—	@@ -96,7 +92,6 @@
97	93	/* obsolete, for old $wgLinkCacheMemcached stuff */
98	94	public function clearLink( $title ) {}
99	95
100		~~- public function getPageLinks() { return $this->mPageLinks; }~~
101	96	public function getGoodLinks() { return $this->mGoodLinks; }
102	97	public function getBadLinks() { return array_keys( $this->mBadLinks ); }
103	98
—	—	@@ -181,7 +176,6 @@
182	177	* Clears cache
183	178	*/
184	179	public function clear() {
185		~~- $this->mPageLinks = array();~~
186	180	$this->mGoodLinks = array();
187	181	$this->mGoodLinkFields = array();
188	182	$this->mBadLinks = array();
Index: trunk/phase3/includes/StringUtils.php
—	—	@@ -167,6 +167,18 @@
168	168	$string = str_replace( '$', '\\$', $string );
169	169	return $string;
170	170	}
	171	+
	172	+ /**
	173	+ * Workalike for explode() with limited memory usage.
	174	+ * Returns an Iterator
	175	+ */
	176	+ static function explode( $separator, $subject ) {
	177	+ if ( substr_count( $subject, $separator ) > 1000 ) {
	178	+ return new ExplodeIterator( $separator, $subject );
	179	+ } else {
	180	+ return new ArrayIterator( explode( $separator, $subject ) );
	181	+ }
	182	+ }
171	183	}
172	184
173	185	/**
—	—	@@ -310,3 +322,90 @@
311	323	return $result;
312	324	}
313	325	}
	326	+
	327	+/**
	328	+ * An iterator which works exactly like:
	329	+ *
	330	+ * foreach ( explode( $delim, $s ) as $element ) {
	331	+ * ...
	332	+ * }
	333	+ *
	334	+ * Except it doesn't use 193 byte per element
	335	+ */
	336	+class ExplodeIterator implements Iterator {
	337	+ // The subject string
	338	+ var $subject, $subjectLength;
	339	+
	340	+ // The delimiter
	341	+ var $delim, $delimLength;
	342	+
	343	+ // The position of the start of the line
	344	+ var $curPos;
	345	+
	346	+ // The position after the end of the next delimiter
	347	+ var $endPos;
	348	+
	349	+ // The current token
	350	+ var $current;
	351	+
	352	+ /**
	353	+ * Construct a DelimIterator
	354	+ */
	355	+ function __construct( $delim, $s ) {
	356	+ $this->subject = $s;
	357	+ $this->delim = $delim;
	358	+
	359	+ // Micro-optimisation (theoretical)
	360	+ $this->subjectLength = strlen( $s );
	361	+ $this->delimLength = strlen( $delim );
	362	+
	363	+ $this->rewind();
	364	+ }
	365	+
	366	+ function rewind() {
	367	+ $this->curPos = 0;
	368	+ $this->endPos = strpos( $this->subject, $this->delim );
	369	+ $this->refreshCurrent();
	370	+ }
	371	+
	372	+
	373	+ function refreshCurrent() {
	374	+ if ( $this->curPos === false ) {
	375	+ $this->current = false;
	376	+ } elseif ( $this->curPos >= $this->subjectLength ) {
	377	+ $this->current = '';
	378	+ } elseif ( $this->endPos === false ) {
	379	+ $this->current = substr( $this->subject, $this->curPos );
	380	+ } else {
	381	+ $this->current = substr( $this->subject, $this->curPos, $this->endPos - $this->curPos );
	382	+ }
	383	+ }
	384	+
	385	+ function current() {
	386	+ return $this->current;
	387	+ }
	388	+
	389	+ function key() {
	390	+ return $this->curPos;
	391	+ }
	392	+
	393	+ function next() {
	394	+ if ( $this->endPos === false ) {
	395	+ $this->curPos = false;
	396	+ } else {
	397	+ $this->curPos = $this->endPos + $this->delimLength;
	398	+ if ( $this->curPos >= $this->subjectLength ) {
	399	+ $this->endPos = false;
	400	+ } else {
	401	+ $this->endPos = strpos( $this->subject, $this->delim, $this->curPos );
	402	+ }
	403	+ }
	404	+ $this->refreshCurrent();
	405	+ return $this->current;
	406	+ }
	407	+
	408	+ function valid() {
	409	+ return $this->curPos !== false;
	410	+ }
	411	+}
	412	+
Index: trunk/phase3/RELEASE-NOTES
—	—	@@ -130,6 +130,7 @@
131	131	gives results
132	132	* Avoid recursive crazy expansions in section edit comments for pages which
133	133	contain '/*' in the title
	134	+* Fix excessive memory usage when parsing pages with lots of links
134	135
135	136	=== API changes in 1.14 ===
136	137

Follow-up revisions

Revision	Commit summary	Author	Date
r39663	2 new PASSING test(s) :)...	brion	21:10, 19 August 2008

Past revisions this follows-up on

Revision	Commit summary	Author	Date
r39414	* In the parser: do link existence tests in batches of 1000. Avoids using exc...	tstarling	16:35, 15 August 2008

Status & tagging log

15:30, 12 September 2011 Meno25 (talk | contribs) changed the status of r39467 [removed: ok added: old]