includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @addtogroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         /**#@+
  80          * @private
  81          */
  82         # Persistent:
  83         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  84                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  85                 $mExtLinkBracketedRegex, $mPreprocessor;
  86
  87         # Cleared with clearState():
  88         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  89         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  90         var $mInterwikiLinkHolders, $mLinkHolders;
  91         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  92         var $mTplExpandCache; // empty-frame expansion cache
  93         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  94
  95         # Temporary
  96         # These are variables reset at least once per parse regardless of $clearState
  97         var $mOptions,      // ParserOptions object
  98                 $mTitle,        // Title context, used for self-link rendering and similar things
  99                 $mOutputType,   // Output type, one of the OT_xxx constants
 100                 $ot,            // Shortcut alias, see setOutputType()
 101                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 102                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 103                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 104
 105         /**#@-*/
 106
 107         /**
 108          * Constructor
 109          *
 110          * @public
 111          */
 112         function __construct( $conf = array() ) {
 113                 $this->mTagHooks = array();
 114                 $this->mTransparentTagHooks = array();
 115                 $this->mFunctionHooks = array();
 116                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 117                 $this->mStripList = array( 'nowiki', 'gallery' );
 118                 $this->mMarkerSuffix = "-QINU\x7f";
 119                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 120                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 121                 if ( isset( $conf['preprocessorClass'] ) ) {
 122                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 123                 } else {
 124                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 125                 }
 126                 $this->mFirstCall = true;
 127         }
 128
 129         /**
 130          * Do various kinds of initialisation on the first call of the parser
 131          */
 132         function firstCallInit() {
 133                 if ( !$this->mFirstCall ) {
 134                         return;
 135                 }
 136                 $this->mFirstCall = false;
 137
 138                 wfProfileIn( __METHOD__ );
 139                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 140
 141                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 142
 143                 # Syntax for arguments (see self::setFunctionHook):
 144                 #  "name for lookup in localized magic words array",
 145                 #  function callback,
 146                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 147                 #    instead of {{#int:...}})
 148                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 170                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 171                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 172                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 173                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 174                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 175                 $this->setFunctionHook( 'tag',              array( 'CoreParserFunctions', 'tagObj'           ), SFH_OBJECT_ARGS );
 176
 177                 if ( $wgAllowDisplayTitle ) {
 178                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 179                 }
 180                 if ( $wgAllowSlowParserFunctions ) {
 181                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 182                 }
 183
 184                 $this->initialiseVariables();
 185
 186                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 187                 wfProfileOut( __METHOD__ );
 188         }
 189
 190         /**
 191          * Clear Parser state
 192          *
 193          * @private
 194          */
 195         function clearState() {
 196                 wfProfileIn( __METHOD__ );
 197                 if ( $this->mFirstCall ) {
 198                         $this->firstCallInit();
 199                 }
 200                 $this->mOutput = new ParserOutput;
 201                 $this->mAutonumber = 0;
 202                 $this->mLastSection = '';
 203                 $this->mDTopen = false;
 204                 $this->mIncludeCount = array();
 205                 $this->mStripState = new StripState;
 206                 $this->mArgStack = false;
 207                 $this->mInPre = false;
 208                 $this->mInterwikiLinkHolders = array(
 209                         'texts' => array(),
 210                         'titles' => array()
 211                 );
 212                 $this->mLinkHolders = array(
 213                         'namespaces' => array(),
 214                         'dbkeys' => array(),
 215                         'queries' => array(),
 216                         'texts' => array(),
 217                         'titles' => array()
 218                 );
 219                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 220
 221                 /**
 222                  * Prefix for temporary replacement strings for the multipass parser.
 223                  * \x07 should never appear in input as it's disallowed in XML.
 224                  * Using it at the front also gives us a little extra robustness
 225                  * since it shouldn't match when butted up against identifier-like
 226                  * string constructs.
 227                  *
 228                  * Must not consist of all title characters, or else it will change
 229                  * the behaviour of <nowiki> in a link.
 230                  */
 231                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 232                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 233
 234                 # Clear these on every parse, bug 4549
 235                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 236
 237                 $this->mShowToc = true;
 238                 $this->mForceTocPosition = false;
 239                 $this->mIncludeSizes = array(
 240                         'post-expand' => 0,
 241                         'arg' => 0,
 242                 );
 243                 $this->mPPNodeCount = 0;
 244                 $this->mDefaultSort = false;
 245                 $this->mHeadings = array();
 246
 247                 wfRunHooks( 'ParserClearState', array( &$this ) );
 248                 wfProfileOut( __METHOD__ );
 249         }
 250
 251         function setOutputType( $ot ) {
 252                 $this->mOutputType = $ot;
 253                 // Shortcut alias
 254                 $this->ot = array(
 255                         'html' => $ot == OT_HTML,
 256                         'wiki' => $ot == OT_WIKI,
 257                         'pre' => $ot == OT_PREPROCESS,
 258                 );
 259         }
 260
 261         /**
 262          * Set the context title
 263          */
 264         function setTitle( $t ) {
 265                 if ( !$t || $t instanceof FakeTitle ) {
 266                         $t = Title::newFromText( 'NO TITLE' );
 267                 }
 268                 if ( strval( $t->getFragment() ) !== '' ) {
 269                         # Strip the fragment to avoid various odd effects
 270                         $this->mTitle = clone $t;
 271                         $this->mTitle->setFragment( '' );
 272                 } else {
 273                         $this->mTitle = $t;
 274                 }
 275         }
 276
 277         /**
 278          * Accessor for mUniqPrefix.
 279          *
 280          * @public
 281          */
 282         function uniqPrefix() {
 283                 if( !isset( $this->mUniqPrefix ) ) {
 284                         // @fixme this is probably *horribly wrong*
 285                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 286                         // if this is called for a parser cache hit, the parser may not
 287                         // have ever been initialized in the first place.
 288                         // Not really sure what the heck is supposed to be going on here.
 289                         return '';
 290                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 291                 }
 292                 return $this->mUniqPrefix;
 293         }
 294
 295         /**
 296          * Convert wikitext to HTML
 297          * Do not call this function recursively.
 298          *
 299          * @param string $text Text we want to parse
 300          * @param Title &$title A title object
 301          * @param array $options
 302          * @param boolean $linestart
 303          * @param boolean $clearState
 304          * @param int $revid number to pass in {{REVISIONID}}
 305          * @return ParserOutput a ParserOutput
 306          */
 307         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 308                 /**
 309                  * First pass--just handle <nowiki> sections, pass the rest off
 310                  * to internalParse() which does all the real work.
 311                  */
 312
 313                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 314                 $fname = 'Parser::parse-' . wfGetCaller();
 315                 wfProfileIn( __METHOD__ );
 316                 wfProfileIn( $fname );
 317
 318                 if ( $clearState ) {
 319                         $this->clearState();
 320                 }
 321
 322                 $this->mOptions = $options;
 323                 $this->setTitle( $title );
 324                 $oldRevisionId = $this->mRevisionId;
 325                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 326                 if( $revid !== null ) {
 327                         $this->mRevisionId = $revid;
 328                         $this->mRevisionTimestamp = null;
 329                 }
 330                 $this->setOutputType( OT_HTML );
 331                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 332                 # No more strip!
 333                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 334                 $text = $this->internalParse( $text );
 335                 $text = $this->mStripState->unstripGeneral( $text );
 336
 337                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 338                 $fixtags = array(
 339                         # french spaces, last one Guillemet-left
 340                         # only if there is something before the space
 341                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 342                         # french spaces, Guillemet-right
 343                         '/(\\302\\253) /' => '\\1&nbsp;',
 344                 );
 345                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 346
 347                 # only once and last
 348                 $text = $this->doBlockLevels( $text, $linestart );
 349
 350                 $this->replaceLinkHolders( $text );
 351
 352                 # the position of the parserConvert() call should not be changed. it
 353                 # assumes that the links are all replaced and the only thing left
 354                 # is the <nowiki> mark.
 355                 # Side-effects: this calls $this->mOutput->setTitleText()
 356                 $text = $wgContLang->parserConvert( $text, $this );
 357
 358                 $text = $this->mStripState->unstripNoWiki( $text );
 359
 360                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 361
 362 //!JF Move to its own function
 363
 364                 $uniq_prefix = $this->mUniqPrefix;
 365                 $matches = array();
 366                 $elements = array_keys( $this->mTransparentTagHooks );
 367                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 368
 369                 foreach( $matches as $marker => $data ) {
 370                         list( $element, $content, $params, $tag ) = $data;
 371                         $tagName = strtolower( $element );
 372                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 373                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 374                                         array( $content, $params, $this ) );
 375                         } else {
 376                                 $output = $tag;
 377                         }
 378                         $this->mStripState->general->setPair( $marker, $output );
 379                 }
 380                 $text = $this->mStripState->unstripGeneral( $text );
 381
 382                 $text = Sanitizer::normalizeCharReferences( $text );
 383
 384                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 385                         $text = Parser::tidy($text);
 386                 } else {
 387                         # attempt to sanitize at least some nesting problems
 388                         # (bug #2702 and quite a few others)
 389                         $tidyregs = array(
 390                                 # ''Something [http://www.cool.com cool''] -->
 391                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 392                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 393                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 394                                 # fix up an anchor inside another anchor, only
 395                                 # at least for a single single nested link (bug 3695)
 396                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 397                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 398                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 399                                 # contains a div, so fix it up here; replace
 400                                 # div with escaped text
 401                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 402                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 403                                 # remove empty italic or bold tag pairs, some
 404                                 # introduced by rules above
 405                                 '/<([bi])><\/\\1>/' => '',
 406                         );
 407
 408                         $text = preg_replace(
 409                                 array_keys( $tidyregs ),
 410                                 array_values( $tidyregs ),
 411                                 $text );
 412                 }
 413
 414                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 415
 416                 # Information on include size limits, for the benefit of users who try to skirt them
 417                 if ( $this->mOptions->getEnableLimitReport() ) {
 418                         $max = $this->mOptions->getMaxIncludeSize();
 419                         $limitReport =
 420                                 "NewPP limit report\n" .
 421                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 422                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 423                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          */
 440         function recursiveTagParse( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Get a random string
 472          *
 473          * @private
 474          * @static
 475          */
 476         function getRandomString() {
 477                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 478         }
 479
 480         function &getTitle() { return $this->mTitle; }
 481         function getOptions() { return $this->mOptions; }
 482
 483         function getFunctionLang() {
 484                 global $wgLang, $wgContLang;
 485                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 486         }
 487
 488         /**
 489          * Get a preprocessor object
 490          */
 491         function getPreprocessor() {
 492                 if ( !isset( $this->mPreprocessor ) ) {
 493                         $class = $this->mPreprocessorClass;
 494                         $this->mPreprocessor = new $class( $this );
 495                 }
 496                 return $this->mPreprocessor;
 497         }
 498
 499         /**
 500          * Replaces all occurrences of HTML-style comments and the given tags
 501          * in the text with a random marker and returns teh next text. The output
 502          * parameter $matches will be an associative array filled with data in
 503          * the form:
 504          *   'UNIQ-xxxxx' => array(
 505          *     'element',
 506          *     'tag content',
 507          *     array( 'param' => 'x' ),
 508          *     '<element param="x">tag content</element>' ) )
 509          *
 510          * @param $elements list of element names. Comments are always extracted.
 511          * @param $text Source text string.
 512          * @param $uniq_prefix
 513          *
 514          * @public
 515          * @static
 516          */
 517         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 518                 static $n = 1;
 519                 $stripped = '';
 520                 $matches = array();
 521
 522                 $taglist = implode( '|', $elements );
 523                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 524
 525                 while ( '' != $text ) {
 526                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 527                         $stripped .= $p[0];
 528                         if( count( $p ) < 5 ) {
 529                                 break;
 530                         }
 531                         if( count( $p ) > 5 ) {
 532                                 // comment
 533                                 $element    = $p[4];
 534                                 $attributes = '';
 535                                 $close      = '';
 536                                 $inside     = $p[5];
 537                         } else {
 538                                 // tag
 539                                 $element    = $p[1];
 540                                 $attributes = $p[2];
 541                                 $close      = $p[3];
 542                                 $inside     = $p[4];
 543                         }
 544
 545                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 546                         $stripped .= $marker;
 547
 548                         if ( $close === '/>' ) {
 549                                 // Empty element tag, <tag />
 550                                 $content = null;
 551                                 $text = $inside;
 552                                 $tail = null;
 553                         } else {
 554                                 if( $element == '!--' ) {
 555                                         $end = '/(-->)/';
 556                                 } else {
 557                                         $end = "/(<\\/$element\\s*>)/i";
 558                                 }
 559                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 560                                 $content = $q[0];
 561                                 if( count( $q ) < 3 ) {
 562                                         # No end tag -- let it run out to the end of the text.
 563                                         $tail = '';
 564                                         $text = '';
 565                                 } else {
 566                                         $tail = $q[1];
 567                                         $text = $q[2];
 568                                 }
 569                         }
 570
 571                         $matches[$marker] = array( $element,
 572                                 $content,
 573                                 Sanitizer::decodeTagAttributes( $attributes ),
 574                                 "<$element$attributes$close$content$tail" );
 575                 }
 576                 return $stripped;
 577         }
 578
 579         /**
 580          * Get a list of strippable XML-like elements
 581          */
 582         function getStripList() {
 583                 global $wgRawHtml;
 584                 $elements = $this->mStripList;
 585                 if( $wgRawHtml ) {
 586                         $elements[] = 'html';
 587                 }
 588                 if( $this->mOptions->getUseTeX() ) {
 589                         $elements[] = 'math';
 590                 }
 591                 return $elements;
 592         }
 593
 594         /**
 595          * @deprecated use replaceVariables
 596          */
 597         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 598                 return $text;
 599         }
 600
 601         /**
 602          * Restores pre, math, and other extensions removed by strip()
 603          *
 604          * always call unstripNoWiki() after this one
 605          * @private
 606          * @deprecated use $this->mStripState->unstrip()
 607          */
 608         function unstrip( $text, $state ) {
 609                 return $state->unstripGeneral( $text );
 610         }
 611
 612         /**
 613          * Always call this after unstrip() to preserve the order
 614          *
 615          * @private
 616          * @deprecated use $this->mStripState->unstrip()
 617          */
 618         function unstripNoWiki( $text, $state ) {
 619                 return $state->unstripNoWiki( $text );
 620         }
 621
 622         /**
 623          * @deprecated use $this->mStripState->unstripBoth()
 624          */
 625         function unstripForHTML( $text ) {
 626                 return $this->mStripState->unstripBoth( $text );
 627         }
 628
 629         /**
 630          * Add an item to the strip state
 631          * Returns the unique tag which must be inserted into the stripped text
 632          * The tag will be replaced with the original text in unstrip()
 633          *
 634          * @private
 635          */
 636         function insertStripItem( $text ) {
 637                 static $n = 0;
 638                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 639                 ++$n;
 640                 $this->mStripState->general->setPair( $rnd, $text );
 641                 return $rnd;
 642         }
 643
 644         /**
 645          * Interface with html tidy, used if $wgUseTidy = true.
 646          * If tidy isn't able to correct the markup, the original will be
 647          * returned in all its glory with a warning comment appended.
 648          *
 649          * Either the external tidy program or the in-process tidy extension
 650          * will be used depending on availability. Override the default
 651          * $wgTidyInternal setting to disable the internal if it's not working.
 652          *
 653          * @param string $text Hideous HTML input
 654          * @return string Corrected HTML output
 655          * @public
 656          * @static
 657          */
 658         function tidy( $text ) {
 659                 global $wgTidyInternal;
 660                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 661 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 662 '<head><title>test</title></head><body>'.$text.'</body></html>';
 663                 if( $wgTidyInternal ) {
 664                         $correctedtext = Parser::internalTidy( $wrappedtext );
 665                 } else {
 666                         $correctedtext = Parser::externalTidy( $wrappedtext );
 667                 }
 668                 if( is_null( $correctedtext ) ) {
 669                         wfDebug( "Tidy error detected!\n" );
 670                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 671                 }
 672                 return $correctedtext;
 673         }
 674
 675         /**
 676          * Spawn an external HTML tidy process and get corrected markup back from it.
 677          *
 678          * @private
 679          * @static
 680          */
 681         function externalTidy( $text ) {
 682                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 683                 $fname = 'Parser::externalTidy';
 684                 wfProfileIn( $fname );
 685
 686                 $cleansource = '';
 687                 $opts = ' -utf8';
 688
 689                 $descriptorspec = array(
 690                         0 => array('pipe', 'r'),
 691                         1 => array('pipe', 'w'),
 692                         2 => array('file', wfGetNull(), 'a')
 693                 );
 694                 $pipes = array();
 695                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 696                 if (is_resource($process)) {
 697                         // Theoretically, this style of communication could cause a deadlock
 698                         // here. If the stdout buffer fills up, then writes to stdin could
 699                         // block. This doesn't appear to happen with tidy, because tidy only
 700                         // writes to stdout after it's finished reading from stdin. Search
 701                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 702                         fwrite($pipes[0], $text);
 703                         fclose($pipes[0]);
 704                         while (!feof($pipes[1])) {
 705                                 $cleansource .= fgets($pipes[1], 1024);
 706                         }
 707                         fclose($pipes[1]);
 708                         proc_close($process);
 709                 }
 710
 711                 wfProfileOut( $fname );
 712
 713                 if( $cleansource == '' && $text != '') {
 714                         // Some kind of error happened, so we couldn't get the corrected text.
 715                         // Just give up; we'll use the source text and append a warning.
 716                         return null;
 717                 } else {
 718                         return $cleansource;
 719                 }
 720         }
 721
 722         /**
 723          * Use the HTML tidy PECL extension to use the tidy library in-process,
 724          * saving the overhead of spawning a new process.
 725          *
 726          * 'pear install tidy' should be able to compile the extension module.
 727          *
 728          * @private
 729          * @static
 730          */
 731         function internalTidy( $text ) {
 732                 global $wgTidyConf, $IP, $wgDebugTidy;
 733                 $fname = 'Parser::internalTidy';
 734                 wfProfileIn( $fname );
 735
 736                 $tidy = new tidy;
 737                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 738                 $tidy->cleanRepair();
 739                 if( $tidy->getStatus() == 2 ) {
 740                         // 2 is magic number for fatal error
 741                         // http://www.php.net/manual/en/function.tidy-get-status.php
 742                         $cleansource = null;
 743                 } else {
 744                         $cleansource = tidy_get_output( $tidy );
 745                 }
 746                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 747                         $cleansource .= "<!--\nTidy reports:\n" .
 748                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 749                                 "\n-->";
 750                 }
 751
 752                 wfProfileOut( $fname );
 753                 return $cleansource;
 754         }
 755
 756         /**
 757          * parse the wiki syntax used to render tables
 758          *
 759          * @private
 760          */
 761         function doTableStuff ( $text ) {
 762                 $fname = 'Parser::doTableStuff';
 763                 wfProfileIn( $fname );
 764
 765                 $lines = explode ( "\n" , $text );
 766                 $td_history = array (); // Is currently a td tag open?
 767                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 768                 $tr_history = array (); // Is currently a tr tag open?
 769                 $tr_attributes = array (); // history of tr attributes
 770                 $has_opened_tr = array(); // Did this table open a <tr> element?
 771                 $indent_level = 0; // indent level of the table
 772                 foreach ( $lines as $key => $line )
 773                 {
 774                         $line = trim ( $line );
 775
 776                         if( $line == '' ) { // empty line, go to next line
 777                                 continue;
 778                         }
 779                         $first_character = $line{0};
 780                         $matches = array();
 781
 782                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 783                                 // First check if we are starting a new table
 784                                 $indent_level = strlen( $matches[1] );
 785
 786                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 787                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 788
 789                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 790                                 array_push ( $td_history , false );
 791                                 array_push ( $last_tag_history , '' );
 792                                 array_push ( $tr_history , false );
 793                                 array_push ( $tr_attributes , '' );
 794                                 array_push ( $has_opened_tr , false );
 795                         } else if ( count ( $td_history ) == 0 ) {
 796                                 // Don't do any of the following
 797                                 continue;
 798                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 799                                 // We are ending a table
 800                                 $line = '</table>' . substr ( $line , 2 );
 801                                 $last_tag = array_pop ( $last_tag_history );
 802
 803                                 if ( !array_pop ( $has_opened_tr ) ) {
 804                                         $line = "<tr><td></td></tr>{$line}";
 805                                 }
 806
 807                                 if ( array_pop ( $tr_history ) ) {
 808                                         $line = "</tr>{$line}";
 809                                 }
 810
 811                                 if ( array_pop ( $td_history ) ) {
 812                                         $line = "</{$last_tag}>{$line}";
 813                                 }
 814                                 array_pop ( $tr_attributes );
 815                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 816                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 817                                 // Now we have a table row
 818                                 $line = preg_replace( '#^\|-+#', '', $line );
 819
 820                                 // Whats after the tag is now only attributes
 821                                 $attributes = $this->mStripState->unstripBoth( $line );
 822                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 823                                 array_pop ( $tr_attributes );
 824                                 array_push ( $tr_attributes , $attributes );
 825
 826                                 $line = '';
 827                                 $last_tag = array_pop ( $last_tag_history );
 828                                 array_pop ( $has_opened_tr );
 829                                 array_push ( $has_opened_tr , true );
 830
 831                                 if ( array_pop ( $tr_history ) ) {
 832                                         $line = '</tr>';
 833                                 }
 834
 835                                 if ( array_pop ( $td_history ) ) {
 836                                         $line = "</{$last_tag}>{$line}";
 837                                 }
 838
 839                                 $lines[$key] = $line;
 840                                 array_push ( $tr_history , false );
 841                                 array_push ( $td_history , false );
 842                                 array_push ( $last_tag_history , '' );
 843                         }
 844                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 845                                 // This might be cell elements, td, th or captions
 846                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 847                                         $first_character = '+';
 848                                         $line = substr ( $line , 1 );
 849                                 }
 850
 851                                 $line = substr ( $line , 1 );
 852
 853                                 if ( $first_character == '!' ) {
 854                                         $line = str_replace ( '!!' , '||' , $line );
 855                                 }
 856
 857                                 // Split up multiple cells on the same line.
 858                                 // FIXME : This can result in improper nesting of tags processed
 859                                 // by earlier parser steps, but should avoid splitting up eg
 860                                 // attribute values containing literal "||".
 861                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 862
 863                                 $lines[$key] = '';
 864
 865                                 // Loop through each table cell
 866                                 foreach ( $cells as $cell )
 867                                 {
 868                                         $previous = '';
 869                                         if ( $first_character != '+' )
 870                                         {
 871                                                 $tr_after = array_pop ( $tr_attributes );
 872                                                 if ( !array_pop ( $tr_history ) ) {
 873                                                         $previous = "<tr{$tr_after}>\n";
 874                                                 }
 875                                                 array_push ( $tr_history , true );
 876                                                 array_push ( $tr_attributes , '' );
 877                                                 array_pop ( $has_opened_tr );
 878                                                 array_push ( $has_opened_tr , true );
 879                                         }
 880
 881                                         $last_tag = array_pop ( $last_tag_history );
 882
 883                                         if ( array_pop ( $td_history ) ) {
 884                                                 $previous = "</{$last_tag}>{$previous}";
 885                                         }
 886
 887                                         if ( $first_character == '|' ) {
 888                                                 $last_tag = 'td';
 889                                         } else if ( $first_character == '!' ) {
 890                                                 $last_tag = 'th';
 891                                         } else if ( $first_character == '+' ) {
 892                                                 $last_tag = 'caption';
 893                                         } else {
 894                                                 $last_tag = '';
 895                                         }
 896
 897                                         array_push ( $last_tag_history , $last_tag );
 898
 899                                         // A cell could contain both parameters and data
 900                                         $cell_data = explode ( '|' , $cell , 2 );
 901
 902                                         // Bug 553: Note that a '|' inside an invalid link should not
 903                                         // be mistaken as delimiting cell parameters
 904                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 905                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 906                                         } else if ( count ( $cell_data ) == 1 )
 907                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 908                                         else {
 909                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 910                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 911                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 912                                         }
 913
 914                                         $lines[$key] .= $cell;
 915                                         array_push ( $td_history , true );
 916                                 }
 917                         }
 918                 }
 919
 920                 // Closing open td, tr && table
 921                 while ( count ( $td_history ) > 0 )
 922                 {
 923                         if ( array_pop ( $td_history ) ) {
 924                                 $lines[] = '</td>' ;
 925                         }
 926                         if ( array_pop ( $tr_history ) ) {
 927                                 $lines[] = '</tr>' ;
 928                         }
 929                         if ( !array_pop ( $has_opened_tr ) ) {
 930                                 $lines[] = "<tr><td></td></tr>" ;
 931                         }
 932
 933                         $lines[] = '</table>' ;
 934                 }
 935
 936                 $output = implode ( "\n" , $lines ) ;
 937
 938                 // special case: don't return empty table
 939                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 940                         $output = '';
 941                 }
 942
 943                 wfProfileOut( $fname );
 944
 945                 return $output;
 946         }
 947
 948         /**
 949          * Helper function for parse() that transforms wiki markup into
 950          * HTML. Only called for $mOutputType == OT_HTML.
 951          *
 952          * @private
 953          */
 954         function internalParse( $text ) {
 955                 $isMain = true;
 956                 $fname = 'Parser::internalParse';
 957                 wfProfileIn( $fname );
 958
 959                 # Hook to suspend the parser in this state
 960                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 961                         wfProfileOut( $fname );
 962                         return $text ;
 963                 }
 964
 965                 $text = $this->replaceVariables( $text );
 966                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 967                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 968
 969                 // Tables need to come after variable replacement for things to work
 970                 // properly; putting them before other transformations should keep
 971                 // exciting things like link expansions from showing up in surprising
 972                 // places.
 973                 $text = $this->doTableStuff( $text );
 974
 975                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 976
 977                 $text = $this->stripToc( $text );
 978                 $this->stripNoGallery( $text );
 979                 $text = $this->doHeadings( $text );
 980                 if($this->mOptions->getUseDynamicDates()) {
 981                         $df =& DateFormatter::getInstance();
 982                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 983                 }
 984                 $text = $this->doAllQuotes( $text );
 985                 $text = $this->replaceInternalLinks( $text );
 986                 $text = $this->replaceExternalLinks( $text );
 987
 988                 # replaceInternalLinks may sometimes leave behind
 989                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 990                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 991
 992                 $text = $this->doMagicLinks( $text );
 993                 $text = $this->formatHeadings( $text, $isMain );
 994
 995                 wfProfileOut( $fname );
 996                 return $text;
 997         }
 998
 999         /**
1000          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1001          * magic external links.
1002          *
1003          * @private
1004          */
1005         function doMagicLinks( $text ) {
1006                 wfProfileIn( __METHOD__ );
1007                 $text = preg_replace_callback(
1008                         '!(?:                           # Start cases
1009                             <a.*?</a> |                 # Skip link text
1010                             <.*?> |                     # Skip stuff inside HTML elements
1011                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1012                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1013                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1014                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1015                                      [0-9Xx]                 # check digit
1016                                    \b)
1017                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1018                 wfProfileOut( __METHOD__ );
1019                 return $text;
1020         }
1021
1022         function magicLinkCallback( $m ) {
1023                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1024                         # Skip HTML element
1025                         return $m[0];
1026                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1027                         $isbn = $m[2];
1028                         $num = strtr( $isbn, array(
1029                                 '-' => '',
1030                                 ' ' => '',
1031                                 'x' => 'X',
1032                         ));
1033                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1034                         $text = '<a href="' .
1035                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1036                                 "\" class=\"internal\">ISBN $isbn</a>";
1037                 } else {
1038                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1039                                 $keyword = 'RFC';
1040                                 $urlmsg = 'rfcurl';
1041                                 $id = $m[1];
1042                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1043                                 $keyword = 'PMID';
1044                                 $urlmsg = 'pubmedurl';
1045                                 $id = $m[1];
1046                         } else {
1047                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1048                                         substr($m[0], 0, 20 ) . '"' );
1049                         }
1050
1051                         $url = wfMsg( $urlmsg, $id);
1052                         $sk = $this->mOptions->getSkin();
1053                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1054                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1055                 }
1056                 return $text;
1057         }
1058
1059         /**
1060          * Parse headers and return html
1061          *
1062          * @private
1063          */
1064         function doHeadings( $text ) {
1065                 $fname = 'Parser::doHeadings';
1066                 wfProfileIn( $fname );
1067                 for ( $i = 6; $i >= 1; --$i ) {
1068                         $h = str_repeat( '=', $i );
1069                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1070                           "<h$i>\\1</h$i>", $text );
1071                 }
1072                 wfProfileOut( $fname );
1073                 return $text;
1074         }
1075
1076         /**
1077          * Replace single quotes with HTML markup
1078          * @private
1079          * @return string the altered text
1080          */
1081         function doAllQuotes( $text ) {
1082                 $fname = 'Parser::doAllQuotes';
1083                 wfProfileIn( $fname );
1084                 $outtext = '';
1085                 $lines = explode( "\n", $text );
1086                 foreach ( $lines as $line ) {
1087                         $outtext .= $this->doQuotes ( $line ) . "\n";
1088                 }
1089                 $outtext = substr($outtext, 0,-1);
1090                 wfProfileOut( $fname );
1091                 return $outtext;
1092         }
1093
1094         /**
1095          * Helper function for doAllQuotes()
1096          */
1097         public function doQuotes( $text ) {
1098                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1099                 if ( count( $arr ) == 1 )
1100                         return $text;
1101                 else
1102                 {
1103                         # First, do some preliminary work. This may shift some apostrophes from
1104                         # being mark-up to being text. It also counts the number of occurrences
1105                         # of bold and italics mark-ups.
1106                         $i = 0;
1107                         $numbold = 0;
1108                         $numitalics = 0;
1109                         foreach ( $arr as $r )
1110                         {
1111                                 if ( ( $i % 2 ) == 1 )
1112                                 {
1113                                         # If there are ever four apostrophes, assume the first is supposed to
1114                                         # be text, and the remaining three constitute mark-up for bold text.
1115                                         if ( strlen( $arr[$i] ) == 4 )
1116                                         {
1117                                                 $arr[$i-1] .= "'";
1118                                                 $arr[$i] = "'''";
1119                                         }
1120                                         # If there are more than 5 apostrophes in a row, assume they're all
1121                                         # text except for the last 5.
1122                                         else if ( strlen( $arr[$i] ) > 5 )
1123                                         {
1124                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1125                                                 $arr[$i] = "'''''";
1126                                         }
1127                                         # Count the number of occurrences of bold and italics mark-ups.
1128                                         # We are not counting sequences of five apostrophes.
1129                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1130                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1131                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1132                                 }
1133                                 $i++;
1134                         }
1135
1136                         # If there is an odd number of both bold and italics, it is likely
1137                         # that one of the bold ones was meant to be an apostrophe followed
1138                         # by italics. Which one we cannot know for certain, but it is more
1139                         # likely to be one that has a single-letter word before it.
1140                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1141                         {
1142                                 $i = 0;
1143                                 $firstsingleletterword = -1;
1144                                 $firstmultiletterword = -1;
1145                                 $firstspace = -1;
1146                                 foreach ( $arr as $r )
1147                                 {
1148                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1149                                         {
1150                                                 $x1 = substr ($arr[$i-1], -1);
1151                                                 $x2 = substr ($arr[$i-1], -2, 1);
1152                                                 if ($x1 == ' ') {
1153                                                         if ($firstspace == -1) $firstspace = $i;
1154                                                 } else if ($x2 == ' ') {
1155                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1156                                                 } else {
1157                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1158                                                 }
1159                                         }
1160                                         $i++;
1161                                 }
1162
1163                                 # If there is a single-letter word, use it!
1164                                 if ($firstsingleletterword > -1)
1165                                 {
1166                                         $arr [ $firstsingleletterword ] = "''";
1167                                         $arr [ $firstsingleletterword-1 ] .= "'";
1168                                 }
1169                                 # If not, but there's a multi-letter word, use that one.
1170                                 else if ($firstmultiletterword > -1)
1171                                 {
1172                                         $arr [ $firstmultiletterword ] = "''";
1173                                         $arr [ $firstmultiletterword-1 ] .= "'";
1174                                 }
1175                                 # ... otherwise use the first one that has neither.
1176                                 # (notice that it is possible for all three to be -1 if, for example,
1177                                 # there is only one pentuple-apostrophe in the line)
1178                                 else if ($firstspace > -1)
1179                                 {
1180                                         $arr [ $firstspace ] = "''";
1181                                         $arr [ $firstspace-1 ] .= "'";
1182                                 }
1183                         }
1184
1185                         # Now let's actually convert our apostrophic mush to HTML!
1186                         $output = '';
1187                         $buffer = '';
1188                         $state = '';
1189                         $i = 0;
1190                         foreach ($arr as $r)
1191                         {
1192                                 if (($i % 2) == 0)
1193                                 {
1194                                         if ($state == 'both')
1195                                                 $buffer .= $r;
1196                                         else
1197                                                 $output .= $r;
1198                                 }
1199                                 else
1200                                 {
1201                                         if (strlen ($r) == 2)
1202                                         {
1203                                                 if ($state == 'i')
1204                                                 { $output .= '</i>'; $state = ''; }
1205                                                 else if ($state == 'bi')
1206                                                 { $output .= '</i>'; $state = 'b'; }
1207                                                 else if ($state == 'ib')
1208                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1209                                                 else if ($state == 'both')
1210                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1211                                                 else # $state can be 'b' or ''
1212                                                 { $output .= '<i>'; $state .= 'i'; }
1213                                         }
1214                                         else if (strlen ($r) == 3)
1215                                         {
1216                                                 if ($state == 'b')
1217                                                 { $output .= '</b>'; $state = ''; }
1218                                                 else if ($state == 'bi')
1219                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1220                                                 else if ($state == 'ib')
1221                                                 { $output .= '</b>'; $state = 'i'; }
1222                                                 else if ($state == 'both')
1223                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1224                                                 else # $state can be 'i' or ''
1225                                                 { $output .= '<b>'; $state .= 'b'; }
1226                                         }
1227                                         else if (strlen ($r) == 5)
1228                                         {
1229                                                 if ($state == 'b')
1230                                                 { $output .= '</b><i>'; $state = 'i'; }
1231                                                 else if ($state == 'i')
1232                                                 { $output .= '</i><b>'; $state = 'b'; }
1233                                                 else if ($state == 'bi')
1234                                                 { $output .= '</i></b>'; $state = ''; }
1235                                                 else if ($state == 'ib')
1236                                                 { $output .= '</b></i>'; $state = ''; }
1237                                                 else if ($state == 'both')
1238                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1239                                                 else # ($state == '')
1240                                                 { $buffer = ''; $state = 'both'; }
1241                                         }
1242                                 }
1243                                 $i++;
1244                         }
1245                         # Now close all remaining tags.  Notice that the order is important.
1246                         if ($state == 'b' || $state == 'ib')
1247                                 $output .= '</b>';
1248                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1249                                 $output .= '</i>';
1250                         if ($state == 'bi')
1251                                 $output .= '</b>';
1252                         # There might be lonely ''''', so make sure we have a buffer
1253                         if ($state == 'both' && $buffer)
1254                                 $output .= '<b><i>'.$buffer.'</i></b>';
1255                         return $output;
1256                 }
1257         }
1258
1259         /**
1260          * Replace external links
1261          *
1262          * Note: this is all very hackish and the order of execution matters a lot.
1263          * Make sure to run maintenance/parserTests.php if you change this code.
1264          *
1265          * @private
1266          */
1267         function replaceExternalLinks( $text ) {
1268                 global $wgContLang;
1269                 $fname = 'Parser::replaceExternalLinks';
1270                 wfProfileIn( $fname );
1271
1272                 $sk = $this->mOptions->getSkin();
1273
1274                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1275
1276                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1277
1278                 $i = 0;
1279                 while ( $i<count( $bits ) ) {
1280                         $url = $bits[$i++];
1281                         $protocol = $bits[$i++];
1282                         $text = $bits[$i++];
1283                         $trail = $bits[$i++];
1284
1285                         # The characters '<' and '>' (which were escaped by
1286                         # removeHTMLtags()) should not be included in
1287                         # URLs, per RFC 2396.
1288                         $m2 = array();
1289                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1290                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1291                                 $url = substr($url, 0, $m2[0][1]);
1292                         }
1293
1294                         # If the link text is an image URL, replace it with an <img> tag
1295                         # This happened by accident in the original parser, but some people used it extensively
1296                         $img = $this->maybeMakeExternalImage( $text );
1297                         if ( $img !== false ) {
1298                                 $text = $img;
1299                         }
1300
1301                         $dtrail = '';
1302
1303                         # Set linktype for CSS - if URL==text, link is essentially free
1304                         $linktype = ($text == $url) ? 'free' : 'text';
1305
1306                         # No link text, e.g. [http://domain.tld/some.link]
1307                         if ( $text == '' ) {
1308                                 # Autonumber if allowed. See bug #5918
1309                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1310                                         $text = '[' . ++$this->mAutonumber . ']';
1311                                         $linktype = 'autonumber';
1312                                 } else {
1313                                         # Otherwise just use the URL
1314                                         $text = htmlspecialchars( $url );
1315                                         $linktype = 'free';
1316                                 }
1317                         } else {
1318                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1319                                 # Check for trail
1320                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1321                         }
1322
1323                         $text = $wgContLang->markNoConversion($text);
1324
1325                         $url = Sanitizer::cleanUrl( $url );
1326
1327                         # Process the trail (i.e. everything after this link up until start of the next link),
1328                         # replacing any non-bracketed links
1329                         $trail = $this->replaceFreeExternalLinks( $trail );
1330
1331                         # Use the encoded URL
1332                         # This means that users can paste URLs directly into the text
1333                         # Funny characters like &ouml; aren't valid in URLs anyway
1334                         # This was changed in August 2004
1335                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1336
1337                         # Register link in the output object.
1338                         # Replace unnecessary URL escape codes with the referenced character
1339                         # This prevents spammers from hiding links from the filters
1340                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1341                         $this->mOutput->addExternalLink( $pasteurized );
1342                 }
1343
1344                 wfProfileOut( $fname );
1345                 return $s;
1346         }
1347
1348         /**
1349          * Replace anything that looks like a URL with a link
1350          * @private
1351          */
1352         function replaceFreeExternalLinks( $text ) {
1353                 global $wgContLang;
1354                 $fname = 'Parser::replaceFreeExternalLinks';
1355                 wfProfileIn( $fname );
1356
1357                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1358                 $s = array_shift( $bits );
1359                 $i = 0;
1360
1361                 $sk = $this->mOptions->getSkin();
1362
1363                 while ( $i < count( $bits ) ){
1364                         $protocol = $bits[$i++];
1365                         $remainder = $bits[$i++];
1366
1367                         $m = array();
1368                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1369                                 # Found some characters after the protocol that look promising
1370                                 $url = $protocol . $m[1];
1371                                 $trail = $m[2];
1372
1373                                 # special case: handle urls as url args:
1374                                 # http://www.example.com/foo?=http://www.example.com/bar
1375                                 if(strlen($trail) == 0 &&
1376                                         isset($bits[$i]) &&
1377                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1378                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1379                                 {
1380                                         # add protocol, arg
1381                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1382                                         $i += 2;
1383                                         $trail = $m[2];
1384                                 }
1385
1386                                 # The characters '<' and '>' (which were escaped by
1387                                 # removeHTMLtags()) should not be included in
1388                                 # URLs, per RFC 2396.
1389                                 $m2 = array();
1390                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1391                                         $trail = substr($url, $m2[0][1]) . $trail;
1392                                         $url = substr($url, 0, $m2[0][1]);
1393                                 }
1394
1395                                 # Move trailing punctuation to $trail
1396                                 $sep = ',;\.:!?';
1397                                 # If there is no left bracket, then consider right brackets fair game too
1398                                 if ( strpos( $url, '(' ) === false ) {
1399                                         $sep .= ')';
1400                                 }
1401
1402                                 $numSepChars = strspn( strrev( $url ), $sep );
1403                                 if ( $numSepChars ) {
1404                                         $trail = substr( $url, -$numSepChars ) . $trail;
1405                                         $url = substr( $url, 0, -$numSepChars );
1406                                 }
1407
1408                                 $url = Sanitizer::cleanUrl( $url );
1409
1410                                 # Is this an external image?
1411                                 $text = $this->maybeMakeExternalImage( $url );
1412                                 if ( $text === false ) {
1413                                         # Not an image, make a link
1414                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1415                                         # Register it in the output object...
1416                                         # Replace unnecessary URL escape codes with their equivalent characters
1417                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1418                                         $this->mOutput->addExternalLink( $pasteurized );
1419                                 }
1420                                 $s .= $text . $trail;
1421                         } else {
1422                                 $s .= $protocol . $remainder;
1423                         }
1424                 }
1425                 wfProfileOut( $fname );
1426                 return $s;
1427         }
1428
1429         /**
1430          * Replace unusual URL escape codes with their equivalent characters
1431          * @param string
1432          * @return string
1433          * @static
1434          * @todo  This can merge genuinely required bits in the path or query string,
1435          *        breaking legit URLs. A proper fix would treat the various parts of
1436          *        the URL differently; as a workaround, just use the output for
1437          *        statistical records, not for actual linking/output.
1438          */
1439         static function replaceUnusualEscapes( $url ) {
1440                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1441                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1442         }
1443
1444         /**
1445          * Callback function used in replaceUnusualEscapes().
1446          * Replaces unusual URL escape codes with their equivalent character
1447          * @static
1448          * @private
1449          */
1450         private static function replaceUnusualEscapesCallback( $matches ) {
1451                 $char = urldecode( $matches[0] );
1452                 $ord = ord( $char );
1453                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1454                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1455                         // No, shouldn't be escaped
1456                         return $char;
1457                 } else {
1458                         // Yes, leave it escaped
1459                         return $matches[0];
1460                 }
1461         }
1462
1463         /**
1464          * make an image if it's allowed, either through the global
1465          * option or through the exception
1466          * @private
1467          */
1468         function maybeMakeExternalImage( $url ) {
1469                 $sk = $this->mOptions->getSkin();
1470                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1471                 $imagesexception = !empty($imagesfrom);
1472                 $text = false;
1473                 if ( $this->mOptions->getAllowExternalImages()
1474                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1475                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1476                                 # Image found
1477                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1478                         }
1479                 }
1480                 return $text;
1481         }
1482
1483         /**
1484          * Process [[ ]] wikilinks
1485          *
1486          * @private
1487          */
1488         function replaceInternalLinks( $s ) {
1489                 global $wgContLang;
1490                 static $fname = 'Parser::replaceInternalLinks' ;
1491
1492                 wfProfileIn( $fname );
1493
1494                 wfProfileIn( $fname.'-setup' );
1495                 static $tc = FALSE;
1496                 # the % is needed to support urlencoded titles as well
1497                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1498
1499                 $sk = $this->mOptions->getSkin();
1500
1501                 #split the entire text string on occurences of [[
1502                 $a = explode( '[[', ' ' . $s );
1503                 #get the first element (all text up to first [[), and remove the space we added
1504                 $s = array_shift( $a );
1505                 $s = substr( $s, 1 );
1506
1507                 # Match a link having the form [[namespace:link|alternate]]trail
1508                 static $e1 = FALSE;
1509                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1510                 # Match cases where there is no "]]", which might still be images
1511                 static $e1_img = FALSE;
1512                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1513
1514                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1515                 $e2 = null;
1516                 if ( $useLinkPrefixExtension ) {
1517                         # Match the end of a line for a word that's not followed by whitespace,
1518                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1519                         $e2 = wfMsgForContent( 'linkprefix' );
1520                 }
1521
1522                 if( is_null( $this->mTitle ) ) {
1523                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1524                 }
1525                 $nottalk = !$this->mTitle->isTalkPage();
1526
1527                 if ( $useLinkPrefixExtension ) {
1528                         $m = array();
1529                         if ( preg_match( $e2, $s, $m ) ) {
1530                                 $first_prefix = $m[2];
1531                         } else {
1532                                 $first_prefix = false;
1533                         }
1534                 } else {
1535                         $prefix = '';
1536                 }
1537
1538                 if($wgContLang->hasVariants()) {
1539                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1540                 } else {
1541                         $selflink = array($this->mTitle->getPrefixedText());
1542                 }
1543                 $useSubpages = $this->areSubpagesAllowed();
1544                 wfProfileOut( $fname.'-setup' );
1545
1546                 # Loop for each link
1547                 for ($k = 0; isset( $a[$k] ); $k++) {
1548                         $line = $a[$k];
1549                         if ( $useLinkPrefixExtension ) {
1550                                 wfProfileIn( $fname.'-prefixhandling' );
1551                                 if ( preg_match( $e2, $s, $m ) ) {
1552                                         $prefix = $m[2];
1553                                         $s = $m[1];
1554                                 } else {
1555                                         $prefix='';
1556                                 }
1557                                 # first link
1558                                 if($first_prefix) {
1559                                         $prefix = $first_prefix;
1560                                         $first_prefix = false;
1561                                 }
1562                                 wfProfileOut( $fname.'-prefixhandling' );
1563                         }
1564
1565                         $might_be_img = false;
1566
1567                         wfProfileIn( "$fname-e1" );
1568                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1569                                 $text = $m[2];
1570                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1571                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1572                                 # the real problem is with the $e1 regex
1573                                 # See bug 1300.
1574                                 #
1575                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1576                                 # and no image is in sight. See bug 2095.
1577                                 #
1578                                 if( $text !== '' &&
1579                                         substr( $m[3], 0, 1 ) === ']' &&
1580                                         strpos($text, '[') !== false
1581                                 )
1582                                 {
1583                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1584                                         $m[3] = substr( $m[3], 1 );
1585                                 }
1586                                 # fix up urlencoded title texts
1587                                 if( strpos( $m[1], '%' ) !== false ) {
1588                                         # Should anchors '#' also be rejected?
1589                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1590                                 }
1591                                 $trail = $m[3];
1592                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1593                                 $might_be_img = true;
1594                                 $text = $m[2];
1595                                 if ( strpos( $m[1], '%' ) !== false ) {
1596                                         $m[1] = urldecode($m[1]);
1597                                 }
1598                                 $trail = "";
1599                         } else { # Invalid form; output directly
1600                                 $s .= $prefix . '[[' . $line ;
1601                                 wfProfileOut( "$fname-e1" );
1602                                 continue;
1603                         }
1604                         wfProfileOut( "$fname-e1" );
1605                         wfProfileIn( "$fname-misc" );
1606
1607                         # Don't allow internal links to pages containing
1608                         # PROTO: where PROTO is a valid URL protocol; these
1609                         # should be external links.
1610                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1611                                 $s .= $prefix . '[[' . $line ;
1612                                 continue;
1613                         }
1614
1615                         # Make subpage if necessary
1616                         if( $useSubpages ) {
1617                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1618                         } else {
1619                                 $link = $m[1];
1620                         }
1621
1622                         $noforce = (substr($m[1], 0, 1) != ':');
1623                         if (!$noforce) {
1624                                 # Strip off leading ':'
1625                                 $link = substr($link, 1);
1626                         }
1627
1628                         wfProfileOut( "$fname-misc" );
1629                         wfProfileIn( "$fname-title" );
1630                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1631                         if( !$nt ) {
1632                                 $s .= $prefix . '[[' . $line;
1633                                 wfProfileOut( "$fname-title" );
1634                                 continue;
1635                         }
1636
1637                         $ns = $nt->getNamespace();
1638                         $iw = $nt->getInterWiki();
1639                         wfProfileOut( "$fname-title" );
1640
1641                         if ($might_be_img) { # if this is actually an invalid link
1642                                 wfProfileIn( "$fname-might_be_img" );
1643                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1644                                         $found = false;
1645                                         while (isset ($a[$k+1]) ) {
1646                                                 #look at the next 'line' to see if we can close it there
1647                                                 $spliced = array_splice( $a, $k + 1, 1 );
1648                                                 $next_line = array_shift( $spliced );
1649                                                 $m = explode( ']]', $next_line, 3 );
1650                                                 if ( count( $m ) == 3 ) {
1651                                                         # the first ]] closes the inner link, the second the image
1652                                                         $found = true;
1653                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1654                                                         $trail = $m[2];
1655                                                         break;
1656                                                 } elseif ( count( $m ) == 2 ) {
1657                                                         #if there's exactly one ]] that's fine, we'll keep looking
1658                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1659                                                 } else {
1660                                                         #if $next_line is invalid too, we need look no further
1661                                                         $text .= '[[' . $next_line;
1662                                                         break;
1663                                                 }
1664                                         }
1665                                         if ( !$found ) {
1666                                                 # we couldn't find the end of this imageLink, so output it raw
1667                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1668                                                 $text = $this->replaceInternalLinks($text);
1669                                                 $s .= "{$prefix}[[$link|$text";
1670                                                 # note: no $trail, because without an end, there *is* no trail
1671                                                 wfProfileOut( "$fname-might_be_img" );
1672                                                 continue;
1673                                         }
1674                                 } else { #it's not an image, so output it raw
1675                                         $s .= "{$prefix}[[$link|$text";
1676                                         # note: no $trail, because without an end, there *is* no trail
1677                                         wfProfileOut( "$fname-might_be_img" );
1678                                         continue;
1679                                 }
1680                                 wfProfileOut( "$fname-might_be_img" );
1681                         }
1682
1683                         $wasblank = ( '' == $text );
1684                         if( $wasblank ) $text = $link;
1685
1686                         # Link not escaped by : , create the various objects
1687                         if( $noforce ) {
1688
1689                                 # Interwikis
1690                                 wfProfileIn( "$fname-interwiki" );
1691                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1692                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1693                                         $s = rtrim($s . $prefix);
1694                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1695                                         wfProfileOut( "$fname-interwiki" );
1696                                         continue;
1697                                 }
1698                                 wfProfileOut( "$fname-interwiki" );
1699
1700                                 if ( $ns == NS_IMAGE ) {
1701                                         wfProfileIn( "$fname-image" );
1702                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1703                                                 # recursively parse links inside the image caption
1704                                                 # actually, this will parse them in any other parameters, too,
1705                                                 # but it might be hard to fix that, and it doesn't matter ATM
1706                                                 $text = $this->replaceExternalLinks($text);
1707                                                 $text = $this->replaceInternalLinks($text);
1708
1709                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1710                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1711                                                 $this->mOutput->addImage( $nt->getDBkey() );
1712
1713                                                 wfProfileOut( "$fname-image" );
1714                                                 continue;
1715                                         } else {
1716                                                 # We still need to record the image's presence on the page
1717                                                 $this->mOutput->addImage( $nt->getDBkey() );
1718                                         }
1719                                         wfProfileOut( "$fname-image" );
1720
1721                                 }
1722
1723                                 if ( $ns == NS_CATEGORY ) {
1724                                         wfProfileIn( "$fname-category" );
1725                                         $s = rtrim($s . "\n"); # bug 87
1726
1727                                         if ( $wasblank ) {
1728                                                 $sortkey = $this->getDefaultSort();
1729                                         } else {
1730                                                 $sortkey = $text;
1731                                         }
1732                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1733                                         $sortkey = str_replace( "\n", '', $sortkey );
1734                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1735                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1736
1737                                         /**
1738                                          * Strip the whitespace Category links produce, see bug 87
1739                                          * @todo We might want to use trim($tmp, "\n") here.
1740                                          */
1741                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1742
1743                                         wfProfileOut( "$fname-category" );
1744                                         continue;
1745                                 }
1746                         }
1747
1748                         # Self-link checking
1749                         if( $nt->getFragment() === '' ) {
1750                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1751                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1752                                         continue;
1753                                 }
1754                         }
1755
1756                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1757                         if( $ns == NS_MEDIA ) {
1758                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1759                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1760                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1761                                 $this->mOutput->addImage( $nt->getDBkey() );
1762                                 continue;
1763                         } elseif( $ns == NS_SPECIAL ) {
1764                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1765                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1766                                 } else {
1767                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1768                                 }
1769                                 continue;
1770                         } elseif( $ns == NS_IMAGE ) {
1771                                 $img = wfFindFile( $nt );
1772                                 if( $img ) {
1773                                         // Force a blue link if the file exists; may be a remote
1774                                         // upload on the shared repository, and we want to see its
1775                                         // auto-generated page.
1776                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1777                                         $this->mOutput->addLink( $nt );
1778                                         continue;
1779                                 }
1780                         }
1781                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1782                 }
1783                 wfProfileOut( $fname );
1784                 return $s;
1785         }
1786
1787         /**
1788          * Make a link placeholder. The text returned can be later resolved to a real link with
1789          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1790          * parsing of interwiki links, and secondly to allow all existence checks and
1791          * article length checks (for stub links) to be bundled into a single query.
1792          *
1793          */
1794         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1795                 wfProfileIn( __METHOD__ );
1796                 if ( ! is_object($nt) ) {
1797                         # Fail gracefully
1798                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1799                 } else {
1800                         # Separate the link trail from the rest of the link
1801                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1802
1803                         if ( $nt->isExternal() ) {
1804                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1805                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1806                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1807                         } else {
1808                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1809                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1810                                 $this->mLinkHolders['queries'][] = $query;
1811                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1812                                 $this->mLinkHolders['titles'][] = $nt;
1813
1814                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1815                         }
1816                 }
1817                 wfProfileOut( __METHOD__ );
1818                 return $retVal;
1819         }
1820
1821         /**
1822          * Render a forced-blue link inline; protect against double expansion of
1823          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1824          * Since this little disaster has to split off the trail text to avoid
1825          * breaking URLs in the following text without breaking trails on the
1826          * wiki links, it's been made into a horrible function.
1827          *
1828          * @param Title $nt
1829          * @param string $text
1830          * @param string $query
1831          * @param string $trail
1832          * @param string $prefix
1833          * @return string HTML-wikitext mix oh yuck
1834          */
1835         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1836                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1837                 $sk = $this->mOptions->getSkin();
1838                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1839                 return $this->armorLinks( $link ) . $trail;
1840         }
1841
1842         /**
1843          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1844          * going to go through further parsing steps before inline URL expansion.
1845          *
1846          * In particular this is important when using action=render, which causes
1847          * full URLs to be included.
1848          *
1849          * Oh man I hate our multi-layer parser!
1850          *
1851          * @param string more-or-less HTML
1852          * @return string less-or-more HTML with NOPARSE bits
1853          */
1854         function armorLinks( $text ) {
1855                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1856                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1857         }
1858
1859         /**
1860          * Return true if subpage links should be expanded on this page.
1861          * @return bool
1862          */
1863         function areSubpagesAllowed() {
1864                 # Some namespaces don't allow subpages
1865                 global $wgNamespacesWithSubpages;
1866                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1867         }
1868
1869         /**
1870          * Handle link to subpage if necessary
1871          * @param string $target the source of the link
1872          * @param string &$text the link text, modified as necessary
1873          * @return string the full name of the link
1874          * @private
1875          */
1876         function maybeDoSubpageLink($target, &$text) {
1877                 # Valid link forms:
1878                 # Foobar -- normal
1879                 # :Foobar -- override special treatment of prefix (images, language links)
1880                 # /Foobar -- convert to CurrentPage/Foobar
1881                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1882                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1883                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1884
1885                 $fname = 'Parser::maybeDoSubpageLink';
1886                 wfProfileIn( $fname );
1887                 $ret = $target; # default return value is no change
1888
1889                 # Some namespaces don't allow subpages,
1890                 # so only perform processing if subpages are allowed
1891                 if( $this->areSubpagesAllowed() ) {
1892                         $hash = strpos( $target, '#' );
1893                         if( $hash !== false ) {
1894                                 $suffix = substr( $target, $hash );
1895                                 $target = substr( $target, 0, $hash );
1896                         } else {
1897                                 $suffix = '';
1898                         }
1899                         # bug 7425
1900                         $target = trim( $target );
1901                         # Look at the first character
1902                         if( $target != '' && $target{0} == '/' ) {
1903                                 # / at end means we don't want the slash to be shown
1904                                 $m = array();
1905                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1906                                 if( $trailingSlashes ) {
1907                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1908                                 } else {
1909                                         $noslash = substr( $target, 1 );
1910                                 }
1911
1912                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1913                                 if( '' === $text ) {
1914                                         $text = $target . $suffix;
1915                                 } # this might be changed for ugliness reasons
1916                         } else {
1917                                 # check for .. subpage backlinks
1918                                 $dotdotcount = 0;
1919                                 $nodotdot = $target;
1920                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1921                                         ++$dotdotcount;
1922                                         $nodotdot = substr( $nodotdot, 3 );
1923                                 }
1924                                 if($dotdotcount > 0) {
1925                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1926                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1927                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1928                                                 # / at the end means don't show full path
1929                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1930                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1931                                                         if( '' === $text ) {
1932                                                                 $text = $nodotdot . $suffix;
1933                                                         }
1934                                                 }
1935                                                 $nodotdot = trim( $nodotdot );
1936                                                 if( $nodotdot != '' ) {
1937                                                         $ret .= '/' . $nodotdot;
1938                                                 }
1939                                                 $ret .= $suffix;
1940                                         }
1941                                 }
1942                         }
1943                 }
1944
1945                 wfProfileOut( $fname );
1946                 return $ret;
1947         }
1948
1949         /**#@+
1950          * Used by doBlockLevels()
1951          * @private
1952          */
1953         /* private */ function closeParagraph() {
1954                 $result = '';
1955                 if ( '' != $this->mLastSection ) {
1956                         $result = '</' . $this->mLastSection  . ">\n";
1957                 }
1958                 $this->mInPre = false;
1959                 $this->mLastSection = '';
1960                 return $result;
1961         }
1962         # getCommon() returns the length of the longest common substring
1963         # of both arguments, starting at the beginning of both.
1964         #
1965         /* private */ function getCommon( $st1, $st2 ) {
1966                 $fl = strlen( $st1 );
1967                 $shorter = strlen( $st2 );
1968                 if ( $fl < $shorter ) { $shorter = $fl; }
1969
1970                 for ( $i = 0; $i < $shorter; ++$i ) {
1971                         if ( $st1{$i} != $st2{$i} ) { break; }
1972                 }
1973                 return $i;
1974         }
1975         # These next three functions open, continue, and close the list
1976         # element appropriate to the prefix character passed into them.
1977         #
1978         /* private */ function openList( $char ) {
1979                 $result = $this->closeParagraph();
1980
1981                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1982                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1983                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1984                 else if ( ';' == $char ) {
1985                         $result .= '<dl><dt>';
1986                         $this->mDTopen = true;
1987                 }
1988                 else { $result = '<!-- ERR 1 -->'; }
1989
1990                 return $result;
1991         }
1992
1993         /* private */ function nextItem( $char ) {
1994                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1995                 else if ( ':' == $char || ';' == $char ) {
1996                         $close = '</dd>';
1997                         if ( $this->mDTopen ) { $close = '</dt>'; }
1998                         if ( ';' == $char ) {
1999                                 $this->mDTopen = true;
2000                                 return $close . '<dt>';
2001                         } else {
2002                                 $this->mDTopen = false;
2003                                 return $close . '<dd>';
2004                         }
2005                 }
2006                 return '<!-- ERR 2 -->';
2007         }
2008
2009         /* private */ function closeList( $char ) {
2010                 if ( '*' == $char ) { $text = '</li></ul>'; }
2011                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2012                 else if ( ':' == $char ) {
2013                         if ( $this->mDTopen ) {
2014                                 $this->mDTopen = false;
2015                                 $text = '</dt></dl>';
2016                         } else {
2017                                 $text = '</dd></dl>';
2018                         }
2019                 }
2020                 else {  return '<!-- ERR 3 -->'; }
2021                 return $text."\n";
2022         }
2023         /**#@-*/
2024
2025         /**
2026          * Make lists from lines starting with ':', '*', '#', etc.
2027          *
2028          * @private
2029          * @return string the lists rendered as HTML
2030          */
2031         function doBlockLevels( $text, $linestart ) {
2032                 $fname = 'Parser::doBlockLevels';
2033                 wfProfileIn( $fname );
2034
2035                 # Parsing through the text line by line.  The main thing
2036                 # happening here is handling of block-level elements p, pre,
2037                 # and making lists from lines starting with * # : etc.
2038                 #
2039                 $textLines = explode( "\n", $text );
2040
2041                 $lastPrefix = $output = '';
2042                 $this->mDTopen = $inBlockElem = false;
2043                 $prefixLength = 0;
2044                 $paragraphStack = false;
2045
2046                 if ( !$linestart ) {
2047                         $output .= array_shift( $textLines );
2048                 }
2049                 foreach ( $textLines as $oLine ) {
2050                         $lastPrefixLength = strlen( $lastPrefix );
2051                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2052                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2053                         if ( !$this->mInPre ) {
2054                                 # Multiple prefixes may abut each other for nested lists.
2055                                 $prefixLength = strspn( $oLine, '*#:;' );
2056                                 $pref = substr( $oLine, 0, $prefixLength );
2057
2058                                 # eh?
2059                                 $pref2 = str_replace( ';', ':', $pref );
2060                                 $t = substr( $oLine, $prefixLength );
2061                                 $this->mInPre = !empty($preOpenMatch);
2062                         } else {
2063                                 # Don't interpret any other prefixes in preformatted text
2064                                 $prefixLength = 0;
2065                                 $pref = $pref2 = '';
2066                                 $t = $oLine;
2067                         }
2068
2069                         # List generation
2070                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2071                                 # Same as the last item, so no need to deal with nesting or opening stuff
2072                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2073                                 $paragraphStack = false;
2074
2075                                 if ( substr( $pref, -1 ) == ';') {
2076                                         # The one nasty exception: definition lists work like this:
2077                                         # ; title : definition text
2078                                         # So we check for : in the remainder text to split up the
2079                                         # title and definition, without b0rking links.
2080                                         $term = $t2 = '';
2081                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2082                                                 $t = $t2;
2083                                                 $output .= $term . $this->nextItem( ':' );
2084                                         }
2085                                 }
2086                         } elseif( $prefixLength || $lastPrefixLength ) {
2087                                 # Either open or close a level...
2088                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2089                                 $paragraphStack = false;
2090
2091                                 while( $commonPrefixLength < $lastPrefixLength ) {
2092                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2093                                         --$lastPrefixLength;
2094                                 }
2095                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2096                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2097                                 }
2098                                 while ( $prefixLength > $commonPrefixLength ) {
2099                                         $char = substr( $pref, $commonPrefixLength, 1 );
2100                                         $output .= $this->openList( $char );
2101
2102                                         if ( ';' == $char ) {
2103                                                 # FIXME: This is dupe of code above
2104                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2105                                                         $t = $t2;
2106                                                         $output .= $term . $this->nextItem( ':' );
2107                                                 }
2108                                         }
2109                                         ++$commonPrefixLength;
2110                                 }
2111                                 $lastPrefix = $pref2;
2112                         }
2113                         if( 0 == $prefixLength ) {
2114                                 wfProfileIn( "$fname-paragraph" );
2115                                 # No prefix (not in list)--go to paragraph mode
2116                                 // XXX: use a stack for nestable elements like span, table and div
2117                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2118                                 $closematch = preg_match(
2119                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2120                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2121                                 if ( $openmatch or $closematch ) {
2122                                         $paragraphStack = false;
2123                                         # TODO bug 5718: paragraph closed
2124                                         $output .= $this->closeParagraph();
2125                                         if ( $preOpenMatch and !$preCloseMatch ) {
2126                                                 $this->mInPre = true;
2127                                         }
2128                                         if ( $closematch ) {
2129                                                 $inBlockElem = false;
2130                                         } else {
2131                                                 $inBlockElem = true;
2132                                         }
2133                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2134                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2135                                                 // pre
2136                                                 if ($this->mLastSection != 'pre') {
2137                                                         $paragraphStack = false;
2138                                                         $output .= $this->closeParagraph().'<pre>';
2139                                                         $this->mLastSection = 'pre';
2140                                                 }
2141                                                 $t = substr( $t, 1 );
2142                                         } else {
2143                                                 // paragraph
2144                                                 if ( '' == trim($t) ) {
2145                                                         if ( $paragraphStack ) {
2146                                                                 $output .= $paragraphStack.'<br />';
2147                                                                 $paragraphStack = false;
2148                                                                 $this->mLastSection = 'p';
2149                                                         } else {
2150                                                                 if ($this->mLastSection != 'p' ) {
2151                                                                         $output .= $this->closeParagraph();
2152                                                                         $this->mLastSection = '';
2153                                                                         $paragraphStack = '<p>';
2154                                                                 } else {
2155                                                                         $paragraphStack = '</p><p>';
2156                                                                 }
2157                                                         }
2158                                                 } else {
2159                                                         if ( $paragraphStack ) {
2160                                                                 $output .= $paragraphStack;
2161                                                                 $paragraphStack = false;
2162                                                                 $this->mLastSection = 'p';
2163                                                         } else if ($this->mLastSection != 'p') {
2164                                                                 $output .= $this->closeParagraph().'<p>';
2165                                                                 $this->mLastSection = 'p';
2166                                                         }
2167                                                 }
2168                                         }
2169                                 }
2170                                 wfProfileOut( "$fname-paragraph" );
2171                         }
2172                         // somewhere above we forget to get out of pre block (bug 785)
2173                         if($preCloseMatch && $this->mInPre) {
2174                                 $this->mInPre = false;
2175                         }
2176                         if ($paragraphStack === false) {
2177                                 $output .= $t."\n";
2178                         }
2179                 }
2180                 while ( $prefixLength ) {
2181                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2182                         --$prefixLength;
2183                 }
2184                 if ( '' != $this->mLastSection ) {
2185                         $output .= '</' . $this->mLastSection . '>';
2186                         $this->mLastSection = '';
2187                 }
2188
2189                 wfProfileOut( $fname );
2190                 return $output;
2191         }
2192
2193         /**
2194          * Split up a string on ':', ignoring any occurences inside tags
2195          * to prevent illegal overlapping.
2196          * @param string $str the string to split
2197          * @param string &$before set to everything before the ':'
2198          * @param string &$after set to everything after the ':'
2199          * return string the position of the ':', or false if none found
2200          */
2201         function findColonNoLinks($str, &$before, &$after) {
2202                 $fname = 'Parser::findColonNoLinks';
2203                 wfProfileIn( $fname );
2204
2205                 $pos = strpos( $str, ':' );
2206                 if( $pos === false ) {
2207                         // Nothing to find!
2208                         wfProfileOut( $fname );
2209                         return false;
2210                 }
2211
2212                 $lt = strpos( $str, '<' );
2213                 if( $lt === false || $lt > $pos ) {
2214                         // Easy; no tag nesting to worry about
2215                         $before = substr( $str, 0, $pos );
2216                         $after = substr( $str, $pos+1 );
2217                         wfProfileOut( $fname );
2218                         return $pos;
2219                 }
2220
2221                 // Ugly state machine to walk through avoiding tags.
2222                 $state = self::COLON_STATE_TEXT;
2223                 $stack = 0;
2224                 $len = strlen( $str );
2225                 for( $i = 0; $i < $len; $i++ ) {
2226                         $c = $str{$i};
2227
2228                         switch( $state ) {
2229                         // (Using the number is a performance hack for common cases)
2230                         case 0: // self::COLON_STATE_TEXT:
2231                                 switch( $c ) {
2232                                 case "<":
2233                                         // Could be either a <start> tag or an </end> tag
2234                                         $state = self::COLON_STATE_TAGSTART;
2235                                         break;
2236                                 case ":":
2237                                         if( $stack == 0 ) {
2238                                                 // We found it!
2239                                                 $before = substr( $str, 0, $i );
2240                                                 $after = substr( $str, $i + 1 );
2241                                                 wfProfileOut( $fname );
2242                                                 return $i;
2243                                         }
2244                                         // Embedded in a tag; don't break it.
2245                                         break;
2246                                 default:
2247                                         // Skip ahead looking for something interesting
2248                                         $colon = strpos( $str, ':', $i );
2249                                         if( $colon === false ) {
2250                                                 // Nothing else interesting
2251                                                 wfProfileOut( $fname );
2252                                                 return false;
2253                                         }
2254                                         $lt = strpos( $str, '<', $i );
2255                                         if( $stack === 0 ) {
2256                                                 if( $lt === false || $colon < $lt ) {
2257                                                         // We found it!
2258                                                         $before = substr( $str, 0, $colon );
2259                                                         $after = substr( $str, $colon + 1 );
2260                                                         wfProfileOut( $fname );
2261                                                         return $i;
2262                                                 }
2263                                         }
2264                                         if( $lt === false ) {
2265                                                 // Nothing else interesting to find; abort!
2266                                                 // We're nested, but there's no close tags left. Abort!
2267                                                 break 2;
2268                                         }
2269                                         // Skip ahead to next tag start
2270                                         $i = $lt;
2271                                         $state = self::COLON_STATE_TAGSTART;
2272                                 }
2273                                 break;
2274                         case 1: // self::COLON_STATE_TAG:
2275                                 // In a <tag>
2276                                 switch( $c ) {
2277                                 case ">":
2278                                         $stack++;
2279                                         $state = self::COLON_STATE_TEXT;
2280                                         break;
2281                                 case "/":
2282                                         // Slash may be followed by >?
2283                                         $state = self::COLON_STATE_TAGSLASH;
2284                                         break;
2285                                 default:
2286                                         // ignore
2287                                 }
2288                                 break;
2289                         case 2: // self::COLON_STATE_TAGSTART:
2290                                 switch( $c ) {
2291                                 case "/":
2292                                         $state = self::COLON_STATE_CLOSETAG;
2293                                         break;
2294                                 case "!":
2295                                         $state = self::COLON_STATE_COMMENT;
2296                                         break;
2297                                 case ">":
2298                                         // Illegal early close? This shouldn't happen D:
2299                                         $state = self::COLON_STATE_TEXT;
2300                                         break;
2301                                 default:
2302                                         $state = self::COLON_STATE_TAG;
2303                                 }
2304                                 break;
2305                         case 3: // self::COLON_STATE_CLOSETAG:
2306                                 // In a </tag>
2307                                 if( $c == ">" ) {
2308                                         $stack--;
2309                                         if( $stack < 0 ) {
2310                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2311                                                 wfProfileOut( $fname );
2312                                                 return false;
2313                                         }
2314                                         $state = self::COLON_STATE_TEXT;
2315                                 }
2316                                 break;
2317                         case self::COLON_STATE_TAGSLASH:
2318                                 if( $c == ">" ) {
2319                                         // Yes, a self-closed tag <blah/>
2320                                         $state = self::COLON_STATE_TEXT;
2321                                 } else {
2322                                         // Probably we're jumping the gun, and this is an attribute
2323                                         $state = self::COLON_STATE_TAG;
2324                                 }
2325                                 break;
2326                         case 5: // self::COLON_STATE_COMMENT:
2327                                 if( $c == "-" ) {
2328                                         $state = self::COLON_STATE_COMMENTDASH;
2329                                 }
2330                                 break;
2331                         case self::COLON_STATE_COMMENTDASH:
2332                                 if( $c == "-" ) {
2333                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2334                                 } else {
2335                                         $state = self::COLON_STATE_COMMENT;
2336                                 }
2337                                 break;
2338                         case self::COLON_STATE_COMMENTDASHDASH:
2339                                 if( $c == ">" ) {
2340                                         $state = self::COLON_STATE_TEXT;
2341                                 } else {
2342                                         $state = self::COLON_STATE_COMMENT;
2343                                 }
2344                                 break;
2345                         default:
2346                                 throw new MWException( "State machine error in $fname" );
2347                         }
2348                 }
2349                 if( $stack > 0 ) {
2350                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2351                         return false;
2352                 }
2353                 wfProfileOut( $fname );
2354                 return false;
2355         }
2356
2357         /**
2358          * Return value of a magic variable (like PAGENAME)
2359          *
2360          * @private
2361          */
2362         function getVariableValue( $index ) {
2363                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2364
2365                 /**
2366                  * Some of these require message or data lookups and can be
2367                  * expensive to check many times.
2368                  */
2369                 static $varCache = array();
2370                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2371                         if ( isset( $varCache[$index] ) ) {
2372                                 return $varCache[$index];
2373                         }
2374                 }
2375
2376                 $ts = time();
2377                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2378
2379                 # Use the time zone
2380                 global $wgLocaltimezone;
2381                 if ( isset( $wgLocaltimezone ) ) {
2382                         $oldtz = getenv( 'TZ' );
2383                         putenv( 'TZ='.$wgLocaltimezone );
2384                 }
2385
2386                 wfSuppressWarnings(); // E_STRICT system time bitching
2387                 $localTimestamp = date( 'YmdHis', $ts );
2388                 $localMonth = date( 'm', $ts );
2389                 $localMonthName = date( 'n', $ts );
2390                 $localDay = date( 'j', $ts );
2391                 $localDay2 = date( 'd', $ts );
2392                 $localDayOfWeek = date( 'w', $ts );
2393                 $localWeek = date( 'W', $ts );
2394                 $localYear = date( 'Y', $ts );
2395                 $localHour = date( 'H', $ts );
2396                 if ( isset( $wgLocaltimezone ) ) {
2397                         putenv( 'TZ='.$oldtz );
2398                 }
2399                 wfRestoreWarnings();
2400
2401                 switch ( $index ) {
2402                         case 'currentmonth':
2403                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2404                         case 'currentmonthname':
2405                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2406                         case 'currentmonthnamegen':
2407                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2408                         case 'currentmonthabbrev':
2409                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2410                         case 'currentday':
2411                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2412                         case 'currentday2':
2413                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2414                         case 'localmonth':
2415                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2416                         case 'localmonthname':
2417                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2418                         case 'localmonthnamegen':
2419                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2420                         case 'localmonthabbrev':
2421                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2422                         case 'localday':
2423                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2424                         case 'localday2':
2425                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2426                         case 'pagename':
2427                                 return wfEscapeWikiText( $this->mTitle->getText() );
2428                         case 'pagenamee':
2429                                 return $this->mTitle->getPartialURL();
2430                         case 'fullpagename':
2431                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2432                         case 'fullpagenamee':
2433                                 return $this->mTitle->getPrefixedURL();
2434                         case 'subpagename':
2435                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2436                         case 'subpagenamee':
2437                                 return $this->mTitle->getSubpageUrlForm();
2438                         case 'basepagename':
2439                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2440                         case 'basepagenamee':
2441                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2442                         case 'talkpagename':
2443                                 if( $this->mTitle->canTalk() ) {
2444                                         $talkPage = $this->mTitle->getTalkPage();
2445                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2446                                 } else {
2447                                         return '';
2448                                 }
2449                         case 'talkpagenamee':
2450                                 if( $this->mTitle->canTalk() ) {
2451                                         $talkPage = $this->mTitle->getTalkPage();
2452                                         return $talkPage->getPrefixedUrl();
2453                                 } else {
2454                                         return '';
2455                                 }
2456                         case 'subjectpagename':
2457                                 $subjPage = $this->mTitle->getSubjectPage();
2458                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2459                         case 'subjectpagenamee':
2460                                 $subjPage = $this->mTitle->getSubjectPage();
2461                                 return $subjPage->getPrefixedUrl();
2462                         case 'revisionid':
2463                                 // Let the edit saving system know we should parse the page
2464                                 // *after* a revision ID has been assigned.
2465                                 $this->mOutput->setFlag( 'vary-revision' );
2466                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2467                                 return $this->mRevisionId;
2468                         case 'revisionday':
2469                                 // Let the edit saving system know we should parse the page
2470                                 // *after* a revision ID has been assigned. This is for null edits.
2471                                 $this->mOutput->setFlag( 'vary-revision' );
2472                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2473                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2474                         case 'revisionday2':
2475                                 // Let the edit saving system know we should parse the page
2476                                 // *after* a revision ID has been assigned. This is for null edits.
2477                                 $this->mOutput->setFlag( 'vary-revision' );
2478                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2479                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2480                         case 'revisionmonth':
2481                                 // Let the edit saving system know we should parse the page
2482                                 // *after* a revision ID has been assigned. This is for null edits.
2483                                 $this->mOutput->setFlag( 'vary-revision' );
2484                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2485                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2486                         case 'revisionyear':
2487                                 // Let the edit saving system know we should parse the page
2488                                 // *after* a revision ID has been assigned. This is for null edits.
2489                                 $this->mOutput->setFlag( 'vary-revision' );
2490                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2491                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2492                         case 'revisiontimestamp':
2493                                 // Let the edit saving system know we should parse the page
2494                                 // *after* a revision ID has been assigned. This is for null edits.
2495                                 $this->mOutput->setFlag( 'vary-revision' );
2496                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2497                                 return $this->getRevisionTimestamp();
2498                         case 'namespace':
2499                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2500                         case 'namespacee':
2501                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2502                         case 'talkspace':
2503                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2504                         case 'talkspacee':
2505                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2506                         case 'subjectspace':
2507                                 return $this->mTitle->getSubjectNsText();
2508                         case 'subjectspacee':
2509                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2510                         case 'currentdayname':
2511                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2512                         case 'currentyear':
2513                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2514                         case 'currenttime':
2515                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2516                         case 'currenthour':
2517                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2518                         case 'currentweek':
2519                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2520                                 // int to remove the padding
2521                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2522                         case 'currentdow':
2523                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2524                         case 'localdayname':
2525                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2526                         case 'localyear':
2527                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2528                         case 'localtime':
2529                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2530                         case 'localhour':
2531                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2532                         case 'localweek':
2533                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2534                                 // int to remove the padding
2535                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2536                         case 'localdow':
2537                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2538                         case 'numberofarticles':
2539                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2540                         case 'numberoffiles':
2541                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2542                         case 'numberofusers':
2543                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2544                         case 'numberofpages':
2545                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2546                         case 'numberofadmins':
2547                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2548                         case 'numberofedits':
2549                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2550                         case 'currenttimestamp':
2551                                 return $varCache[$index] = wfTimestampNow();
2552                         case 'localtimestamp':
2553                                 return $varCache[$index] = $localTimestamp;
2554                         case 'currentversion':
2555                                 return $varCache[$index] = SpecialVersion::getVersion();
2556                         case 'sitename':
2557                                 return $wgSitename;
2558                         case 'server':
2559                                 return $wgServer;
2560                         case 'servername':
2561                                 return $wgServerName;
2562                         case 'scriptpath':
2563                                 return $wgScriptPath;
2564                         case 'directionmark':
2565                                 return $wgContLang->getDirMark();
2566                         case 'contentlanguage':
2567                                 global $wgContLanguageCode;
2568                                 return $wgContLanguageCode;
2569                         default:
2570                                 $ret = null;
2571                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2572                                         return $ret;
2573                                 else
2574                                         return null;
2575                 }
2576         }
2577
2578         /**
2579          * initialise the magic variables (like CURRENTMONTHNAME)
2580          *
2581          * @private
2582          */
2583         function initialiseVariables() {
2584                 $fname = 'Parser::initialiseVariables';
2585                 wfProfileIn( $fname );
2586                 $variableIDs = MagicWord::getVariableIDs();
2587
2588                 $this->mVariables = new MagicWordArray( $variableIDs );
2589                 wfProfileOut( $fname );
2590         }
2591
2592         /**
2593          * Preprocess some wikitext and return the document tree.
2594          * This is the ghost of replace_variables().
2595          *
2596          * @param string $text The text to parse
2597          * @param integer flags Bitwise combination of:
2598          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2599          *                                     included. Default is to assume a direct page view.
2600          *
2601          * The generated DOM tree must depend only on the input text and the flags.
2602          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2603          *
2604          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2605          * change in the DOM tree for a given text, must be passed through the section identifier
2606          * in the section edit link and thus back to extractSections().
2607          *
2608          * The output of this function is currently only cached in process memory, but a persistent
2609          * cache may be implemented at a later date which takes further advantage of these strict
2610          * dependency requirements.
2611          *
2612          * @private
2613          */
2614         function preprocessToDom ( $text, $flags = 0 ) {
2615                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2616                 return $dom;
2617         }
2618
2619         /*
2620          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2621          */
2622         public static function splitWhitespace( $s ) {
2623                 $ltrimmed = ltrim( $s );
2624                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2625                 $trimmed = rtrim( $ltrimmed );
2626                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2627                 if ( $diff > 0 ) {
2628                         $w2 = substr( $ltrimmed, -$diff );
2629                 } else {
2630                         $w2 = '';
2631                 }
2632                 return array( $w1, $trimmed, $w2 );
2633         }
2634
2635         /**
2636          * Replace magic variables, templates, and template arguments
2637          * with the appropriate text. Templates are substituted recursively,
2638          * taking care to avoid infinite loops.
2639          *
2640          * Note that the substitution depends on value of $mOutputType:
2641          *  OT_WIKI: only {{subst:}} templates
2642          *  OT_PREPROCESS: templates but not extension tags
2643          *  OT_HTML: all templates and extension tags
2644          *
2645          * @param string $tex The text to transform
2646          * @param PPFrame $frame Object describing the arguments passed to the template
2647          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2648          * @private
2649          */
2650         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2651                 # Prevent too big inclusions
2652                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2653                         return $text;
2654                 }
2655
2656                 $fname = __METHOD__;
2657                 wfProfileIn( $fname );
2658
2659                 if ( $frame === false ) {
2660                         $frame = $this->getPreprocessor()->newFrame();
2661                 } elseif ( !( $frame instanceof PPFrame ) ) {
2662                         throw new MWException( __METHOD__ . ' called using the old argument format' );
2663                 }
2664
2665                 $dom = $this->preprocessToDom( $text );
2666                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2667                 $text = $frame->expand( $dom, $flags );
2668
2669                 wfProfileOut( $fname );
2670                 return $text;
2671         }
2672
2673         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2674         static function createAssocArgs( $args ) {
2675                 $assocArgs = array();
2676                 $index = 1;
2677                 foreach( $args as $arg ) {
2678                         $eqpos = strpos( $arg, '=' );
2679                         if ( $eqpos === false ) {
2680                                 $assocArgs[$index++] = $arg;
2681                         } else {
2682                                 $name = trim( substr( $arg, 0, $eqpos ) );
2683                                 $value = trim( substr( $arg, $eqpos+1 ) );
2684                                 if ( $value === false ) {
2685                                         $value = '';
2686                                 }
2687                                 if ( $name !== false ) {
2688                                         $assocArgs[$name] = $value;
2689                                 }
2690                         }
2691                 }
2692
2693                 return $assocArgs;
2694         }
2695
2696         /**
2697          * Return the text of a template, after recursively
2698          * replacing any variables or templates within the template.
2699          *
2700          * @param array $piece The parts of the template
2701          *  $piece['title']: the title, i.e. the part before the |
2702          *  $piece['parts']: the parameter array
2703          *  $piece['lineStart']: whether the brace was at the start of a line
2704          * @param PPFrame The current frame, contains template arguments
2705          * @return string the text of the template
2706          * @private
2707          */
2708         function braceSubstitution( $piece, $frame ) {
2709                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2710                 $fname = __METHOD__;
2711                 wfProfileIn( $fname );
2712                 wfProfileIn( __METHOD__.'-setup' );
2713
2714                 # Flags
2715                 $found = false;             # $text has been filled
2716                 $nowiki = false;            # wiki markup in $text should be escaped
2717                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2718                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2719                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2720                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2721
2722                 # Title object, where $text came from
2723                 $title = NULL;
2724
2725                 # $part1 is the bit before the first |, and must contain only title characters.
2726                 # Various prefixes will be stripped from it later.
2727                 $titleWithSpaces = $frame->expand( $piece['title'] );
2728                 $part1 = trim( $titleWithSpaces );
2729                 $titleText = false;
2730
2731                 # Original title text preserved for various purposes
2732                 $originalTitle = $part1;
2733
2734                 # $args is a list of argument nodes, starting from index 0, not including $part1
2735                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2736                 wfProfileOut( __METHOD__.'-setup' );
2737
2738                 # SUBST
2739                 wfProfileIn( __METHOD__.'-modifiers' );
2740                 if ( !$found ) {
2741                         $mwSubst =& MagicWord::get( 'subst' );
2742                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2743                                 # One of two possibilities is true:
2744                                 # 1) Found SUBST but not in the PST phase
2745                                 # 2) Didn't find SUBST and in the PST phase
2746                                 # In either case, return without further processing
2747                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2748                                 $isLocalObj = true;
2749                                 $found = true;
2750                         }
2751                 }
2752
2753                 # Variables
2754                 if ( !$found && $args->getLength() == 0 ) {
2755                         $id = $this->mVariables->matchStartToEnd( $part1 );
2756                         if ( $id !== false ) {
2757                                 $text = $this->getVariableValue( $id );
2758                                 if (MagicWord::getCacheTTL($id)>-1)
2759                                         $this->mOutput->mContainsOldMagic = true;
2760                                 $found = true;
2761                         }
2762                 }
2763
2764                 # MSG, MSGNW and RAW
2765                 if ( !$found ) {
2766                         # Check for MSGNW:
2767                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2768                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2769                                 $nowiki = true;
2770                         } else {
2771                                 # Remove obsolete MSG:
2772                                 $mwMsg =& MagicWord::get( 'msg' );
2773                                 $mwMsg->matchStartAndRemove( $part1 );
2774                         }
2775
2776                         # Check for RAW:
2777                         $mwRaw =& MagicWord::get( 'raw' );
2778                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2779                                 $forceRawInterwiki = true;
2780                         }
2781                 }
2782                 wfProfileOut( __METHOD__.'-modifiers' );
2783
2784                 # Parser functions
2785                 if ( !$found ) {
2786                         wfProfileIn( __METHOD__ . '-pfunc' );
2787
2788                         $colonPos = strpos( $part1, ':' );
2789                         if ( $colonPos !== false ) {
2790                                 # Case sensitive functions
2791                                 $function = substr( $part1, 0, $colonPos );
2792                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2793                                         $function = $this->mFunctionSynonyms[1][$function];
2794                                 } else {
2795                                         # Case insensitive functions
2796                                         $function = strtolower( $function );
2797                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2798                                                 $function = $this->mFunctionSynonyms[0][$function];
2799                                         } else {
2800                                                 $function = false;
2801                                         }
2802                                 }
2803                                 if ( $function ) {
2804                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2805                                         $initialArgs = array( &$this );
2806                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2807                                         if ( $flags & SFH_OBJECT_ARGS ) {
2808                                                 # Add a frame parameter, and pass the arguments as an array
2809                                                 $allArgs = $initialArgs;
2810                                                 $allArgs[] = $frame;
2811                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2812                                                         $funcArgs[] = $args->item( $i );
2813                                                 }
2814                                                 $allArgs[] = $funcArgs;
2815                                         } else {
2816                                                 # Convert arguments to plain text
2817                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2818                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2819                                                 }
2820                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2821                                         }
2822
2823                                         $result = call_user_func_array( $callback, $allArgs );
2824                                         $found = true;
2825
2826                                         if ( is_array( $result ) ) {
2827                                                 if ( isset( $result[0] ) ) {
2828                                                         $text = $result[0];
2829                                                         unset( $result[0] );
2830                                                 }
2831
2832                                                 // Extract flags into the local scope
2833                                                 // This allows callers to set flags such as nowiki, found, etc.
2834                                                 extract( $result );
2835                                         } else {
2836                                                 $text = $result;
2837                                         }
2838                                 }
2839                         }
2840                         wfProfileOut( __METHOD__ . '-pfunc' );
2841                 }
2842
2843                 # Finish mangling title and then check for loops.
2844                 # Set $title to a Title object and $titleText to the PDBK
2845                 if ( !$found ) {
2846                         $ns = NS_TEMPLATE;
2847                         # Split the title into page and subpage
2848                         $subpage = '';
2849                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2850                         if ($subpage !== '') {
2851                                 $ns = $this->mTitle->getNamespace();
2852                         }
2853                         $title = Title::newFromText( $part1, $ns );
2854                         if ( $title ) {
2855                                 $titleText = $title->getPrefixedText();
2856                                 # Check for language variants if the template is not found
2857                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2858                                         $wgContLang->findVariantLink($part1, $title);
2859                                 }
2860                                 # Do infinite loop check
2861                                 if ( !$frame->loopCheck( $title ) ) {
2862                                         $found = true;
2863                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2864                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2865                                 }
2866                                 # Do recursion depth check
2867                                 $limit = $this->mOptions->getMaxTemplateDepth();
2868                                 if ( $frame->depth >= $limit ) {
2869                                         $found = true;
2870                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2871                                 }
2872                         }
2873                 }
2874
2875                 # Load from database
2876                 if ( !$found && $title ) {
2877                         wfProfileIn( __METHOD__ . '-loadtpl' );
2878                         if ( !$title->isExternal() ) {
2879                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2880                                         $text = SpecialPage::capturePath( $title );
2881                                         if ( is_string( $text ) ) {
2882                                                 $found = true;
2883                                                 $isHTML = true;
2884                                                 $this->disableCache();
2885                                         }
2886                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2887                                         $found = false; //access denied
2888                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
2889                                 } else {
2890                                         list( $text, $title ) = $this->getTemplateDom( $title );
2891                                         if ( $text !== false ) {
2892                                                 $found = true;
2893                                                 $isChildObj = true;
2894                                         }
2895                                 }
2896
2897                                 # If the title is valid but undisplayable, make a link to it
2898                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2899                                         $text = "[[:$titleText]]";
2900                                         $found = true;
2901                                 }
2902                         } elseif ( $title->isTrans() ) {
2903                                 // Interwiki transclusion
2904                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2905                                         $text = $this->interwikiTransclude( $title, 'render' );
2906                                         $isHTML = true;
2907                                 } else {
2908                                         $text = $this->interwikiTransclude( $title, 'raw' );
2909                                         // Preprocess it like a template
2910                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2911                                         $isChildObj = true;
2912                                 }
2913                                 $found = true;
2914                         }
2915                         wfProfileOut( __METHOD__ . '-loadtpl' );
2916                 }
2917
2918                 # If we haven't found text to substitute by now, we're done
2919                 # Recover the source wikitext and return it
2920                 if ( !$found ) {
2921                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2922                         wfProfileOut( $fname );
2923                         return array( 'object' => $text );
2924                 }
2925
2926                 # Expand DOM-style return values in a child frame
2927                 if ( $isChildObj ) {
2928                         # Clean up argument array
2929                         $newFrame = $frame->newChild( $args, $title );
2930
2931                         if ( $titleText !== false && $newFrame->isEmpty() ) {
2932                                 # Expansion is eligible for the empty-frame cache
2933                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2934                                         $text = $this->mTplExpandCache[$titleText];
2935                                 } else {
2936                                         $text = $newFrame->expand( $text );
2937                                         $this->mTplExpandCache[$titleText] = $text;
2938                                 }
2939                         } else {
2940                                 # Uncached expansion
2941                                 $text = $newFrame->expand( $text );
2942                         }
2943                 }
2944
2945                 # Replace raw HTML by a placeholder
2946                 # Add a blank line preceding, to prevent it from mucking up
2947                 # immediately preceding headings
2948                 if ( $isHTML ) {
2949                         $text = "\n\n" . $this->insertStripItem( $text );
2950                 }
2951                 # Escape nowiki-style return values
2952                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2953                         $text = wfEscapeWikiText( $text );
2954                 }
2955                 # Bug 529: if the template begins with a table or block-level
2956                 # element, it should be treated as beginning a new line.
2957                 # This behaviour is somewhat controversial.
2958                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2959                         $text = "\n" . $text;
2960                 }
2961
2962                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2963                         # Error, oversize inclusion
2964                         $text = "[[$originalTitle]]" .
2965                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2966                 }
2967
2968                 if ( $isLocalObj ) {
2969                         $ret = array( 'object' => $text );
2970                 } else {
2971                         $ret = array( 'text' => $text );
2972                 }
2973
2974                 wfProfileOut( $fname );
2975                 return $ret;
2976         }
2977
2978         /**
2979          * Get the semi-parsed DOM representation of a template with a given title,
2980          * and its redirect destination title. Cached.
2981          */
2982         function getTemplateDom( $title ) {
2983                 $cacheTitle = $title;
2984                 $titleText = $title->getPrefixedDBkey();
2985
2986                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
2987                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
2988                         $title = Title::makeTitle( $ns, $dbk );
2989                         $titleText = $title->getPrefixedDBkey();
2990                 }
2991                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
2992                         return array( $this->mTplDomCache[$titleText], $title );
2993                 }
2994
2995                 // Cache miss, go to the database
2996                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
2997
2998                 if ( $text === false ) {
2999                         $this->mTplDomCache[$titleText] = false;
3000                         return array( false, $title );
3001                 }
3002
3003                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3004                 $this->mTplDomCache[ $titleText ] = $dom;
3005
3006                 if (! $title->equals($cacheTitle)) {
3007                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3008                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3009                 }
3010
3011                 return array( $dom, $title );
3012         }
3013
3014         /**
3015          * Fetch the unparsed text of a template and register a reference to it.
3016          */
3017         function fetchTemplateAndTitle( $title ) {
3018                 $templateCb = $this->mOptions->getTemplateCallback();
3019                 $stuff = call_user_func( $templateCb, $title );
3020                 $text = $stuff['text'];
3021                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3022                 if ( isset( $stuff['deps'] ) ) {
3023                         foreach ( $stuff['deps'] as $dep ) {
3024                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3025                         }
3026                 }
3027                 return array($text,$finalTitle);
3028         }
3029
3030         function fetchTemplate( $title ) {
3031                 $rv = $this->fetchTemplateAndTitle($title);
3032                 return $rv[0];
3033         }
3034
3035         /**
3036          * Static function to get a template
3037          * Can be overridden via ParserOptions::setTemplateCallback().
3038          */
3039         static function statelessFetchTemplate( $title ) {
3040                 $text = $skip = false;
3041                 $finalTitle = $title;
3042                 $deps = array();
3043
3044                 // Loop to fetch the article, with up to 1 redirect
3045                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3046                         # Give extensions a chance to select the revision instead
3047                         $id = false; // Assume current
3048                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3049
3050                         if( $skip ) {
3051                                 $text = false;
3052                                 $deps[] = array(
3053                                         'title' => $title,
3054                                         'page_id' => $title->getArticleID(),
3055                                         'rev_id' => null );
3056                                 break;
3057                         }
3058                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3059                         $rev_id = $rev ? $rev->getId() : 0;
3060
3061                         $deps[] = array(
3062                                 'title' => $title,
3063                                 'page_id' => $title->getArticleID(),
3064                                 'rev_id' => $rev_id );
3065
3066                         if( $rev ) {
3067                                 $text = $rev->getText();
3068                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3069                                 global $wgLang;
3070                                 $message = $wgLang->lcfirst( $title->getText() );
3071                                 $text = wfMsgForContentNoTrans( $message );
3072                                 if( wfEmptyMsg( $message, $text ) ) {
3073                                         $text = false;
3074                                         break;
3075                                 }
3076                         } else {
3077                                 break;
3078                         }
3079                         if ( $text === false ) {
3080                                 break;
3081                         }
3082                         // Redirect?
3083                         $finalTitle = $title;
3084                         $title = Title::newFromRedirect( $text );
3085                 }
3086                 return array(
3087                         'text' => $text,
3088                         'finalTitle' => $finalTitle,
3089                         'deps' => $deps );
3090         }
3091
3092         /**
3093          * Transclude an interwiki link.
3094          */
3095         function interwikiTransclude( $title, $action ) {
3096                 global $wgEnableScaryTranscluding;
3097
3098                 if (!$wgEnableScaryTranscluding)
3099                         return wfMsg('scarytranscludedisabled');
3100
3101                 $url = $title->getFullUrl( "action=$action" );
3102
3103                 if (strlen($url) > 255)
3104                         return wfMsg('scarytranscludetoolong');
3105                 return $this->fetchScaryTemplateMaybeFromCache($url);
3106         }
3107
3108         function fetchScaryTemplateMaybeFromCache($url) {
3109                 global $wgTranscludeCacheExpiry;
3110                 $dbr = wfGetDB(DB_SLAVE);
3111                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3112                                 array('tc_url' => $url));
3113                 if ($obj) {
3114                         $time = $obj->tc_time;
3115                         $text = $obj->tc_contents;
3116                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3117                                 return $text;
3118                         }
3119                 }
3120
3121                 $text = Http::get($url);
3122                 if (!$text)
3123                         return wfMsg('scarytranscludefailed', $url);
3124
3125                 $dbw = wfGetDB(DB_MASTER);
3126                 $dbw->replace('transcache', array('tc_url'), array(
3127                         'tc_url' => $url,
3128                         'tc_time' => time(),
3129                         'tc_contents' => $text));
3130                 return $text;
3131         }
3132
3133
3134         /**
3135          * Triple brace replacement -- used for template arguments
3136          * @private
3137          */
3138         function argSubstitution( $piece, $frame ) {
3139                 wfProfileIn( __METHOD__ );
3140
3141                 $error = false;
3142                 $parts = $piece['parts'];
3143                 $nameWithSpaces = $frame->expand( $piece['title'] );
3144                 $argName = trim( $nameWithSpaces );
3145                 $object = false;
3146                 $text = $frame->getArgument( $argName );
3147                 if (  $text === false && ( $this->ot['html'] || $this->ot['pre'] ) && $parts->getLength() > 0 ) {
3148                         # No match in frame, use the supplied default
3149                         $object = $parts->item( 0 )->getChildren();
3150                 }
3151                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3152                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3153                 }
3154
3155                 if ( $text === false && $object === false ) {
3156                         # No match anywhere
3157                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3158                 }
3159                 if ( $error !== false ) {
3160                         $text .= $error;
3161                 }
3162                 if ( $object !== false ) {
3163                         $ret = array( 'object' => $object );
3164                 } else {
3165                         $ret = array( 'text' => $text );
3166                 }
3167
3168                 wfProfileOut( __METHOD__ );
3169                 return $ret;
3170         }
3171
3172         /**
3173          * Return the text to be used for a given extension tag.
3174          * This is the ghost of strip().
3175          *
3176          * @param array $params Associative array of parameters:
3177          *     name       PPNode for the tag name
3178          *     attr       PPNode for unparsed text where tag attributes are thought to be
3179          *     attributes Optional associative array of parsed attributes
3180          *     inner      Contents of extension element
3181          *     noClose    Original text did not have a close tag
3182          * @param PPFrame $frame
3183          */
3184         function extensionSubstitution( $params, $frame ) {
3185                 global $wgRawHtml, $wgContLang;
3186                 static $n = 1;
3187
3188                 $name = $frame->expand( $params['name'] );
3189                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3190                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3191
3192                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3193
3194                 if ( $this->ot['html'] ) {
3195                         $name = strtolower( $name );
3196
3197                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3198                         if ( isset( $params['attributes'] ) ) {
3199                                 $attributes = $attributes + $params['attributes'];
3200                         }
3201                         switch ( $name ) {
3202                                 case 'html':
3203                                         if( $wgRawHtml ) {
3204                                                 $output = $content;
3205                                                 break;
3206                                         } else {
3207                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3208                                         }
3209                                 case 'nowiki':
3210                                         $output = Xml::escapeTagsOnly( $content );
3211                                         break;
3212                                 case 'math':
3213                                         $output = $wgContLang->armourMath(
3214                                                 MathRenderer::renderMath( $content, $attributes ) );
3215                                         break;
3216                                 case 'gallery':
3217                                         $output = $this->renderImageGallery( $content, $attributes );
3218                                         break;
3219                                 default:
3220                                         if( isset( $this->mTagHooks[$name] ) ) {
3221                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3222                                                         array( $content, $attributes, $this ) );
3223                                         } else {
3224                                                 throw new MWException( "Invalid call hook $name" );
3225                                         }
3226                         }
3227                 } else {
3228                         if ( is_null( $attrText ) ) {
3229                                 $attrText = '';
3230                         }
3231                         if ( isset( $params['attributes'] ) ) {
3232                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3233                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3234                                                 htmlspecialchars( $attrValue ) . '"';
3235                                 }
3236                         }
3237                         if ( $content === null ) {
3238                                 $output = "<$name$attrText/>";
3239                         } else {
3240                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3241                                 $output = "<$name$attrText>$content$close";
3242                         }
3243                 }
3244
3245                 if ( $name == 'html' || $name == 'nowiki' ) {
3246                         $this->mStripState->nowiki->setPair( $marker, $output );
3247                 } else {
3248                         $this->mStripState->general->setPair( $marker, $output );
3249                 }
3250                 return $marker;
3251         }
3252
3253         /**
3254          * Increment an include size counter
3255          *
3256          * @param string $type The type of expansion
3257          * @param integer $size The size of the text
3258          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3259          */
3260         function incrementIncludeSize( $type, $size ) {
3261                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3262                         return false;
3263                 } else {
3264                         $this->mIncludeSizes[$type] += $size;
3265                         return true;
3266                 }
3267         }
3268
3269         /**
3270          * Detect __NOGALLERY__ magic word and set a placeholder
3271          */
3272         function stripNoGallery( &$text ) {
3273                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3274                 # do not add TOC
3275                 $mw = MagicWord::get( 'nogallery' );
3276                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3277         }
3278
3279         /**
3280          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3281          * placeholder that will then be replaced by the real TOC in
3282          * ->formatHeadings, this works because at this points real
3283          * comments will have already been discarded by the sanitizer.
3284          *
3285          * Any additional __TOC__ magic words left over will be discarded
3286          * as there can only be one TOC on the page.
3287          */
3288         function stripToc( $text ) {
3289                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3290                 # do not add TOC
3291                 $mw = MagicWord::get( 'notoc' );
3292                 if( $mw->matchAndRemove( $text ) ) {
3293                         $this->mShowToc = false;
3294                 }
3295
3296                 $mw = MagicWord::get( 'toc' );
3297                 if( $mw->match( $text ) ) {
3298                         $this->mShowToc = true;
3299                         $this->mForceTocPosition = true;
3300
3301                         // Set a placeholder. At the end we'll fill it in with the TOC.
3302                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3303
3304                         // Only keep the first one.
3305                         $text = $mw->replace( '', $text );
3306                 }
3307                 return $text;
3308         }
3309
3310         /**
3311          * This function accomplishes several tasks:
3312          * 1) Auto-number headings if that option is enabled
3313          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3314          * 3) Add a Table of contents on the top for users who have enabled the option
3315          * 4) Auto-anchor headings
3316          *
3317          * It loops through all headlines, collects the necessary data, then splits up the
3318          * string and re-inserts the newly formatted headlines.
3319          *
3320          * @param string $text
3321          * @param boolean $isMain
3322          * @private
3323          */
3324         function formatHeadings( $text, $isMain=true ) {
3325                 global $wgMaxTocLevel, $wgContLang;
3326
3327                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3328                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3329                         $showEditLink = 0;
3330                 } else {
3331                         $showEditLink = $this->mOptions->getEditSection();
3332                 }
3333
3334                 # Inhibit editsection links if requested in the page
3335                 $esw =& MagicWord::get( 'noeditsection' );
3336                 if( $esw->matchAndRemove( $text ) ) {
3337                         $showEditLink = 0;
3338                 }
3339
3340                 # Get all headlines for numbering them and adding funky stuff like [edit]
3341                 # links - this is for later, but we need the number of headlines right now
3342                 $matches = array();
3343                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3344
3345                 # if there are fewer than 4 headlines in the article, do not show TOC
3346                 # unless it's been explicitly enabled.
3347                 $enoughToc = $this->mShowToc &&
3348                         (($numMatches >= 4) || $this->mForceTocPosition);
3349
3350                 # Allow user to stipulate that a page should have a "new section"
3351                 # link added via __NEWSECTIONLINK__
3352                 $mw =& MagicWord::get( 'newsectionlink' );
3353                 if( $mw->matchAndRemove( $text ) )
3354                         $this->mOutput->setNewSection( true );
3355
3356                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3357                 # override above conditions and always show TOC above first header
3358                 $mw =& MagicWord::get( 'forcetoc' );
3359                 if ($mw->matchAndRemove( $text ) ) {
3360                         $this->mShowToc = true;
3361                         $enoughToc = true;
3362                 }
3363
3364                 # We need this to perform operations on the HTML
3365                 $sk = $this->mOptions->getSkin();
3366
3367                 # headline counter
3368                 $headlineCount = 0;
3369                 $numVisible = 0;
3370
3371                 # Ugh .. the TOC should have neat indentation levels which can be
3372                 # passed to the skin functions. These are determined here
3373                 $toc = '';
3374                 $full = '';
3375                 $head = array();
3376                 $sublevelCount = array();
3377                 $levelCount = array();
3378                 $toclevel = 0;
3379                 $level = 0;
3380                 $prevlevel = 0;
3381                 $toclevel = 0;
3382                 $prevtoclevel = 0;
3383                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3384                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3385                 $tocraw = array();
3386
3387                 foreach( $matches[3] as $headline ) {
3388                         $isTemplate = false;
3389                         $titleText = false;
3390                         $sectionIndex = false;
3391                         $numbering = '';
3392                         $markerMatches = array();
3393                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3394                                 $serial = $markerMatches[1];
3395                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3396                                 $isTemplate = ($titleText != $baseTitleText);
3397                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3398                         }
3399
3400                         if( $toclevel ) {
3401                                 $prevlevel = $level;
3402                                 $prevtoclevel = $toclevel;
3403                         }
3404                         $level = $matches[1][$headlineCount];
3405
3406                         if( $doNumberHeadings || $enoughToc ) {
3407
3408                                 if ( $level > $prevlevel ) {
3409                                         # Increase TOC level
3410                                         $toclevel++;
3411                                         $sublevelCount[$toclevel] = 0;
3412                                         if( $toclevel<$wgMaxTocLevel ) {
3413                                                 $prevtoclevel = $toclevel;
3414                                                 $toc .= $sk->tocIndent();
3415                                                 $numVisible++;
3416                                         }
3417                                 }
3418                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3419                                         # Decrease TOC level, find level to jump to
3420
3421                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3422                                                 # Can only go down to level 1
3423                                                 $toclevel = 1;
3424                                         } else {
3425                                                 for ($i = $toclevel; $i > 0; $i--) {
3426                                                         if ( $levelCount[$i] == $level ) {
3427                                                                 # Found last matching level
3428                                                                 $toclevel = $i;
3429                                                                 break;
3430                                                         }
3431                                                         elseif ( $levelCount[$i] < $level ) {
3432                                                                 # Found first matching level below current level
3433                                                                 $toclevel = $i + 1;
3434                                                                 break;
3435                                                         }
3436                                                 }
3437                                         }
3438                                         if( $toclevel<$wgMaxTocLevel ) {
3439                                                 if($prevtoclevel < $wgMaxTocLevel) {
3440                                                         # Unindent only if the previous toc level was shown :p
3441                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3442                                                 } else {
3443                                                         $toc .= $sk->tocLineEnd();
3444                                                 }
3445                                         }
3446                                 }
3447                                 else {
3448                                         # No change in level, end TOC line
3449                                         if( $toclevel<$wgMaxTocLevel ) {
3450                                                 $toc .= $sk->tocLineEnd();
3451                                         }
3452                                 }
3453
3454                                 $levelCount[$toclevel] = $level;
3455
3456                                 # count number of headlines for each level
3457                                 @$sublevelCount[$toclevel]++;
3458                                 $dot = 0;
3459                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3460                                         if( !empty( $sublevelCount[$i] ) ) {
3461                                                 if( $dot ) {
3462                                                         $numbering .= '.';
3463                                                 }
3464                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3465                                                 $dot = 1;
3466                                         }
3467                                 }
3468                         }
3469
3470                         # The safe header is a version of the header text safe to use for links
3471                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3472                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3473
3474                         # Remove link placeholders by the link text.
3475                         #     <!--LINK number-->
3476                         # turns into
3477                         #     link text with suffix
3478                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3479                                                             "\$this->mLinkHolders['texts'][\$1]",
3480                                                             $safeHeadline );
3481                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3482                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3483                                                             $safeHeadline );
3484
3485                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3486                         $tocline = preg_replace(
3487                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3488                                 array( '',                          '<$1>'),
3489                                 $safeHeadline
3490                         );
3491                         $tocline = trim( $tocline );
3492
3493                         # For the anchor, strip out HTML-y stuff period
3494                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3495                         $safeHeadline = trim( $safeHeadline );
3496
3497                         # Save headline for section edit hint before it's escaped
3498                         $headlineHint = $safeHeadline;
3499                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3500                         $refers[$headlineCount] = $safeHeadline;
3501
3502                         # count how many in assoc. array so we can track dupes in anchors
3503                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3504                         $refcount[$headlineCount] = $refers[$safeHeadline];
3505
3506                         # Don't number the heading if it is the only one (looks silly)
3507                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3508                                 # the two are different if the line contains a link
3509                                 $headline=$numbering . ' ' . $headline;
3510                         }
3511
3512                         # Create the anchor for linking from the TOC to the section
3513                         $anchor = $safeHeadline;
3514                         if($refcount[$headlineCount] > 1 ) {
3515                                 $anchor .= '_' . $refcount[$headlineCount];
3516                         }
3517                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3518                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3519                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3520                         }
3521                         # give headline the correct <h#> tag
3522                         if( $showEditLink && $sectionIndex !== false ) {
3523                                 if( $isTemplate ) {
3524                                         # Put a T flag in the section identifier, to indicate to extractSections()
3525                                         # that sections inside <includeonly> should be counted.
3526                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
3527                                 } else {
3528                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3529                                 }
3530                         } else {
3531                                 $editlink = '';
3532                         }
3533                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3534
3535                         $headlineCount++;
3536                 }
3537
3538                 $this->mOutput->setSections( $tocraw );
3539
3540                 # Never ever show TOC if no headers
3541                 if( $numVisible < 1 ) {
3542                         $enoughToc = false;
3543                 }
3544
3545                 if( $enoughToc ) {
3546                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3547                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3548                         }
3549                         $toc = $sk->tocList( $toc );
3550                 }
3551
3552                 # split up and insert constructed headlines
3553
3554                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3555                 $i = 0;
3556
3557                 foreach( $blocks as $block ) {
3558                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3559                                 # This is the [edit] link that appears for the top block of text when
3560                                 # section editing is enabled
3561
3562                                 # Disabled because it broke block formatting
3563                                 # For example, a bullet point in the top line
3564                                 # $full .= $sk->editSectionLink(0);
3565                         }
3566                         $full .= $block;
3567                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3568                                 # Top anchor now in skin
3569                                 $full = $full.$toc;
3570                         }
3571
3572                         if( !empty( $head[$i] ) ) {
3573                                 $full .= $head[$i];
3574                         }
3575                         $i++;
3576                 }
3577                 if( $this->mForceTocPosition ) {
3578                         return str_replace( '<!--MWTOC-->', $toc, $full );
3579                 } else {
3580                         return $full;
3581                 }
3582         }
3583
3584         /**
3585          * Transform wiki markup when saving a page by doing \r\n -> \n
3586          * conversion, substitting signatures, {{subst:}} templates, etc.
3587          *
3588          * @param string $text the text to transform
3589          * @param Title &$title the Title object for the current article
3590          * @param User &$user the User object describing the current user
3591          * @param ParserOptions $options parsing options
3592          * @param bool $clearState whether to clear the parser state first
3593          * @return string the altered wiki markup
3594          * @public
3595          */
3596         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3597                 $this->mOptions = $options;
3598                 $this->setTitle( $title );
3599                 $this->setOutputType( OT_WIKI );
3600
3601                 if ( $clearState ) {
3602                         $this->clearState();
3603                 }
3604
3605                 $pairs = array(
3606                         "\r\n" => "\n",
3607                 );
3608                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3609                 $text = $this->pstPass2( $text, $user );
3610                 $text = $this->mStripState->unstripBoth( $text );
3611                 return $text;
3612         }
3613
3614         /**
3615          * Pre-save transform helper function
3616          * @private
3617          */
3618         function pstPass2( $text, $user ) {
3619                 global $wgContLang, $wgLocaltimezone;
3620
3621                 /* Note: This is the timestamp saved as hardcoded wikitext to
3622                  * the database, we use $wgContLang here in order to give
3623                  * everyone the same signature and use the default one rather
3624                  * than the one selected in each user's preferences.
3625                  */
3626                 if ( isset( $wgLocaltimezone ) ) {
3627                         $oldtz = getenv( 'TZ' );
3628                         putenv( 'TZ='.$wgLocaltimezone );
3629                 }
3630                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3631                   ' (' . date( 'T' ) . ')';
3632                 if ( isset( $wgLocaltimezone ) ) {
3633                         putenv( 'TZ='.$oldtz );
3634                 }
3635
3636                 # Variable replacement
3637                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3638                 $text = $this->replaceVariables( $text );
3639
3640                 # Strip out <nowiki> etc. added via replaceVariables
3641                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
3642
3643                 # Signatures
3644                 $sigText = $this->getUserSig( $user );
3645                 $text = strtr( $text, array(
3646                         '~~~~~' => $d,
3647                         '~~~~' => "$sigText $d",
3648                         '~~~' => $sigText
3649                 ) );
3650
3651                 # Context links: [[|name]] and [[name (context)|]]
3652                 #
3653                 global $wgLegalTitleChars;
3654                 $tc = "[$wgLegalTitleChars]";
3655                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3656
3657                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3658                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3659                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3660
3661                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3662                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3663                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3664
3665                 $t = $this->mTitle->getText();
3666                 $m = array();
3667                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3668                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3669                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3670                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3671                 } else {
3672                         # if there's no context, don't bother duplicating the title
3673                         $text = preg_replace( $p2, '[[\\1]]', $text );
3674                 }
3675
3676                 # Trim trailing whitespace
3677                 $text = rtrim( $text );
3678
3679                 return $text;
3680         }
3681
3682         /**
3683          * Fetch the user's signature text, if any, and normalize to
3684          * validated, ready-to-insert wikitext.
3685          *
3686          * @param User $user
3687          * @return string
3688          * @private
3689          */
3690         function getUserSig( &$user ) {
3691                 global $wgMaxSigChars;
3692
3693                 $username = $user->getName();
3694                 $nickname = $user->getOption( 'nickname' );
3695                 $nickname = $nickname === '' ? $username : $nickname;
3696
3697                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3698                         $nickname = $username;
3699                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3700                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3701                         # Sig. might contain markup; validate this
3702                         if( $this->validateSig( $nickname ) !== false ) {
3703                                 # Validated; clean up (if needed) and return it
3704                                 return $this->cleanSig( $nickname, true );
3705                         } else {
3706                                 # Failed to validate; fall back to the default
3707                                 $nickname = $username;
3708                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3709                         }
3710                 }
3711
3712                 // Make sure nickname doesnt get a sig in a sig
3713                 $nickname = $this->cleanSigInSig( $nickname );
3714
3715                 # If we're still here, make it a link to the user page
3716                 $userText = wfEscapeWikiText( $username );
3717                 $nickText = wfEscapeWikiText( $nickname );
3718                 if ( $user->isAnon() )  {
3719                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3720                 } else {
3721                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3722                 }
3723         }
3724
3725         /**
3726          * Check that the user's signature contains no bad XML
3727          *
3728          * @param string $text
3729          * @return mixed An expanded string, or false if invalid.
3730          */
3731         function validateSig( $text ) {
3732                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3733         }
3734
3735         /**
3736          * Clean up signature text
3737          *
3738          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3739          * 2) Substitute all transclusions
3740          *
3741          * @param string $text
3742          * @param $parsing Whether we're cleaning (preferences save) or parsing
3743          * @return string Signature text
3744          */
3745         function cleanSig( $text, $parsing = false ) {
3746                 if ( !$parsing ) {
3747                         global $wgTitle;
3748                         $this->clearState();
3749                         $this->setTitle( $wgTitle );
3750                         $this->mOptions = new ParserOptions;
3751                         $this->setOutputType = OT_PREPROCESS;
3752                 }
3753
3754                 # FIXME: regex doesn't respect extension tags or nowiki
3755                 #  => Move this logic to braceSubstitution()
3756                 $substWord = MagicWord::get( 'subst' );
3757                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3758                 $substText = '{{' . $substWord->getSynonym( 0 );
3759
3760                 $text = preg_replace( $substRegex, $substText, $text );
3761                 $text = $this->cleanSigInSig( $text );
3762                 $dom = $this->preprocessToDom( $text );
3763                 $frame = $this->getPreprocessor()->newFrame();
3764                 $text = $frame->expand( $dom );
3765
3766                 if ( !$parsing ) {
3767                         $text = $this->mStripState->unstripBoth( $text );
3768                 }
3769
3770                 return $text;
3771         }
3772
3773         /**
3774          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3775          * @param string $text
3776          * @return string Signature text with /~{3,5}/ removed
3777          */
3778         function cleanSigInSig( $text ) {
3779                 $text = preg_replace( '/~{3,5}/', '', $text );
3780                 return $text;
3781         }
3782
3783         /**
3784          * Set up some variables which are usually set up in parse()
3785          * so that an external function can call some class members with confidence
3786          * @public
3787          */
3788         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3789                 $this->setTitle( $title );
3790                 $this->mOptions = $options;
3791                 $this->setOutputType( $outputType );
3792                 if ( $clearState ) {
3793                         $this->clearState();
3794                 }
3795         }
3796
3797         /**
3798          * Wrapper for preprocess()
3799          *
3800          * @param string $text the text to preprocess
3801          * @param ParserOptions $options  options
3802          * @return string
3803          * @public
3804          */
3805         function transformMsg( $text, $options ) {
3806                 global $wgTitle;
3807                 static $executing = false;
3808
3809                 $fname = "Parser::transformMsg";
3810
3811                 # Guard against infinite recursion
3812                 if ( $executing ) {
3813                         return $text;
3814                 }
3815                 $executing = true;
3816
3817                 wfProfileIn($fname);
3818                 $text = $this->preprocess( $text, $wgTitle, $options );
3819
3820                 $executing = false;
3821                 wfProfileOut($fname);
3822                 return $text;
3823         }
3824
3825         /**
3826          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3827          * The callback should have the following form:
3828          *    function myParserHook( $text, $params, &$parser ) { ... }
3829          *
3830          * Transform and return $text. Use $parser for any required context, e.g. use
3831          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3832          *
3833          * @public
3834          *
3835          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3836          * @param mixed $callback The callback function (and object) to use for the tag
3837          *
3838          * @return The old value of the mTagHooks array associated with the hook
3839          */
3840         function setHook( $tag, $callback ) {
3841                 $tag = strtolower( $tag );
3842                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3843                 $this->mTagHooks[$tag] = $callback;
3844                 $this->mStripList[] = $tag;
3845
3846                 return $oldVal;
3847         }
3848
3849         function setTransparentTagHook( $tag, $callback ) {
3850                 $tag = strtolower( $tag );
3851                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3852                 $this->mTransparentTagHooks[$tag] = $callback;
3853
3854                 return $oldVal;
3855         }
3856
3857         /**
3858          * Create a function, e.g. {{sum:1|2|3}}
3859          * The callback function should have the form:
3860          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3861          *
3862          * The callback may either return the text result of the function, or an array with the text
3863          * in element 0, and a number of flags in the other elements. The names of the flags are
3864          * specified in the keys. Valid flags are:
3865          *   found                     The text returned is valid, stop processing the template. This
3866          *                             is on by default.
3867          *   nowiki                    Wiki markup in the return value should be escaped
3868          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3869          *
3870          * @public
3871          *
3872          * @param string $id The magic word ID
3873          * @param mixed $callback The callback function (and object) to use
3874          * @param integer $flags a combination of the following flags:
3875          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3876          *
3877          * @return The old callback function for this name, if any
3878          */
3879         function setFunctionHook( $id, $callback, $flags = 0 ) {
3880                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
3881                 $this->mFunctionHooks[$id] = array( $callback, $flags );
3882
3883                 # Add to function cache
3884                 $mw = MagicWord::get( $id );
3885                 if( !$mw )
3886                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3887
3888                 $synonyms = $mw->getSynonyms();
3889                 $sensitive = intval( $mw->isCaseSensitive() );
3890
3891                 foreach ( $synonyms as $syn ) {
3892                         # Case
3893                         if ( !$sensitive ) {
3894                                 $syn = strtolower( $syn );
3895                         }
3896                         # Add leading hash
3897                         if ( !( $flags & SFH_NO_HASH ) ) {
3898                                 $syn = '#' . $syn;
3899                         }
3900                         # Remove trailing colon
3901                         if ( substr( $syn, -1, 1 ) == ':' ) {
3902                                 $syn = substr( $syn, 0, -1 );
3903                         }
3904                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3905                 }
3906                 return $oldVal;
3907         }
3908
3909         /**
3910          * Get all registered function hook identifiers
3911          *
3912          * @return array
3913          */
3914         function getFunctionHooks() {
3915                 return array_keys( $this->mFunctionHooks );
3916         }
3917
3918         /**
3919          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3920          * Placeholders created in Skin::makeLinkObj()
3921          * Returns an array of link CSS classes, indexed by PDBK.
3922          * $options is a bit field, RLH_FOR_UPDATE to select for update
3923          */
3924         function replaceLinkHolders( &$text, $options = 0 ) {
3925                 global $wgUser;
3926                 global $wgContLang;
3927
3928                 $fname = 'Parser::replaceLinkHolders';
3929                 wfProfileIn( $fname );
3930
3931                 $pdbks = array();
3932                 $colours = array();
3933                 $linkcolour_ids = array();
3934                 $sk = $this->mOptions->getSkin();
3935                 $linkCache =& LinkCache::singleton();
3936
3937                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3938                         wfProfileIn( $fname.'-check' );
3939                         $dbr = wfGetDB( DB_SLAVE );
3940                         $page = $dbr->tableName( 'page' );
3941                         $threshold = $wgUser->getOption('stubthreshold');
3942
3943                         # Sort by namespace
3944                         asort( $this->mLinkHolders['namespaces'] );
3945
3946                         # Generate query
3947                         $query = false;
3948                         $current = null;
3949                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3950                                 # Make title object
3951                                 $title = $this->mLinkHolders['titles'][$key];
3952
3953                                 # Skip invalid entries.
3954                                 # Result will be ugly, but prevents crash.
3955                                 if ( is_null( $title ) ) {
3956                                         continue;
3957                                 }
3958                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3959
3960                                 # Check if it's a static known link, e.g. interwiki
3961                                 if ( $title->isAlwaysKnown() ) {
3962                                         $colours[$pdbk] = '';
3963                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3964                                         $colours[$pdbk] = '';
3965                                         $this->mOutput->addLink( $title, $id );
3966                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3967                                         $colours[$pdbk] = 'new';
3968                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
3969                                         $colours[$pdbk] = 'new';
3970                                 } else {
3971                                         # Not in the link cache, add it to the query
3972                                         if ( !isset( $current ) ) {
3973                                                 $current = $ns;
3974                                                 $query =  "SELECT page_id, page_namespace, page_title";
3975                                                 if ( $threshold > 0 ) {
3976                                                         $query .= ', page_len, page_is_redirect';
3977                                                 }
3978                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3979                                         } elseif ( $current != $ns ) {
3980                                                 $current = $ns;
3981                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3982                                         } else {
3983                                                 $query .= ', ';
3984                                         }
3985
3986                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3987                                 }
3988                         }
3989                         if ( $query ) {
3990                                 $query .= '))';
3991                                 if ( $options & RLH_FOR_UPDATE ) {
3992                                         $query .= ' FOR UPDATE';
3993                                 }
3994
3995                                 $res = $dbr->query( $query, $fname );
3996
3997                                 # Fetch data and form into an associative array
3998                                 # non-existent = broken
3999                                 while ( $s = $dbr->fetchObject($res) ) {
4000                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4001                                         $pdbk = $title->getPrefixedDBkey();
4002                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4003                                         $this->mOutput->addLink( $title, $s->page_id );
4004                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4005                                         //add id to the extension todolist
4006                                         $linkcolour_ids[$s->page_id] = $pdbk;
4007                                 }
4008                                 //pass an array of page_ids to an extension
4009                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4010                         }
4011                         wfProfileOut( $fname.'-check' );
4012
4013                         # Do a second query for different language variants of links and categories
4014                         if($wgContLang->hasVariants()){
4015                                 $linkBatch = new LinkBatch();
4016                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4017                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4018                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4019
4020                                 $categories = $this->mOutput->getCategoryLinks();
4021
4022                                 // Add variants of links to link batch
4023                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4024                                         $title = $this->mLinkHolders['titles'][$key];
4025                                         if ( is_null( $title ) )
4026                                                 continue;
4027
4028                                         $pdbk = $title->getPrefixedDBkey();
4029                                         $titleText = $title->getText();
4030
4031                                         // generate all variants of the link title text
4032                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4033
4034                                         // if link was not found (in first query), add all variants to query
4035                                         if ( !isset($colours[$pdbk]) ){
4036                                                 foreach($allTextVariants as $textVariant){
4037                                                         if($textVariant != $titleText){
4038                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4039                                                                 if(is_null($variantTitle)) continue;
4040                                                                 $linkBatch->addObj( $variantTitle );
4041                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4042                                                         }
4043                                                 }
4044                                         }
4045                                 }
4046
4047                                 // process categories, check if a category exists in some variant
4048                                 foreach( $categories as $category ){
4049                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4050                                         foreach($variants as $variant){
4051                                                 if($variant != $category){
4052                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4053                                                         if(is_null($variantTitle)) continue;
4054                                                         $linkBatch->addObj( $variantTitle );
4055                                                         $categoryMap[$variant] = $category;
4056                                                 }
4057                                         }
4058                                 }
4059
4060
4061                                 if(!$linkBatch->isEmpty()){
4062                                         // construct query
4063                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4064
4065                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4066                                         if ( $threshold > 0 ) {
4067                                                 $variantQuery .= ', page_len, page_is_redirect';
4068                                         }
4069
4070                                         $variantQuery .= " FROM $page WHERE $titleClause";
4071                                         if ( $options & RLH_FOR_UPDATE ) {
4072                                                 $variantQuery .= ' FOR UPDATE';
4073                                         }
4074
4075                                         $varRes = $dbr->query( $variantQuery, $fname );
4076
4077                                         // for each found variants, figure out link holders and replace
4078                                         while ( $s = $dbr->fetchObject($varRes) ) {
4079
4080                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4081                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4082                                                 $vardbk = $variantTitle->getDBkey();
4083
4084                                                 $holderKeys = array();
4085                                                 if(isset($variantMap[$varPdbk])){
4086                                                         $holderKeys = $variantMap[$varPdbk];
4087                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4088                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4089                                                 }
4090
4091                                                 // loop over link holders
4092                                                 foreach($holderKeys as $key){
4093                                                         $title = $this->mLinkHolders['titles'][$key];
4094                                                         if ( is_null( $title ) ) continue;
4095
4096                                                         $pdbk = $title->getPrefixedDBkey();
4097
4098                                                         if(!isset($colours[$pdbk])){
4099                                                                 // found link in some of the variants, replace the link holder data
4100                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4101                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4102
4103                                                                 // set pdbk and colour
4104                                                                 $pdbks[$key] = $varPdbk;
4105                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4106                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4107                                                         }
4108                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4109                                                 }
4110
4111                                                 // check if the object is a variant of a category
4112                                                 if(isset($categoryMap[$vardbk])){
4113                                                         $oldkey = $categoryMap[$vardbk];
4114                                                         if($oldkey != $vardbk)
4115                                                                 $varCategories[$oldkey]=$vardbk;
4116                                                 }
4117                                         }
4118
4119                                         // rebuild the categories in original order (if there are replacements)
4120                                         if(count($varCategories)>0){
4121                                                 $newCats = array();
4122                                                 $originalCats = $this->mOutput->getCategories();
4123                                                 foreach($originalCats as $cat => $sortkey){
4124                                                         // make the replacement
4125                                                         if( array_key_exists($cat,$varCategories) )
4126                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4127                                                         else $newCats[$cat] = $sortkey;
4128                                                 }
4129                                                 $this->mOutput->setCategoryLinks($newCats);
4130                                         }
4131                                 }
4132                         }
4133
4134                         # Construct search and replace arrays
4135                         wfProfileIn( $fname.'-construct' );
4136                         $replacePairs = array();
4137                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4138                                 $pdbk = $pdbks[$key];
4139                                 $searchkey = "<!--LINK $key-->";
4140                                 $title = $this->mLinkHolders['titles'][$key];
4141                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4142                                         $linkCache->addBadLinkObj( $title );
4143                                         $colours[$pdbk] = 'new';
4144                                         $this->mOutput->addLink( $title, 0 );
4145                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4146                                                                         $this->mLinkHolders['texts'][$key],
4147                                                                         $this->mLinkHolders['queries'][$key] );
4148                                 } else {
4149                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4150                                                                         $this->mLinkHolders['texts'][$key],
4151                                                                         $this->mLinkHolders['queries'][$key] );
4152                                 }
4153                         }
4154                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4155                         wfProfileOut( $fname.'-construct' );
4156
4157                         # Do the thing
4158                         wfProfileIn( $fname.'-replace' );
4159                         $text = preg_replace_callback(
4160                                 '/(<!--LINK .*?-->)/',
4161                                 $replacer->cb(),
4162                                 $text);
4163
4164                         wfProfileOut( $fname.'-replace' );
4165                 }
4166
4167                 # Now process interwiki link holders
4168                 # This is quite a bit simpler than internal links
4169                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4170                         wfProfileIn( $fname.'-interwiki' );
4171                         # Make interwiki link HTML
4172                         $replacePairs = array();
4173                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4174                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4175                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4176                         }
4177                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4178
4179                         $text = preg_replace_callback(
4180                                 '/<!--IWLINK (.*?)-->/',
4181                                 $replacer->cb(),
4182                                 $text );
4183                         wfProfileOut( $fname.'-interwiki' );
4184                 }
4185
4186                 wfProfileOut( $fname );
4187                 return $colours;
4188         }
4189
4190         /**
4191          * Replace <!--LINK--> link placeholders with plain text of links
4192          * (not HTML-formatted).
4193          * @param string $text
4194          * @return string
4195          */
4196         function replaceLinkHoldersText( $text ) {
4197                 $fname = 'Parser::replaceLinkHoldersText';
4198                 wfProfileIn( $fname );
4199
4200                 $text = preg_replace_callback(
4201                         '/<!--(LINK|IWLINK) (.*?)-->/',
4202                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4203                         $text );
4204
4205                 wfProfileOut( $fname );
4206                 return $text;
4207         }
4208
4209         /**
4210          * @param array $matches
4211          * @return string
4212          * @private
4213          */
4214         function replaceLinkHoldersTextCallback( $matches ) {
4215                 $type = $matches[1];
4216                 $key  = $matches[2];
4217                 if( $type == 'LINK' ) {
4218                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4219                                 return $this->mLinkHolders['texts'][$key];
4220                         }
4221                 } elseif( $type == 'IWLINK' ) {
4222                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4223                                 return $this->mInterwikiLinkHolders['texts'][$key];
4224                         }
4225                 }
4226                 return $matches[0];
4227         }
4228
4229         /**
4230          * Tag hook handler for 'pre'.
4231          */
4232         function renderPreTag( $text, $attribs ) {
4233                 // Backwards-compatibility hack
4234                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4235
4236                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4237                 return wfOpenElement( 'pre', $attribs ) .
4238                         Xml::escapeTagsOnly( $content ) .
4239                         '</pre>';
4240         }
4241
4242         /**
4243          * Renders an image gallery from a text with one line per image.
4244          * text labels may be given by using |-style alternative text. E.g.
4245          *   Image:one.jpg|The number "1"
4246          *   Image:tree.jpg|A tree
4247          * given as text will return the HTML of a gallery with two images,
4248          * labeled 'The number "1"' and
4249          * 'A tree'.
4250          */
4251         function renderImageGallery( $text, $params ) {
4252                 $ig = new ImageGallery();
4253                 $ig->setContextTitle( $this->mTitle );
4254                 $ig->setShowBytes( false );
4255                 $ig->setShowFilename( false );
4256                 $ig->setParser( $this );
4257                 $ig->setHideBadImages();
4258                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4259                 $ig->useSkin( $this->mOptions->getSkin() );
4260                 $ig->mRevisionId = $this->mRevisionId;
4261
4262                 if( isset( $params['caption'] ) ) {
4263                         $caption = $params['caption'];
4264                         $caption = htmlspecialchars( $caption );
4265                         $caption = $this->replaceInternalLinks( $caption );
4266                         $ig->setCaptionHtml( $caption );
4267                 }
4268                 if( isset( $params['perrow'] ) ) {
4269                         $ig->setPerRow( $params['perrow'] );
4270                 }
4271                 if( isset( $params['widths'] ) ) {
4272                         $ig->setWidths( $params['widths'] );
4273                 }
4274                 if( isset( $params['heights'] ) ) {
4275                         $ig->setHeights( $params['heights'] );
4276                 }
4277
4278                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4279
4280                 $lines = explode( "\n", $text );
4281                 foreach ( $lines as $line ) {
4282                         # match lines like these:
4283                         # Image:someimage.jpg|This is some image
4284                         $matches = array();
4285                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4286                         # Skip empty lines
4287                         if ( count( $matches ) == 0 ) {
4288                                 continue;
4289                         }
4290                         $tp = Title::newFromText( $matches[1] );
4291                         $nt =& $tp;
4292                         if( is_null( $nt ) ) {
4293                                 # Bogus title. Ignore these so we don't bomb out later.
4294                                 continue;
4295                         }
4296                         if ( isset( $matches[3] ) ) {
4297                                 $label = $matches[3];
4298                         } else {
4299                                 $label = '';
4300                         }
4301
4302                         $html = $this->recursiveTagParse( trim( $label ) );
4303
4304                         $ig->add( $nt, $html );
4305
4306                         # Only add real images (bug #5586)
4307                         if ( $nt->getNamespace() == NS_IMAGE ) {
4308                                 $this->mOutput->addImage( $nt->getDBkey() );
4309                         }
4310                 }
4311                 return $ig->toHTML();
4312         }
4313
4314         function getImageParams( $handler ) {
4315                 if ( $handler ) {
4316                         $handlerClass = get_class( $handler );
4317                 } else {
4318                         $handlerClass = '';
4319                 }
4320                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4321                         // Initialise static lists
4322                         static $internalParamNames = array(
4323                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4324                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4325                                         'bottom', 'text-bottom' ),
4326                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4327                                         'upright', 'border' ),
4328                         );
4329                         static $internalParamMap;
4330                         if ( !$internalParamMap ) {
4331                                 $internalParamMap = array();
4332                                 foreach ( $internalParamNames as $type => $names ) {
4333                                         foreach ( $names as $name ) {
4334                                                 $magicName = str_replace( '-', '_', "img_$name" );
4335                                                 $internalParamMap[$magicName] = array( $type, $name );
4336                                         }
4337                                 }
4338                         }
4339
4340                         // Add handler params
4341                         $paramMap = $internalParamMap;
4342                         if ( $handler ) {
4343                                 $handlerParamMap = $handler->getParamMap();
4344                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4345                                         $paramMap[$magic] = array( 'handler', $paramName );
4346                                 }
4347                         }
4348                         $this->mImageParams[$handlerClass] = $paramMap;
4349                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4350                 }
4351                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4352         }
4353
4354         /**
4355          * Parse image options text and use it to make an image
4356          */
4357         function makeImage( $title, $options ) {
4358                 # @TODO: let the MediaHandler specify its transform parameters
4359                 #
4360                 # Check if the options text is of the form "options|alt text"
4361                 # Options are:
4362                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4363                 #  * left               no resizing, just left align. label is used for alt= only
4364                 #  * right              same, but right aligned
4365                 #  * none               same, but not aligned
4366                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4367                 #  * center             center the image
4368                 #  * framed             Keep original image size, no magnify-button.
4369                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4370                 #  * upright            reduce width for upright images, rounded to full __0 px
4371                 #  * border             draw a 1px border around the image
4372                 # vertical-align values (no % or length right now):
4373                 #  * baseline
4374                 #  * sub
4375                 #  * super
4376                 #  * top
4377                 #  * text-top
4378                 #  * middle
4379                 #  * bottom
4380                 #  * text-bottom
4381
4382                 $parts = array_map( 'trim', explode( '|', $options) );
4383                 $sk = $this->mOptions->getSkin();
4384
4385                 # Give extensions a chance to select the file revision for us
4386                 $skip = $time = false;
4387                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4388
4389                 if ( $skip ) {
4390                         return $sk->makeLinkObj( $title );
4391                 }
4392
4393                 # Get parameter map
4394                 $file = wfFindFile( $title, $time );
4395                 $handler = $file ? $file->getHandler() : false;
4396
4397                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4398
4399                 # Process the input parameters
4400                 $caption = '';
4401                 $params = array( 'frame' => array(), 'handler' => array(),
4402                         'horizAlign' => array(), 'vertAlign' => array() );
4403                 foreach( $parts as $part ) {
4404                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4405                         if ( isset( $paramMap[$magicName] ) ) {
4406                                 list( $type, $paramName ) = $paramMap[$magicName];
4407                                 $params[$type][$paramName] = $value;
4408
4409                                 // Special case; width and height come in one variable together
4410                                 if( $type == 'handler' && $paramName == 'width' ) {
4411                                         $m = array();
4412                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4413                                                 $params[$type]['width'] = intval( $m[1] );
4414                                                 $params[$type]['height'] = intval( $m[2] );
4415                                         } else {
4416                                                 $params[$type]['width'] = intval( $value );
4417                                         }
4418                                 }
4419                         } else {
4420                                 $caption = $part;
4421                         }
4422                 }
4423
4424                 # Process alignment parameters
4425                 if ( $params['horizAlign'] ) {
4426                         $params['frame']['align'] = key( $params['horizAlign'] );
4427                 }
4428                 if ( $params['vertAlign'] ) {
4429                         $params['frame']['valign'] = key( $params['vertAlign'] );
4430                 }
4431
4432                 # Validate the handler parameters
4433                 if ( $handler ) {
4434                         foreach ( $params['handler'] as $name => $value ) {
4435                                 if ( !$handler->validateParam( $name, $value ) ) {
4436                                         unset( $params['handler'][$name] );
4437                                 }
4438                         }
4439                 }
4440
4441                 # Strip bad stuff out of the alt text
4442                 $alt = $this->replaceLinkHoldersText( $caption );
4443
4444                 # make sure there are no placeholders in thumbnail attributes
4445                 # that are later expanded to html- so expand them now and
4446                 # remove the tags
4447                 $alt = $this->mStripState->unstripBoth( $alt );
4448                 $alt = Sanitizer::stripAllTags( $alt );
4449
4450                 $params['frame']['alt'] = $alt;
4451                 $params['frame']['caption'] = $caption;
4452
4453                 # Linker does the rest
4454                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4455
4456                 # Give the handler a chance to modify the parser object
4457                 if ( $handler ) {
4458                         $handler->parserTransformHook( $this, $file );
4459                 }
4460
4461                 return $ret;
4462         }
4463
4464         /**
4465          * Set a flag in the output object indicating that the content is dynamic and
4466          * shouldn't be cached.
4467          */
4468         function disableCache() {
4469                 wfDebug( "Parser output marked as uncacheable.\n" );
4470                 $this->mOutput->mCacheTime = -1;
4471         }
4472
4473         /**#@+
4474          * Callback from the Sanitizer for expanding items found in HTML attribute
4475          * values, so they can be safely tested and escaped.
4476          * @param string $text
4477          * @param PPFrame $frame
4478          * @return string
4479          * @private
4480          */
4481         function attributeStripCallback( &$text, $frame = false ) {
4482                 $text = $this->replaceVariables( $text, $frame );
4483                 $text = $this->mStripState->unstripBoth( $text );
4484                 return $text;
4485         }
4486
4487         /**#@-*/
4488
4489         /**#@+
4490          * Accessor/mutator
4491          */
4492         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4493         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4494         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4495         /**#@-*/
4496
4497         /**#@+
4498          * Accessor
4499          */
4500         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4501         /**#@-*/
4502
4503
4504         /**
4505          * Break wikitext input into sections, and either pull or replace
4506          * some particular section's text.
4507          *
4508          * External callers should use the getSection and replaceSection methods.
4509          *
4510          * @param string $text Page wikitext
4511          * @param string $section A section identifier string of the form:
4512          *   <flag1> - <flag2> - ... - <section number>
4513          *
4514          * Currently the only recognised flag is "T", which means the target section number
4515          * was derived during a template inclusion parse, in other words this is a template
4516          * section edit link. If no flags are given, it was an ordinary section edit link.
4517          * This flag is required to avoid a section numbering mismatch when a section is
4518          * enclosed by <includeonly> (bug 6563).
4519          *
4520          * The section number 0 pulls the text before the first heading; other numbers will
4521          * pull the given section along with its lower-level subsections. If the section is
4522          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4523          *
4524          * @param string $mode One of "get" or "replace"
4525          * @param string $newText Replacement text for section data.
4526          * @return string for "get", the extracted section text.
4527          *                for "replace", the whole page with the section replaced.
4528          */
4529         private function extractSections( $text, $section, $mode, $newText='' ) {
4530                 global $wgTitle;
4531                 $this->clearState();
4532                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4533                 $this->mOptions = new ParserOptions;
4534                 $this->setOutputType( OT_WIKI );
4535                 $curIndex = 0;
4536                 $outText = '';
4537                 $frame = $this->getPreprocessor()->newFrame();
4538
4539                 // Process section extraction flags
4540                 $flags = 0;
4541                 $sectionParts = explode( '-', $section );
4542                 $sectionIndex = array_pop( $sectionParts );
4543                 foreach ( $sectionParts as $part ) {
4544                         if ( $part == 'T' ) {
4545                                 $flags |= self::PTD_FOR_INCLUSION;
4546                         }
4547                 }
4548                 // Preprocess the text
4549                 $root = $this->preprocessToDom( $text, $flags );
4550
4551                 // <h> nodes indicate section breaks
4552                 // They can only occur at the top level, so we can find them by iterating the root's children
4553                 $node = $root->getFirstChild();
4554
4555                 // Find the target section
4556                 if ( $sectionIndex == 0 ) {
4557                         // Section zero doesn't nest, level=big
4558                         $targetLevel = 1000;
4559                 } else {
4560                         while ( $node ) {
4561                                 if ( $node->getName() == 'h' ) {
4562                                         if ( $curIndex + 1 == $sectionIndex ) {
4563                                                 break;
4564                                         }
4565                                         $curIndex++;
4566                                 }
4567                                 if ( $mode == 'replace' ) {
4568                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4569                                 }
4570                                 $node = $node->getNextSibling();
4571                         }
4572                         if ( $node ) {
4573                                 $bits = $node->splitHeading();
4574                                 $targetLevel = $bits['level'];
4575                         }
4576                 }
4577
4578                 if ( !$node ) {
4579                         // Not found
4580                         if ( $mode == 'get' ) {
4581                                 return $newText;
4582                         } else {
4583                                 return $text;
4584                         }
4585                 }
4586
4587                 // Find the end of the section, including nested sections
4588                 do {
4589                         if ( $node->getName() == 'h' ) {
4590                                 $curIndex++;
4591                                 $bits = $node->splitHeading();
4592                                 $curLevel = $bits['level'];
4593                                 if ( $curIndex != $sectionIndex && $curLevel <= $targetLevel ) {
4594                                         break;
4595                                 }
4596                         }
4597                         if ( $mode == 'get' ) {
4598                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4599                         }
4600                         $node = $node->getNextSibling();
4601                 } while ( $node );
4602
4603                 // Write out the remainder (in replace mode only)
4604                 if ( $mode == 'replace' ) {
4605                         // Output the replacement text
4606                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4607                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4608                         $outText .= $newText . "\n\n";
4609                         while ( $node ) {
4610                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4611                                 $node = $node->getNextSibling();
4612                         }
4613                 }
4614
4615                 if ( is_string( $outText ) ) {
4616                         // Re-insert stripped tags
4617                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
4618                 }
4619
4620                 return $outText;
4621         }
4622
4623         /**
4624          * This function returns the text of a section, specified by a number ($section).
4625          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4626          * the first section before any such heading (section 0).
4627          *
4628          * If a section contains subsections, these are also returned.
4629          *
4630          * @param string $text text to look in
4631          * @param string $section section identifier
4632          * @param string $deftext default to return if section is not found
4633          * @return string text of the requested section
4634          */
4635         public function getSection( $text, $section, $deftext='' ) {
4636                 return $this->extractSections( $text, $section, "get", $deftext );
4637         }
4638
4639         public function replaceSection( $oldtext, $section, $text ) {
4640                 return $this->extractSections( $oldtext, $section, "replace", $text );
4641         }
4642
4643         /**
4644          * Get the timestamp associated with the current revision, adjusted for
4645          * the default server-local timestamp
4646          */
4647         function getRevisionTimestamp() {
4648                 if ( is_null( $this->mRevisionTimestamp ) ) {
4649                         wfProfileIn( __METHOD__ );
4650                         global $wgContLang;
4651                         $dbr = wfGetDB( DB_SLAVE );
4652                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4653                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4654
4655                         // Normalize timestamp to internal MW format for timezone processing.
4656                         // This has the added side-effect of replacing a null value with
4657                         // the current time, which gives us more sensible behavior for
4658                         // previews.
4659                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4660
4661                         // The cryptic '' timezone parameter tells to use the site-default
4662                         // timezone offset instead of the user settings.
4663                         //
4664                         // Since this value will be saved into the parser cache, served
4665                         // to other users, and potentially even used inside links and such,
4666                         // it needs to be consistent for all visitors.
4667                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4668
4669                         wfProfileOut( __METHOD__ );
4670                 }
4671                 return $this->mRevisionTimestamp;
4672         }
4673
4674         /**
4675          * Mutator for $mDefaultSort
4676          *
4677          * @param $sort New value
4678          */
4679         public function setDefaultSort( $sort ) {
4680                 $this->mDefaultSort = $sort;
4681         }
4682
4683         /**
4684          * Accessor for $mDefaultSort
4685          * Will use the title/prefixed title if none is set
4686          *
4687          * @return string
4688          */
4689         public function getDefaultSort() {
4690                 if( $this->mDefaultSort !== false ) {
4691                         return $this->mDefaultSort;
4692                 } else {
4693                         return $this->mTitle->getNamespace() == NS_CATEGORY
4694                                         ? $this->mTitle->getText()
4695                                         : $this->mTitle->getPrefixedText();
4696                 }
4697         }
4698
4699         /**
4700          * Try to guess the section anchor name based on a wikitext fragment
4701          * presumably extracted from a heading, for example "Header" from
4702          * "== Header ==".
4703          */
4704         public function guessSectionNameFromWikiText( $text ) {
4705                 # Strip out wikitext links(they break the anchor)
4706                 $text = $this->stripSectionName( $text );
4707                 $headline = Sanitizer::decodeCharReferences( $text );
4708                 # strip out HTML
4709                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4710                 $headline = trim( $headline );
4711                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4712                 $replacearray = array(
4713                         '%3A' => ':',
4714                         '%' => '.'
4715                 );
4716                 return str_replace(
4717                         array_keys( $replacearray ),
4718                         array_values( $replacearray ),
4719                         $sectionanchor );
4720         }
4721
4722         /**
4723          * Strips a text string of wikitext for use in a section anchor
4724          *
4725          * Accepts a text string and then removes all wikitext from the
4726          * string and leaves only the resultant text (i.e. the result of
4727          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4728          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4729          * to create valid section anchors by mimicing the output of the
4730          * parser when headings are parsed.
4731          *
4732          * @param $text string Text string to be stripped of wikitext
4733          * for use in a Section anchor
4734          * @return Filtered text string
4735          */
4736         public function stripSectionName( $text ) {
4737                 # Strip internal link markup
4738                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4739                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4740
4741                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4742                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4743                 # on how many empty links there are on the page - need to figure that out.
4744                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4745
4746                 # Parse wikitext quotes (italics & bold)
4747                 $text = $this->doQuotes($text);
4748
4749                 # Strip HTML tags
4750                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4751                 return $text;
4752         }
4753
4754         function srvus( $text ) {
4755                 return $this->testSrvus( $text, $this->mOutputType );
4756         }
4757
4758         /**
4759          * strip/replaceVariables/unstrip for preprocessor regression testing
4760          */
4761         function testSrvus( $text, $title, $options, $outputType = OT_HTML ) {
4762                 $this->clearState();
4763                 $this->mTitle = $title;
4764                 $this->mOptions = $options;
4765                 $this->setOutputType( $outputType );
4766                 $text = $this->replaceVariables( $text );
4767                 $text = $this->mStripState->unstripBoth( $text );
4768                 $text = Sanitizer::removeHTMLtags( $text );
4769                 return $text;
4770         }
4771
4772         function testPst( $text, $title, $options ) {
4773                 global $wgUser;
4774                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4775         }
4776
4777         function testPreprocess( $text, $title, $options ) {
4778                 return $this->testSrvus( $text, $title, $options, OT_PREPROCESS );
4779         }
4780 }
4781
4782 /**
4783  * @todo document, briefly.
4784  * @addtogroup Parser
4785  */
4786 class StripState {
4787         var $general, $nowiki;
4788
4789         function __construct() {
4790                 $this->general = new ReplacementArray;
4791                 $this->nowiki = new ReplacementArray;
4792         }
4793
4794         function unstripGeneral( $text ) {
4795                 wfProfileIn( __METHOD__ );
4796                 do {
4797                         $oldText = $text;
4798                         $text = $this->general->replace( $text );
4799                 } while ( $text != $oldText );
4800                 wfProfileOut( __METHOD__ );
4801                 return $text;
4802         }
4803
4804         function unstripNoWiki( $text ) {
4805                 wfProfileIn( __METHOD__ );
4806                 do {
4807                         $oldText = $text;
4808                         $text = $this->nowiki->replace( $text );
4809                 } while ( $text != $oldText );
4810                 wfProfileOut( __METHOD__ );
4811                 return $text;
4812         }
4813
4814         function unstripBoth( $text ) {
4815                 wfProfileIn( __METHOD__ );
4816                 do {
4817                         $oldText = $text;
4818                         $text = $this->general->replace( $text );
4819                         $text = $this->nowiki->replace( $text );
4820                 } while ( $text != $oldText );
4821                 wfProfileOut( __METHOD__ );
4822                 return $text;
4823         }
4824 }
4825
4826 /**
4827  * @todo document, briefly.
4828  * @addtogroup Parser
4829  */
4830 class OnlyIncludeReplacer {
4831         var $output = '';
4832
4833         function replace( $matches ) {
4834                 if ( substr( $matches[1], -1 ) == "\n" ) {
4835                         $this->output .= substr( $matches[1], 0, -1 );
4836                 } else {
4837                         $this->output .= $matches[1];
4838                 }
4839         }
4840 }
4841