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 four main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * transformMsg()
  24  *   performs brace substitution on MediaWiki messages
  25  * preprocess()
  26  *   removes HTML comments and expands templates
  27  *
  28  * Globals used:
  29  *    objects:   $wgLang, $wgContLang
  30  *
  31  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  32  *
  33  * settings:
  34  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  35  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  36  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  37  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.2';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  62                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  63
  64         // State constants for the definition list colon extraction
  65         const COLON_STATE_TEXT = 0;
  66         const COLON_STATE_TAG = 1;
  67         const COLON_STATE_TAGSTART = 2;
  68         const COLON_STATE_CLOSETAG = 3;
  69         const COLON_STATE_TAGSLASH = 4;
  70         const COLON_STATE_COMMENT = 5;
  71         const COLON_STATE_COMMENTDASH = 6;
  72         const COLON_STATE_COMMENTDASHDASH = 7;
  73
  74         /**#@+
  75          * @private
  76          */
  77         # Persistent:
  78         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  79                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  80                 $mExtLinkBracketedRegex;
  81
  82         # Cleared with clearState():
  83         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  84         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  85         var $mInterwikiLinkHolders, $mLinkHolders;
  86         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  87         var $mTplExpandCache,// empty-frame expansion cache
  88             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  89                                 // in this path. Used for loop detection.
  90         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  91
  92         # Temporary
  93         # These are variables reset at least once per parse regardless of $clearState
  94         var $mOptions,      // ParserOptions object
  95                 $mTitle,        // Title context, used for self-link rendering and similar things
  96                 $mOutputType,   // Output type, one of the OT_xxx constants
  97                 $ot,            // Shortcut alias, see setOutputType()
  98                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  99                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 100                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 101
 102         /**#@-*/
 103
 104         /**
 105          * Constructor
 106          *
 107          * @public
 108          */
 109         function __construct( $conf = array() ) {
 110                 $this->mTagHooks = array();
 111                 $this->mTransparentTagHooks = array();
 112                 $this->mFunctionHooks = array();
 113                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 114                 $this->mStripList = array( 'nowiki', 'gallery' );
 115                 $this->mMarkerSuffix = "-QINU\x7f";
 116                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 117                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 118                 $this->mFirstCall = true;
 119         }
 120
 121         /**
 122          * Do various kinds of initialisation on the first call of the parser
 123          */
 124         function firstCallInit() {
 125                 if ( !$this->mFirstCall ) {
 126                         return;
 127                 }
 128
 129                 wfProfileIn( __METHOD__ );
 130                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 131
 132                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 133
 134                 # Syntax for arguments (see self::setFunctionHook):
 135                 #  "name for lookup in localized magic words array",
 136                 #  function callback,
 137                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 138                 #    instead of {{#int:...}})
 139                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 140                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 141                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 142                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 164                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 166
 167                 if ( $wgAllowDisplayTitle ) {
 168                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 169                 }
 170                 if ( $wgAllowSlowParserFunctions ) {
 171                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 172                 }
 173
 174                 $this->initialiseVariables();
 175                 $this->mFirstCall = false;
 176                 wfProfileOut( __METHOD__ );
 177         }
 178
 179         /**
 180          * Clear Parser state
 181          *
 182          * @private
 183          */
 184         function clearState() {
 185                 wfProfileIn( __METHOD__ );
 186                 if ( $this->mFirstCall ) {
 187                         $this->firstCallInit();
 188                 }
 189                 $this->mOutput = new ParserOutput;
 190                 $this->mAutonumber = 0;
 191                 $this->mLastSection = '';
 192                 $this->mDTopen = false;
 193                 $this->mIncludeCount = array();
 194                 $this->mStripState = new StripState;
 195                 $this->mArgStack = false;
 196                 $this->mInPre = false;
 197                 $this->mInterwikiLinkHolders = array(
 198                         'texts' => array(),
 199                         'titles' => array()
 200                 );
 201                 $this->mLinkHolders = array(
 202                         'namespaces' => array(),
 203                         'dbkeys' => array(),
 204                         'queries' => array(),
 205                         'texts' => array(),
 206                         'titles' => array()
 207                 );
 208                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 209
 210                 /**
 211                  * Prefix for temporary replacement strings for the multipass parser.
 212                  * \x07 should never appear in input as it's disallowed in XML.
 213                  * Using it at the front also gives us a little extra robustness
 214                  * since it shouldn't match when butted up against identifier-like
 215                  * string constructs.
 216                  *
 217                  * Must not consist of all title characters, or else it will change
 218                  * the behaviour of <nowiki> in a link.
 219                  */
 220                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 221                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTemplatePath = array();
 225                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'post-expand' => 0,
 231                         'arg' => 0,
 232                 );
 233                 $this->mPPNodeCount = 0;
 234                 $this->mDefaultSort = false;
 235                 $this->mHeadings = array();
 236
 237                 wfRunHooks( 'ParserClearState', array( &$this ) );
 238                 wfProfileOut( __METHOD__ );
 239         }
 240
 241         function setOutputType( $ot ) {
 242                 $this->mOutputType = $ot;
 243                 // Shortcut alias
 244                 $this->ot = array(
 245                         'html' => $ot == OT_HTML,
 246                         'wiki' => $ot == OT_WIKI,
 247                         'msg' => $ot == OT_MSG,
 248                         'pre' => $ot == OT_PREPROCESS,
 249                 );
 250         }
 251
 252         /**
 253          * Accessor for mUniqPrefix.
 254          *
 255          * @public
 256          */
 257         function uniqPrefix() {
 258                 return $this->mUniqPrefix;
 259         }
 260
 261         /**
 262          * Convert wikitext to HTML
 263          * Do not call this function recursively.
 264          *
 265          * @param string $text Text we want to parse
 266          * @param Title &$title A title object
 267          * @param array $options
 268          * @param boolean $linestart
 269          * @param boolean $clearState
 270          * @param int $revid number to pass in {{REVISIONID}}
 271          * @return ParserOutput a ParserOutput
 272          */
 273         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 274                 /**
 275                  * First pass--just handle <nowiki> sections, pass the rest off
 276                  * to internalParse() which does all the real work.
 277                  */
 278
 279                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 280                 $fname = 'Parser::parse-' . wfGetCaller();
 281                 wfProfileIn( __METHOD__ );
 282                 wfProfileIn( $fname );
 283
 284                 if ( $clearState ) {
 285                         $this->clearState();
 286                 }
 287
 288                 $this->mOptions = $options;
 289                 $this->mTitle =& $title;
 290                 $oldRevisionId = $this->mRevisionId;
 291                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 292                 if( $revid !== null ) {
 293                         $this->mRevisionId = $revid;
 294                         $this->mRevisionTimestamp = null;
 295                 }
 296                 $this->setOutputType( OT_HTML );
 297                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 298                 # No more strip!
 299                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 300                 $text = $this->internalParse( $text );
 301                 $text = $this->mStripState->unstripGeneral( $text );
 302
 303                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 304                 $fixtags = array(
 305                         # french spaces, last one Guillemet-left
 306                         # only if there is something before the space
 307                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 308                         # french spaces, Guillemet-right
 309                         '/(\\302\\253) /' => '\\1&nbsp;',
 310                 );
 311                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 312
 313                 # only once and last
 314                 $text = $this->doBlockLevels( $text, $linestart );
 315
 316                 $this->replaceLinkHolders( $text );
 317
 318                 # the position of the parserConvert() call should not be changed. it
 319                 # assumes that the links are all replaced and the only thing left
 320                 # is the <nowiki> mark.
 321                 # Side-effects: this calls $this->mOutput->setTitleText()
 322                 $text = $wgContLang->parserConvert( $text, $this );
 323
 324                 $text = $this->mStripState->unstripNoWiki( $text );
 325
 326                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 327
 328 //!JF Move to its own function
 329
 330                 $uniq_prefix = $this->mUniqPrefix;
 331                 $matches = array();
 332                 $elements = array_keys( $this->mTransparentTagHooks );
 333                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 334
 335                 foreach( $matches as $marker => $data ) {
 336                         list( $element, $content, $params, $tag ) = $data;
 337                         $tagName = strtolower( $element );
 338                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 339                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 340                                         array( $content, $params, $this ) );
 341                         } else {
 342                                 $output = $tag;
 343                         }
 344                         $this->mStripState->general->setPair( $marker, $output );
 345                 }
 346                 $text = $this->mStripState->unstripGeneral( $text );
 347
 348                 $text = Sanitizer::normalizeCharReferences( $text );
 349
 350                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 351                         $text = Parser::tidy($text);
 352                 } else {
 353                         # attempt to sanitize at least some nesting problems
 354                         # (bug #2702 and quite a few others)
 355                         $tidyregs = array(
 356                                 # ''Something [http://www.cool.com cool''] -->
 357                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 358                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 359                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 360                                 # fix up an anchor inside another anchor, only
 361                                 # at least for a single single nested link (bug 3695)
 362                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 363                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 364                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 365                                 # contains a div, so fix it up here; replace
 366                                 # div with escaped text
 367                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 368                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 369                                 # remove empty italic or bold tag pairs, some
 370                                 # introduced by rules above
 371                                 '/<([bi])><\/\\1>/' => '',
 372                         );
 373
 374                         $text = preg_replace(
 375                                 array_keys( $tidyregs ),
 376                                 array_values( $tidyregs ),
 377                                 $text );
 378                 }
 379
 380                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 381
 382                 # Information on include size limits, for the benefit of users who try to skirt them
 383                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 384                         $max = $this->mOptions->getMaxIncludeSize();
 385                         $text .= "<!-- \n" .
 386                                 "Preprocessor node count: {$this->mPPNodeCount}\n" .
 387                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 388                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 389                                 "Maximum: $max bytes\n" .
 390                                 "-->\n";
 391                 }
 392                 $this->mOutput->setText( $text );
 393                 $this->mRevisionId = $oldRevisionId;
 394                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 395                 wfProfileOut( $fname );
 396                 wfProfileOut( __METHOD__ );
 397
 398                 return $this->mOutput;
 399         }
 400
 401         /**
 402          * Recursive parser entry point that can be called from an extension tag
 403          * hook.
 404          */
 405         function recursiveTagParse( $text ) {
 406                 wfProfileIn( __METHOD__ );
 407                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 408                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 409                 $text = $this->internalParse( $text );
 410                 wfProfileOut( __METHOD__ );
 411                 return $text;
 412         }
 413
 414         /**
 415          * Expand templates and variables in the text, producing valid, static wikitext.
 416          * Also removes comments.
 417          */
 418         function preprocess( $text, $title, $options, $revid = null ) {
 419                 wfProfileIn( __METHOD__ );
 420                 $this->clearState();
 421                 $this->setOutputType( OT_PREPROCESS );
 422                 $this->mOptions = $options;
 423                 $this->mTitle = $title;
 424                 if( $revid !== null ) {
 425                         $this->mRevisionId = $revid;
 426                 }
 427                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 428                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 429                 $text = $this->replaceVariables( $text );
 430                 if ( $this->mOptions->getRemoveComments() ) {
 431                         $text = Sanitizer::removeHTMLcomments( $text );
 432                 }
 433                 $text = $this->mStripState->unstripBoth( $text );
 434                 wfProfileOut( __METHOD__ );
 435                 return $text;
 436         }
 437
 438         /**
 439          * Get a random string
 440          *
 441          * @private
 442          * @static
 443          */
 444         function getRandomString() {
 445                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 446         }
 447
 448         function &getTitle() { return $this->mTitle; }
 449         function getOptions() { return $this->mOptions; }
 450
 451         function getFunctionLang() {
 452                 global $wgLang, $wgContLang;
 453                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 454         }
 455
 456         /**
 457          * Replaces all occurrences of HTML-style comments and the given tags
 458          * in the text with a random marker and returns teh next text. The output
 459          * parameter $matches will be an associative array filled with data in
 460          * the form:
 461          *   'UNIQ-xxxxx' => array(
 462          *     'element',
 463          *     'tag content',
 464          *     array( 'param' => 'x' ),
 465          *     '<element param="x">tag content</element>' ) )
 466          *
 467          * @param $elements list of element names. Comments are always extracted.
 468          * @param $text Source text string.
 469          * @param $uniq_prefix
 470          *
 471          * @public
 472          * @static
 473          */
 474         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 475                 static $n = 1;
 476                 $stripped = '';
 477                 $matches = array();
 478
 479                 $taglist = implode( '|', $elements );
 480                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 481
 482                 while ( '' != $text ) {
 483                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 484                         $stripped .= $p[0];
 485                         if( count( $p ) < 5 ) {
 486                                 break;
 487                         }
 488                         if( count( $p ) > 5 ) {
 489                                 // comment
 490                                 $element    = $p[4];
 491                                 $attributes = '';
 492                                 $close      = '';
 493                                 $inside     = $p[5];
 494                         } else {
 495                                 // tag
 496                                 $element    = $p[1];
 497                                 $attributes = $p[2];
 498                                 $close      = $p[3];
 499                                 $inside     = $p[4];
 500                         }
 501
 502                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 503                         $stripped .= $marker;
 504
 505                         if ( $close === '/>' ) {
 506                                 // Empty element tag, <tag />
 507                                 $content = null;
 508                                 $text = $inside;
 509                                 $tail = null;
 510                         } else {
 511                                 if( $element == '!--' ) {
 512                                         $end = '/(-->)/';
 513                                 } else {
 514                                         $end = "/(<\\/$element\\s*>)/i";
 515                                 }
 516                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 517                                 $content = $q[0];
 518                                 if( count( $q ) < 3 ) {
 519                                         # No end tag -- let it run out to the end of the text.
 520                                         $tail = '';
 521                                         $text = '';
 522                                 } else {
 523                                         $tail = $q[1];
 524                                         $text = $q[2];
 525                                 }
 526                         }
 527
 528                         $matches[$marker] = array( $element,
 529                                 $content,
 530                                 Sanitizer::decodeTagAttributes( $attributes ),
 531                                 "<$element$attributes$close$content$tail" );
 532                 }
 533                 return $stripped;
 534         }
 535
 536         /**
 537          * Get a list of strippable XML-like elements
 538          */
 539         function getStripList() {
 540                 global $wgRawHtml;
 541                 $elements = $this->mStripList;
 542                 if( $wgRawHtml ) {
 543                         $elements[] = 'html';
 544                 }
 545                 if( $this->mOptions->getUseTeX() ) {
 546                         $elements[] = 'math';
 547                 }
 548                 return $elements;
 549         }
 550
 551         /**
 552          * @deprecated use replaceVariables
 553          */
 554         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 555                 return $text;
 556         }
 557
 558         /**
 559          * Restores pre, math, and other extensions removed by strip()
 560          *
 561          * always call unstripNoWiki() after this one
 562          * @private
 563          * @deprecated use $this->mStripState->unstrip()
 564          */
 565         function unstrip( $text, $state ) {
 566                 return $state->unstripGeneral( $text );
 567         }
 568
 569         /**
 570          * Always call this after unstrip() to preserve the order
 571          *
 572          * @private
 573          * @deprecated use $this->mStripState->unstrip()
 574          */
 575         function unstripNoWiki( $text, $state ) {
 576                 return $state->unstripNoWiki( $text );
 577         }
 578
 579         /**
 580          * @deprecated use $this->mStripState->unstripBoth()
 581          */
 582         function unstripForHTML( $text ) {
 583                 return $this->mStripState->unstripBoth( $text );
 584         }
 585
 586         /**
 587          * Add an item to the strip state
 588          * Returns the unique tag which must be inserted into the stripped text
 589          * The tag will be replaced with the original text in unstrip()
 590          *
 591          * @private
 592          */
 593         function insertStripItem( $text ) {
 594                 static $n = 0;
 595                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 596                 $this->mStripState->general->setPair( $rnd, $text );
 597                 return $rnd;
 598         }
 599
 600         /**
 601          * Interface with html tidy, used if $wgUseTidy = true.
 602          * If tidy isn't able to correct the markup, the original will be
 603          * returned in all its glory with a warning comment appended.
 604          *
 605          * Either the external tidy program or the in-process tidy extension
 606          * will be used depending on availability. Override the default
 607          * $wgTidyInternal setting to disable the internal if it's not working.
 608          *
 609          * @param string $text Hideous HTML input
 610          * @return string Corrected HTML output
 611          * @public
 612          * @static
 613          */
 614         function tidy( $text ) {
 615                 global $wgTidyInternal;
 616                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 617 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 618 '<head><title>test</title></head><body>'.$text.'</body></html>';
 619                 if( $wgTidyInternal ) {
 620                         $correctedtext = Parser::internalTidy( $wrappedtext );
 621                 } else {
 622                         $correctedtext = Parser::externalTidy( $wrappedtext );
 623                 }
 624                 if( is_null( $correctedtext ) ) {
 625                         wfDebug( "Tidy error detected!\n" );
 626                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 627                 }
 628                 return $correctedtext;
 629         }
 630
 631         /**
 632          * Spawn an external HTML tidy process and get corrected markup back from it.
 633          *
 634          * @private
 635          * @static
 636          */
 637         function externalTidy( $text ) {
 638                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 639                 $fname = 'Parser::externalTidy';
 640                 wfProfileIn( $fname );
 641
 642                 $cleansource = '';
 643                 $opts = ' -utf8';
 644
 645                 $descriptorspec = array(
 646                         0 => array('pipe', 'r'),
 647                         1 => array('pipe', 'w'),
 648                         2 => array('file', wfGetNull(), 'a')
 649                 );
 650                 $pipes = array();
 651                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 652                 if (is_resource($process)) {
 653                         // Theoretically, this style of communication could cause a deadlock
 654                         // here. If the stdout buffer fills up, then writes to stdin could
 655                         // block. This doesn't appear to happen with tidy, because tidy only
 656                         // writes to stdout after it's finished reading from stdin. Search
 657                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 658                         fwrite($pipes[0], $text);
 659                         fclose($pipes[0]);
 660                         while (!feof($pipes[1])) {
 661                                 $cleansource .= fgets($pipes[1], 1024);
 662                         }
 663                         fclose($pipes[1]);
 664                         proc_close($process);
 665                 }
 666
 667                 wfProfileOut( $fname );
 668
 669                 if( $cleansource == '' && $text != '') {
 670                         // Some kind of error happened, so we couldn't get the corrected text.
 671                         // Just give up; we'll use the source text and append a warning.
 672                         return null;
 673                 } else {
 674                         return $cleansource;
 675                 }
 676         }
 677
 678         /**
 679          * Use the HTML tidy PECL extension to use the tidy library in-process,
 680          * saving the overhead of spawning a new process.
 681          *
 682          * 'pear install tidy' should be able to compile the extension module.
 683          *
 684          * @private
 685          * @static
 686          */
 687         function internalTidy( $text ) {
 688                 global $wgTidyConf, $IP, $wgDebugTidy;
 689                 $fname = 'Parser::internalTidy';
 690                 wfProfileIn( $fname );
 691
 692                 $tidy = new tidy;
 693                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 694                 $tidy->cleanRepair();
 695                 if( $tidy->getStatus() == 2 ) {
 696                         // 2 is magic number for fatal error
 697                         // http://www.php.net/manual/en/function.tidy-get-status.php
 698                         $cleansource = null;
 699                 } else {
 700                         $cleansource = tidy_get_output( $tidy );
 701                 }
 702                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 703                         $cleansource .= "<!--\nTidy reports:\n" .
 704                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 705                                 "\n-->";
 706                 }
 707
 708                 wfProfileOut( $fname );
 709                 return $cleansource;
 710         }
 711
 712         /**
 713          * parse the wiki syntax used to render tables
 714          *
 715          * @private
 716          */
 717         function doTableStuff ( $text ) {
 718                 $fname = 'Parser::doTableStuff';
 719                 wfProfileIn( $fname );
 720
 721                 $lines = explode ( "\n" , $text );
 722                 $td_history = array (); // Is currently a td tag open?
 723                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 724                 $tr_history = array (); // Is currently a tr tag open?
 725                 $tr_attributes = array (); // history of tr attributes
 726                 $has_opened_tr = array(); // Did this table open a <tr> element?
 727                 $indent_level = 0; // indent level of the table
 728                 foreach ( $lines as $key => $line )
 729                 {
 730                         $line = trim ( $line );
 731
 732                         if( $line == '' ) { // empty line, go to next line
 733                                 continue;
 734                         }
 735                         $first_character = $line{0};
 736                         $matches = array();
 737
 738                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 739                                 // First check if we are starting a new table
 740                                 $indent_level = strlen( $matches[1] );
 741
 742                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 743                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 744
 745                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 746                                 array_push ( $td_history , false );
 747                                 array_push ( $last_tag_history , '' );
 748                                 array_push ( $tr_history , false );
 749                                 array_push ( $tr_attributes , '' );
 750                                 array_push ( $has_opened_tr , false );
 751                         } else if ( count ( $td_history ) == 0 ) {
 752                                 // Don't do any of the following
 753                                 continue;
 754                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 755                                 // We are ending a table
 756                                 $line = '</table>' . substr ( $line , 2 );
 757                                 $last_tag = array_pop ( $last_tag_history );
 758
 759                                 if ( !array_pop ( $has_opened_tr ) ) {
 760                                         $line = "<tr><td></td></tr>{$line}";
 761                                 }
 762
 763                                 if ( array_pop ( $tr_history ) ) {
 764                                         $line = "</tr>{$line}";
 765                                 }
 766
 767                                 if ( array_pop ( $td_history ) ) {
 768                                         $line = "</{$last_tag}>{$line}";
 769                                 }
 770                                 array_pop ( $tr_attributes );
 771                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 772                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 773                                 // Now we have a table row
 774                                 $line = preg_replace( '#^\|-+#', '', $line );
 775
 776                                 // Whats after the tag is now only attributes
 777                                 $attributes = $this->mStripState->unstripBoth( $line );
 778                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 779                                 array_pop ( $tr_attributes );
 780                                 array_push ( $tr_attributes , $attributes );
 781
 782                                 $line = '';
 783                                 $last_tag = array_pop ( $last_tag_history );
 784                                 array_pop ( $has_opened_tr );
 785                                 array_push ( $has_opened_tr , true );
 786
 787                                 if ( array_pop ( $tr_history ) ) {
 788                                         $line = '</tr>';
 789                                 }
 790
 791                                 if ( array_pop ( $td_history ) ) {
 792                                         $line = "</{$last_tag}>{$line}";
 793                                 }
 794
 795                                 $lines[$key] = $line;
 796                                 array_push ( $tr_history , false );
 797                                 array_push ( $td_history , false );
 798                                 array_push ( $last_tag_history , '' );
 799                         }
 800                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 801                                 // This might be cell elements, td, th or captions
 802                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 803                                         $first_character = '+';
 804                                         $line = substr ( $line , 1 );
 805                                 }
 806
 807                                 $line = substr ( $line , 1 );
 808
 809                                 if ( $first_character == '!' ) {
 810                                         $line = str_replace ( '!!' , '||' , $line );
 811                                 }
 812
 813                                 // Split up multiple cells on the same line.
 814                                 // FIXME : This can result in improper nesting of tags processed
 815                                 // by earlier parser steps, but should avoid splitting up eg
 816                                 // attribute values containing literal "||".
 817                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 818
 819                                 $lines[$key] = '';
 820
 821                                 // Loop through each table cell
 822                                 foreach ( $cells as $cell )
 823                                 {
 824                                         $previous = '';
 825                                         if ( $first_character != '+' )
 826                                         {
 827                                                 $tr_after = array_pop ( $tr_attributes );
 828                                                 if ( !array_pop ( $tr_history ) ) {
 829                                                         $previous = "<tr{$tr_after}>\n";
 830                                                 }
 831                                                 array_push ( $tr_history , true );
 832                                                 array_push ( $tr_attributes , '' );
 833                                                 array_pop ( $has_opened_tr );
 834                                                 array_push ( $has_opened_tr , true );
 835                                         }
 836
 837                                         $last_tag = array_pop ( $last_tag_history );
 838
 839                                         if ( array_pop ( $td_history ) ) {
 840                                                 $previous = "</{$last_tag}>{$previous}";
 841                                         }
 842
 843                                         if ( $first_character == '|' ) {
 844                                                 $last_tag = 'td';
 845                                         } else if ( $first_character == '!' ) {
 846                                                 $last_tag = 'th';
 847                                         } else if ( $first_character == '+' ) {
 848                                                 $last_tag = 'caption';
 849                                         } else {
 850                                                 $last_tag = '';
 851                                         }
 852
 853                                         array_push ( $last_tag_history , $last_tag );
 854
 855                                         // A cell could contain both parameters and data
 856                                         $cell_data = explode ( '|' , $cell , 2 );
 857
 858                                         // Bug 553: Note that a '|' inside an invalid link should not
 859                                         // be mistaken as delimiting cell parameters
 860                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 861                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 862                                         } else if ( count ( $cell_data ) == 1 )
 863                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 864                                         else {
 865                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 866                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 867                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 868                                         }
 869
 870                                         $lines[$key] .= $cell;
 871                                         array_push ( $td_history , true );
 872                                 }
 873                         }
 874                 }
 875
 876                 // Closing open td, tr && table
 877                 while ( count ( $td_history ) > 0 )
 878                 {
 879                         if ( array_pop ( $td_history ) ) {
 880                                 $lines[] = '</td>' ;
 881                         }
 882                         if ( array_pop ( $tr_history ) ) {
 883                                 $lines[] = '</tr>' ;
 884                         }
 885                         if ( !array_pop ( $has_opened_tr ) ) {
 886                                 $lines[] = "<tr><td></td></tr>" ;
 887                         }
 888
 889                         $lines[] = '</table>' ;
 890                 }
 891
 892                 $output = implode ( "\n" , $lines ) ;
 893
 894                 // special case: don't return empty table
 895                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 896                         $output = '';
 897                 }
 898
 899                 wfProfileOut( $fname );
 900
 901                 return $output;
 902         }
 903
 904         /**
 905          * Helper function for parse() that transforms wiki markup into
 906          * HTML. Only called for $mOutputType == OT_HTML.
 907          *
 908          * @private
 909          */
 910         function internalParse( $text ) {
 911                 $isMain = true;
 912                 $fname = 'Parser::internalParse';
 913                 wfProfileIn( $fname );
 914
 915                 # Hook to suspend the parser in this state
 916                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 917                         wfProfileOut( $fname );
 918                         return $text ;
 919                 }
 920
 921                 # Remove <noinclude> tags and <includeonly> sections
 922                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 923                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 924                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 925
 926                 $text = $this->replaceVariables( $text );
 927                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 928                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 929
 930                 // Tables need to come after variable replacement for things to work
 931                 // properly; putting them before other transformations should keep
 932                 // exciting things like link expansions from showing up in surprising
 933                 // places.
 934                 $text = $this->doTableStuff( $text );
 935
 936                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 937
 938                 $text = $this->stripToc( $text );
 939                 $this->stripNoGallery( $text );
 940                 $text = $this->doHeadings( $text );
 941                 if($this->mOptions->getUseDynamicDates()) {
 942                         $df =& DateFormatter::getInstance();
 943                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 944                 }
 945                 $text = $this->doAllQuotes( $text );
 946                 $text = $this->replaceInternalLinks( $text );
 947                 $text = $this->replaceExternalLinks( $text );
 948
 949                 # replaceInternalLinks may sometimes leave behind
 950                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 951                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 952
 953                 $text = $this->doMagicLinks( $text );
 954                 $text = $this->formatHeadings( $text, $isMain );
 955
 956                 wfProfileOut( $fname );
 957                 return $text;
 958         }
 959
 960         /**
 961          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 962          * magic external links.
 963          *
 964          * @private
 965          */
 966         function doMagicLinks( $text ) {
 967                 wfProfileIn( __METHOD__ );
 968                 $text = preg_replace_callback(
 969                         '!(?:                           # Start cases
 970                             <a.*?</a> |                 # Skip link text
 971                             <.*?> |                     # Skip stuff inside HTML elements
 972                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 973                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 974                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 975                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 976                                      [0-9Xx]                 # check digit
 977                                    \b)
 978                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 979                 wfProfileOut( __METHOD__ );
 980                 return $text;
 981         }
 982
 983         function magicLinkCallback( $m ) {
 984                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 985                         # Skip HTML element
 986                         return $m[0];
 987                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 988                         $isbn = $m[2];
 989                         $num = strtr( $isbn, array(
 990                                 '-' => '',
 991                                 ' ' => '',
 992                                 'x' => 'X',
 993                         ));
 994                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
 995                         $text = '<a href="' .
 996                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
 997                                 "\" class=\"internal\">ISBN $isbn</a>";
 998                 } else {
 999                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1000                                 $keyword = 'RFC';
1001                                 $urlmsg = 'rfcurl';
1002                                 $id = $m[1];
1003                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1004                                 $keyword = 'PMID';
1005                                 $urlmsg = 'pubmedurl';
1006                                 $id = $m[1];
1007                         } else {
1008                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1009                                         substr($m[0], 0, 20 ) . '"' );
1010                         }
1011
1012                         $url = wfMsg( $urlmsg, $id);
1013                         $sk = $this->mOptions->getSkin();
1014                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1015                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1016                 }
1017                 return $text;
1018         }
1019
1020         /**
1021          * Parse headers and return html
1022          *
1023          * @private
1024          */
1025         function doHeadings( $text ) {
1026                 $fname = 'Parser::doHeadings';
1027                 wfProfileIn( $fname );
1028                 for ( $i = 6; $i >= 1; --$i ) {
1029                         $h = str_repeat( '=', $i );
1030                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1031                           "<h$i>\\1</h$i>", $text );
1032                 }
1033                 wfProfileOut( $fname );
1034                 return $text;
1035         }
1036
1037         /**
1038          * Replace single quotes with HTML markup
1039          * @private
1040          * @return string the altered text
1041          */
1042         function doAllQuotes( $text ) {
1043                 $fname = 'Parser::doAllQuotes';
1044                 wfProfileIn( $fname );
1045                 $outtext = '';
1046                 $lines = explode( "\n", $text );
1047                 foreach ( $lines as $line ) {
1048                         $outtext .= $this->doQuotes ( $line ) . "\n";
1049                 }
1050                 $outtext = substr($outtext, 0,-1);
1051                 wfProfileOut( $fname );
1052                 return $outtext;
1053         }
1054
1055         /**
1056          * Helper function for doAllQuotes()
1057          */
1058         public function doQuotes( $text ) {
1059                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1060                 if ( count( $arr ) == 1 )
1061                         return $text;
1062                 else
1063                 {
1064                         # First, do some preliminary work. This may shift some apostrophes from
1065                         # being mark-up to being text. It also counts the number of occurrences
1066                         # of bold and italics mark-ups.
1067                         $i = 0;
1068                         $numbold = 0;
1069                         $numitalics = 0;
1070                         foreach ( $arr as $r )
1071                         {
1072                                 if ( ( $i % 2 ) == 1 )
1073                                 {
1074                                         # If there are ever four apostrophes, assume the first is supposed to
1075                                         # be text, and the remaining three constitute mark-up for bold text.
1076                                         if ( strlen( $arr[$i] ) == 4 )
1077                                         {
1078                                                 $arr[$i-1] .= "'";
1079                                                 $arr[$i] = "'''";
1080                                         }
1081                                         # If there are more than 5 apostrophes in a row, assume they're all
1082                                         # text except for the last 5.
1083                                         else if ( strlen( $arr[$i] ) > 5 )
1084                                         {
1085                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1086                                                 $arr[$i] = "'''''";
1087                                         }
1088                                         # Count the number of occurrences of bold and italics mark-ups.
1089                                         # We are not counting sequences of five apostrophes.
1090                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1091                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1092                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1093                                 }
1094                                 $i++;
1095                         }
1096
1097                         # If there is an odd number of both bold and italics, it is likely
1098                         # that one of the bold ones was meant to be an apostrophe followed
1099                         # by italics. Which one we cannot know for certain, but it is more
1100                         # likely to be one that has a single-letter word before it.
1101                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1102                         {
1103                                 $i = 0;
1104                                 $firstsingleletterword = -1;
1105                                 $firstmultiletterword = -1;
1106                                 $firstspace = -1;
1107                                 foreach ( $arr as $r )
1108                                 {
1109                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1110                                         {
1111                                                 $x1 = substr ($arr[$i-1], -1);
1112                                                 $x2 = substr ($arr[$i-1], -2, 1);
1113                                                 if ($x1 == ' ') {
1114                                                         if ($firstspace == -1) $firstspace = $i;
1115                                                 } else if ($x2 == ' ') {
1116                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1117                                                 } else {
1118                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1119                                                 }
1120                                         }
1121                                         $i++;
1122                                 }
1123
1124                                 # If there is a single-letter word, use it!
1125                                 if ($firstsingleletterword > -1)
1126                                 {
1127                                         $arr [ $firstsingleletterword ] = "''";
1128                                         $arr [ $firstsingleletterword-1 ] .= "'";
1129                                 }
1130                                 # If not, but there's a multi-letter word, use that one.
1131                                 else if ($firstmultiletterword > -1)
1132                                 {
1133                                         $arr [ $firstmultiletterword ] = "''";
1134                                         $arr [ $firstmultiletterword-1 ] .= "'";
1135                                 }
1136                                 # ... otherwise use the first one that has neither.
1137                                 # (notice that it is possible for all three to be -1 if, for example,
1138                                 # there is only one pentuple-apostrophe in the line)
1139                                 else if ($firstspace > -1)
1140                                 {
1141                                         $arr [ $firstspace ] = "''";
1142                                         $arr [ $firstspace-1 ] .= "'";
1143                                 }
1144                         }
1145
1146                         # Now let's actually convert our apostrophic mush to HTML!
1147                         $output = '';
1148                         $buffer = '';
1149                         $state = '';
1150                         $i = 0;
1151                         foreach ($arr as $r)
1152                         {
1153                                 if (($i % 2) == 0)
1154                                 {
1155                                         if ($state == 'both')
1156                                                 $buffer .= $r;
1157                                         else
1158                                                 $output .= $r;
1159                                 }
1160                                 else
1161                                 {
1162                                         if (strlen ($r) == 2)
1163                                         {
1164                                                 if ($state == 'i')
1165                                                 { $output .= '</i>'; $state = ''; }
1166                                                 else if ($state == 'bi')
1167                                                 { $output .= '</i>'; $state = 'b'; }
1168                                                 else if ($state == 'ib')
1169                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1170                                                 else if ($state == 'both')
1171                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1172                                                 else # $state can be 'b' or ''
1173                                                 { $output .= '<i>'; $state .= 'i'; }
1174                                         }
1175                                         else if (strlen ($r) == 3)
1176                                         {
1177                                                 if ($state == 'b')
1178                                                 { $output .= '</b>'; $state = ''; }
1179                                                 else if ($state == 'bi')
1180                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1181                                                 else if ($state == 'ib')
1182                                                 { $output .= '</b>'; $state = 'i'; }
1183                                                 else if ($state == 'both')
1184                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1185                                                 else # $state can be 'i' or ''
1186                                                 { $output .= '<b>'; $state .= 'b'; }
1187                                         }
1188                                         else if (strlen ($r) == 5)
1189                                         {
1190                                                 if ($state == 'b')
1191                                                 { $output .= '</b><i>'; $state = 'i'; }
1192                                                 else if ($state == 'i')
1193                                                 { $output .= '</i><b>'; $state = 'b'; }
1194                                                 else if ($state == 'bi')
1195                                                 { $output .= '</i></b>'; $state = ''; }
1196                                                 else if ($state == 'ib')
1197                                                 { $output .= '</b></i>'; $state = ''; }
1198                                                 else if ($state == 'both')
1199                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1200                                                 else # ($state == '')
1201                                                 { $buffer = ''; $state = 'both'; }
1202                                         }
1203                                 }
1204                                 $i++;
1205                         }
1206                         # Now close all remaining tags.  Notice that the order is important.
1207                         if ($state == 'b' || $state == 'ib')
1208                                 $output .= '</b>';
1209                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1210                                 $output .= '</i>';
1211                         if ($state == 'bi')
1212                                 $output .= '</b>';
1213                         # There might be lonely ''''', so make sure we have a buffer
1214                         if ($state == 'both' && $buffer)
1215                                 $output .= '<b><i>'.$buffer.'</i></b>';
1216                         return $output;
1217                 }
1218         }
1219
1220         /**
1221          * Replace external links
1222          *
1223          * Note: this is all very hackish and the order of execution matters a lot.
1224          * Make sure to run maintenance/parserTests.php if you change this code.
1225          *
1226          * @private
1227          */
1228         function replaceExternalLinks( $text ) {
1229                 global $wgContLang;
1230                 $fname = 'Parser::replaceExternalLinks';
1231                 wfProfileIn( $fname );
1232
1233                 $sk = $this->mOptions->getSkin();
1234
1235                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1236
1237                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1238
1239                 $i = 0;
1240                 while ( $i<count( $bits ) ) {
1241                         $url = $bits[$i++];
1242                         $protocol = $bits[$i++];
1243                         $text = $bits[$i++];
1244                         $trail = $bits[$i++];
1245
1246                         # The characters '<' and '>' (which were escaped by
1247                         # removeHTMLtags()) should not be included in
1248                         # URLs, per RFC 2396.
1249                         $m2 = array();
1250                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1251                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1252                                 $url = substr($url, 0, $m2[0][1]);
1253                         }
1254
1255                         # If the link text is an image URL, replace it with an <img> tag
1256                         # This happened by accident in the original parser, but some people used it extensively
1257                         $img = $this->maybeMakeExternalImage( $text );
1258                         if ( $img !== false ) {
1259                                 $text = $img;
1260                         }
1261
1262                         $dtrail = '';
1263
1264                         # Set linktype for CSS - if URL==text, link is essentially free
1265                         $linktype = ($text == $url) ? 'free' : 'text';
1266
1267                         # No link text, e.g. [http://domain.tld/some.link]
1268                         if ( $text == '' ) {
1269                                 # Autonumber if allowed. See bug #5918
1270                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1271                                         $text = '[' . ++$this->mAutonumber . ']';
1272                                         $linktype = 'autonumber';
1273                                 } else {
1274                                         # Otherwise just use the URL
1275                                         $text = htmlspecialchars( $url );
1276                                         $linktype = 'free';
1277                                 }
1278                         } else {
1279                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1280                                 # Check for trail
1281                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1282                         }
1283
1284                         $text = $wgContLang->markNoConversion($text);
1285
1286                         $url = Sanitizer::cleanUrl( $url );
1287
1288                         # Process the trail (i.e. everything after this link up until start of the next link),
1289                         # replacing any non-bracketed links
1290                         $trail = $this->replaceFreeExternalLinks( $trail );
1291
1292                         # Use the encoded URL
1293                         # This means that users can paste URLs directly into the text
1294                         # Funny characters like &ouml; aren't valid in URLs anyway
1295                         # This was changed in August 2004
1296                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1297
1298                         # Register link in the output object.
1299                         # Replace unnecessary URL escape codes with the referenced character
1300                         # This prevents spammers from hiding links from the filters
1301                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1302                         $this->mOutput->addExternalLink( $pasteurized );
1303                 }
1304
1305                 wfProfileOut( $fname );
1306                 return $s;
1307         }
1308
1309         /**
1310          * Replace anything that looks like a URL with a link
1311          * @private
1312          */
1313         function replaceFreeExternalLinks( $text ) {
1314                 global $wgContLang;
1315                 $fname = 'Parser::replaceFreeExternalLinks';
1316                 wfProfileIn( $fname );
1317
1318                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1319                 $s = array_shift( $bits );
1320                 $i = 0;
1321
1322                 $sk = $this->mOptions->getSkin();
1323
1324                 while ( $i < count( $bits ) ){
1325                         $protocol = $bits[$i++];
1326                         $remainder = $bits[$i++];
1327
1328                         $m = array();
1329                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1330                                 # Found some characters after the protocol that look promising
1331                                 $url = $protocol . $m[1];
1332                                 $trail = $m[2];
1333
1334                                 # special case: handle urls as url args:
1335                                 # http://www.example.com/foo?=http://www.example.com/bar
1336                                 if(strlen($trail) == 0 &&
1337                                         isset($bits[$i]) &&
1338                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1339                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1340                                 {
1341                                         # add protocol, arg
1342                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1343                                         $i += 2;
1344                                         $trail = $m[2];
1345                                 }
1346
1347                                 # The characters '<' and '>' (which were escaped by
1348                                 # removeHTMLtags()) should not be included in
1349                                 # URLs, per RFC 2396.
1350                                 $m2 = array();
1351                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1352                                         $trail = substr($url, $m2[0][1]) . $trail;
1353                                         $url = substr($url, 0, $m2[0][1]);
1354                                 }
1355
1356                                 # Move trailing punctuation to $trail
1357                                 $sep = ',;\.:!?';
1358                                 # If there is no left bracket, then consider right brackets fair game too
1359                                 if ( strpos( $url, '(' ) === false ) {
1360                                         $sep .= ')';
1361                                 }
1362
1363                                 $numSepChars = strspn( strrev( $url ), $sep );
1364                                 if ( $numSepChars ) {
1365                                         $trail = substr( $url, -$numSepChars ) . $trail;
1366                                         $url = substr( $url, 0, -$numSepChars );
1367                                 }
1368
1369                                 $url = Sanitizer::cleanUrl( $url );
1370
1371                                 # Is this an external image?
1372                                 $text = $this->maybeMakeExternalImage( $url );
1373                                 if ( $text === false ) {
1374                                         # Not an image, make a link
1375                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1376                                         # Register it in the output object...
1377                                         # Replace unnecessary URL escape codes with their equivalent characters
1378                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1379                                         $this->mOutput->addExternalLink( $pasteurized );
1380                                 }
1381                                 $s .= $text . $trail;
1382                         } else {
1383                                 $s .= $protocol . $remainder;
1384                         }
1385                 }
1386                 wfProfileOut( $fname );
1387                 return $s;
1388         }
1389
1390         /**
1391          * Replace unusual URL escape codes with their equivalent characters
1392          * @param string
1393          * @return string
1394          * @static
1395          * @todo  This can merge genuinely required bits in the path or query string,
1396          *        breaking legit URLs. A proper fix would treat the various parts of
1397          *        the URL differently; as a workaround, just use the output for
1398          *        statistical records, not for actual linking/output.
1399          */
1400         static function replaceUnusualEscapes( $url ) {
1401                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1402                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1403         }
1404
1405         /**
1406          * Callback function used in replaceUnusualEscapes().
1407          * Replaces unusual URL escape codes with their equivalent character
1408          * @static
1409          * @private
1410          */
1411         private static function replaceUnusualEscapesCallback( $matches ) {
1412                 $char = urldecode( $matches[0] );
1413                 $ord = ord( $char );
1414                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1415                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1416                         // No, shouldn't be escaped
1417                         return $char;
1418                 } else {
1419                         // Yes, leave it escaped
1420                         return $matches[0];
1421                 }
1422         }
1423
1424         /**
1425          * make an image if it's allowed, either through the global
1426          * option or through the exception
1427          * @private
1428          */
1429         function maybeMakeExternalImage( $url ) {
1430                 $sk = $this->mOptions->getSkin();
1431                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1432                 $imagesexception = !empty($imagesfrom);
1433                 $text = false;
1434                 if ( $this->mOptions->getAllowExternalImages()
1435                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1436                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1437                                 # Image found
1438                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1439                         }
1440                 }
1441                 return $text;
1442         }
1443
1444         /**
1445          * Process [[ ]] wikilinks
1446          *
1447          * @private
1448          */
1449         function replaceInternalLinks( $s ) {
1450                 global $wgContLang;
1451                 static $fname = 'Parser::replaceInternalLinks' ;
1452
1453                 wfProfileIn( $fname );
1454
1455                 wfProfileIn( $fname.'-setup' );
1456                 static $tc = FALSE;
1457                 # the % is needed to support urlencoded titles as well
1458                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1459
1460                 $sk = $this->mOptions->getSkin();
1461
1462                 #split the entire text string on occurences of [[
1463                 $a = explode( '[[', ' ' . $s );
1464                 #get the first element (all text up to first [[), and remove the space we added
1465                 $s = array_shift( $a );
1466                 $s = substr( $s, 1 );
1467
1468                 # Match a link having the form [[namespace:link|alternate]]trail
1469                 static $e1 = FALSE;
1470                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1471                 # Match cases where there is no "]]", which might still be images
1472                 static $e1_img = FALSE;
1473                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1474
1475                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1476                 $e2 = null;
1477                 if ( $useLinkPrefixExtension ) {
1478                         # Match the end of a line for a word that's not followed by whitespace,
1479                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1480                         $e2 = wfMsgForContent( 'linkprefix' );
1481                 }
1482
1483                 if( is_null( $this->mTitle ) ) {
1484                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1485                 }
1486                 $nottalk = !$this->mTitle->isTalkPage();
1487
1488                 if ( $useLinkPrefixExtension ) {
1489                         $m = array();
1490                         if ( preg_match( $e2, $s, $m ) ) {
1491                                 $first_prefix = $m[2];
1492                         } else {
1493                                 $first_prefix = false;
1494                         }
1495                 } else {
1496                         $prefix = '';
1497                 }
1498
1499                 if($wgContLang->hasVariants()) {
1500                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1501                 } else {
1502                         $selflink = array($this->mTitle->getPrefixedText());
1503                 }
1504                 $useSubpages = $this->areSubpagesAllowed();
1505                 wfProfileOut( $fname.'-setup' );
1506
1507                 # Loop for each link
1508                 for ($k = 0; isset( $a[$k] ); $k++) {
1509                         $line = $a[$k];
1510                         if ( $useLinkPrefixExtension ) {
1511                                 wfProfileIn( $fname.'-prefixhandling' );
1512                                 if ( preg_match( $e2, $s, $m ) ) {
1513                                         $prefix = $m[2];
1514                                         $s = $m[1];
1515                                 } else {
1516                                         $prefix='';
1517                                 }
1518                                 # first link
1519                                 if($first_prefix) {
1520                                         $prefix = $first_prefix;
1521                                         $first_prefix = false;
1522                                 }
1523                                 wfProfileOut( $fname.'-prefixhandling' );
1524                         }
1525
1526                         $might_be_img = false;
1527
1528                         wfProfileIn( "$fname-e1" );
1529                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1530                                 $text = $m[2];
1531                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1532                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1533                                 # the real problem is with the $e1 regex
1534                                 # See bug 1300.
1535                                 #
1536                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1537                                 # and no image is in sight. See bug 2095.
1538                                 #
1539                                 if( $text !== '' &&
1540                                         substr( $m[3], 0, 1 ) === ']' &&
1541                                         strpos($text, '[') !== false
1542                                 )
1543                                 {
1544                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1545                                         $m[3] = substr( $m[3], 1 );
1546                                 }
1547                                 # fix up urlencoded title texts
1548                                 if( strpos( $m[1], '%' ) !== false ) {
1549                                         # Should anchors '#' also be rejected?
1550                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1551                                 }
1552                                 $trail = $m[3];
1553                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1554                                 $might_be_img = true;
1555                                 $text = $m[2];
1556                                 if ( strpos( $m[1], '%' ) !== false ) {
1557                                         $m[1] = urldecode($m[1]);
1558                                 }
1559                                 $trail = "";
1560                         } else { # Invalid form; output directly
1561                                 $s .= $prefix . '[[' . $line ;
1562                                 wfProfileOut( "$fname-e1" );
1563                                 continue;
1564                         }
1565                         wfProfileOut( "$fname-e1" );
1566                         wfProfileIn( "$fname-misc" );
1567
1568                         # Don't allow internal links to pages containing
1569                         # PROTO: where PROTO is a valid URL protocol; these
1570                         # should be external links.
1571                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1572                                 $s .= $prefix . '[[' . $line ;
1573                                 continue;
1574                         }
1575
1576                         # Make subpage if necessary
1577                         if( $useSubpages ) {
1578                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1579                         } else {
1580                                 $link = $m[1];
1581                         }
1582
1583                         $noforce = (substr($m[1], 0, 1) != ':');
1584                         if (!$noforce) {
1585                                 # Strip off leading ':'
1586                                 $link = substr($link, 1);
1587                         }
1588
1589                         wfProfileOut( "$fname-misc" );
1590                         wfProfileIn( "$fname-title" );
1591                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1592                         if( !$nt ) {
1593                                 $s .= $prefix . '[[' . $line;
1594                                 wfProfileOut( "$fname-title" );
1595                                 continue;
1596                         }
1597
1598                         $ns = $nt->getNamespace();
1599                         $iw = $nt->getInterWiki();
1600                         wfProfileOut( "$fname-title" );
1601
1602                         if ($might_be_img) { # if this is actually an invalid link
1603                                 wfProfileIn( "$fname-might_be_img" );
1604                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1605                                         $found = false;
1606                                         while (isset ($a[$k+1]) ) {
1607                                                 #look at the next 'line' to see if we can close it there
1608                                                 $spliced = array_splice( $a, $k + 1, 1 );
1609                                                 $next_line = array_shift( $spliced );
1610                                                 $m = explode( ']]', $next_line, 3 );
1611                                                 if ( count( $m ) == 3 ) {
1612                                                         # the first ]] closes the inner link, the second the image
1613                                                         $found = true;
1614                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1615                                                         $trail = $m[2];
1616                                                         break;
1617                                                 } elseif ( count( $m ) == 2 ) {
1618                                                         #if there's exactly one ]] that's fine, we'll keep looking
1619                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1620                                                 } else {
1621                                                         #if $next_line is invalid too, we need look no further
1622                                                         $text .= '[[' . $next_line;
1623                                                         break;
1624                                                 }
1625                                         }
1626                                         if ( !$found ) {
1627                                                 # we couldn't find the end of this imageLink, so output it raw
1628                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1629                                                 $text = $this->replaceInternalLinks($text);
1630                                                 $s .= "{$prefix}[[$link|$text";
1631                                                 # note: no $trail, because without an end, there *is* no trail
1632                                                 wfProfileOut( "$fname-might_be_img" );
1633                                                 continue;
1634                                         }
1635                                 } else { #it's not an image, so output it raw
1636                                         $s .= "{$prefix}[[$link|$text";
1637                                         # note: no $trail, because without an end, there *is* no trail
1638                                         wfProfileOut( "$fname-might_be_img" );
1639                                         continue;
1640                                 }
1641                                 wfProfileOut( "$fname-might_be_img" );
1642                         }
1643
1644                         $wasblank = ( '' == $text );
1645                         if( $wasblank ) $text = $link;
1646
1647                         # Link not escaped by : , create the various objects
1648                         if( $noforce ) {
1649
1650                                 # Interwikis
1651                                 wfProfileIn( "$fname-interwiki" );
1652                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1653                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1654                                         $s = rtrim($s . $prefix);
1655                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1656                                         wfProfileOut( "$fname-interwiki" );
1657                                         continue;
1658                                 }
1659                                 wfProfileOut( "$fname-interwiki" );
1660
1661                                 if ( $ns == NS_IMAGE ) {
1662                                         wfProfileIn( "$fname-image" );
1663                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1664                                                 # recursively parse links inside the image caption
1665                                                 # actually, this will parse them in any other parameters, too,
1666                                                 # but it might be hard to fix that, and it doesn't matter ATM
1667                                                 $text = $this->replaceExternalLinks($text);
1668                                                 $text = $this->replaceInternalLinks($text);
1669
1670                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1671                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1672                                                 $this->mOutput->addImage( $nt->getDBkey() );
1673
1674                                                 wfProfileOut( "$fname-image" );
1675                                                 continue;
1676                                         } else {
1677                                                 # We still need to record the image's presence on the page
1678                                                 $this->mOutput->addImage( $nt->getDBkey() );
1679                                         }
1680                                         wfProfileOut( "$fname-image" );
1681
1682                                 }
1683
1684                                 if ( $ns == NS_CATEGORY ) {
1685                                         wfProfileIn( "$fname-category" );
1686                                         $s = rtrim($s . "\n"); # bug 87
1687
1688                                         if ( $wasblank ) {
1689                                                 $sortkey = $this->getDefaultSort();
1690                                         } else {
1691                                                 $sortkey = $text;
1692                                         }
1693                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1694                                         $sortkey = str_replace( "\n", '', $sortkey );
1695                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1696                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1697
1698                                         /**
1699                                          * Strip the whitespace Category links produce, see bug 87
1700                                          * @todo We might want to use trim($tmp, "\n") here.
1701                                          */
1702                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1703
1704                                         wfProfileOut( "$fname-category" );
1705                                         continue;
1706                                 }
1707                         }
1708
1709                         # Self-link checking
1710                         if( $nt->getFragment() === '' ) {
1711                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1712                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1713                                         continue;
1714                                 }
1715                         }
1716
1717                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1718                         if( $ns == NS_MEDIA ) {
1719                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1720                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1721                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1722                                 $this->mOutput->addImage( $nt->getDBkey() );
1723                                 continue;
1724                         } elseif( $ns == NS_SPECIAL ) {
1725                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1726                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1727                                 } else {
1728                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1729                                 }
1730                                 continue;
1731                         } elseif( $ns == NS_IMAGE ) {
1732                                 $img = wfFindFile( $nt );
1733                                 if( $img ) {
1734                                         // Force a blue link if the file exists; may be a remote
1735                                         // upload on the shared repository, and we want to see its
1736                                         // auto-generated page.
1737                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1738                                         $this->mOutput->addLink( $nt );
1739                                         continue;
1740                                 }
1741                         }
1742                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1743                 }
1744                 wfProfileOut( $fname );
1745                 return $s;
1746         }
1747
1748         /**
1749          * Make a link placeholder. The text returned can be later resolved to a real link with
1750          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1751          * parsing of interwiki links, and secondly to allow all existence checks and
1752          * article length checks (for stub links) to be bundled into a single query.
1753          *
1754          */
1755         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1756                 wfProfileIn( __METHOD__ );
1757                 if ( ! is_object($nt) ) {
1758                         # Fail gracefully
1759                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1760                 } else {
1761                         # Separate the link trail from the rest of the link
1762                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1763
1764                         if ( $nt->isExternal() ) {
1765                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1766                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1767                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1768                         } else {
1769                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1770                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1771                                 $this->mLinkHolders['queries'][] = $query;
1772                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1773                                 $this->mLinkHolders['titles'][] = $nt;
1774
1775                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1776                         }
1777                 }
1778                 wfProfileOut( __METHOD__ );
1779                 return $retVal;
1780         }
1781
1782         /**
1783          * Render a forced-blue link inline; protect against double expansion of
1784          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1785          * Since this little disaster has to split off the trail text to avoid
1786          * breaking URLs in the following text without breaking trails on the
1787          * wiki links, it's been made into a horrible function.
1788          *
1789          * @param Title $nt
1790          * @param string $text
1791          * @param string $query
1792          * @param string $trail
1793          * @param string $prefix
1794          * @return string HTML-wikitext mix oh yuck
1795          */
1796         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1797                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1798                 $sk = $this->mOptions->getSkin();
1799                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1800                 return $this->armorLinks( $link ) . $trail;
1801         }
1802
1803         /**
1804          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1805          * going to go through further parsing steps before inline URL expansion.
1806          *
1807          * In particular this is important when using action=render, which causes
1808          * full URLs to be included.
1809          *
1810          * Oh man I hate our multi-layer parser!
1811          *
1812          * @param string more-or-less HTML
1813          * @return string less-or-more HTML with NOPARSE bits
1814          */
1815         function armorLinks( $text ) {
1816                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1817                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1818         }
1819
1820         /**
1821          * Return true if subpage links should be expanded on this page.
1822          * @return bool
1823          */
1824         function areSubpagesAllowed() {
1825                 # Some namespaces don't allow subpages
1826                 global $wgNamespacesWithSubpages;
1827                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1828         }
1829
1830         /**
1831          * Handle link to subpage if necessary
1832          * @param string $target the source of the link
1833          * @param string &$text the link text, modified as necessary
1834          * @return string the full name of the link
1835          * @private
1836          */
1837         function maybeDoSubpageLink($target, &$text) {
1838                 # Valid link forms:
1839                 # Foobar -- normal
1840                 # :Foobar -- override special treatment of prefix (images, language links)
1841                 # /Foobar -- convert to CurrentPage/Foobar
1842                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1843                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1844                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1845
1846                 $fname = 'Parser::maybeDoSubpageLink';
1847                 wfProfileIn( $fname );
1848                 $ret = $target; # default return value is no change
1849
1850                 # Some namespaces don't allow subpages,
1851                 # so only perform processing if subpages are allowed
1852                 if( $this->areSubpagesAllowed() ) {
1853                         $hash = strpos( $target, '#' );
1854                         if( $hash !== false ) {
1855                                 $suffix = substr( $target, $hash );
1856                                 $target = substr( $target, 0, $hash );
1857                         } else {
1858                                 $suffix = '';
1859                         }
1860                         # bug 7425
1861                         $target = trim( $target );
1862                         # Look at the first character
1863                         if( $target != '' && $target{0} == '/' ) {
1864                                 # / at end means we don't want the slash to be shown
1865                                 $m = array();
1866                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1867                                 if( $trailingSlashes ) {
1868                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1869                                 } else {
1870                                         $noslash = substr( $target, 1 );
1871                                 }
1872
1873                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1874                                 if( '' === $text ) {
1875                                         $text = $target . $suffix;
1876                                 } # this might be changed for ugliness reasons
1877                         } else {
1878                                 # check for .. subpage backlinks
1879                                 $dotdotcount = 0;
1880                                 $nodotdot = $target;
1881                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1882                                         ++$dotdotcount;
1883                                         $nodotdot = substr( $nodotdot, 3 );
1884                                 }
1885                                 if($dotdotcount > 0) {
1886                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1887                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1888                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1889                                                 # / at the end means don't show full path
1890                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1891                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1892                                                         if( '' === $text ) {
1893                                                                 $text = $nodotdot . $suffix;
1894                                                         }
1895                                                 }
1896                                                 $nodotdot = trim( $nodotdot );
1897                                                 if( $nodotdot != '' ) {
1898                                                         $ret .= '/' . $nodotdot;
1899                                                 }
1900                                                 $ret .= $suffix;
1901                                         }
1902                                 }
1903                         }
1904                 }
1905
1906                 wfProfileOut( $fname );
1907                 return $ret;
1908         }
1909
1910         /**#@+
1911          * Used by doBlockLevels()
1912          * @private
1913          */
1914         /* private */ function closeParagraph() {
1915                 $result = '';
1916                 if ( '' != $this->mLastSection ) {
1917                         $result = '</' . $this->mLastSection  . ">\n";
1918                 }
1919                 $this->mInPre = false;
1920                 $this->mLastSection = '';
1921                 return $result;
1922         }
1923         # getCommon() returns the length of the longest common substring
1924         # of both arguments, starting at the beginning of both.
1925         #
1926         /* private */ function getCommon( $st1, $st2 ) {
1927                 $fl = strlen( $st1 );
1928                 $shorter = strlen( $st2 );
1929                 if ( $fl < $shorter ) { $shorter = $fl; }
1930
1931                 for ( $i = 0; $i < $shorter; ++$i ) {
1932                         if ( $st1{$i} != $st2{$i} ) { break; }
1933                 }
1934                 return $i;
1935         }
1936         # These next three functions open, continue, and close the list
1937         # element appropriate to the prefix character passed into them.
1938         #
1939         /* private */ function openList( $char ) {
1940                 $result = $this->closeParagraph();
1941
1942                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1943                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1944                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1945                 else if ( ';' == $char ) {
1946                         $result .= '<dl><dt>';
1947                         $this->mDTopen = true;
1948                 }
1949                 else { $result = '<!-- ERR 1 -->'; }
1950
1951                 return $result;
1952         }
1953
1954         /* private */ function nextItem( $char ) {
1955                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1956                 else if ( ':' == $char || ';' == $char ) {
1957                         $close = '</dd>';
1958                         if ( $this->mDTopen ) { $close = '</dt>'; }
1959                         if ( ';' == $char ) {
1960                                 $this->mDTopen = true;
1961                                 return $close . '<dt>';
1962                         } else {
1963                                 $this->mDTopen = false;
1964                                 return $close . '<dd>';
1965                         }
1966                 }
1967                 return '<!-- ERR 2 -->';
1968         }
1969
1970         /* private */ function closeList( $char ) {
1971                 if ( '*' == $char ) { $text = '</li></ul>'; }
1972                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1973                 else if ( ':' == $char ) {
1974                         if ( $this->mDTopen ) {
1975                                 $this->mDTopen = false;
1976                                 $text = '</dt></dl>';
1977                         } else {
1978                                 $text = '</dd></dl>';
1979                         }
1980                 }
1981                 else {  return '<!-- ERR 3 -->'; }
1982                 return $text."\n";
1983         }
1984         /**#@-*/
1985
1986         /**
1987          * Make lists from lines starting with ':', '*', '#', etc.
1988          *
1989          * @private
1990          * @return string the lists rendered as HTML
1991          */
1992         function doBlockLevels( $text, $linestart ) {
1993                 $fname = 'Parser::doBlockLevels';
1994                 wfProfileIn( $fname );
1995
1996                 # Parsing through the text line by line.  The main thing
1997                 # happening here is handling of block-level elements p, pre,
1998                 # and making lists from lines starting with * # : etc.
1999                 #
2000                 $textLines = explode( "\n", $text );
2001
2002                 $lastPrefix = $output = '';
2003                 $this->mDTopen = $inBlockElem = false;
2004                 $prefixLength = 0;
2005                 $paragraphStack = false;
2006
2007                 if ( !$linestart ) {
2008                         $output .= array_shift( $textLines );
2009                 }
2010                 foreach ( $textLines as $oLine ) {
2011                         $lastPrefixLength = strlen( $lastPrefix );
2012                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2013                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2014                         if ( !$this->mInPre ) {
2015                                 # Multiple prefixes may abut each other for nested lists.
2016                                 $prefixLength = strspn( $oLine, '*#:;' );
2017                                 $pref = substr( $oLine, 0, $prefixLength );
2018
2019                                 # eh?
2020                                 $pref2 = str_replace( ';', ':', $pref );
2021                                 $t = substr( $oLine, $prefixLength );
2022                                 $this->mInPre = !empty($preOpenMatch);
2023                         } else {
2024                                 # Don't interpret any other prefixes in preformatted text
2025                                 $prefixLength = 0;
2026                                 $pref = $pref2 = '';
2027                                 $t = $oLine;
2028                         }
2029
2030                         # List generation
2031                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2032                                 # Same as the last item, so no need to deal with nesting or opening stuff
2033                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2034                                 $paragraphStack = false;
2035
2036                                 if ( substr( $pref, -1 ) == ';') {
2037                                         # The one nasty exception: definition lists work like this:
2038                                         # ; title : definition text
2039                                         # So we check for : in the remainder text to split up the
2040                                         # title and definition, without b0rking links.
2041                                         $term = $t2 = '';
2042                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2043                                                 $t = $t2;
2044                                                 $output .= $term . $this->nextItem( ':' );
2045                                         }
2046                                 }
2047                         } elseif( $prefixLength || $lastPrefixLength ) {
2048                                 # Either open or close a level...
2049                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2050                                 $paragraphStack = false;
2051
2052                                 while( $commonPrefixLength < $lastPrefixLength ) {
2053                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2054                                         --$lastPrefixLength;
2055                                 }
2056                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2057                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2058                                 }
2059                                 while ( $prefixLength > $commonPrefixLength ) {
2060                                         $char = substr( $pref, $commonPrefixLength, 1 );
2061                                         $output .= $this->openList( $char );
2062
2063                                         if ( ';' == $char ) {
2064                                                 # FIXME: This is dupe of code above
2065                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2066                                                         $t = $t2;
2067                                                         $output .= $term . $this->nextItem( ':' );
2068                                                 }
2069                                         }
2070                                         ++$commonPrefixLength;
2071                                 }
2072                                 $lastPrefix = $pref2;
2073                         }
2074                         if( 0 == $prefixLength ) {
2075                                 wfProfileIn( "$fname-paragraph" );
2076                                 # No prefix (not in list)--go to paragraph mode
2077                                 // XXX: use a stack for nestable elements like span, table and div
2078                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2079                                 $closematch = preg_match(
2080                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2081                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2082                                 if ( $openmatch or $closematch ) {
2083                                         $paragraphStack = false;
2084                                         # TODO bug 5718: paragraph closed
2085                                         $output .= $this->closeParagraph();
2086                                         if ( $preOpenMatch and !$preCloseMatch ) {
2087                                                 $this->mInPre = true;
2088                                         }
2089                                         if ( $closematch ) {
2090                                                 $inBlockElem = false;
2091                                         } else {
2092                                                 $inBlockElem = true;
2093                                         }
2094                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2095                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2096                                                 // pre
2097                                                 if ($this->mLastSection != 'pre') {
2098                                                         $paragraphStack = false;
2099                                                         $output .= $this->closeParagraph().'<pre>';
2100                                                         $this->mLastSection = 'pre';
2101                                                 }
2102                                                 $t = substr( $t, 1 );
2103                                         } else {
2104                                                 // paragraph
2105                                                 if ( '' == trim($t) ) {
2106                                                         if ( $paragraphStack ) {
2107                                                                 $output .= $paragraphStack.'<br />';
2108                                                                 $paragraphStack = false;
2109                                                                 $this->mLastSection = 'p';
2110                                                         } else {
2111                                                                 if ($this->mLastSection != 'p' ) {
2112                                                                         $output .= $this->closeParagraph();
2113                                                                         $this->mLastSection = '';
2114                                                                         $paragraphStack = '<p>';
2115                                                                 } else {
2116                                                                         $paragraphStack = '</p><p>';
2117                                                                 }
2118                                                         }
2119                                                 } else {
2120                                                         if ( $paragraphStack ) {
2121                                                                 $output .= $paragraphStack;
2122                                                                 $paragraphStack = false;
2123                                                                 $this->mLastSection = 'p';
2124                                                         } else if ($this->mLastSection != 'p') {
2125                                                                 $output .= $this->closeParagraph().'<p>';
2126                                                                 $this->mLastSection = 'p';
2127                                                         }
2128                                                 }
2129                                         }
2130                                 }
2131                                 wfProfileOut( "$fname-paragraph" );
2132                         }
2133                         // somewhere above we forget to get out of pre block (bug 785)
2134                         if($preCloseMatch && $this->mInPre) {
2135                                 $this->mInPre = false;
2136                         }
2137                         if ($paragraphStack === false) {
2138                                 $output .= $t."\n";
2139                         }
2140                 }
2141                 while ( $prefixLength ) {
2142                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2143                         --$prefixLength;
2144                 }
2145                 if ( '' != $this->mLastSection ) {
2146                         $output .= '</' . $this->mLastSection . '>';
2147                         $this->mLastSection = '';
2148                 }
2149
2150                 wfProfileOut( $fname );
2151                 return $output;
2152         }
2153
2154         /**
2155          * Split up a string on ':', ignoring any occurences inside tags
2156          * to prevent illegal overlapping.
2157          * @param string $str the string to split
2158          * @param string &$before set to everything before the ':'
2159          * @param string &$after set to everything after the ':'
2160          * return string the position of the ':', or false if none found
2161          */
2162         function findColonNoLinks($str, &$before, &$after) {
2163                 $fname = 'Parser::findColonNoLinks';
2164                 wfProfileIn( $fname );
2165
2166                 $pos = strpos( $str, ':' );
2167                 if( $pos === false ) {
2168                         // Nothing to find!
2169                         wfProfileOut( $fname );
2170                         return false;
2171                 }
2172
2173                 $lt = strpos( $str, '<' );
2174                 if( $lt === false || $lt > $pos ) {
2175                         // Easy; no tag nesting to worry about
2176                         $before = substr( $str, 0, $pos );
2177                         $after = substr( $str, $pos+1 );
2178                         wfProfileOut( $fname );
2179                         return $pos;
2180                 }
2181
2182                 // Ugly state machine to walk through avoiding tags.
2183                 $state = self::COLON_STATE_TEXT;
2184                 $stack = 0;
2185                 $len = strlen( $str );
2186                 for( $i = 0; $i < $len; $i++ ) {
2187                         $c = $str{$i};
2188
2189                         switch( $state ) {
2190                         // (Using the number is a performance hack for common cases)
2191                         case 0: // self::COLON_STATE_TEXT:
2192                                 switch( $c ) {
2193                                 case "<":
2194                                         // Could be either a <start> tag or an </end> tag
2195                                         $state = self::COLON_STATE_TAGSTART;
2196                                         break;
2197                                 case ":":
2198                                         if( $stack == 0 ) {
2199                                                 // We found it!
2200                                                 $before = substr( $str, 0, $i );
2201                                                 $after = substr( $str, $i + 1 );
2202                                                 wfProfileOut( $fname );
2203                                                 return $i;
2204                                         }
2205                                         // Embedded in a tag; don't break it.
2206                                         break;
2207                                 default:
2208                                         // Skip ahead looking for something interesting
2209                                         $colon = strpos( $str, ':', $i );
2210                                         if( $colon === false ) {
2211                                                 // Nothing else interesting
2212                                                 wfProfileOut( $fname );
2213                                                 return false;
2214                                         }
2215                                         $lt = strpos( $str, '<', $i );
2216                                         if( $stack === 0 ) {
2217                                                 if( $lt === false || $colon < $lt ) {
2218                                                         // We found it!
2219                                                         $before = substr( $str, 0, $colon );
2220                                                         $after = substr( $str, $colon + 1 );
2221                                                         wfProfileOut( $fname );
2222                                                         return $i;
2223                                                 }
2224                                         }
2225                                         if( $lt === false ) {
2226                                                 // Nothing else interesting to find; abort!
2227                                                 // We're nested, but there's no close tags left. Abort!
2228                                                 break 2;
2229                                         }
2230                                         // Skip ahead to next tag start
2231                                         $i = $lt;
2232                                         $state = self::COLON_STATE_TAGSTART;
2233                                 }
2234                                 break;
2235                         case 1: // self::COLON_STATE_TAG:
2236                                 // In a <tag>
2237                                 switch( $c ) {
2238                                 case ">":
2239                                         $stack++;
2240                                         $state = self::COLON_STATE_TEXT;
2241                                         break;
2242                                 case "/":
2243                                         // Slash may be followed by >?
2244                                         $state = self::COLON_STATE_TAGSLASH;
2245                                         break;
2246                                 default:
2247                                         // ignore
2248                                 }
2249                                 break;
2250                         case 2: // self::COLON_STATE_TAGSTART:
2251                                 switch( $c ) {
2252                                 case "/":
2253                                         $state = self::COLON_STATE_CLOSETAG;
2254                                         break;
2255                                 case "!":
2256                                         $state = self::COLON_STATE_COMMENT;
2257                                         break;
2258                                 case ">":
2259                                         // Illegal early close? This shouldn't happen D:
2260                                         $state = self::COLON_STATE_TEXT;
2261                                         break;
2262                                 default:
2263                                         $state = self::COLON_STATE_TAG;
2264                                 }
2265                                 break;
2266                         case 3: // self::COLON_STATE_CLOSETAG:
2267                                 // In a </tag>
2268                                 if( $c == ">" ) {
2269                                         $stack--;
2270                                         if( $stack < 0 ) {
2271                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2272                                                 wfProfileOut( $fname );
2273                                                 return false;
2274                                         }
2275                                         $state = self::COLON_STATE_TEXT;
2276                                 }
2277                                 break;
2278                         case self::COLON_STATE_TAGSLASH:
2279                                 if( $c == ">" ) {
2280                                         // Yes, a self-closed tag <blah/>
2281                                         $state = self::COLON_STATE_TEXT;
2282                                 } else {
2283                                         // Probably we're jumping the gun, and this is an attribute
2284                                         $state = self::COLON_STATE_TAG;
2285                                 }
2286                                 break;
2287                         case 5: // self::COLON_STATE_COMMENT:
2288                                 if( $c == "-" ) {
2289                                         $state = self::COLON_STATE_COMMENTDASH;
2290                                 }
2291                                 break;
2292                         case self::COLON_STATE_COMMENTDASH:
2293                                 if( $c == "-" ) {
2294                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2295                                 } else {
2296                                         $state = self::COLON_STATE_COMMENT;
2297                                 }
2298                                 break;
2299                         case self::COLON_STATE_COMMENTDASHDASH:
2300                                 if( $c == ">" ) {
2301                                         $state = self::COLON_STATE_TEXT;
2302                                 } else {
2303                                         $state = self::COLON_STATE_COMMENT;
2304                                 }
2305                                 break;
2306                         default:
2307                                 throw new MWException( "State machine error in $fname" );
2308                         }
2309                 }
2310                 if( $stack > 0 ) {
2311                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2312                         return false;
2313                 }
2314                 wfProfileOut( $fname );
2315                 return false;
2316         }
2317
2318         /**
2319          * Return value of a magic variable (like PAGENAME)
2320          *
2321          * @private
2322          */
2323         function getVariableValue( $index ) {
2324                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2325
2326                 /**
2327                  * Some of these require message or data lookups and can be
2328                  * expensive to check many times.
2329                  */
2330                 static $varCache = array();
2331                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2332                         if ( isset( $varCache[$index] ) ) {
2333                                 return $varCache[$index];
2334                         }
2335                 }
2336
2337                 $ts = time();
2338                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2339
2340                 # Use the time zone
2341                 global $wgLocaltimezone;
2342                 if ( isset( $wgLocaltimezone ) ) {
2343                         $oldtz = getenv( 'TZ' );
2344                         putenv( 'TZ='.$wgLocaltimezone );
2345                 }
2346
2347                 wfSuppressWarnings(); // E_STRICT system time bitching
2348                 $localTimestamp = date( 'YmdHis', $ts );
2349                 $localMonth = date( 'm', $ts );
2350                 $localMonthName = date( 'n', $ts );
2351                 $localDay = date( 'j', $ts );
2352                 $localDay2 = date( 'd', $ts );
2353                 $localDayOfWeek = date( 'w', $ts );
2354                 $localWeek = date( 'W', $ts );
2355                 $localYear = date( 'Y', $ts );
2356                 $localHour = date( 'H', $ts );
2357                 if ( isset( $wgLocaltimezone ) ) {
2358                         putenv( 'TZ='.$oldtz );
2359                 }
2360                 wfRestoreWarnings();
2361
2362                 switch ( $index ) {
2363                         case 'currentmonth':
2364                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2365                         case 'currentmonthname':
2366                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2367                         case 'currentmonthnamegen':
2368                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2369                         case 'currentmonthabbrev':
2370                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2371                         case 'currentday':
2372                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2373                         case 'currentday2':
2374                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2375                         case 'localmonth':
2376                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2377                         case 'localmonthname':
2378                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2379                         case 'localmonthnamegen':
2380                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2381                         case 'localmonthabbrev':
2382                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2383                         case 'localday':
2384                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2385                         case 'localday2':
2386                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2387                         case 'pagename':
2388                                 return wfEscapeWikiText( $this->mTitle->getText() );
2389                         case 'pagenamee':
2390                                 return $this->mTitle->getPartialURL();
2391                         case 'fullpagename':
2392                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2393                         case 'fullpagenamee':
2394                                 return $this->mTitle->getPrefixedURL();
2395                         case 'subpagename':
2396                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2397                         case 'subpagenamee':
2398                                 return $this->mTitle->getSubpageUrlForm();
2399                         case 'basepagename':
2400                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2401                         case 'basepagenamee':
2402                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2403                         case 'talkpagename':
2404                                 if( $this->mTitle->canTalk() ) {
2405                                         $talkPage = $this->mTitle->getTalkPage();
2406                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2407                                 } else {
2408                                         return '';
2409                                 }
2410                         case 'talkpagenamee':
2411                                 if( $this->mTitle->canTalk() ) {
2412                                         $talkPage = $this->mTitle->getTalkPage();
2413                                         return $talkPage->getPrefixedUrl();
2414                                 } else {
2415                                         return '';
2416                                 }
2417                         case 'subjectpagename':
2418                                 $subjPage = $this->mTitle->getSubjectPage();
2419                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2420                         case 'subjectpagenamee':
2421                                 $subjPage = $this->mTitle->getSubjectPage();
2422                                 return $subjPage->getPrefixedUrl();
2423                         case 'revisionid':
2424                                 // Let the edit saving system know we should parse the page
2425                                 // *after* a revision ID has been assigned.
2426                                 $this->mOutput->setFlag( 'vary-revision' );
2427                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2428                                 return $this->mRevisionId;
2429                         case 'revisionday':
2430                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2431                         case 'revisionday2':
2432                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2433                         case 'revisionmonth':
2434                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2435                         case 'revisionyear':
2436                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2437                         case 'revisiontimestamp':
2438                                 return $this->getRevisionTimestamp();
2439                         case 'namespace':
2440                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2441                         case 'namespacee':
2442                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2443                         case 'talkspace':
2444                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2445                         case 'talkspacee':
2446                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2447                         case 'subjectspace':
2448                                 return $this->mTitle->getSubjectNsText();
2449                         case 'subjectspacee':
2450                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2451                         case 'currentdayname':
2452                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2453                         case 'currentyear':
2454                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2455                         case 'currenttime':
2456                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2457                         case 'currenthour':
2458                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2459                         case 'currentweek':
2460                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2461                                 // int to remove the padding
2462                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2463                         case 'currentdow':
2464                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2465                         case 'localdayname':
2466                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2467                         case 'localyear':
2468                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2469                         case 'localtime':
2470                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2471                         case 'localhour':
2472                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2473                         case 'localweek':
2474                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2475                                 // int to remove the padding
2476                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2477                         case 'localdow':
2478                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2479                         case 'numberofarticles':
2480                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2481                         case 'numberoffiles':
2482                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2483                         case 'numberofusers':
2484                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2485                         case 'numberofpages':
2486                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2487                         case 'numberofadmins':
2488                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2489                         case 'numberofedits':
2490                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2491                         case 'currenttimestamp':
2492                                 return $varCache[$index] = wfTimestampNow();
2493                         case 'localtimestamp':
2494                                 return $varCache[$index] = $localTimestamp;
2495                         case 'currentversion':
2496                                 return $varCache[$index] = SpecialVersion::getVersion();
2497                         case 'sitename':
2498                                 return $wgSitename;
2499                         case 'server':
2500                                 return $wgServer;
2501                         case 'servername':
2502                                 return $wgServerName;
2503                         case 'scriptpath':
2504                                 return $wgScriptPath;
2505                         case 'directionmark':
2506                                 return $wgContLang->getDirMark();
2507                         case 'contentlanguage':
2508                                 global $wgContLanguageCode;
2509                                 return $wgContLanguageCode;
2510                         default:
2511                                 $ret = null;
2512                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2513                                         return $ret;
2514                                 else
2515                                         return null;
2516                 }
2517         }
2518
2519         /**
2520          * initialise the magic variables (like CURRENTMONTHNAME)
2521          *
2522          * @private
2523          */
2524         function initialiseVariables() {
2525                 $fname = 'Parser::initialiseVariables';
2526                 wfProfileIn( $fname );
2527                 $variableIDs = MagicWord::getVariableIDs();
2528
2529                 $this->mVariables = new MagicWordArray( $variableIDs );
2530                 wfProfileOut( $fname );
2531         }
2532
2533         /**
2534          * Parse any parentheses in format ((title|part|part)} and return the document tree
2535          * This is the ghost of replace_variables().
2536          *
2537          * @param string $text The text to parse
2538          * @private
2539          */
2540         function preprocessToDom ( $text ) {
2541                 wfProfileIn( __METHOD__ );
2542                 wfProfileIn( __METHOD__.'-makexml' );
2543
2544                 static $msgRules, $normalRules;
2545                 if ( !$msgRules ) {
2546                         $msgRules = array(
2547                                 '{' => array(
2548                                         'end' => '}',
2549                                         'names' => array(
2550                                                 2 => 'template',
2551                                         ),
2552                                         'min' => 2,
2553                                         'max' => 2,
2554                                 ),
2555                                 '[' => array(
2556                                         'end' => ']',
2557                                         'names' => array( 2 => null ),
2558                                         'min' => 2,
2559                                         'max' => 2,
2560                                 )
2561                         );
2562                         $normalRules = array(
2563                                 '{' => array(
2564                                         'end' => '}',
2565                                         'names' => array(
2566                                                 2 => 'template',
2567                                                 3 => 'tplarg',
2568                                         ),
2569                                         'min' => 2,
2570                                         'max' => 3,
2571                                 ),
2572                                 '[' => array(
2573                                         'end' => ']',
2574                                         'names' => array( 2 => null ),
2575                                         'min' => 2,
2576                                         'max' => 2,
2577                                 )
2578                         );
2579                 }
2580                 if ( $this->ot['msg'] ) {
2581                         $rules = $msgRules;
2582                 } else {
2583                         $rules = $normalRules;
2584                 }
2585
2586                 $extElements = implode( '|', $this->getStripList() );
2587                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2588                 $extElementsRegex = "/($extElements)(?:\s|\/>|>)|(!--)/iA";
2589
2590                 $stack = array();      # Stack of unclosed parentheses
2591                 $stackIndex = -1;      # Stack read pointer
2592
2593                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2594
2595                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2596                 $topAccum = '<root>';      # Top level text accumulator
2597                 $accum =& $topAccum;            # Current text accumulator
2598                 $findEquals = false;            # True to find equals signs in arguments
2599                 $findHeading = false;           # True to look at LF characters for possible headings
2600                 $findPipe = false;              # True to take notice of pipe characters
2601                 $headingIndex = 1;
2602                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2603
2604                 while ( $i < strlen( $text ) ) {
2605                         if ( $i == -1 ) {
2606                                 $found = 'line-start';
2607                                 $curChar = '';
2608                         } else {
2609                                 # Find next opening brace, closing brace or pipe
2610                                 $search = $searchBase;
2611                                 if ( $stackIndex == -1 ) {
2612                                         $currentClosing = '';
2613                                         // Look for headings only at the top stack level
2614                                         // Among other things, this resolves the ambiguity between =
2615                                         // for headings and = for template arguments
2616                                         $search .= "\n";
2617                                 } else {
2618                                         $currentClosing = $stack[$stackIndex]['close'];
2619                                         $search .= $currentClosing;
2620                                 }
2621                                 if ( $findPipe ) {
2622                                         $search .= '|';
2623                                 }
2624                                 if ( $findEquals ) {
2625                                         $search .= '=';
2626                                 }
2627                                 $rule = null;
2628                                 # Output literal section, advance input counter
2629                                 $literalLength = strcspn( $text, $search, $i );
2630                                 if ( $literalLength > 0 ) {
2631                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2632                                         $i += $literalLength;
2633                                 }
2634                                 if ( $i >= strlen( $text ) ) {
2635                                         if ( $currentClosing == "\n" ) {
2636                                                 // Do a past-the-end run to finish off the heading
2637                                                 $curChar = '';
2638                                                 $found = 'line-end';
2639                                         } else {
2640                                                 # All done
2641                                                 break;
2642                                         }
2643                                 } else {
2644                                         $curChar = $text[$i];
2645                                         if ( $curChar == '|' ) {
2646                                                 $found = 'pipe';
2647                                         } elseif ( $curChar == '=' ) {
2648                                                 $found = 'equals';
2649                                         } elseif ( $curChar == '<' ) {
2650                                                 $found = 'angle';
2651                                         } elseif ( $curChar == "\n" ) {
2652                                                 if ( $stackIndex == -1 ) {
2653                                                         $found = 'line-start';
2654                                                 } else {
2655                                                         $found = 'line-end';
2656                                                 }
2657                                         } elseif ( $curChar == $currentClosing ) {
2658                                                 $found = 'close';
2659                                         } elseif ( isset( $rules[$curChar] ) ) {
2660                                                 $found = 'open';
2661                                                 $rule = $rules[$curChar];
2662                                         } else {
2663                                                 # Some versions of PHP have a strcspn which stops on null characters
2664                                                 # Ignore and continue
2665                                                 ++$i;
2666                                                 continue;
2667                                         }
2668                                 }
2669                         }
2670
2671                         if ( $found == 'angle' ) {
2672                                 $matches = false;
2673                                 // Determine element name
2674                                 if ( !preg_match( $extElementsRegex, $text, $matches, 0, $i + 1 ) ) {
2675                                         // Element name missing or not listed
2676                                         $accum .= '&lt;';
2677                                         ++$i;
2678                                         continue;
2679                                 }
2680                                 // Handle comments
2681                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2682                                         // HTML comment, scan to end
2683                                         $endpos = strpos( $text, '-->', $i + 4 );
2684                                         if ( $endpos === false ) {
2685                                                 // Unclosed comment in input, runs to end
2686                                                 $accum .= htmlspecialchars( substr( $text, $i ) );
2687                                                 if ( $this->ot['html'] ) {
2688                                                         // Close it so later stripping can remove it
2689                                                         $accum .= htmlspecialchars( '-->' );
2690                                                 }
2691                                                 $i = strlen( $text );
2692                                                 continue;
2693                                         }
2694                                         $accum .= htmlspecialchars( substr( $text, $i, $endpos - $i + 3 ) );
2695                                         #$inner = substr( $text, $i + 4, $endpos - $i - 4 );
2696                                         #$accum .= '<ext><name>!--</name><inner>' . htmlspecialchars( $inner ) . '</inner></ext>';
2697                                         $i = $endpos + 3;
2698                                         continue;
2699                                 }
2700                                 $name = $matches[1];
2701                                 $attrStart = $i + strlen( $name ) + 1;
2702
2703                                 // Find end of tag
2704                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2705                                 if ( $tagEndPos === false ) {
2706                                         // Infinite backtrack
2707                                         // Disable tag search to prevent worst-case O(N^2) performance
2708                                         $noMoreGT = true;
2709                                         $accum .= '&lt;';
2710                                         ++$i;
2711                                         continue;
2712                                 }
2713                                 if ( $text[$tagEndPos-1] == '/' ) {
2714                                         $attrEnd = $tagEndPos - 1;
2715                                         $inner = null;
2716                                         $i = $tagEndPos + 1;
2717                                         $close = '';
2718                                 } else {
2719                                         $attrEnd = $tagEndPos;
2720                                         // Find closing tag
2721                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2722                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2723                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2724                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2725                                         } else {
2726                                                 // No end tag -- let it run out to the end of the text.
2727                                                 $inner = substr( $text, $tagEndPos + 1 );
2728                                                 $i = strlen( $text );
2729                                                 $close = '';
2730                                         }
2731                                 }
2732                                 $accum .= '<ext>';
2733                                 if ( $attrEnd <= $attrStart ) {
2734                                         $attr = '';
2735                                 } else {
2736                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2737                                 }
2738                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2739                                         // Note that the attr element contains the whitespace between name and attribute,
2740                                         // this is necessary for precise reconstruction during pre-save transform.
2741                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2742                                 if ( $inner !== null ) {
2743                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2744                                 }
2745                                 $accum .= $close . '</ext>';
2746                         }
2747
2748                         elseif ( $found == 'line-start' ) {
2749                                 // Is this the start of a heading?
2750                                 // Line break belongs before the heading element in any case
2751                                 $accum .= $curChar;
2752                                 $i++;
2753
2754                                 $count = strspn( $text, '=', $i, 6 );
2755                                 if ( $count > 0 ) {
2756                                         $piece = array(
2757                                                 'open' => "\n",
2758                                                 'close' => "\n",
2759                                                 'parts' => array( str_repeat( '=', $count ) ),
2760                                                 'count' => $count );
2761                                         $stack[++$stackIndex] = $piece;
2762                                         $i += $count;
2763                                         $accum =& $stack[$stackIndex]['parts'][0];
2764                                         $findPipe = false;
2765                                 }
2766                         }
2767
2768                         elseif ( $found == 'line-end' ) {
2769                                 $piece = $stack[$stackIndex];
2770                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2771                                 assert( $piece['open'] == "\n" );
2772                                 assert( $stackIndex == 0 );
2773                                 // Search back through the accumulator to see if it has a proper close
2774                                 // No efficient way to do this in PHP AFAICT: strrev, PCRE search with $ anchor
2775                                 // and rtrim are all O(N) in total size. Optimal would be O(N) in trailing
2776                                 // whitespace size only.
2777                                 $m = false;
2778                                 $count = $piece['count'];
2779                                 if ( preg_match( "/(={{$count}})\s*$/", $accum, $m, 0, $count ) ) {
2780                                         // Found match, output <h>
2781                                         $count = min( strlen( $m[1] ), $count );
2782                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2783                                         $headingIndex++;
2784                                 } else {
2785                                         // No match, no <h>, just pass down the inner text
2786                                         $element = $accum;
2787                                 }
2788                                 // Unwind the stack
2789                                 // Headings can only occur on the top level, so this is a bit simpler than the
2790                                 // generic stack unwind operation in the close case
2791                                 unset( $stack[$stackIndex--] );
2792                                 $accum =& $topAccum;
2793                                 $findEquals = false;
2794                                 $findPipe = false;
2795
2796                                 // Append the result to the enclosing accumulator
2797                                 $accum .= $element;
2798                                 // Note that we do NOT increment the input pointer.
2799                                 // This is because the closing linebreak could be the opening linebreak of
2800                                 // another heading. Infinite loops are avoided because the next iteration MUST
2801                                 // hit the heading open case above, which unconditionally increments the
2802                                 // input pointer.
2803                         }
2804
2805                         elseif ( $found == 'open' ) {
2806                                 # count opening brace characters
2807                                 $count = strspn( $text, $curChar, $i );
2808
2809                                 # we need to add to stack only if opening brace count is enough for one of the rules
2810                                 if ( $count >= $rule['min'] ) {
2811                                         # Add it to the stack
2812                                         $piece = array(
2813                                                 'open' => $curChar,
2814                                                 'close' => $rule['end'],
2815                                                 'count' => $count,
2816                                                 'parts' => array( '' ),
2817                                                 'eqpos' => array(),
2818                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2819                                         );
2820
2821                                         $stackIndex ++;
2822                                         $stack[$stackIndex] = $piece;
2823                                         $accum =& $stack[$stackIndex]['parts'][0];
2824                                         $findEquals = false;
2825                                         $findPipe = true;
2826                                 } else {
2827                                         # Add literal brace(s)
2828                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2829                                 }
2830                                 $i += $count;
2831                         }
2832
2833                         elseif ( $found == 'close' ) {
2834                                 $piece = $stack[$stackIndex];
2835                                 # lets check if there are enough characters for closing brace
2836                                 $maxCount = $piece['count'];
2837                                 $count = strspn( $text, $curChar, $i, $maxCount );
2838
2839                                 # check for maximum matching characters (if there are 5 closing
2840                                 # characters, we will probably need only 3 - depending on the rules)
2841                                 $matchingCount = 0;
2842                                 $rule = $rules[$piece['open']];
2843                                 if ( $count > $rule['max'] ) {
2844                                         # The specified maximum exists in the callback array, unless the caller
2845                                         # has made an error
2846                                         $matchingCount = $rule['max'];
2847                                 } else {
2848                                         # Count is less than the maximum
2849                                         # Skip any gaps in the callback array to find the true largest match
2850                                         # Need to use array_key_exists not isset because the callback can be null
2851                                         $matchingCount = $count;
2852                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2853                                                 --$matchingCount;
2854                                         }
2855                                 }
2856
2857                                 if ($matchingCount <= 0) {
2858                                         # No matching element found in callback array
2859                                         # Output a literal closing brace and continue
2860                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2861                                         $i += $count;
2862                                         continue;
2863                                 }
2864                                 $name = $rule['names'][$matchingCount];
2865                                 if ( $name === null ) {
2866                                         // No element, just literal text
2867                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2868                                                 implode( '|', $piece['parts'] ) .
2869                                                 str_repeat( $rule['end'], $matchingCount );
2870                                 } else {
2871                                         # Create XML element
2872                                         # Note: $parts is already XML, does not need to be encoded further
2873                                         $parts = $piece['parts'];
2874                                         $title = $parts[0];
2875                                         unset( $parts[0] );
2876
2877                                         # The invocation is at the start of the line if lineStart is set in
2878                                         # the stack, and all opening brackets are used up.
2879                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
2880                                                 $attr = ' lineStart="1"';
2881                                         } else {
2882                                                 $attr = '';
2883                                         }
2884
2885                                         $element = "<$name$attr>";
2886                                         $element .= "<title>$title</title>";
2887                                         $argIndex = 1;
2888                                         foreach ( $parts as $partIndex => $part ) {
2889                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
2890                                                         $eqpos = $piece['eqpos'][$partIndex];
2891                                                         list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
2892                                                         list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
2893                                                         $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
2894                                                 } else {
2895                                                         list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
2896                                                         $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
2897                                                         $argIndex++;
2898                                                 }
2899                                         }
2900                                         $element .= "</$name>";
2901                                 }
2902
2903                                 # Advance input pointer
2904                                 $i += $matchingCount;
2905
2906                                 # Unwind the stack
2907                                 unset( $stack[$stackIndex--] );
2908                                 if ( $stackIndex == -1 ) {
2909                                         $accum =& $topAccum;
2910                                         $findEquals = false;
2911                                         $findPipe = false;
2912                                 } else {
2913                                         $partCount = count( $stack[$stackIndex]['parts'] );
2914                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
2915                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
2916                                         $findEquals = $findPipe && $partCount > 1
2917                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
2918                                 }
2919
2920                                 # Re-add the old stack element if it still has unmatched opening characters remaining
2921                                 if ($matchingCount < $piece['count']) {
2922                                         $piece['parts'] = array( '' );
2923                                         $piece['count'] -= $matchingCount;
2924                                         $piece['eqpos'] = array();
2925                                         # do we still qualify for any callback with remaining count?
2926                                         $names = $rules[$piece['open']]['names'];
2927                                         $skippedBraces = 0;
2928                                         $enclosingAccum =& $accum;
2929                                         while ( $piece['count'] ) {
2930                                                 if ( array_key_exists( $piece['count'], $names ) ) {
2931                                                         $stackIndex++;
2932                                                         $stack[$stackIndex] = $piece;
2933                                                         $accum =& $stack[$stackIndex]['parts'][0];
2934                                                         $findEquals = true;
2935                                                         $findPipe = true;
2936                                                         break;
2937                                                 }
2938                                                 --$piece['count'];
2939                                                 $skippedBraces ++;
2940                                         }
2941                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
2942                                 }
2943
2944                                 # Add XML element to the enclosing accumulator
2945                                 $accum .= $element;
2946                         }
2947
2948                         elseif ( $found == 'pipe' ) {
2949                                 $stack[$stackIndex]['parts'][] = '';
2950                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2951                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
2952                                 $findEquals = true;
2953                                 ++$i;
2954                         }
2955
2956                         elseif ( $found == 'equals' ) {
2957                                 $findEquals = false;
2958                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2959                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
2960                                 $accum .= '=';
2961                                 ++$i;
2962                         }
2963                 }
2964
2965                 # Output any remaining unclosed brackets
2966                 foreach ( $stack as $piece ) {
2967                         if ( $piece['open'] == "\n" ) {
2968                                 $topAccum .= $piece['parts'][0];
2969                         } else {
2970                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
2971                         }
2972                 }
2973                 $topAccum .= '</root>';
2974
2975                 wfProfileOut( __METHOD__.'-makexml' );
2976                 wfProfileIn( __METHOD__.'-loadXML' );
2977                 $dom = new DOMDocument;
2978                 if ( !$dom->loadXML( $topAccum ) ) {
2979                         throw new MWException( __METHOD__.' generated invalid XML' );
2980                 }
2981                 wfProfileOut( __METHOD__.'-loadXML' );
2982                 wfProfileOut( __METHOD__ );
2983                 return $dom;
2984         }
2985
2986         /*
2987          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2988          */
2989         public static function splitWhitespace( $s ) {
2990                 $ltrimmed = ltrim( $s );
2991                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2992                 $trimmed = rtrim( $ltrimmed );
2993                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2994                 if ( $diff > 0 ) {
2995                         $w2 = substr( $ltrimmed, -$diff );
2996                 } else {
2997                         $w2 = '';
2998                 }
2999                 return array( $w1, $trimmed, $w2 );
3000         }
3001
3002         /**
3003          * Replace magic variables, templates, and template arguments
3004          * with the appropriate text. Templates are substituted recursively,
3005          * taking care to avoid infinite loops.
3006          *
3007          * Note that the substitution depends on value of $mOutputType:
3008          *  OT_WIKI: only {{subst:}} templates
3009          *  OT_MSG: only magic variables
3010          *  OT_HTML: all templates and magic variables
3011          *
3012          * @param string $tex The text to transform
3013          * @param PPFrame $frame Object describing the arguments passed to the template
3014          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3015          * @private
3016          */
3017         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3018                 # Prevent too big inclusions
3019                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3020                         return $text;
3021                 }
3022
3023                 $fname = __METHOD__;
3024                 wfProfileIn( $fname );
3025
3026                 if ( $frame === false ) {
3027                         $frame = new PPFrame( $this );
3028                 } elseif ( !( $frame instanceof PPFrame ) ) {
3029                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3030                 }
3031
3032                 $dom = $this->preprocessToDom( $text );
3033                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3034                 $text = $frame->expand( $dom, 0, $flags );
3035
3036                 wfProfileOut( $fname );
3037                 return $text;
3038         }
3039
3040         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3041         static function createAssocArgs( $args ) {
3042                 $assocArgs = array();
3043                 $index = 1;
3044                 foreach( $args as $arg ) {
3045                         $eqpos = strpos( $arg, '=' );
3046                         if ( $eqpos === false ) {
3047                                 $assocArgs[$index++] = $arg;
3048                         } else {
3049                                 $name = trim( substr( $arg, 0, $eqpos ) );
3050                                 $value = trim( substr( $arg, $eqpos+1 ) );
3051                                 if ( $value === false ) {
3052                                         $value = '';
3053                                 }
3054                                 if ( $name !== false ) {
3055                                         $assocArgs[$name] = $value;
3056                                 }
3057                         }
3058                 }
3059
3060                 return $assocArgs;
3061         }
3062
3063         /**
3064          * Return the text of a template, after recursively
3065          * replacing any variables or templates within the template.
3066          *
3067          * @param array $piece The parts of the template
3068          *  $piece['text']: matched text
3069          *  $piece['title']: the title, i.e. the part before the |
3070          *  $piece['parts']: the parameter array
3071          * @param PPFrame The current frame, contains template arguments
3072          * @return string the text of the template
3073          * @private
3074          */
3075         function braceSubstitution( $piece, $frame ) {
3076                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3077                 $fname = __METHOD__;
3078                 wfProfileIn( $fname );
3079                 wfProfileIn( __METHOD__.'-setup' );
3080
3081                 # Flags
3082                 $found = false;             # $text has been filled
3083                 $nowiki = false;            # wiki markup in $text should be escaped
3084                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
3085                 $noargs = false;            # Don't replace triple-brace arguments in $text
3086                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3087                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3088                 $isDOM = false;             # $text is a DOM node needing expansion
3089
3090                 # Title object, where $text came from
3091                 $title = NULL;
3092
3093                 # $part1 is the bit before the first |, and must contain only title characters.
3094                 # Various prefixes will be stripped from it later.
3095                 $titleWithSpaces = $frame->expand( $piece['title'] );
3096                 $part1 = trim( $titleWithSpaces );
3097                 $titleText = false;
3098
3099                 # Original title text preserved for various purposes
3100                 $originalTitle = $part1;
3101
3102                 # $args is a list of argument nodes, starting from index 0, not including $part1
3103                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3104                 wfProfileOut( __METHOD__.'-setup' );
3105
3106                 # SUBST
3107                 wfProfileIn( __METHOD__.'-modifiers' );
3108                 if ( !$found ) {
3109                         $mwSubst =& MagicWord::get( 'subst' );
3110                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3111                                 # One of two possibilities is true:
3112                                 # 1) Found SUBST but not in the PST phase
3113                                 # 2) Didn't find SUBST and in the PST phase
3114                                 # In either case, return without further processing
3115                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3116                                 $found = true;
3117                                 $noparse = true;
3118                                 $noargs = true;
3119                         }
3120                 }
3121
3122                 # Variables
3123                 if ( !$found && $args->length == 0 ) {
3124                         $id = $this->mVariables->matchStartToEnd( $part1 );
3125                         if ( $id !== false ) {
3126                                 $text = $this->getVariableValue( $id );
3127                                 $this->mOutput->mContainsOldMagic = true;
3128                                 $found = true;
3129                                 $noparse = true;
3130                                 $noargs = true;
3131                         }
3132                 }
3133
3134                 # MSG, MSGNW and RAW
3135                 if ( !$found ) {
3136                         # Check for MSGNW:
3137                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3138                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3139                                 $nowiki = true;
3140                         } else {
3141                                 # Remove obsolete MSG:
3142                                 $mwMsg =& MagicWord::get( 'msg' );
3143                                 $mwMsg->matchStartAndRemove( $part1 );
3144                         }
3145
3146                         # Check for RAW:
3147                         $mwRaw =& MagicWord::get( 'raw' );
3148                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3149                                 $forceRawInterwiki = true;
3150                         }
3151                 }
3152                 wfProfileOut( __METHOD__.'-modifiers' );
3153
3154                 # Save path level before recursing into functions & templates.
3155                 $lastPathLevel = $this->mTemplatePath;
3156
3157                 # Parser functions
3158                 if ( !$found ) {
3159                         wfProfileIn( __METHOD__ . '-pfunc' );
3160
3161                         $colonPos = strpos( $part1, ':' );
3162                         if ( $colonPos !== false ) {
3163                                 # Case sensitive functions
3164                                 $function = substr( $part1, 0, $colonPos );
3165                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3166                                         $function = $this->mFunctionSynonyms[1][$function];
3167                                 } else {
3168                                         # Case insensitive functions
3169                                         $function = strtolower( $function );
3170                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3171                                                 $function = $this->mFunctionSynonyms[0][$function];
3172                                         } else {
3173                                                 $function = false;
3174                                         }
3175                                 }
3176                                 if ( $function ) {
3177                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3178                                         $initialArgs = array( &$this );
3179                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3180                                         if ( $flags & SFH_OBJECT_ARGS ) {
3181                                                 # Add a frame parameter, and pass the arguments as an array
3182                                                 $allArgs = $initialArgs;
3183                                                 $allArgs[] = $frame;
3184                                                 foreach ( $args as $arg ) {
3185                                                         $funcArgs[] = $arg;
3186                                                 }
3187                                                 $allArgs[] = $funcArgs;
3188                                         } else {
3189                                                 # Convert arguments to plain text
3190                                                 foreach ( $args as $arg ) {
3191                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3192                                                 }
3193                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3194                                         }
3195
3196                                         $result = call_user_func_array( $callback, $allArgs );
3197                                         $found = true;
3198
3199                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3200                                         $noargs = true;
3201                                         $noparse = true;
3202
3203                                         if ( is_array( $result ) ) {
3204                                                 if ( isset( $result[0] ) ) {
3205                                                         $text = $result[0];
3206                                                         unset( $result[0] );
3207                                                 }
3208
3209                                                 // Extract flags into the local scope
3210                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3211                                                 extract( $result );
3212                                         } else {
3213                                                 $text = $result;
3214                                         }
3215                                 }
3216                         }
3217                         wfProfileOut( __METHOD__ . '-pfunc' );
3218                 }
3219
3220                 # Finish mangling title and then check for loops.
3221                 # Set $title to a Title object and $titleText to the PDBK
3222                 if ( !$found ) {
3223                         $ns = NS_TEMPLATE;
3224                         # Split the title into page and subpage
3225                         $subpage = '';
3226                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3227                         if ($subpage !== '') {
3228                                 $ns = $this->mTitle->getNamespace();
3229                         }
3230                         $title = Title::newFromText( $part1, $ns );
3231                         if ( $title ) {
3232                                 $titleText = $title->getPrefixedText();
3233                                 # Check for language variants if the template is not found
3234                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3235                                         $wgContLang->findVariantLink($part1, $title);
3236                                 }
3237                                 # Do infinite loop check
3238                                 if ( isset( $this->mTemplatePath[$titleText] ) ) {
3239                                         $noparse = true;
3240                                         $noargs = true;
3241                                         $found = true;
3242                                         $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
3243                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3244                                 }
3245                         }
3246                 }
3247
3248                 # Load from database
3249                 if ( !$found && $title ) {
3250                         wfProfileIn( __METHOD__ . '-loadtpl' );
3251                         if ( !$title->isExternal() ) {
3252                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3253                                         $text = SpecialPage::capturePath( $title );
3254                                         if ( is_string( $text ) ) {
3255                                                 $found = true;
3256                                                 $noparse = true;
3257                                                 $noargs = true;
3258                                                 $isHTML = true;
3259                                                 $this->disableCache();
3260                                         }
3261                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3262                                         $found = false; //access denied
3263                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3264                                 } else {
3265                                         list( $text, $title ) = $this->getTemplateDom( $title );
3266                                         if ( $text !== false ) {
3267                                                 $found = true;
3268                                                 $isDOM = true;
3269                                         }
3270                                 }
3271
3272                                 # If the title is valid but undisplayable, make a link to it
3273                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3274                                         $text = "[[:$titleText]]";
3275                                         $found = true;
3276                                 }
3277                         } elseif ( $title->isTrans() ) {
3278                                 // Interwiki transclusion
3279                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3280                                         $text = $this->interwikiTransclude( $title, 'render' );
3281                                         $isHTML = true;
3282                                         $noparse = true;
3283                                 } else {
3284                                         $text = $this->interwikiTransclude( $title, 'raw' );
3285                                 }
3286                                 $found = true;
3287                         }
3288                         wfProfileOut( __METHOD__ . '-loadtpl' );
3289                 }
3290
3291                 # Recursive parsing, escaping and link table handling
3292                 # Only for HTML output
3293                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3294                         if ( $isDOM ) {
3295                                 $text = $frame->expand( $text );
3296                         }
3297                         $text = wfEscapeWikiText( $text );
3298                 } elseif ( !$this->ot['msg'] && $found ) {
3299                         if ( $noargs ) {
3300                                 $newFrame = $frame->newChild();
3301                         } else {
3302                                 # Clean up argument array
3303                                 $newFrame = $frame->newChild( $args, $title );
3304                                 # Add a new element to the templace recursion path
3305                                 $this->mTemplatePath[$titleText] = 1;
3306                         }
3307
3308                         if ( !$noparse ) {
3309                                 if ( $isDOM ) {
3310                                         if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
3311                                                 # Expansion is eligible for the empty-frame cache
3312                                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3313                                                         $text = $this->mTplExpandCache[$titleText];
3314                                                 } else {
3315                                                         $text = $newFrame->expand( $text );
3316                                                         $this->mTplExpandCache[$titleText] = $text;
3317                                                 }
3318                                         } else {
3319                                                 $text = $newFrame->expand( $text );
3320                                         }
3321                                 } else {
3322                                         $text = $this->replaceVariables( $text, $newFrame );
3323                                 }
3324
3325                                 # strip woz 'ere 2004-07
3326
3327                                 # Bug 529: if the template begins with a table or block-level
3328                                 # element, it should be treated as beginning a new line.
3329                                 # This behaviour is somewhat controversial.
3330                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3331                                         $text = "\n" . $text;
3332                                 }
3333                         } elseif ( !$noargs ) {
3334                                 # $noparse and !$noargs
3335                                 # Just replace the arguments, not any double-brace items
3336                                 # This is used for rendered interwiki transclusion
3337                                 if ( $isDOM ) {
3338                                         $text = $newFrame->expand( $text, 0, PPFrame::NO_TEMPLATES );
3339                                 } else {
3340                                         $text = $this->replaceVariables( $text, $newFrame, true );
3341                                 }
3342                         } elseif ( $isDOM ) {
3343                                 $text = $frame->expand( $text );
3344                         }
3345                 } elseif ( $isDOM ) {
3346                         $text = $frame->expand( $text, 0, PPFrame::NO_TEMPLATES | PPFrame::NO_ARGS );
3347                 }
3348
3349                 # Prune lower levels off the recursion check path
3350                 $this->mTemplatePath = $lastPathLevel;
3351
3352                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3353                         # Error, oversize inclusion
3354                         $text = "[[$originalTitle]]" .
3355                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3356                         $noparse = true;
3357                         $noargs = true;
3358                 }
3359
3360                 if ( !$found ) {
3361                         wfProfileOut( $fname );
3362                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3363                 } else {
3364                         wfProfileIn( __METHOD__ . '-placeholders' );
3365                         if ( $isHTML ) {
3366                                 # Replace raw HTML by a placeholder
3367                                 # Add a blank line preceding, to prevent it from mucking up
3368                                 # immediately preceding headings
3369                                 $text = "\n\n" . $this->insertStripItem( $text );
3370                         }
3371                         wfProfileOut( __METHOD__ . '-placeholders' );
3372                 }
3373
3374                 # Prune lower levels off the recursion check path
3375                 $this->mTemplatePath = $lastPathLevel;
3376
3377                 if ( !$found ) {
3378                         wfProfileOut( $fname );
3379                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3380                 } else {
3381                         wfProfileOut( $fname );
3382                         return $text;
3383                 }
3384         }
3385
3386         /**
3387          * Get the semi-parsed DOM representation of a template with a given title,
3388          * and its redirect destination title. Cached.
3389          */
3390         function getTemplateDom( $title ) {
3391                 $cacheTitle = $title;
3392                 $titleText = $title->getPrefixedDBkey();
3393
3394                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3395                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3396                         $title = Title::makeTitle( $ns, $dbk );
3397                         $titleText = $title->getPrefixedDBkey();
3398                 }
3399                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3400                         return array( $this->mTplDomCache[$titleText], $title );
3401                 }
3402
3403                 // Cache miss, go to the database
3404                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3405
3406                 if ( $text === false ) {
3407                         $this->mTplDomCache[$titleText] = false;
3408                         return array( false, $title );
3409                 }
3410
3411                 # If there are any <onlyinclude> tags, only include them
3412                 if ( !$this->ot['msg'] ) {
3413                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3414                                 $replacer = new OnlyIncludeReplacer;
3415                                 StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3416                                         array( &$replacer, 'replace' ), $text );
3417                                 $text = $replacer->output;
3418                         }
3419                         # Remove <noinclude> sections and <includeonly> tags
3420                         $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3421                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3422                 }
3423
3424                 $dom = $this->preprocessToDom( $text );
3425
3426                 $this->mTplDomCache[ $titleText ] = $dom;
3427
3428                 if (! $title->equals($cacheTitle)) {
3429                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3430                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3431                 }
3432
3433                 return array( $dom, $title );
3434         }
3435
3436         /**
3437          * Fetch the unparsed text of a template and register a reference to it.
3438          */
3439         function fetchTemplateAndTitle( $title ) {
3440                 $templateCb = $this->mOptions->getTemplateCallback();
3441                 $stuff = call_user_func( $templateCb, $title );
3442                 $text = $stuff['text'];
3443                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3444                 if ( isset( $stuff['deps'] ) ) {
3445                         foreach ( $stuff['deps'] as $dep ) {
3446                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3447                         }
3448                 }
3449                 return array($text,$finalTitle);
3450         }
3451
3452         function fetchTemplate( $title ) {
3453                 $rv = $this->fetchTemplateAndTitle($title);
3454                 return $rv[0];
3455         }
3456
3457         /**
3458          * Static function to get a template
3459          * Can be overridden via ParserOptions::setTemplateCallback().
3460          */
3461         static function statelessFetchTemplate( $title ) {
3462                 $text = $skip = false;
3463                 $finalTitle = $title;
3464                 $deps = array();
3465
3466                 // Loop to fetch the article, with up to 1 redirect
3467                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3468                         # Give extensions a chance to select the revision instead
3469                         $id = false; // Assume current
3470                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3471
3472                         if( $skip ) {
3473                                 $text = false;
3474                                 $deps[] = array(
3475                                         'title' => $title,
3476                                         'page_id' => $title->getArticleID(),
3477                                         'rev_id' => null );
3478                                 break;
3479                         }
3480                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3481                         $rev_id = $rev ? $rev->getId() : 0;
3482
3483                         $deps[] = array(
3484                                 'title' => $title,
3485                                 'page_id' => $title->getArticleID(),
3486                                 'rev_id' => $rev_id );
3487
3488                         if( $rev ) {
3489                                 $text = $rev->getText();
3490                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3491                                 global $wgLang;
3492                                 $message = $wgLang->lcfirst( $title->getText() );
3493                                 $text = wfMsgForContentNoTrans( $message );
3494                                 if( wfEmptyMsg( $message, $text ) ) {
3495                                         $text = false;
3496                                         break;
3497                                 }
3498                         } else {
3499                                 break;
3500                         }
3501                         if ( $text === false ) {
3502                                 break;
3503                         }
3504                         // Redirect?
3505                         $finalTitle = $title;
3506                         $title = Title::newFromRedirect( $text );
3507                 }
3508                 return array(
3509                         'text' => $text,
3510                         'finalTitle' => $finalTitle,
3511                         'deps' => $deps );
3512         }
3513
3514         /**
3515          * Transclude an interwiki link.
3516          */
3517         function interwikiTransclude( $title, $action ) {
3518                 global $wgEnableScaryTranscluding;
3519
3520                 if (!$wgEnableScaryTranscluding)
3521                         return wfMsg('scarytranscludedisabled');
3522
3523                 $url = $title->getFullUrl( "action=$action" );
3524
3525                 if (strlen($url) > 255)
3526                         return wfMsg('scarytranscludetoolong');
3527                 return $this->fetchScaryTemplateMaybeFromCache($url);
3528         }
3529
3530         function fetchScaryTemplateMaybeFromCache($url) {
3531                 global $wgTranscludeCacheExpiry;
3532                 $dbr = wfGetDB(DB_SLAVE);
3533                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3534                                 array('tc_url' => $url));
3535                 if ($obj) {
3536                         $time = $obj->tc_time;
3537                         $text = $obj->tc_contents;
3538                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3539                                 return $text;
3540                         }
3541                 }
3542
3543                 $text = Http::get($url);
3544                 if (!$text)
3545                         return wfMsg('scarytranscludefailed', $url);
3546
3547                 $dbw = wfGetDB(DB_MASTER);
3548                 $dbw->replace('transcache', array('tc_url'), array(
3549                         'tc_url' => $url,
3550                         'tc_time' => time(),
3551                         'tc_contents' => $text));
3552                 return $text;
3553         }
3554
3555
3556         /**
3557          * Triple brace replacement -- used for template arguments
3558          * @private
3559          */
3560         function argSubstitution( $piece, $frame ) {
3561                 wfProfileIn( __METHOD__ );
3562
3563                 $text = false;
3564                 $error = false;
3565                 $parts = $piece['parts'];
3566                 $argWithSpaces = $frame->expand( $piece['title'] );
3567                 $arg = trim( $argWithSpaces );
3568
3569                 if ( isset( $frame->args[$arg] ) ) {
3570                         $text = $frame->parent->expand( $frame->args[$arg] );
3571                 } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3572                         $text = $frame->expand( $parts->item( 0 ) );
3573                 }
3574                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3575                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3576                 }
3577
3578                 if ( $text === false ) {
3579                         $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
3580                 }
3581                 if ( $error !== false ) {
3582                         $text .= $error;
3583                 }
3584
3585                 wfProfileOut( __METHOD__ );
3586                 return $text;
3587         }
3588
3589         /**
3590          * Return the text to be used for a given extension tag.
3591          * This is the ghost of strip().
3592          *
3593          * @param array $params Associative array of parameters:
3594          *     name       DOMNode for the tag name
3595          *     attrText   DOMNode for unparsed text where tag attributes are thought to be
3596          *     inner      Contents of extension element
3597          *     noClose    Original text did not have a close tag
3598          * @param PPFrame $frame
3599          */
3600         function extensionSubstitution( $params, $frame ) {
3601                 global $wgRawHtml, $wgContLang;
3602                 static $n = 1;
3603
3604                 $name = $frame->expand( $params['name'] );
3605                 $attrText = is_null( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3606                 $content = is_null( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3607
3608                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3609
3610                 if ( $this->ot['html'] ) {
3611                         if ( $name == '!--' ) {
3612                                 return '';
3613                         }
3614                         $name = strtolower( $name );
3615
3616                         $params = Sanitizer::decodeTagAttributes( $attrText );
3617                         switch ( $name ) {
3618                                 case 'html':
3619                                         if( $wgRawHtml ) {
3620                                                 $output = $content;
3621                                                 break;
3622                                         } else {
3623                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3624                                         }
3625                                 case 'nowiki':
3626                                         $output = Xml::escapeTagsOnly( $content );
3627                                         break;
3628                                 case 'math':
3629                                         $output = $wgContLang->armourMath(
3630                                                 MathRenderer::renderMath( $content, $params ) );
3631                                         break;
3632                                 case 'gallery':
3633                                         $output = $this->renderImageGallery( $content, $params );
3634                                         break;
3635                                 default:
3636                                         if( isset( $this->mTagHooks[$name] ) ) {
3637                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3638                                                         array( $content, $params, $this ) );
3639                                         } else {
3640                                                 throw new MWException( "Invalid call hook $name" );
3641                                         }
3642                         }
3643                 } else {
3644                         if ( $name == '!--' ) {
3645                                 $output = '<!--' . $content . '-->';
3646                         } else {
3647                                 if ( $content === null ) {
3648                                         $output = "<$name$attrText/>";
3649                                 } else {
3650                                         $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3651                                         $output = "<$name$attrText>$content$close";
3652                                 }
3653                         }
3654                 }
3655
3656                 if ( $name == 'html' || $name == 'nowiki' ) {
3657                         $this->mStripState->nowiki->setPair( $marker, $output );
3658                 } else {
3659                         $this->mStripState->general->setPair( $marker, $output );
3660                 }
3661                 return $marker;
3662         }
3663
3664         /**
3665          * Increment an include size counter
3666          *
3667          * @param string $type The type of expansion
3668          * @param integer $size The size of the text
3669          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3670          */
3671         function incrementIncludeSize( $type, $size ) {
3672                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3673                         return false;
3674                 } else {
3675                         $this->mIncludeSizes[$type] += $size;
3676                         return true;
3677                 }
3678         }
3679
3680         /**
3681          * Detect __NOGALLERY__ magic word and set a placeholder
3682          */
3683         function stripNoGallery( &$text ) {
3684                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3685                 # do not add TOC
3686                 $mw = MagicWord::get( 'nogallery' );
3687                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3688         }
3689
3690         /**
3691          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3692          * placeholder that will then be replaced by the real TOC in
3693          * ->formatHeadings, this works because at this points real
3694          * comments will have already been discarded by the sanitizer.
3695          *
3696          * Any additional __TOC__ magic words left over will be discarded
3697          * as there can only be one TOC on the page.
3698          */
3699         function stripToc( $text ) {
3700                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3701                 # do not add TOC
3702                 $mw = MagicWord::get( 'notoc' );
3703                 if( $mw->matchAndRemove( $text ) ) {
3704                         $this->mShowToc = false;
3705                 }
3706
3707                 $mw = MagicWord::get( 'toc' );
3708                 if( $mw->match( $text ) ) {
3709                         $this->mShowToc = true;
3710                         $this->mForceTocPosition = true;
3711
3712                         // Set a placeholder. At the end we'll fill it in with the TOC.
3713                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3714
3715                         // Only keep the first one.
3716                         $text = $mw->replace( '', $text );
3717                 }
3718                 return $text;
3719         }
3720
3721         /**
3722          * This function accomplishes several tasks:
3723          * 1) Auto-number headings if that option is enabled
3724          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3725          * 3) Add a Table of contents on the top for users who have enabled the option
3726          * 4) Auto-anchor headings
3727          *
3728          * It loops through all headlines, collects the necessary data, then splits up the
3729          * string and re-inserts the newly formatted headlines.
3730          *
3731          * @param string $text
3732          * @param boolean $isMain
3733          * @private
3734          */
3735         function formatHeadings( $text, $isMain=true ) {
3736                 global $wgMaxTocLevel, $wgContLang;
3737
3738                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3739                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3740                         $showEditLink = 0;
3741                 } else {
3742                         $showEditLink = $this->mOptions->getEditSection();
3743                 }
3744
3745                 # Inhibit editsection links if requested in the page
3746                 $esw =& MagicWord::get( 'noeditsection' );
3747                 if( $esw->matchAndRemove( $text ) ) {
3748                         $showEditLink = 0;
3749                 }
3750
3751                 # Get all headlines for numbering them and adding funky stuff like [edit]
3752                 # links - this is for later, but we need the number of headlines right now
3753                 $matches = array();
3754                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3755
3756                 # if there are fewer than 4 headlines in the article, do not show TOC
3757                 # unless it's been explicitly enabled.
3758                 $enoughToc = $this->mShowToc &&
3759                         (($numMatches >= 4) || $this->mForceTocPosition);
3760
3761                 # Allow user to stipulate that a page should have a "new section"
3762                 # link added via __NEWSECTIONLINK__
3763                 $mw =& MagicWord::get( 'newsectionlink' );
3764                 if( $mw->matchAndRemove( $text ) )
3765                         $this->mOutput->setNewSection( true );
3766
3767                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3768                 # override above conditions and always show TOC above first header
3769                 $mw =& MagicWord::get( 'forcetoc' );
3770                 if ($mw->matchAndRemove( $text ) ) {
3771                         $this->mShowToc = true;
3772                         $enoughToc = true;
3773                 }
3774
3775                 # We need this to perform operations on the HTML
3776                 $sk = $this->mOptions->getSkin();
3777
3778                 # headline counter
3779                 $headlineCount = 0;
3780                 $numVisible = 0;
3781
3782                 # Ugh .. the TOC should have neat indentation levels which can be
3783                 # passed to the skin functions. These are determined here
3784                 $toc = '';
3785                 $full = '';
3786                 $head = array();
3787                 $sublevelCount = array();
3788                 $levelCount = array();
3789                 $toclevel = 0;
3790                 $level = 0;
3791                 $prevlevel = 0;
3792                 $toclevel = 0;
3793                 $prevtoclevel = 0;
3794                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3795                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3796
3797                 foreach( $matches[3] as $headline ) {
3798                         $isTemplate = false;
3799                         $titleText = false;
3800                         $sectionIndex = false;
3801                         $numbering = '';
3802                         $markerMatches = array();
3803                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3804                                 $serial = $markerMatches[1];
3805                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3806                                 $isTemplate = ($titleText != $baseTitleText);
3807                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3808                         }
3809
3810                         if( $toclevel ) {
3811                                 $prevlevel = $level;
3812                                 $prevtoclevel = $toclevel;
3813                         }
3814                         $level = $matches[1][$headlineCount];
3815
3816                         if( $doNumberHeadings || $enoughToc ) {
3817
3818                                 if ( $level > $prevlevel ) {
3819                                         # Increase TOC level
3820                                         $toclevel++;
3821                                         $sublevelCount[$toclevel] = 0;
3822                                         if( $toclevel<$wgMaxTocLevel ) {
3823                                                 $prevtoclevel = $toclevel;
3824                                                 $toc .= $sk->tocIndent();
3825                                                 $numVisible++;
3826                                         }
3827                                 }
3828                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3829                                         # Decrease TOC level, find level to jump to
3830
3831                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3832                                                 # Can only go down to level 1
3833                                                 $toclevel = 1;
3834                                         } else {
3835                                                 for ($i = $toclevel; $i > 0; $i--) {
3836                                                         if ( $levelCount[$i] == $level ) {
3837                                                                 # Found last matching level
3838                                                                 $toclevel = $i;
3839                                                                 break;
3840                                                         }
3841                                                         elseif ( $levelCount[$i] < $level ) {
3842                                                                 # Found first matching level below current level
3843                                                                 $toclevel = $i + 1;
3844                                                                 break;
3845                                                         }
3846                                                 }
3847                                         }
3848                                         if( $toclevel<$wgMaxTocLevel ) {
3849                                                 if($prevtoclevel < $wgMaxTocLevel) {
3850                                                         # Unindent only if the previous toc level was shown :p
3851                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3852                                                 } else {
3853                                                         $toc .= $sk->tocLineEnd();
3854                                                 }
3855                                         }
3856                                 }
3857                                 else {
3858                                         # No change in level, end TOC line
3859                                         if( $toclevel<$wgMaxTocLevel ) {
3860                                                 $toc .= $sk->tocLineEnd();
3861                                         }
3862                                 }
3863
3864                                 $levelCount[$toclevel] = $level;
3865
3866                                 # count number of headlines for each level
3867                                 @$sublevelCount[$toclevel]++;
3868                                 $dot = 0;
3869                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3870                                         if( !empty( $sublevelCount[$i] ) ) {
3871                                                 if( $dot ) {
3872                                                         $numbering .= '.';
3873                                                 }
3874                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3875                                                 $dot = 1;
3876                                         }
3877                                 }
3878                         }
3879
3880                         # The safe header is a version of the header text safe to use for links
3881                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3882                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3883
3884                         # Remove link placeholders by the link text.
3885                         #     <!--LINK number-->
3886                         # turns into
3887                         #     link text with suffix
3888                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3889                                                             "\$this->mLinkHolders['texts'][\$1]",
3890                                                             $safeHeadline );
3891                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3892                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3893                                                             $safeHeadline );
3894
3895                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3896                         $tocline = preg_replace(
3897                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3898                                 array( '',                          '<$1>'),
3899                                 $safeHeadline
3900                         );
3901                         $tocline = trim( $tocline );
3902
3903                         # For the anchor, strip out HTML-y stuff period
3904                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3905                         $safeHeadline = trim( $safeHeadline );
3906
3907                         # Save headline for section edit hint before it's escaped
3908                         $headlineHint = $safeHeadline;
3909                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3910                         $refers[$headlineCount] = $safeHeadline;
3911
3912                         # count how many in assoc. array so we can track dupes in anchors
3913                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3914                         $refcount[$headlineCount] = $refers[$safeHeadline];
3915
3916                         # Don't number the heading if it is the only one (looks silly)
3917                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3918                                 # the two are different if the line contains a link
3919                                 $headline=$numbering . ' ' . $headline;
3920                         }
3921
3922                         # Create the anchor for linking from the TOC to the section
3923                         $anchor = $safeHeadline;
3924                         if($refcount[$headlineCount] > 1 ) {
3925                                 $anchor .= '_' . $refcount[$headlineCount];
3926                         }
3927                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3928                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3929                         }
3930                         # give headline the correct <h#> tag
3931                         if( $showEditLink && $sectionIndex !== false ) {
3932                                 if( $isTemplate )
3933                                         $editlink = $sk->editSectionLinkForOther($titleText, $sectionIndex);
3934                                 else
3935                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3936                         } else {
3937                                 $editlink = '';
3938                         }
3939                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3940
3941                         $headlineCount++;
3942                 }
3943
3944                 # Never ever show TOC if no headers
3945                 if( $numVisible < 1 ) {
3946                         $enoughToc = false;
3947                 }
3948
3949                 if( $enoughToc ) {
3950                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3951                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3952                         }
3953                         $toc = $sk->tocList( $toc );
3954                 }
3955
3956                 # split up and insert constructed headlines
3957
3958                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3959                 $i = 0;
3960
3961                 foreach( $blocks as $block ) {
3962                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3963                                 # This is the [edit] link that appears for the top block of text when
3964                                 # section editing is enabled
3965
3966                                 # Disabled because it broke block formatting
3967                                 # For example, a bullet point in the top line
3968                                 # $full .= $sk->editSectionLink(0);
3969                         }
3970                         $full .= $block;
3971                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3972                                 # Top anchor now in skin
3973                                 $full = $full.$toc;
3974                         }
3975
3976                         if( !empty( $head[$i] ) ) {
3977                                 $full .= $head[$i];
3978                         }
3979                         $i++;
3980                 }
3981                 if( $this->mForceTocPosition ) {
3982                         return str_replace( '<!--MWTOC-->', $toc, $full );
3983                 } else {
3984                         return $full;
3985                 }
3986         }
3987
3988         /**
3989          * Transform wiki markup when saving a page by doing \r\n -> \n
3990          * conversion, substitting signatures, {{subst:}} templates, etc.
3991          *
3992          * @param string $text the text to transform
3993          * @param Title &$title the Title object for the current article
3994          * @param User &$user the User object describing the current user
3995          * @param ParserOptions $options parsing options
3996          * @param bool $clearState whether to clear the parser state first
3997          * @return string the altered wiki markup
3998          * @public
3999          */
4000         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
4001                 $this->mOptions = $options;
4002                 $this->mTitle =& $title;
4003                 $this->setOutputType( OT_WIKI );
4004
4005                 if ( $clearState ) {
4006                         $this->clearState();
4007                 }
4008
4009                 $pairs = array(
4010                         "\r\n" => "\n",
4011                 );
4012                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4013                 $text = $this->pstPass2( $text, $user );
4014                 $text = $this->mStripState->unstripBoth( $text );
4015                 return $text;
4016         }
4017
4018         /**
4019          * Pre-save transform helper function
4020          * @private
4021          */
4022         function pstPass2( $text, $user ) {
4023                 global $wgContLang, $wgLocaltimezone;
4024
4025                 /* Note: This is the timestamp saved as hardcoded wikitext to
4026                  * the database, we use $wgContLang here in order to give
4027                  * everyone the same signature and use the default one rather
4028                  * than the one selected in each user's preferences.
4029                  */
4030                 if ( isset( $wgLocaltimezone ) ) {
4031                         $oldtz = getenv( 'TZ' );
4032                         putenv( 'TZ='.$wgLocaltimezone );
4033                 }
4034                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4035                   ' (' . date( 'T' ) . ')';
4036                 if ( isset( $wgLocaltimezone ) ) {
4037                         putenv( 'TZ='.$oldtz );
4038                 }
4039
4040                 # Variable replacement
4041                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4042                 $text = $this->replaceVariables( $text );
4043
4044                 # Strip out <nowiki> etc. added via replaceVariables
4045                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4046
4047                 # Signatures
4048                 $sigText = $this->getUserSig( $user );
4049                 $text = strtr( $text, array(
4050                         '~~~~~' => $d,
4051                         '~~~~' => "$sigText $d",
4052                         '~~~' => $sigText
4053                 ) );
4054
4055                 # Context links: [[|name]] and [[name (context)|]]
4056                 #
4057                 global $wgLegalTitleChars;
4058                 $tc = "[$wgLegalTitleChars]";
4059                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4060
4061                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4062                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4063                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4064
4065                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4066                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4067                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4068
4069                 $t = $this->mTitle->getText();
4070                 $m = array();
4071                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4072                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4073                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4074                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4075                 } else {
4076                         # if there's no context, don't bother duplicating the title
4077                         $text = preg_replace( $p2, '[[\\1]]', $text );
4078                 }
4079
4080                 # Trim trailing whitespace
4081                 $text = rtrim( $text );
4082
4083                 return $text;
4084         }
4085
4086         /**
4087          * Fetch the user's signature text, if any, and normalize to
4088          * validated, ready-to-insert wikitext.
4089          *
4090          * @param User $user
4091          * @return string
4092          * @private
4093          */
4094         function getUserSig( &$user ) {
4095                 global $wgMaxSigChars;
4096
4097                 $username = $user->getName();
4098                 $nickname = $user->getOption( 'nickname' );
4099                 $nickname = $nickname === '' ? $username : $nickname;
4100
4101                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4102                         $nickname = $username;
4103                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4104                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4105                         # Sig. might contain markup; validate this
4106                         if( $this->validateSig( $nickname ) !== false ) {
4107                                 # Validated; clean up (if needed) and return it
4108                                 return $this->cleanSig( $nickname, true );
4109                         } else {
4110                                 # Failed to validate; fall back to the default
4111                                 $nickname = $username;
4112                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4113                         }
4114                 }
4115
4116                 // Make sure nickname doesnt get a sig in a sig
4117                 $nickname = $this->cleanSigInSig( $nickname );
4118
4119                 # If we're still here, make it a link to the user page
4120                 $userText = wfEscapeWikiText( $username );
4121                 $nickText = wfEscapeWikiText( $nickname );
4122                 if ( $user->isAnon() )  {
4123                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4124                 } else {
4125                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4126                 }
4127         }
4128
4129         /**
4130          * Check that the user's signature contains no bad XML
4131          *
4132          * @param string $text
4133          * @return mixed An expanded string, or false if invalid.
4134          */
4135         function validateSig( $text ) {
4136                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4137         }
4138
4139         /**
4140          * Clean up signature text
4141          *
4142          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4143          * 2) Substitute all transclusions
4144          *
4145          * @param string $text
4146          * @param $parsing Whether we're cleaning (preferences save) or parsing
4147          * @return string Signature text
4148          */
4149         function cleanSig( $text, $parsing = false ) {
4150                 global $wgTitle;
4151                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
4152
4153                 $substWord = MagicWord::get( 'subst' );
4154                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4155                 $substText = '{{' . $substWord->getSynonym( 0 );
4156
4157                 $text = preg_replace( $substRegex, $substText, $text );
4158                 $text = $this->cleanSigInSig( $text );
4159                 $text = $this->replaceVariables( $text );
4160
4161                 $this->clearState();
4162                 return $text;
4163         }
4164
4165         /**
4166          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4167          * @param string $text
4168          * @return string Signature text with /~{3,5}/ removed
4169          */
4170         function cleanSigInSig( $text ) {
4171                 $text = preg_replace( '/~{3,5}/', '', $text );
4172                 return $text;
4173         }
4174
4175         /**
4176          * Set up some variables which are usually set up in parse()
4177          * so that an external function can call some class members with confidence
4178          * @public
4179          */
4180         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4181                 $this->mTitle =& $title;
4182                 $this->mOptions = $options;
4183                 $this->setOutputType( $outputType );
4184                 if ( $clearState ) {
4185                         $this->clearState();
4186                 }
4187         }
4188
4189         /**
4190          * Transform a MediaWiki message by replacing magic variables.
4191          *
4192          * For some unknown reason, it also expands templates, but only to the
4193          * first recursion level. This is wrong and broken, probably introduced
4194          * accidentally during refactoring, but probably relied upon by thousands
4195          * of users.
4196          *
4197          * @param string $text the text to transform
4198          * @param ParserOptions $options  options
4199          * @return string the text with variables substituted
4200          * @public
4201          */
4202         function transformMsg( $text, $options ) {
4203                 global $wgTitle;
4204                 static $executing = false;
4205
4206                 $fname = "Parser::transformMsg";
4207
4208                 # Guard against infinite recursion
4209                 if ( $executing ) {
4210                         return $text;
4211                 }
4212                 $executing = true;
4213
4214                 wfProfileIn($fname);
4215
4216                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4217                         $this->mTitle = $wgTitle;
4218                 } else {
4219                         $this->mTitle = Title::newFromText('msg');
4220                 }
4221                 $this->mOptions = $options;
4222                 $this->setOutputType( OT_MSG );
4223                 $this->clearState();
4224                 $text = $this->replaceVariables( $text );
4225                 $text = $this->mStripState->unstripBoth( $text );
4226
4227                 $executing = false;
4228                 wfProfileOut($fname);
4229                 return $text;
4230         }
4231
4232         /**
4233          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4234          * The callback should have the following form:
4235          *    function myParserHook( $text, $params, &$parser ) { ... }
4236          *
4237          * Transform and return $text. Use $parser for any required context, e.g. use
4238          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4239          *
4240          * @public
4241          *
4242          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4243          * @param mixed $callback The callback function (and object) to use for the tag
4244          *
4245          * @return The old value of the mTagHooks array associated with the hook
4246          */
4247         function setHook( $tag, $callback ) {
4248                 $tag = strtolower( $tag );
4249                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4250                 $this->mTagHooks[$tag] = $callback;
4251                 $this->mStripList[] = $tag;
4252
4253                 return $oldVal;
4254         }
4255
4256         function setTransparentTagHook( $tag, $callback ) {
4257                 $tag = strtolower( $tag );
4258                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4259                 $this->mTransparentTagHooks[$tag] = $callback;
4260
4261                 return $oldVal;
4262         }
4263
4264         /**
4265          * Create a function, e.g. {{sum:1|2|3}}
4266          * The callback function should have the form:
4267          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4268          *
4269          * The callback may either return the text result of the function, or an array with the text
4270          * in element 0, and a number of flags in the other elements. The names of the flags are
4271          * specified in the keys. Valid flags are:
4272          *   found                     The text returned is valid, stop processing the template. This
4273          *                             is on by default.
4274          *   nowiki                    Wiki markup in the return value should be escaped
4275          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4276          *   noargs                    Don't replace triple-brace arguments in the return value
4277          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4278          *
4279          * @public
4280          *
4281          * @param string $id The magic word ID
4282          * @param mixed $callback The callback function (and object) to use
4283          * @param integer $flags a combination of the following flags:
4284          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4285          *
4286          * @return The old callback function for this name, if any
4287          */
4288         function setFunctionHook( $id, $callback, $flags = 0 ) {
4289                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4290                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4291
4292                 # Add to function cache
4293                 $mw = MagicWord::get( $id );
4294                 if( !$mw )
4295                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4296
4297                 $synonyms = $mw->getSynonyms();
4298                 $sensitive = intval( $mw->isCaseSensitive() );
4299
4300                 foreach ( $synonyms as $syn ) {
4301                         # Case
4302                         if ( !$sensitive ) {
4303                                 $syn = strtolower( $syn );
4304                         }
4305                         # Add leading hash
4306                         if ( !( $flags & SFH_NO_HASH ) ) {
4307                                 $syn = '#' . $syn;
4308                         }
4309                         # Remove trailing colon
4310                         if ( substr( $syn, -1, 1 ) == ':' ) {
4311                                 $syn = substr( $syn, 0, -1 );
4312                         }
4313                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4314                 }
4315                 return $oldVal;
4316         }
4317
4318         /**
4319          * Get all registered function hook identifiers
4320          *
4321          * @return array
4322          */
4323         function getFunctionHooks() {
4324                 return array_keys( $this->mFunctionHooks );
4325         }
4326
4327         /**
4328          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4329          * Placeholders created in Skin::makeLinkObj()
4330          * Returns an array of link CSS classes, indexed by PDBK.
4331          * $options is a bit field, RLH_FOR_UPDATE to select for update
4332          */
4333         function replaceLinkHolders( &$text, $options = 0 ) {
4334                 global $wgUser;
4335                 global $wgContLang;
4336
4337                 $fname = 'Parser::replaceLinkHolders';
4338                 wfProfileIn( $fname );
4339
4340                 $pdbks = array();
4341                 $colours = array();
4342                 $linkcolour_ids = array();
4343                 $sk = $this->mOptions->getSkin();
4344                 $linkCache =& LinkCache::singleton();
4345
4346                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4347                         wfProfileIn( $fname.'-check' );
4348                         $dbr = wfGetDB( DB_SLAVE );
4349                         $page = $dbr->tableName( 'page' );
4350                         $threshold = $wgUser->getOption('stubthreshold');
4351
4352                         # Sort by namespace
4353                         asort( $this->mLinkHolders['namespaces'] );
4354
4355                         # Generate query
4356                         $query = false;
4357                         $current = null;
4358                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4359                                 # Make title object
4360                                 $title = $this->mLinkHolders['titles'][$key];
4361
4362                                 # Skip invalid entries.
4363                                 # Result will be ugly, but prevents crash.
4364                                 if ( is_null( $title ) ) {
4365                                         continue;
4366                                 }
4367                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4368
4369                                 # Check if it's a static known link, e.g. interwiki
4370                                 if ( $title->isAlwaysKnown() ) {
4371                                         $colours[$pdbk] = '';
4372                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4373                                         $colours[$pdbk] = '';
4374                                         $this->mOutput->addLink( $title, $id );
4375                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4376                                         $colours[$pdbk] = 'new';
4377                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4378                                         $colours[$pdbk] = 'new';
4379                                 } else {
4380                                         # Not in the link cache, add it to the query
4381                                         if ( !isset( $current ) ) {
4382                                                 $current = $ns;
4383                                                 $query =  "SELECT page_id, page_namespace, page_title";
4384                                                 if ( $threshold > 0 ) {
4385                                                         $query .= ', page_len, page_is_redirect';
4386                                                 }
4387                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4388                                         } elseif ( $current != $ns ) {
4389                                                 $current = $ns;
4390                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4391                                         } else {
4392                                                 $query .= ', ';
4393                                         }
4394
4395                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4396                                 }
4397                         }
4398                         if ( $query ) {
4399                                 $query .= '))';
4400                                 if ( $options & RLH_FOR_UPDATE ) {
4401                                         $query .= ' FOR UPDATE';
4402                                 }
4403
4404                                 $res = $dbr->query( $query, $fname );
4405
4406                                 # Fetch data and form into an associative array
4407                                 # non-existent = broken
4408                                 while ( $s = $dbr->fetchObject($res) ) {
4409                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4410                                         $pdbk = $title->getPrefixedDBkey();
4411                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4412                                         $this->mOutput->addLink( $title, $s->page_id );
4413                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4414                                         //add id to the extension todolist
4415                                         $linkcolour_ids[$s->page_id] = $pdbk;
4416                                 }
4417                                 //pass an array of page_ids to an extension
4418                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4419                         }
4420                         wfProfileOut( $fname.'-check' );
4421
4422                         # Do a second query for different language variants of links and categories
4423                         if($wgContLang->hasVariants()){
4424                                 $linkBatch = new LinkBatch();
4425                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4426                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4427                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4428
4429                                 $categories = $this->mOutput->getCategoryLinks();
4430
4431                                 // Add variants of links to link batch
4432                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4433                                         $title = $this->mLinkHolders['titles'][$key];
4434                                         if ( is_null( $title ) )
4435                                                 continue;
4436
4437                                         $pdbk = $title->getPrefixedDBkey();
4438                                         $titleText = $title->getText();
4439
4440                                         // generate all variants of the link title text
4441                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4442
4443                                         // if link was not found (in first query), add all variants to query
4444                                         if ( !isset($colours[$pdbk]) ){
4445                                                 foreach($allTextVariants as $textVariant){
4446                                                         if($textVariant != $titleText){
4447                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4448                                                                 if(is_null($variantTitle)) continue;
4449                                                                 $linkBatch->addObj( $variantTitle );
4450                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4451                                                         }
4452                                                 }
4453                                         }
4454                                 }
4455
4456                                 // process categories, check if a category exists in some variant
4457                                 foreach( $categories as $category ){
4458                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4459                                         foreach($variants as $variant){
4460                                                 if($variant != $category){
4461                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4462                                                         if(is_null($variantTitle)) continue;
4463                                                         $linkBatch->addObj( $variantTitle );
4464                                                         $categoryMap[$variant] = $category;
4465                                                 }
4466                                         }
4467                                 }
4468
4469
4470                                 if(!$linkBatch->isEmpty()){
4471                                         // construct query
4472                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4473
4474                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4475                                         if ( $threshold > 0 ) {
4476                                                 $variantQuery .= ', page_len, page_is_redirect';
4477                                         }
4478
4479                                         $variantQuery .= " FROM $page WHERE $titleClause";
4480                                         if ( $options & RLH_FOR_UPDATE ) {
4481                                                 $variantQuery .= ' FOR UPDATE';
4482                                         }
4483
4484                                         $varRes = $dbr->query( $variantQuery, $fname );
4485
4486                                         // for each found variants, figure out link holders and replace
4487                                         while ( $s = $dbr->fetchObject($varRes) ) {
4488
4489                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4490                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4491                                                 $vardbk = $variantTitle->getDBkey();
4492
4493                                                 $holderKeys = array();
4494                                                 if(isset($variantMap[$varPdbk])){
4495                                                         $holderKeys = $variantMap[$varPdbk];
4496                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4497                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4498                                                 }
4499
4500                                                 // loop over link holders
4501                                                 foreach($holderKeys as $key){
4502                                                         $title = $this->mLinkHolders['titles'][$key];
4503                                                         if ( is_null( $title ) ) continue;
4504
4505                                                         $pdbk = $title->getPrefixedDBkey();
4506
4507                                                         if(!isset($colours[$pdbk])){
4508                                                                 // found link in some of the variants, replace the link holder data
4509                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4510                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4511
4512                                                                 // set pdbk and colour
4513                                                                 $pdbks[$key] = $varPdbk;
4514                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4515                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4516                                                         }
4517                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4518                                                 }
4519
4520                                                 // check if the object is a variant of a category
4521                                                 if(isset($categoryMap[$vardbk])){
4522                                                         $oldkey = $categoryMap[$vardbk];
4523                                                         if($oldkey != $vardbk)
4524                                                                 $varCategories[$oldkey]=$vardbk;
4525                                                 }
4526                                         }
4527
4528                                         // rebuild the categories in original order (if there are replacements)
4529                                         if(count($varCategories)>0){
4530                                                 $newCats = array();
4531                                                 $originalCats = $this->mOutput->getCategories();
4532                                                 foreach($originalCats as $cat => $sortkey){
4533                                                         // make the replacement
4534                                                         if( array_key_exists($cat,$varCategories) )
4535                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4536                                                         else $newCats[$cat] = $sortkey;
4537                                                 }
4538                                                 $this->mOutput->setCategoryLinks($newCats);
4539                                         }
4540                                 }
4541                         }
4542
4543                         # Construct search and replace arrays
4544                         wfProfileIn( $fname.'-construct' );
4545                         $replacePairs = array();
4546                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4547                                 $pdbk = $pdbks[$key];
4548                                 $searchkey = "<!--LINK $key-->";
4549                                 $title = $this->mLinkHolders['titles'][$key];
4550                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4551                                         $linkCache->addBadLinkObj( $title );
4552                                         $colours[$pdbk] = 'new';
4553                                         $this->mOutput->addLink( $title, 0 );
4554                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4555                                                                         $this->mLinkHolders['texts'][$key],
4556                                                                         $this->mLinkHolders['queries'][$key] );
4557                                 } else {
4558                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4559                                                                         $this->mLinkHolders['texts'][$key],
4560                                                                         $this->mLinkHolders['queries'][$key] );
4561                                 }
4562                         }
4563                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4564                         wfProfileOut( $fname.'-construct' );
4565
4566                         # Do the thing
4567                         wfProfileIn( $fname.'-replace' );
4568                         $text = preg_replace_callback(
4569                                 '/(<!--LINK .*?-->)/',
4570                                 $replacer->cb(),
4571                                 $text);
4572
4573                         wfProfileOut( $fname.'-replace' );
4574                 }
4575
4576                 # Now process interwiki link holders
4577                 # This is quite a bit simpler than internal links
4578                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4579                         wfProfileIn( $fname.'-interwiki' );
4580                         # Make interwiki link HTML
4581                         $replacePairs = array();
4582                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4583                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4584                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4585                         }
4586                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4587
4588                         $text = preg_replace_callback(
4589                                 '/<!--IWLINK (.*?)-->/',
4590                                 $replacer->cb(),
4591                                 $text );
4592                         wfProfileOut( $fname.'-interwiki' );
4593                 }
4594
4595                 wfProfileOut( $fname );
4596                 return $colours;
4597         }
4598
4599         /**
4600          * Replace <!--LINK--> link placeholders with plain text of links
4601          * (not HTML-formatted).
4602          * @param string $text
4603          * @return string
4604          */
4605         function replaceLinkHoldersText( $text ) {
4606                 $fname = 'Parser::replaceLinkHoldersText';
4607                 wfProfileIn( $fname );
4608
4609                 $text = preg_replace_callback(
4610                         '/<!--(LINK|IWLINK) (.*?)-->/',
4611                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4612                         $text );
4613
4614                 wfProfileOut( $fname );
4615                 return $text;
4616         }
4617
4618         /**
4619          * @param array $matches
4620          * @return string
4621          * @private
4622          */
4623         function replaceLinkHoldersTextCallback( $matches ) {
4624                 $type = $matches[1];
4625                 $key  = $matches[2];
4626                 if( $type == 'LINK' ) {
4627                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4628                                 return $this->mLinkHolders['texts'][$key];
4629                         }
4630                 } elseif( $type == 'IWLINK' ) {
4631                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4632                                 return $this->mInterwikiLinkHolders['texts'][$key];
4633                         }
4634                 }
4635                 return $matches[0];
4636         }
4637
4638         /**
4639          * Tag hook handler for 'pre'.
4640          */
4641         function renderPreTag( $text, $attribs ) {
4642                 // Backwards-compatibility hack
4643                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4644
4645                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4646                 return wfOpenElement( 'pre', $attribs ) .
4647                         Xml::escapeTagsOnly( $content ) .
4648                         '</pre>';
4649         }
4650
4651         /**
4652          * Renders an image gallery from a text with one line per image.
4653          * text labels may be given by using |-style alternative text. E.g.
4654          *   Image:one.jpg|The number "1"
4655          *   Image:tree.jpg|A tree
4656          * given as text will return the HTML of a gallery with two images,
4657          * labeled 'The number "1"' and
4658          * 'A tree'.
4659          */
4660         function renderImageGallery( $text, $params ) {
4661                 $ig = new ImageGallery();
4662                 $ig->setContextTitle( $this->mTitle );
4663                 $ig->setShowBytes( false );
4664                 $ig->setShowFilename( false );
4665                 $ig->setParser( $this );
4666                 $ig->setHideBadImages();
4667                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4668                 $ig->useSkin( $this->mOptions->getSkin() );
4669                 $ig->mRevisionId = $this->mRevisionId;
4670
4671                 if( isset( $params['caption'] ) ) {
4672                         $caption = $params['caption'];
4673                         $caption = htmlspecialchars( $caption );
4674                         $caption = $this->replaceInternalLinks( $caption );
4675                         $ig->setCaptionHtml( $caption );
4676                 }
4677                 if( isset( $params['perrow'] ) ) {
4678                         $ig->setPerRow( $params['perrow'] );
4679                 }
4680                 if( isset( $params['widths'] ) ) {
4681                         $ig->setWidths( $params['widths'] );
4682                 }
4683                 if( isset( $params['heights'] ) ) {
4684                         $ig->setHeights( $params['heights'] );
4685                 }
4686
4687                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4688
4689                 $lines = explode( "\n", $text );
4690                 foreach ( $lines as $line ) {
4691                         # match lines like these:
4692                         # Image:someimage.jpg|This is some image
4693                         $matches = array();
4694                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4695                         # Skip empty lines
4696                         if ( count( $matches ) == 0 ) {
4697                                 continue;
4698                         }
4699                         $tp = Title::newFromText( $matches[1] );
4700                         $nt =& $tp;
4701                         if( is_null( $nt ) ) {
4702                                 # Bogus title. Ignore these so we don't bomb out later.
4703                                 continue;
4704                         }
4705                         if ( isset( $matches[3] ) ) {
4706                                 $label = $matches[3];
4707                         } else {
4708                                 $label = '';
4709                         }
4710
4711                         $pout = $this->parse( $label,
4712                                 $this->mTitle,
4713                                 $this->mOptions,
4714                                 false, // Strip whitespace...?
4715                                 false  // Don't clear state!
4716                         );
4717                         $html = $pout->getText();
4718
4719                         $ig->add( $nt, $html );
4720
4721                         # Only add real images (bug #5586)
4722                         if ( $nt->getNamespace() == NS_IMAGE ) {
4723                                 $this->mOutput->addImage( $nt->getDBkey() );
4724                         }
4725                 }
4726                 return $ig->toHTML();
4727         }
4728
4729         function getImageParams( $handler ) {
4730                 if ( $handler ) {
4731                         $handlerClass = get_class( $handler );
4732                 } else {
4733                         $handlerClass = '';
4734                 }
4735                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4736                         // Initialise static lists
4737                         static $internalParamNames = array(
4738                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4739                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4740                                         'bottom', 'text-bottom' ),
4741                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4742                                         'upright', 'border' ),
4743                         );
4744                         static $internalParamMap;
4745                         if ( !$internalParamMap ) {
4746                                 $internalParamMap = array();
4747                                 foreach ( $internalParamNames as $type => $names ) {
4748                                         foreach ( $names as $name ) {
4749                                                 $magicName = str_replace( '-', '_', "img_$name" );
4750                                                 $internalParamMap[$magicName] = array( $type, $name );
4751                                         }
4752                                 }
4753                         }
4754
4755                         // Add handler params
4756                         $paramMap = $internalParamMap;
4757                         if ( $handler ) {
4758                                 $handlerParamMap = $handler->getParamMap();
4759                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4760                                         $paramMap[$magic] = array( 'handler', $paramName );
4761                                 }
4762                         }
4763                         $this->mImageParams[$handlerClass] = $paramMap;
4764                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4765                 }
4766                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4767         }
4768
4769         /**
4770          * Parse image options text and use it to make an image
4771          */
4772         function makeImage( $title, $options ) {
4773                 # @TODO: let the MediaHandler specify its transform parameters
4774                 #
4775                 # Check if the options text is of the form "options|alt text"
4776                 # Options are:
4777                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4778                 #  * left               no resizing, just left align. label is used for alt= only
4779                 #  * right              same, but right aligned
4780                 #  * none               same, but not aligned
4781                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4782                 #  * center             center the image
4783                 #  * framed             Keep original image size, no magnify-button.
4784                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4785                 #  * upright            reduce width for upright images, rounded to full __0 px
4786                 #  * border             draw a 1px border around the image
4787                 # vertical-align values (no % or length right now):
4788                 #  * baseline
4789                 #  * sub
4790                 #  * super
4791                 #  * top
4792                 #  * text-top
4793                 #  * middle
4794                 #  * bottom
4795                 #  * text-bottom
4796
4797                 $parts = array_map( 'trim', explode( '|', $options) );
4798                 $sk = $this->mOptions->getSkin();
4799
4800                 # Give extensions a chance to select the file revision for us
4801                 $skip = $time = false;
4802                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4803
4804                 if ( $skip ) {
4805                         return $sk->makeLinkObj( $title );
4806                 }
4807
4808                 # Get parameter map
4809                 $file = wfFindFile( $title, $time );
4810                 $handler = $file ? $file->getHandler() : false;
4811
4812                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4813
4814                 # Process the input parameters
4815                 $caption = '';
4816                 $params = array( 'frame' => array(), 'handler' => array(),
4817                         'horizAlign' => array(), 'vertAlign' => array() );
4818                 foreach( $parts as $part ) {
4819                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4820                         if ( isset( $paramMap[$magicName] ) ) {
4821                                 list( $type, $paramName ) = $paramMap[$magicName];
4822                                 $params[$type][$paramName] = $value;
4823
4824                                 // Special case; width and height come in one variable together
4825                                 if( $type == 'handler' && $paramName == 'width' ) {
4826                                         $m = array();
4827                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4828                                                 $params[$type]['width'] = intval( $m[1] );
4829                                                 $params[$type]['height'] = intval( $m[2] );
4830                                         } else {
4831                                                 $params[$type]['width'] = intval( $value );
4832                                         }
4833                                 }
4834                         } else {
4835                                 $caption = $part;
4836                         }
4837                 }
4838
4839                 # Process alignment parameters
4840                 if ( $params['horizAlign'] ) {
4841                         $params['frame']['align'] = key( $params['horizAlign'] );
4842                 }
4843                 if ( $params['vertAlign'] ) {
4844                         $params['frame']['valign'] = key( $params['vertAlign'] );
4845                 }
4846
4847                 # Validate the handler parameters
4848                 if ( $handler ) {
4849                         foreach ( $params['handler'] as $name => $value ) {
4850                                 if ( !$handler->validateParam( $name, $value ) ) {
4851                                         unset( $params['handler'][$name] );
4852                                 }
4853                         }
4854                 }
4855
4856                 # Strip bad stuff out of the alt text
4857                 $alt = $this->replaceLinkHoldersText( $caption );
4858
4859                 # make sure there are no placeholders in thumbnail attributes
4860                 # that are later expanded to html- so expand them now and
4861                 # remove the tags
4862                 $alt = $this->mStripState->unstripBoth( $alt );
4863                 $alt = Sanitizer::stripAllTags( $alt );
4864
4865                 $params['frame']['alt'] = $alt;
4866                 $params['frame']['caption'] = $caption;
4867
4868                 # Linker does the rest
4869                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4870
4871                 # Give the handler a chance to modify the parser object
4872                 if ( $handler ) {
4873                         $handler->parserTransformHook( $this, $file );
4874                 }
4875
4876                 return $ret;
4877         }
4878
4879         /**
4880          * Set a flag in the output object indicating that the content is dynamic and
4881          * shouldn't be cached.
4882          */
4883         function disableCache() {
4884                 wfDebug( "Parser output marked as uncacheable.\n" );
4885                 $this->mOutput->mCacheTime = -1;
4886         }
4887
4888         /**#@+
4889          * Callback from the Sanitizer for expanding items found in HTML attribute
4890          * values, so they can be safely tested and escaped.
4891          * @param string $text
4892          * @param PPFrame $frame
4893          * @return string
4894          * @private
4895          */
4896         function attributeStripCallback( &$text, $frame = false ) {
4897                 $text = $this->replaceVariables( $text, $frame );
4898                 $text = $this->mStripState->unstripBoth( $text );
4899                 return $text;
4900         }
4901
4902         /**#@-*/
4903
4904         /**#@+
4905          * Accessor/mutator
4906          */
4907         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4908         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4909         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4910         /**#@-*/
4911
4912         /**#@+
4913          * Accessor
4914          */
4915         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4916         /**#@-*/
4917
4918
4919         /**
4920          * Break wikitext input into sections, and either pull or replace
4921          * some particular section's text.
4922          *
4923          * External callers should use the getSection and replaceSection methods.
4924          *
4925          * @param $text Page wikitext
4926          * @param $section Numbered section. 0 pulls the text before the first
4927          *                 heading; other numbers will pull the given section
4928          *                 along with its lower-level subsections. If the section is
4929          *                 not found, $mode=get will return $newtext, and
4930          *                 $mode=replace will return $text.
4931          * @param $mode One of "get" or "replace"
4932          * @param $newText Replacement text for section data.
4933          * @return string for "get", the extracted section text.
4934          *                for "replace", the whole page with the section replaced.
4935          */
4936         private function extractSections( $text, $section, $mode, $newText='' ) {
4937                 $this->clearState();
4938                 $this->mOptions = new ParserOptions;
4939                 $this->setOutputType( OT_WIKI );
4940                 $curIndex = 0;
4941                 $outText = '';
4942                 $frame = new PPFrame( $this );
4943
4944                 // Preprocess the text
4945                 $dom = $this->preprocessToDom( $text );
4946                 $root = $dom->documentElement;
4947
4948                 // <h> nodes indicate section breaks
4949                 // They can only occur at the top level, so we can find them by iterating the root's children
4950                 $node = $root->firstChild;
4951
4952                 // Find the target section
4953                 if ( $section == 0 ) {
4954                         // Section zero doesn't nest, level=big
4955                         $targetLevel = 1000;
4956                 } else {
4957                         while ( $node ) {
4958                                 if ( $node->nodeName == 'h' ) {
4959                                         if ( $curIndex + 1 == $section ) {
4960                                                 break;
4961                                         }
4962                                         $curIndex++;
4963                                 }
4964                                 if ( $mode == 'replace' ) {
4965                                         $outText .= $frame->expand( $node, 0, PPFrame::RECOVER_ORIG );
4966                                 }
4967                                 $node = $node->nextSibling;
4968                         }
4969                         if ( $node ) {
4970                                 $targetLevel = $node->getAttribute( 'level' );
4971                         }
4972                 }
4973
4974                 if ( !$node ) {
4975                         // Not found
4976                         if ( $mode == 'get' ) {
4977                                 return $newText;
4978                         } else {
4979                                 return $text;
4980                         }
4981                 }
4982
4983                 // Find the end of the section, including nested sections
4984                 do {
4985                         if ( $node->nodeName == 'h' ) {
4986                                 $curIndex++;
4987                                 $curLevel = $node->getAttribute( 'level' );
4988                                 if ( $curIndex != $section && $curLevel <= $targetLevel ) {
4989                                         break;
4990                                 }
4991                         }
4992                         if ( $mode == 'get' ) {
4993                                 $outText .= $frame->expand( $node, 0, PPFrame::RECOVER_ORIG );
4994                         }
4995                         $node = $node->nextSibling;
4996                 } while ( $node );
4997
4998                 // Write out the remainder (in replace mode only)
4999                 if ( $mode == 'replace' ) {
5000                         // Output the replacement text
5001                         // Add two newlines on -- trailing whitespace in $newText is conventionally
5002                         // stripped by the editor, so we need both newlines to restore the paragraph gap
5003                         $outText .= $newText . "\n\n";
5004                         while ( $node ) {
5005                                 $outText .= $frame->expand( $node, 0, PPFrame::RECOVER_ORIG );
5006                                 $node = $node->nextSibling;
5007                         }
5008                 }
5009
5010                 if ( is_string( $outText ) ) {
5011                         // Re-insert stripped tags
5012                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5013                 }
5014
5015                 return $outText;
5016         }
5017
5018         /**
5019          * This function returns the text of a section, specified by a number ($section).
5020          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5021          * the first section before any such heading (section 0).
5022          *
5023          * If a section contains subsections, these are also returned.
5024          *
5025          * @param $text String: text to look in
5026          * @param $section Integer: section number
5027          * @param $deftext: default to return if section is not found
5028          * @return string text of the requested section
5029          */
5030         public function getSection( $text, $section, $deftext='' ) {
5031                 return $this->extractSections( $text, $section, "get", $deftext );
5032         }
5033
5034         public function replaceSection( $oldtext, $section, $text ) {
5035                 return $this->extractSections( $oldtext, $section, "replace", $text );
5036         }
5037
5038         /**
5039          * Get the timestamp associated with the current revision, adjusted for
5040          * the default server-local timestamp
5041          */
5042         function getRevisionTimestamp() {
5043                 if ( is_null( $this->mRevisionTimestamp ) ) {
5044                         wfProfileIn( __METHOD__ );
5045                         global $wgContLang;
5046                         $dbr = wfGetDB( DB_SLAVE );
5047                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5048                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5049
5050                         // Normalize timestamp to internal MW format for timezone processing.
5051                         // This has the added side-effect of replacing a null value with
5052                         // the current time, which gives us more sensible behavior for
5053                         // previews.
5054                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5055
5056                         // The cryptic '' timezone parameter tells to use the site-default
5057                         // timezone offset instead of the user settings.
5058                         //
5059                         // Since this value will be saved into the parser cache, served
5060                         // to other users, and potentially even used inside links and such,
5061                         // it needs to be consistent for all visitors.
5062                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5063
5064                         wfProfileOut( __METHOD__ );
5065                 }
5066                 return $this->mRevisionTimestamp;
5067         }
5068
5069         /**
5070          * Mutator for $mDefaultSort
5071          *
5072          * @param $sort New value
5073          */
5074         public function setDefaultSort( $sort ) {
5075                 $this->mDefaultSort = $sort;
5076         }
5077
5078         /**
5079          * Accessor for $mDefaultSort
5080          * Will use the title/prefixed title if none is set
5081          *
5082          * @return string
5083          */
5084         public function getDefaultSort() {
5085                 if( $this->mDefaultSort !== false ) {
5086                         return $this->mDefaultSort;
5087                 } else {
5088                         return $this->mTitle->getNamespace() == NS_CATEGORY
5089                                         ? $this->mTitle->getText()
5090                                         : $this->mTitle->getPrefixedText();
5091                 }
5092         }
5093
5094         /**
5095          * Try to guess the section anchor name based on a wikitext fragment
5096          * presumably extracted from a heading, for example "Header" from
5097          * "== Header ==".
5098          */
5099         public function guessSectionNameFromWikiText( $text ) {
5100                 # Strip out wikitext links(they break the anchor)
5101                 $text = $this->stripSectionName( $text );
5102                 $headline = Sanitizer::decodeCharReferences( $text );
5103                 # strip out HTML
5104                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5105                 $headline = trim( $headline );
5106                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5107                 $replacearray = array(
5108                         '%3A' => ':',
5109                         '%' => '.'
5110                 );
5111                 return str_replace(
5112                         array_keys( $replacearray ),
5113                         array_values( $replacearray ),
5114                         $sectionanchor );
5115         }
5116
5117         /**
5118          * Strips a text string of wikitext for use in a section anchor
5119          *
5120          * Accepts a text string and then removes all wikitext from the
5121          * string and leaves only the resultant text (i.e. the result of
5122          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5123          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5124          * to create valid section anchors by mimicing the output of the
5125          * parser when headings are parsed.
5126          *
5127          * @param $text string Text string to be stripped of wikitext
5128          * for use in a Section anchor
5129          * @return Filtered text string
5130          */
5131         public function stripSectionName( $text ) {
5132                 # Strip internal link markup
5133                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5134                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5135
5136                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5137                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5138                 # on how many empty links there are on the page - need to figure that out.
5139                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5140
5141                 # Parse wikitext quotes (italics & bold)
5142                 $text = $this->doQuotes($text);
5143
5144                 # Strip HTML tags
5145                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5146                 return $text;
5147         }
5148
5149         /**
5150          * strip/replaceVariables/unstrip for preprocessor regression testing
5151          */
5152         function srvus( $text ) {
5153                 $text = $this->replaceVariables( $text );
5154                 $text = $this->mStripState->unstripBoth( $text );
5155                 return $text;
5156         }
5157 }
5158
5159 /**
5160  * @todo document, briefly.
5161  * @addtogroup Parser
5162  */
5163 class OnlyIncludeReplacer {
5164         var $output = '';
5165
5166         function replace( $matches ) {
5167                 if ( substr( $matches[1], -1 ) == "\n" ) {
5168                         $this->output .= substr( $matches[1], 0, -1 );
5169                 } else {
5170                         $this->output .= $matches[1];
5171                 }
5172         }
5173 }
5174
5175 /**
5176  * @todo document, briefly.
5177  * @addtogroup Parser
5178  */
5179 class StripState {
5180         var $general, $nowiki;
5181
5182         function __construct() {
5183                 $this->general = new ReplacementArray;
5184                 $this->nowiki = new ReplacementArray;
5185         }
5186
5187         function unstripGeneral( $text ) {
5188                 wfProfileIn( __METHOD__ );
5189                 do {
5190                         $oldText = $text;
5191                         $text = $this->general->replace( $text );
5192                 } while ( $text != $oldText );
5193                 wfProfileOut( __METHOD__ );
5194                 return $text;
5195         }
5196
5197         function unstripNoWiki( $text ) {
5198                 wfProfileIn( __METHOD__ );
5199                 do {
5200                         $oldText = $text;
5201                         $text = $this->nowiki->replace( $text );
5202                 } while ( $text != $oldText );
5203                 wfProfileOut( __METHOD__ );
5204                 return $text;
5205         }
5206
5207         function unstripBoth( $text ) {
5208                 wfProfileIn( __METHOD__ );
5209                 do {
5210                         $oldText = $text;
5211                         $text = $this->general->replace( $text );
5212                         $text = $this->nowiki->replace( $text );
5213                 } while ( $text != $oldText );
5214                 wfProfileOut( __METHOD__ );
5215                 return $text;
5216         }
5217 }
5218
5219 /**
5220  * An expansion frame, used as a context to expand the result of preprocessToDom()
5221  */
5222 class PPFrame {
5223         var $parser, $title;
5224
5225         const NO_ARGS = 1;
5226         const NO_TEMPLATES = 2;
5227         const RECOVER_ORIG = 3;
5228
5229         /**
5230          * Construct a new preprocessor frame.
5231          * @param Parser $parser The parent parser
5232          * @param Title $title The context title, or false if there isn't one
5233          */
5234         function __construct( $parser ) {
5235                 $this->parser = $parser;
5236                 $this->title = $parser->mTitle;
5237         }
5238
5239         /**
5240          * Create a new child frame
5241          * $args is optionally a DOMNodeList containing the template arguments
5242          */
5243         function newChild( $args = false, $title = false ) {
5244                 $assocArgs = array();
5245                 if ( $title === false ) {
5246                         $title = $this->title;
5247                 }
5248                 if ( $args !== false ) {
5249                         $xpath = false;
5250                         foreach ( $args as $arg ) {
5251                                 if ( !$xpath ) {
5252                                         $xpath = new DOMXPath( $arg->ownerDocument );
5253                                 }
5254
5255                                 $nameNodes = $xpath->query( 'name', $arg );
5256                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5257                                         // Numbered parameter
5258                                         $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5259                                 } else {
5260                                         // Named parameter
5261                                         $name = $this->expand( $nameNodes->item( 0 ) );
5262                                 }
5263
5264                                 $value = $xpath->query( 'value', $arg );
5265                                 $assocArgs[$name] = $value->item( 0 );
5266                         }
5267                 }
5268                 return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
5269         }
5270
5271         /**
5272          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5273          * using the current context
5274          * @param $root the node
5275          */
5276         function expand( $root, $shallowFlags = 0, $deepFlags = 0 ) {
5277                 if ( is_string( $root ) ) {
5278                         return $root;
5279                 }
5280
5281                 if ( $this->parser->ot['html']
5282                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5283                 {
5284                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5285                 }
5286                 $flags = $shallowFlags | $deepFlags;
5287
5288                 if ( is_array( $root ) ) {
5289                         $s = '';
5290                         foreach ( $root as $node ) {
5291                                 $s .= $this->expand( $node, 0, $deepFlags );
5292                         }
5293                 } elseif ( $root instanceof DOMNodeList ) {
5294                         $s = '';
5295                         foreach ( $root as $node ) {
5296                                 $s .= $this->expand( $node, 0, $deepFlags );
5297                         }
5298                 } elseif ( $root instanceof DOMNode ) {
5299                         if ( $root->nodeType == XML_TEXT_NODE ) {
5300                                 $s = $root->nodeValue;
5301                         } elseif ( $root->nodeName == 'template' ) {
5302                                 # Double-brace expansion
5303                                 $xpath = new DOMXPath( $root->ownerDocument );
5304                                 $titles = $xpath->query( 'title', $root );
5305                                 $title = $titles->item( 0 );
5306                                 $parts = $xpath->query( 'part', $root );
5307                                 if ( $flags & self::NO_TEMPLATES ) {
5308                                         $s = '{{' . $this->implodeWithFlags( '|', 0, $deepFlags, $title, $parts ) . '}}';
5309                                 } else {
5310                                         $lineStart = $root->getAttribute( 'lineStart' );
5311                                         $params = array(
5312                                                 'title' => $title,
5313                                                 'parts' => $parts,
5314                                                 'lineStart' => $lineStart,
5315                                                 'text' => 'FIXME' );
5316                                         $s = $this->parser->braceSubstitution( $params, $this );
5317                                 }
5318                         } elseif ( $root->nodeName == 'tplarg' ) {
5319                                 # Triple-brace expansion
5320                                 $xpath = new DOMXPath( $root->ownerDocument );
5321                                 $titles = $xpath->query( 'title', $root );
5322                                 $title = $titles->item( 0 );
5323                                 $parts = $xpath->query( 'part', $root );
5324                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5325                                         $s = '{{{' . $this->implode( '|', 0, $deepFlags, $title, $parts ) . '}}}';
5326                                 } else {
5327                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5328                                         $s = $this->parser->argSubstitution( $params, $this );
5329                                 }
5330                         } elseif ( $root->nodeName == 'ext' ) {
5331                                 # Extension tag
5332                                 $xpath = new DOMXPath( $root->ownerDocument );
5333                                 $names = $xpath->query( 'name', $root );
5334                                 $attrs = $xpath->query( 'attr', $root );
5335                                 $inners = $xpath->query( 'inner', $root );
5336                                 $closes = $xpath->query( 'close', $root );
5337                                 $params = array(
5338                                         'name' => $names->item( 0 ),
5339                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5340                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5341                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5342                                 );
5343                                 $s = $this->parser->extensionSubstitution( $params, $this );
5344                         } elseif ( $root->nodeName == 'h' ) {
5345                                 # Heading
5346                                 $s = $this->expand( $root->childNodes, 0, $deepFlags );
5347
5348                                 if ( $this->parser->ot['html'] ) {
5349                                         # Insert heading index marker
5350                                         $headingIndex = $root->getAttribute( 'i' );
5351                                         $titleText = $this->title->getPrefixedDBkey();
5352                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5353                                         $serial = count( $this->parser->mHeadings ) - 1;
5354                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5355                                         $count = $root->getAttribute( 'level' );
5356
5357                                         // FIXME: bug-for-bug with old parser
5358                                         // Lose whitespace for no apparent reason
5359                                         // Remove this after differential testing is done
5360                                         if ( true ) {
5361                                                 // Good version
5362                                                 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5363                                         } else {
5364                                                 // Bad version
5365                                                 if ( preg_match( '/^(={1,6})(.*?)(={1,6})\s*?$/', $s, $m ) ) {
5366                                                         if ( $m[2] != '' ) {
5367                                                                 $s = $m[1] . $marker . $m[2] . $m[3];
5368                                                         }
5369                                                 }
5370                                         }
5371                                         $this->parser->mStripState->general->setPair( $marker, '' );
5372                                 }
5373                         } else {
5374                                 # Generic recursive expansion
5375                                 $s = '';
5376                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5377                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5378                                                 $s .= $node->nodeValue;
5379                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5380                                                 $s .= $this->expand( $node, 0, $deepFlags );
5381                                         }
5382                                 }
5383                         }
5384                 } else {
5385                         throw new MWException( __METHOD__.': Invalid parameter type' );
5386                 }
5387                 return $s;
5388         }
5389
5390         function implodeWithFlags( $sep, $shallowFlags, $deepFlags /*, ... */ ) {
5391                 $args = array_slice( func_get_args(), 3 );
5392
5393                 $first = true;
5394                 $s = '';
5395                 foreach ( $args as $root ) {
5396                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5397                                 $root = array( $root );
5398                         }
5399                         foreach ( $root as $node ) {
5400                                 if ( $first ) {
5401                                         $first = false;
5402                                 } else {
5403                                         $s .= $sep;
5404                                 }
5405                                 $s .= $this->expand( $node, $shallowFlags, $deepFlags );
5406                         }
5407                 }
5408                 return $s;
5409         }
5410
5411         function implode( $sep /*, ... */ ) {
5412                 $args = func_get_args();
5413                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0, 0 ), array_slice( $args, 1 ) );
5414                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5415         }
5416
5417         function __toString() {
5418                 return 'frame{}';
5419         }
5420 }
5421
5422 /**
5423  * Expansion frame with template arguments
5424  */
5425 class PPTemplateFrame extends PPFrame {
5426         public $parser, $args, $parent, $serial;
5427
5428         function __construct( $parser, $parent = false, $args = array(), $title = false ) {
5429                 $this->parser = $parser;
5430                 $this->parent = $parent;
5431                 $this->args = $args;
5432                 $this->title = $title;
5433         }
5434
5435         function __toString() {
5436                 $s = 'tplframe{';
5437                 $first = true;
5438                 foreach ( $this->args as $name => $value ) {
5439                         if ( $first ) {
5440                                 $first = false;
5441                         } else {
5442                                 $s .= ', ';
5443                         }
5444                         $s .= "\"$name\":\"" .
5445                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5446                 }
5447                 $s .= '}';
5448                 return $s;
5449         }
5450 }
5451