includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10 /**
  11  * Update this version number when the ParserOutput format
  12  * changes in an incompatible way, so the parser cache
  13  * can automatically discard old data.
  14  */
  15 define( 'MW_PARSER_VERSION', '1.6.1' );
  16
  17 define( 'RLH_FOR_UPDATE', 1 );
  18
  19 # Allowed values for $mOutputType
  20 define( 'OT_HTML', 1 );
  21 define( 'OT_WIKI', 2 );
  22 define( 'OT_MSG' , 3 );
  23 define( 'OT_PREPROCESS', 4 );
  24
  25 # Flags for setFunctionHook
  26 define( 'SFH_NO_HASH', 1 );
  27
  28 # string parameter for extractTags which will cause it
  29 # to strip HTML comments in addition to regular
  30 # <XML>-style tags. This should not be anything we
  31 # may want to use in wikisyntax
  32 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  33
  34 # Constants needed for external link processing
  35 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  36 # Everything except bracket, space, or control characters
  37 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  38 # Including space, but excluding newlines
  39 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  40 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  41 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  42 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  43         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  44 define( 'EXT_IMAGE_REGEX',
  45         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  46         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  47         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  48 );
  49
  50 // State constants for the definition list colon extraction
  51 define( 'MW_COLON_STATE_TEXT', 0 );
  52 define( 'MW_COLON_STATE_TAG', 1 );
  53 define( 'MW_COLON_STATE_TAGSTART', 2 );
  54 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  55 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  56 define( 'MW_COLON_STATE_COMMENT', 5 );
  57 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  58 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  59
  60 /**
  61  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  62  * syntax, such as "[[link]]" for making links), and provides a one-way
  63  * transformation of that wiki markup it into XHTML output / markup
  64  * (which in turn the browser understands, and can display).
  65  *
  66  * <pre>
  67  * There are four main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  * preprocess()
  75  *   removes HTML comments and expands templates
  76  *
  77  * Globals used:
  78  *    objects:   $wgLang, $wgContLang
  79  *
  80  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  81  *
  82  * settings:
  83  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  84  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  85  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  86  *  $wgMaxArticleSize*
  87  *
  88  *  * only within ParserOptions
  89  * </pre>
  90  *
  91  * @addtogroup Parser
  92  */
  93 class Parser
  94 {
  95         const VERSION = MW_PARSER_VERSION;
  96         /**#@+
  97          * @private
  98          */
  99         # Persistent:
 100         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
 101
 102         # Cleared with clearState():
 103         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 104         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 105         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 106         var $mIncludeSizes, $mDefaultSort;
 107         var $mTemplates,        // cache of already loaded templates, avoids
 108                                 // multiple SQL queries for the same string
 109             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 110                                 // in this path. Used for loop detection.
 111
 112         # Temporary
 113         # These are variables reset at least once per parse regardless of $clearState
 114         var $mOptions,      // ParserOptions object
 115                 $mTitle,        // Title context, used for self-link rendering and similar things
 116                 $mOutputType,   // Output type, one of the OT_xxx constants
 117                 $ot,            // Shortcut alias, see setOutputType()
 118                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 119                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 120                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 121
 122         /**#@-*/
 123
 124         /**
 125          * Constructor
 126          *
 127          * @public
 128          */
 129         function Parser() {
 130                 $this->mTagHooks = array();
 131                 $this->mFunctionHooks = array();
 132                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 133                 $this->mFirstCall = true;
 134         }
 135
 136         /**
 137          * Do various kinds of initialisation on the first call of the parser
 138          */
 139         function firstCallInit() {
 140                 if ( !$this->mFirstCall ) {
 141                         return;
 142                 }
 143
 144                 wfProfileIn( __METHOD__ );
 145                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 146
 147                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 148
 149                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 170                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 171                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 172                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 173                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 174                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 175
 176                 if ( $wgAllowDisplayTitle ) {
 177                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 178                 }
 179                 if ( $wgAllowSlowParserFunctions ) {
 180                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 181                 }
 182
 183                 $this->initialiseVariables();
 184                 $this->mFirstCall = false;
 185                 wfProfileOut( __METHOD__ );
 186         }
 187
 188         /**
 189          * Clear Parser state
 190          *
 191          * @private
 192          */
 193         function clearState() {
 194                 wfProfileIn( __METHOD__ );
 195                 if ( $this->mFirstCall ) {
 196                         $this->firstCallInit();
 197                 }
 198                 $this->mOutput = new ParserOutput;
 199                 $this->mAutonumber = 0;
 200                 $this->mLastSection = '';
 201                 $this->mDTopen = false;
 202                 $this->mIncludeCount = array();
 203                 $this->mStripState = new StripState;
 204                 $this->mArgStack = array();
 205                 $this->mInPre = false;
 206                 $this->mInterwikiLinkHolders = array(
 207                         'texts' => array(),
 208                         'titles' => array()
 209                 );
 210                 $this->mLinkHolders = array(
 211                         'namespaces' => array(),
 212                         'dbkeys' => array(),
 213                         'queries' => array(),
 214                         'texts' => array(),
 215                         'titles' => array()
 216                 );
 217                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 218
 219                 /**
 220                  * Prefix for temporary replacement strings for the multipass parser.
 221                  * \x07 should never appear in input as it's disallowed in XML.
 222                  * Using it at the front also gives us a little extra robustness
 223                  * since it shouldn't match when butted up against identifier-like
 224                  * string constructs.
 225                  */
 226                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 227
 228                 # Clear these on every parse, bug 4549
 229                 $this->mTemplates = array();
 230                 $this->mTemplatePath = array();
 231
 232                 $this->mShowToc = true;
 233                 $this->mForceTocPosition = false;
 234                 $this->mIncludeSizes = array(
 235                         'pre-expand' => 0,
 236                         'post-expand' => 0,
 237                         'arg' => 0
 238                 );
 239                 $this->mDefaultSort = false;
 240
 241                 wfRunHooks( 'ParserClearState', array( &$this ) );
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         function setOutputType( $ot ) {
 246                 $this->mOutputType = $ot;
 247                 // Shortcut alias
 248                 $this->ot = array(
 249                         'html' => $ot == OT_HTML,
 250                         'wiki' => $ot == OT_WIKI,
 251                         'msg' => $ot == OT_MSG,
 252                         'pre' => $ot == OT_PREPROCESS,
 253                 );
 254         }
 255
 256         /**
 257          * Accessor for mUniqPrefix.
 258          *
 259          * @public
 260          */
 261         function uniqPrefix() {
 262                 return $this->mUniqPrefix;
 263         }
 264
 265         /**
 266          * Convert wikitext to HTML
 267          * Do not call this function recursively.
 268          *
 269          * @param string $text Text we want to parse
 270          * @param Title &$title A title object
 271          * @param array $options
 272          * @param boolean $linestart
 273          * @param boolean $clearState
 274          * @param int $revid number to pass in {{REVISIONID}}
 275          * @return ParserOutput a ParserOutput
 276          */
 277         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 278                 /**
 279                  * First pass--just handle <nowiki> sections, pass the rest off
 280                  * to internalParse() which does all the real work.
 281                  */
 282
 283                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 284                 $fname = 'Parser::parse-' . wfGetCaller();
 285                 wfProfileIn( __METHOD__ );
 286                 wfProfileIn( $fname );
 287
 288                 if ( $clearState ) {
 289                         $this->clearState();
 290                 }
 291
 292                 $this->mOptions = $options;
 293                 $this->mTitle =& $title;
 294                 $oldRevisionId = $this->mRevisionId;
 295                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 296                 if( $revid !== null ) {
 297                         $this->mRevisionId = $revid;
 298                         $this->mRevisionTimestamp = null;
 299                 }
 300                 $this->setOutputType( OT_HTML );
 301                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 302                 $text = $this->strip( $text, $this->mStripState );
 303                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 304                 $text = $this->internalParse( $text );
 305                 $text = $this->mStripState->unstripGeneral( $text );
 306
 307                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 308                 $fixtags = array(
 309                         # french spaces, last one Guillemet-left
 310                         # only if there is something before the space
 311                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 312                         # french spaces, Guillemet-right
 313                         '/(\\302\\253) /' => '\\1&nbsp;',
 314                 );
 315                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 316
 317                 # only once and last
 318                 $text = $this->doBlockLevels( $text, $linestart );
 319
 320                 $this->replaceLinkHolders( $text );
 321
 322                 # the position of the parserConvert() call should not be changed. it
 323                 # assumes that the links are all replaced and the only thing left
 324                 # is the <nowiki> mark.
 325                 # Side-effects: this calls $this->mOutput->setTitleText()
 326                 $text = $wgContLang->parserConvert( $text, $this );
 327
 328                 $text = $this->mStripState->unstripNoWiki( $text );
 329
 330                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 331
 332                 $text = Sanitizer::normalizeCharReferences( $text );
 333
 334                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 335                         $text = Parser::tidy($text);
 336                 } else {
 337                         # attempt to sanitize at least some nesting problems
 338                         # (bug #2702 and quite a few others)
 339                         $tidyregs = array(
 340                                 # ''Something [http://www.cool.com cool''] -->
 341                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 342                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 343                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 344                                 # fix up an anchor inside another anchor, only
 345                                 # at least for a single single nested link (bug 3695)
 346                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 347                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 348                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 349                                 # contains a div, so fix it up here; replace
 350                                 # div with escaped text
 351                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 352                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 353                                 # remove empty italic or bold tag pairs, some
 354                                 # introduced by rules above
 355                                 '/<([bi])><\/\\1>/' => '',
 356                         );
 357
 358                         $text = preg_replace(
 359                                 array_keys( $tidyregs ),
 360                                 array_values( $tidyregs ),
 361                                 $text );
 362                 }
 363
 364                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 365
 366                 # Information on include size limits, for the benefit of users who try to skirt them
 367                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 368                         $max = $this->mOptions->getMaxIncludeSize();
 369                         $text .= "<!-- \n" .
 370                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 371                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 372                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 373                                 "Maximum: $max bytes\n" .
 374                                 "-->\n";
 375                 }
 376                 $this->mOutput->setText( $text );
 377                 $this->mRevisionId = $oldRevisionId;
 378                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 379                 wfProfileOut( $fname );
 380                 wfProfileOut( __METHOD__ );
 381
 382                 return $this->mOutput;
 383         }
 384
 385         /**
 386          * Recursive parser entry point that can be called from an extension tag
 387          * hook.
 388          */
 389         function recursiveTagParse( $text ) {
 390                 wfProfileIn( __METHOD__ );
 391                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 392                 $text = $this->strip( $text, $this->mStripState );
 393                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 394                 $text = $this->internalParse( $text );
 395                 wfProfileOut( __METHOD__ );
 396                 return $text;
 397         }
 398
 399         /**
 400          * Expand templates and variables in the text, producing valid, static wikitext.
 401          * Also removes comments.
 402          */
 403         function preprocess( $text, $title, $options, $revid = null ) {
 404                 wfProfileIn( __METHOD__ );
 405                 $this->clearState();
 406                 $this->setOutputType( OT_PREPROCESS );
 407                 $this->mOptions = $options;
 408                 $this->mTitle = $title;
 409                 if( $revid !== null ) {
 410                         $this->mRevisionId = $revid;
 411                 }
 412                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 413                 $text = $this->strip( $text, $this->mStripState );
 414                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 415                 if ( $this->mOptions->getRemoveComments() ) {
 416                         $text = Sanitizer::removeHTMLcomments( $text );
 417                 }
 418                 $text = $this->replaceVariables( $text );
 419                 $text = $this->mStripState->unstripBoth( $text );
 420                 wfProfileOut( __METHOD__ );
 421                 return $text;
 422         }
 423
 424         /**
 425          * Get a random string
 426          *
 427          * @private
 428          * @static
 429          */
 430         function getRandomString() {
 431                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 432         }
 433
 434         function &getTitle() { return $this->mTitle; }
 435         function getOptions() { return $this->mOptions; }
 436
 437         function getFunctionLang() {
 438                 global $wgLang, $wgContLang;
 439                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 440         }
 441
 442         /**
 443          * Replaces all occurrences of HTML-style comments and the given tags
 444          * in the text with a random marker and returns teh next text. The output
 445          * parameter $matches will be an associative array filled with data in
 446          * the form:
 447          *   'UNIQ-xxxxx' => array(
 448          *     'element',
 449          *     'tag content',
 450          *     array( 'param' => 'x' ),
 451          *     '<element param="x">tag content</element>' ) )
 452          *
 453          * @param $elements list of element names. Comments are always extracted.
 454          * @param $text Source text string.
 455          * @param $uniq_prefix
 456          *
 457          * @public
 458          * @static
 459          */
 460         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 461                 static $n = 1;
 462                 $stripped = '';
 463                 $matches = array();
 464
 465                 $taglist = implode( '|', $elements );
 466                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 467
 468                 while ( '' != $text ) {
 469                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 470                         $stripped .= $p[0];
 471                         if( count( $p ) < 5 ) {
 472                                 break;
 473                         }
 474                         if( count( $p ) > 5 ) {
 475                                 // comment
 476                                 $element    = $p[4];
 477                                 $attributes = '';
 478                                 $close      = '';
 479                                 $inside     = $p[5];
 480                         } else {
 481                                 // tag
 482                                 $element    = $p[1];
 483                                 $attributes = $p[2];
 484                                 $close      = $p[3];
 485                                 $inside     = $p[4];
 486                         }
 487
 488                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 489                         $stripped .= $marker;
 490
 491                         if ( $close === '/>' ) {
 492                                 // Empty element tag, <tag />
 493                                 $content = null;
 494                                 $text = $inside;
 495                                 $tail = null;
 496                         } else {
 497                                 if( $element == '!--' ) {
 498                                         $end = '/(-->)/';
 499                                 } else {
 500                                         $end = "/(<\\/$element\\s*>)/i";
 501                                 }
 502                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 503                                 $content = $q[0];
 504                                 if( count( $q ) < 3 ) {
 505                                         # No end tag -- let it run out to the end of the text.
 506                                         $tail = '';
 507                                         $text = '';
 508                                 } else {
 509                                         $tail = $q[1];
 510                                         $text = $q[2];
 511                                 }
 512                         }
 513
 514                         $matches[$marker] = array( $element,
 515                                 $content,
 516                                 Sanitizer::decodeTagAttributes( $attributes ),
 517                                 "<$element$attributes$close$content$tail" );
 518                 }
 519                 return $stripped;
 520         }
 521
 522         /**
 523          * Strips and renders nowiki, pre, math, hiero
 524          * If $render is set, performs necessary rendering operations on plugins
 525          * Returns the text, and fills an array with data needed in unstrip()
 526          *
 527          * @param StripState $state
 528          *
 529          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 530          *  will be stripped in addition to other tags. This is important
 531          *  for section editing, where these comments cause confusion when
 532          *  counting the sections in the wikisource
 533          *
 534          * @param array dontstrip contains tags which should not be stripped;
 535          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 536          *
 537          * @private
 538          */
 539         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 540                 global $wgContLang;
 541                 wfProfileIn( __METHOD__ );
 542                 $render = ($this->mOutputType == OT_HTML);
 543
 544                 $uniq_prefix = $this->mUniqPrefix;
 545                 $commentState = new ReplacementArray;
 546                 $nowikiItems = array();
 547                 $generalItems = array();
 548
 549                 $elements = array_merge(
 550                         array( 'nowiki', 'gallery' ),
 551                         array_keys( $this->mTagHooks ) );
 552                 global $wgRawHtml;
 553                 if( $wgRawHtml ) {
 554                         $elements[] = 'html';
 555                 }
 556                 if( $this->mOptions->getUseTeX() ) {
 557                         $elements[] = 'math';
 558                 }
 559
 560                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 561                 foreach ( $elements AS $k => $v ) {
 562                         if ( !in_array ( $v , $dontstrip ) ) continue;
 563                         unset ( $elements[$k] );
 564                 }
 565
 566                 $matches = array();
 567                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 568
 569                 foreach( $matches as $marker => $data ) {
 570                         list( $element, $content, $params, $tag ) = $data;
 571                         if( $render ) {
 572                                 $tagName = strtolower( $element );
 573                                 wfProfileIn( __METHOD__."-render-$tagName" );
 574                                 switch( $tagName ) {
 575                                 case '!--':
 576                                         // Comment
 577                                         if( substr( $tag, -3 ) == '-->' ) {
 578                                                 $output = $tag;
 579                                         } else {
 580                                                 // Unclosed comment in input.
 581                                                 // Close it so later stripping can remove it
 582                                                 $output = "$tag-->";
 583                                         }
 584                                         break;
 585                                 case 'html':
 586                                         if( $wgRawHtml ) {
 587                                                 $output = $content;
 588                                                 break;
 589                                         }
 590                                         // Shouldn't happen otherwise. :)
 591                                 case 'nowiki':
 592                                         $output = Xml::escapeTagsOnly( $content );
 593                                         break;
 594                                 case 'math':
 595                                         $output = $wgContLang->armourMath( MathRenderer::renderMath( $content ) );
 596                                         break;
 597                                 case 'gallery':
 598                                         $output = $this->renderImageGallery( $content, $params );
 599                                         break;
 600                                 default:
 601                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 602                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 603                                                         array( $content, $params, $this ) );
 604                                         } else {
 605                                                 throw new MWException( "Invalid call hook $element" );
 606                                         }
 607                                 }
 608                                 wfProfileOut( __METHOD__."-render-$tagName" );
 609                         } else {
 610                                 // Just stripping tags; keep the source
 611                                 $output = $tag;
 612                         }
 613
 614                         // Unstrip the output, to support recursive strip() calls
 615                         $output = $state->unstripBoth( $output );
 616
 617                         if( !$stripcomments && $element == '!--' ) {
 618                                 $commentState->setPair( $marker, $output );
 619                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 620                                 $nowikiItems[$marker] = $output;
 621                         } else {
 622                                 $generalItems[$marker] = $output;
 623                         }
 624                 }
 625                 # Add the new items to the state
 626                 # We do this after the loop instead of during it to avoid slowing
 627                 # down the recursive unstrip
 628                 $state->nowiki->mergeArray( $nowikiItems );
 629                 $state->general->mergeArray( $generalItems );
 630
 631                 # Unstrip comments unless explicitly told otherwise.
 632                 # (The comments are always stripped prior to this point, so as to
 633                 # not invoke any extension tags / parser hooks contained within
 634                 # a comment.)
 635                 if ( !$stripcomments ) {
 636                         // Put them all back and forget them
 637                         $text = $commentState->replace( $text );
 638                 }
 639
 640                 wfProfileOut( __METHOD__ );
 641                 return $text;
 642         }
 643
 644         /**
 645          * Restores pre, math, and other extensions removed by strip()
 646          *
 647          * always call unstripNoWiki() after this one
 648          * @private
 649          * @deprecated use $this->mStripState->unstrip()
 650          */
 651         function unstrip( $text, $state ) {
 652                 return $state->unstripGeneral( $text );
 653         }
 654
 655         /**
 656          * Always call this after unstrip() to preserve the order
 657          *
 658          * @private
 659          * @deprecated use $this->mStripState->unstrip()
 660          */
 661         function unstripNoWiki( $text, $state ) {
 662                 return $state->unstripNoWiki( $text );
 663         }
 664
 665         /**
 666          * @deprecated use $this->mStripState->unstripBoth()
 667          */
 668         function unstripForHTML( $text ) {
 669                 return $this->mStripState->unstripBoth( $text );
 670         }
 671
 672         /**
 673          * Add an item to the strip state
 674          * Returns the unique tag which must be inserted into the stripped text
 675          * The tag will be replaced with the original text in unstrip()
 676          *
 677          * @private
 678          */
 679         function insertStripItem( $text, &$state ) {
 680                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 681                 $state->general->setPair( $rnd, $text );
 682                 return $rnd;
 683         }
 684
 685         /**
 686          * Interface with html tidy, used if $wgUseTidy = true.
 687          * If tidy isn't able to correct the markup, the original will be
 688          * returned in all its glory with a warning comment appended.
 689          *
 690          * Either the external tidy program or the in-process tidy extension
 691          * will be used depending on availability. Override the default
 692          * $wgTidyInternal setting to disable the internal if it's not working.
 693          *
 694          * @param string $text Hideous HTML input
 695          * @return string Corrected HTML output
 696          * @public
 697          * @static
 698          */
 699         function tidy( $text ) {
 700                 global $wgTidyInternal;
 701                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 702 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 703 '<head><title>test</title></head><body>'.$text.'</body></html>';
 704                 if( $wgTidyInternal ) {
 705                         $correctedtext = Parser::internalTidy( $wrappedtext );
 706                 } else {
 707                         $correctedtext = Parser::externalTidy( $wrappedtext );
 708                 }
 709                 if( is_null( $correctedtext ) ) {
 710                         wfDebug( "Tidy error detected!\n" );
 711                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 712                 }
 713                 return $correctedtext;
 714         }
 715
 716         /**
 717          * Spawn an external HTML tidy process and get corrected markup back from it.
 718          *
 719          * @private
 720          * @static
 721          */
 722         function externalTidy( $text ) {
 723                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 724                 $fname = 'Parser::externalTidy';
 725                 wfProfileIn( $fname );
 726
 727                 $cleansource = '';
 728                 $opts = ' -utf8';
 729
 730                 $descriptorspec = array(
 731                         0 => array('pipe', 'r'),
 732                         1 => array('pipe', 'w'),
 733                         2 => array('file', '/dev/null', 'a')  // FIXME: this line in UNIX-specific, it generates a warning on Windows, because /dev/null is not a valid Windows file.
 734                 );
 735                 $pipes = array();
 736                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 737                 if (is_resource($process)) {
 738                         // Theoretically, this style of communication could cause a deadlock
 739                         // here. If the stdout buffer fills up, then writes to stdin could
 740                         // block. This doesn't appear to happen with tidy, because tidy only
 741                         // writes to stdout after it's finished reading from stdin. Search
 742                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 743                         fwrite($pipes[0], $text);
 744                         fclose($pipes[0]);
 745                         while (!feof($pipes[1])) {
 746                                 $cleansource .= fgets($pipes[1], 1024);
 747                         }
 748                         fclose($pipes[1]);
 749                         proc_close($process);
 750                 }
 751
 752                 wfProfileOut( $fname );
 753
 754                 if( $cleansource == '' && $text != '') {
 755                         // Some kind of error happened, so we couldn't get the corrected text.
 756                         // Just give up; we'll use the source text and append a warning.
 757                         return null;
 758                 } else {
 759                         return $cleansource;
 760                 }
 761         }
 762
 763         /**
 764          * Use the HTML tidy PECL extension to use the tidy library in-process,
 765          * saving the overhead of spawning a new process. Currently written to
 766          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 767          *
 768          * 'pear install tidy' should be able to compile the extension module.
 769          *
 770          * @private
 771          * @static
 772          */
 773         function internalTidy( $text ) {
 774                 global $wgTidyConf;
 775                 $fname = 'Parser::internalTidy';
 776                 wfProfileIn( $fname );
 777
 778                 tidy_load_config( $wgTidyConf );
 779                 tidy_set_encoding( 'utf8' );
 780                 tidy_parse_string( $text );
 781                 tidy_clean_repair();
 782                 if( tidy_get_status() == 2 ) {
 783                         // 2 is magic number for fatal error
 784                         // http://www.php.net/manual/en/function.tidy-get-status.php
 785                         $cleansource = null;
 786                 } else {
 787                         $cleansource = tidy_get_output();
 788                 }
 789                 wfProfileOut( $fname );
 790                 return $cleansource;
 791         }
 792
 793         /**
 794          * parse the wiki syntax used to render tables
 795          *
 796          * @private
 797          */
 798         function doTableStuff ( $text ) {
 799                 $fname = 'Parser::doTableStuff';
 800                 wfProfileIn( $fname );
 801
 802                 $lines = explode ( "\n" , $text );
 803                 $td_history = array (); // Is currently a td tag open?
 804                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 805                 $tr_history = array (); // Is currently a tr tag open?
 806                 $tr_attributes = array (); // history of tr attributes
 807                 $has_opened_tr = array(); // Did this table open a <tr> element?
 808                 $indent_level = 0; // indent level of the table
 809                 foreach ( $lines as $key => $line )
 810                 {
 811                         $line = trim ( $line );
 812
 813                         if( $line == '' ) { // empty line, go to next line
 814                                 continue;
 815                         }
 816                         $first_character = $line{0};
 817                         $matches = array();
 818
 819                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 820                                 // First check if we are starting a new table
 821                                 $indent_level = strlen( $matches[1] );
 822
 823                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 824                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 825
 826                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 827                                 array_push ( $td_history , false );
 828                                 array_push ( $last_tag_history , '' );
 829                                 array_push ( $tr_history , false );
 830                                 array_push ( $tr_attributes , '' );
 831                                 array_push ( $has_opened_tr , false );
 832                         } else if ( count ( $td_history ) == 0 ) {
 833                                 // Don't do any of the following
 834                                 continue;
 835                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 836                                 // We are ending a table
 837                                 $line = '</table>' . substr ( $line , 2 );
 838                                 $last_tag = array_pop ( $last_tag_history );
 839
 840                                 if ( !array_pop ( $has_opened_tr ) ) {
 841                                         $line = "<tr><td></td></tr>{$line}";
 842                                 }
 843
 844                                 if ( array_pop ( $tr_history ) ) {
 845                                         $line = "</tr>{$line}";
 846                                 }
 847
 848                                 if ( array_pop ( $td_history ) ) {
 849                                         $line = "</{$last_tag}>{$line}";
 850                                 }
 851                                 array_pop ( $tr_attributes );
 852                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 853                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 854                                 // Now we have a table row
 855                                 $line = preg_replace( '#^\|-+#', '', $line );
 856
 857                                 // Whats after the tag is now only attributes
 858                                 $attributes = $this->mStripState->unstripBoth( $line );
 859                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 860                                 array_pop ( $tr_attributes );
 861                                 array_push ( $tr_attributes , $attributes );
 862
 863                                 $line = '';
 864                                 $last_tag = array_pop ( $last_tag_history );
 865                                 array_pop ( $has_opened_tr );
 866                                 array_push ( $has_opened_tr , true );
 867
 868                                 if ( array_pop ( $tr_history ) ) {
 869                                         $line = '</tr>';
 870                                 }
 871
 872                                 if ( array_pop ( $td_history ) ) {
 873                                         $line = "</{$last_tag}>{$line}";
 874                                 }
 875
 876                                 $lines[$key] = $line;
 877                                 array_push ( $tr_history , false );
 878                                 array_push ( $td_history , false );
 879                                 array_push ( $last_tag_history , '' );
 880                         }
 881                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 882                                 // This might be cell elements, td, th or captions
 883                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 884                                         $first_character = '+';
 885                                         $line = substr ( $line , 1 );
 886                                 }
 887
 888                                 $line = substr ( $line , 1 );
 889
 890                                 if ( $first_character == '!' ) {
 891                                         $line = str_replace ( '!!' , '||' , $line );
 892                                 }
 893
 894                                 // Split up multiple cells on the same line.
 895                                 // FIXME : This can result in improper nesting of tags processed
 896                                 // by earlier parser steps, but should avoid splitting up eg
 897                                 // attribute values containing literal "||".
 898                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 899
 900                                 $lines[$key] = '';
 901
 902                                 // Loop through each table cell
 903                                 foreach ( $cells as $cell )
 904                                 {
 905                                         $previous = '';
 906                                         if ( $first_character != '+' )
 907                                         {
 908                                                 $tr_after = array_pop ( $tr_attributes );
 909                                                 if ( !array_pop ( $tr_history ) ) {
 910                                                         $previous = "<tr{$tr_after}>\n";
 911                                                 }
 912                                                 array_push ( $tr_history , true );
 913                                                 array_push ( $tr_attributes , '' );
 914                                                 array_pop ( $has_opened_tr );
 915                                                 array_push ( $has_opened_tr , true );
 916                                         }
 917
 918                                         $last_tag = array_pop ( $last_tag_history );
 919
 920                                         if ( array_pop ( $td_history ) ) {
 921                                                 $previous = "</{$last_tag}>{$previous}";
 922                                         }
 923
 924                                         if ( $first_character == '|' ) {
 925                                                 $last_tag = 'td';
 926                                         } else if ( $first_character == '!' ) {
 927                                                 $last_tag = 'th';
 928                                         } else if ( $first_character == '+' ) {
 929                                                 $last_tag = 'caption';
 930                                         } else {
 931                                                 $last_tag = '';
 932                                         }
 933
 934                                         array_push ( $last_tag_history , $last_tag );
 935
 936                                         // A cell could contain both parameters and data
 937                                         $cell_data = explode ( '|' , $cell , 2 );
 938
 939                                         // Bug 553: Note that a '|' inside an invalid link should not
 940                                         // be mistaken as delimiting cell parameters
 941                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 942                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 943                                         } else if ( count ( $cell_data ) == 1 )
 944                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 945                                         else {
 946                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 947                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 948                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 949                                         }
 950
 951                                         $lines[$key] .= $cell;
 952                                         array_push ( $td_history , true );
 953                                 }
 954                         }
 955                 }
 956
 957                 // Closing open td, tr && table
 958                 while ( count ( $td_history ) > 0 )
 959                 {
 960                         if ( array_pop ( $td_history ) ) {
 961                                 $lines[] = '</td>' ;
 962                         }
 963                         if ( array_pop ( $tr_history ) ) {
 964                                 $lines[] = '</tr>' ;
 965                         }
 966                         if ( !array_pop ( $has_opened_tr ) ) {
 967                                 $lines[] = "<tr><td></td></tr>" ;
 968                         }
 969
 970                         $lines[] = '</table>' ;
 971                 }
 972
 973                 $output = implode ( "\n" , $lines ) ;
 974
 975                 // special case: don't return empty table
 976                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 977                         $output = '';
 978                 }
 979
 980                 wfProfileOut( $fname );
 981
 982                 return $output;
 983         }
 984
 985         /**
 986          * Helper function for parse() that transforms wiki markup into
 987          * HTML. Only called for $mOutputType == OT_HTML.
 988          *
 989          * @private
 990          */
 991         function internalParse( $text ) {
 992                 $args = array();
 993                 $isMain = true;
 994                 $fname = 'Parser::internalParse';
 995                 wfProfileIn( $fname );
 996
 997                 # Hook to suspend the parser in this state
 998                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 999                         wfProfileOut( $fname );
1000                         return $text ;
1001                 }
1002
1003                 # Remove <noinclude> tags and <includeonly> sections
1004                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
1005                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
1006                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
1007
1008                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
1009
1010                 $text = $this->replaceVariables( $text, $args );
1011                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1012
1013                 // Tables need to come after variable replacement for things to work
1014                 // properly; putting them before other transformations should keep
1015                 // exciting things like link expansions from showing up in surprising
1016                 // places.
1017                 $text = $this->doTableStuff( $text );
1018
1019                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1020
1021                 $text = $this->stripToc( $text );
1022                 $this->stripNoGallery( $text );
1023                 $text = $this->doHeadings( $text );
1024                 if($this->mOptions->getUseDynamicDates()) {
1025                         $df =& DateFormatter::getInstance();
1026                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1027                 }
1028                 $text = $this->doAllQuotes( $text );
1029                 $text = $this->replaceInternalLinks( $text );
1030                 $text = $this->replaceExternalLinks( $text );
1031
1032                 # replaceInternalLinks may sometimes leave behind
1033                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1034                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1035
1036                 $text = $this->doMagicLinks( $text );
1037                 $text = $this->formatHeadings( $text, $isMain );
1038
1039                 wfProfileOut( $fname );
1040                 return $text;
1041         }
1042
1043         /**
1044          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1045          * magic external links.
1046          *
1047          * @private
1048          */
1049         function &doMagicLinks( &$text ) {
1050                 wfProfileIn( __METHOD__ );
1051                 $text = preg_replace_callback(
1052                         '!(?:                           # Start cases
1053                             <a.*?</a> |                 # Skip link text
1054                             <.*?> |                     # Skip stuff inside HTML elements
1055                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1056                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1057                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1058                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1059                                      [0-9Xx]                 # check digit
1060                                    \b)
1061                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1062                 wfProfileOut( __METHOD__ );
1063                 return $text;
1064         }
1065
1066         function magicLinkCallback( $m ) {
1067                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1068                         # Skip HTML element
1069                         return $m[0];
1070                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1071                         $isbn = $m[2];
1072                         $num = strtr( $isbn, array(
1073                                 '-' => '',
1074                                 ' ' => '',
1075                                 'x' => 'X',
1076                         ));
1077                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1078                         $text = '<a href="' .
1079                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1080                                 "\" class=\"internal\">ISBN $isbn</a>";
1081                 } else {
1082                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1083                                 $keyword = 'RFC';
1084                                 $urlmsg = 'rfcurl';
1085                                 $id = $m[1];
1086                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1087                                 $keyword = 'PMID';
1088                                 $urlmsg = 'pubmedurl';
1089                                 $id = $m[1];
1090                         } else {
1091                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1092                                         substr($m[0], 0, 20 ) . '"' );
1093                         }
1094
1095                         $url = wfMsg( $urlmsg, $id);
1096                         $sk = $this->mOptions->getSkin();
1097                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1098                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1099                 }
1100                 return $text;
1101         }
1102
1103         /**
1104          * Parse headers and return html
1105          *
1106          * @private
1107          */
1108         function doHeadings( $text ) {
1109                 $fname = 'Parser::doHeadings';
1110                 wfProfileIn( $fname );
1111                 for ( $i = 6; $i >= 1; --$i ) {
1112                         $h = str_repeat( '=', $i );
1113                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1114                           "<h{$i}>\\1</h{$i}>\\2", $text );
1115                 }
1116                 wfProfileOut( $fname );
1117                 return $text;
1118         }
1119
1120         /**
1121          * Replace single quotes with HTML markup
1122          * @private
1123          * @return string the altered text
1124          */
1125         function doAllQuotes( $text ) {
1126                 $fname = 'Parser::doAllQuotes';
1127                 wfProfileIn( $fname );
1128                 $outtext = '';
1129                 $lines = explode( "\n", $text );
1130                 foreach ( $lines as $line ) {
1131                         $outtext .= $this->doQuotes ( $line ) . "\n";
1132                 }
1133                 $outtext = substr($outtext, 0,-1);
1134                 wfProfileOut( $fname );
1135                 return $outtext;
1136         }
1137
1138         /**
1139          * Helper function for doAllQuotes()
1140          * @private
1141          */
1142         function doQuotes( $text ) {
1143                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1144                 if ( count( $arr ) == 1 )
1145                         return $text;
1146                 else
1147                 {
1148                         # First, do some preliminary work. This may shift some apostrophes from
1149                         # being mark-up to being text. It also counts the number of occurrences
1150                         # of bold and italics mark-ups.
1151                         $i = 0;
1152                         $numbold = 0;
1153                         $numitalics = 0;
1154                         foreach ( $arr as $r )
1155                         {
1156                                 if ( ( $i % 2 ) == 1 )
1157                                 {
1158                                         # If there are ever four apostrophes, assume the first is supposed to
1159                                         # be text, and the remaining three constitute mark-up for bold text.
1160                                         if ( strlen( $arr[$i] ) == 4 )
1161                                         {
1162                                                 $arr[$i-1] .= "'";
1163                                                 $arr[$i] = "'''";
1164                                         }
1165                                         # If there are more than 5 apostrophes in a row, assume they're all
1166                                         # text except for the last 5.
1167                                         else if ( strlen( $arr[$i] ) > 5 )
1168                                         {
1169                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1170                                                 $arr[$i] = "'''''";
1171                                         }
1172                                         # Count the number of occurrences of bold and italics mark-ups.
1173                                         # We are not counting sequences of five apostrophes.
1174                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1175                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1176                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1177                                 }
1178                                 $i++;
1179                         }
1180
1181                         # If there is an odd number of both bold and italics, it is likely
1182                         # that one of the bold ones was meant to be an apostrophe followed
1183                         # by italics. Which one we cannot know for certain, but it is more
1184                         # likely to be one that has a single-letter word before it.
1185                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1186                         {
1187                                 $i = 0;
1188                                 $firstsingleletterword = -1;
1189                                 $firstmultiletterword = -1;
1190                                 $firstspace = -1;
1191                                 foreach ( $arr as $r )
1192                                 {
1193                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1194                                         {
1195                                                 $x1 = substr ($arr[$i-1], -1);
1196                                                 $x2 = substr ($arr[$i-1], -2, 1);
1197                                                 if ($x1 == ' ') {
1198                                                         if ($firstspace == -1) $firstspace = $i;
1199                                                 } else if ($x2 == ' ') {
1200                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1201                                                 } else {
1202                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1203                                                 }
1204                                         }
1205                                         $i++;
1206                                 }
1207
1208                                 # If there is a single-letter word, use it!
1209                                 if ($firstsingleletterword > -1)
1210                                 {
1211                                         $arr [ $firstsingleletterword ] = "''";
1212                                         $arr [ $firstsingleletterword-1 ] .= "'";
1213                                 }
1214                                 # If not, but there's a multi-letter word, use that one.
1215                                 else if ($firstmultiletterword > -1)
1216                                 {
1217                                         $arr [ $firstmultiletterword ] = "''";
1218                                         $arr [ $firstmultiletterword-1 ] .= "'";
1219                                 }
1220                                 # ... otherwise use the first one that has neither.
1221                                 # (notice that it is possible for all three to be -1 if, for example,
1222                                 # there is only one pentuple-apostrophe in the line)
1223                                 else if ($firstspace > -1)
1224                                 {
1225                                         $arr [ $firstspace ] = "''";
1226                                         $arr [ $firstspace-1 ] .= "'";
1227                                 }
1228                         }
1229
1230                         # Now let's actually convert our apostrophic mush to HTML!
1231                         $output = '';
1232                         $buffer = '';
1233                         $state = '';
1234                         $i = 0;
1235                         foreach ($arr as $r)
1236                         {
1237                                 if (($i % 2) == 0)
1238                                 {
1239                                         if ($state == 'both')
1240                                                 $buffer .= $r;
1241                                         else
1242                                                 $output .= $r;
1243                                 }
1244                                 else
1245                                 {
1246                                         if (strlen ($r) == 2)
1247                                         {
1248                                                 if ($state == 'i')
1249                                                 { $output .= '</i>'; $state = ''; }
1250                                                 else if ($state == 'bi')
1251                                                 { $output .= '</i>'; $state = 'b'; }
1252                                                 else if ($state == 'ib')
1253                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1254                                                 else if ($state == 'both')
1255                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1256                                                 else # $state can be 'b' or ''
1257                                                 { $output .= '<i>'; $state .= 'i'; }
1258                                         }
1259                                         else if (strlen ($r) == 3)
1260                                         {
1261                                                 if ($state == 'b')
1262                                                 { $output .= '</b>'; $state = ''; }
1263                                                 else if ($state == 'bi')
1264                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1265                                                 else if ($state == 'ib')
1266                                                 { $output .= '</b>'; $state = 'i'; }
1267                                                 else if ($state == 'both')
1268                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1269                                                 else # $state can be 'i' or ''
1270                                                 { $output .= '<b>'; $state .= 'b'; }
1271                                         }
1272                                         else if (strlen ($r) == 5)
1273                                         {
1274                                                 if ($state == 'b')
1275                                                 { $output .= '</b><i>'; $state = 'i'; }
1276                                                 else if ($state == 'i')
1277                                                 { $output .= '</i><b>'; $state = 'b'; }
1278                                                 else if ($state == 'bi')
1279                                                 { $output .= '</i></b>'; $state = ''; }
1280                                                 else if ($state == 'ib')
1281                                                 { $output .= '</b></i>'; $state = ''; }
1282                                                 else if ($state == 'both')
1283                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1284                                                 else # ($state == '')
1285                                                 { $buffer = ''; $state = 'both'; }
1286                                         }
1287                                 }
1288                                 $i++;
1289                         }
1290                         # Now close all remaining tags.  Notice that the order is important.
1291                         if ($state == 'b' || $state == 'ib')
1292                                 $output .= '</b>';
1293                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1294                                 $output .= '</i>';
1295                         if ($state == 'bi')
1296                                 $output .= '</b>';
1297                         # There might be lonely ''''', so make sure we have a buffer
1298                         if ($state == 'both' && $buffer)
1299                                 $output .= '<b><i>'.$buffer.'</i></b>';
1300                         return $output;
1301                 }
1302         }
1303
1304         /**
1305          * Replace external links
1306          *
1307          * Note: this is all very hackish and the order of execution matters a lot.
1308          * Make sure to run maintenance/parserTests.php if you change this code.
1309          *
1310          * @private
1311          */
1312         function replaceExternalLinks( $text ) {
1313                 global $wgContLang;
1314                 $fname = 'Parser::replaceExternalLinks';
1315                 wfProfileIn( $fname );
1316
1317                 $sk = $this->mOptions->getSkin();
1318
1319                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1320
1321                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1322
1323                 $i = 0;
1324                 while ( $i<count( $bits ) ) {
1325                         $url = $bits[$i++];
1326                         $protocol = $bits[$i++];
1327                         $text = $bits[$i++];
1328                         $trail = $bits[$i++];
1329
1330                         # The characters '<' and '>' (which were escaped by
1331                         # removeHTMLtags()) should not be included in
1332                         # URLs, per RFC 2396.
1333                         $m2 = array();
1334                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1335                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1336                                 $url = substr($url, 0, $m2[0][1]);
1337                         }
1338
1339                         # If the link text is an image URL, replace it with an <img> tag
1340                         # This happened by accident in the original parser, but some people used it extensively
1341                         $img = $this->maybeMakeExternalImage( $text );
1342                         if ( $img !== false ) {
1343                                 $text = $img;
1344                         }
1345
1346                         $dtrail = '';
1347
1348                         # Set linktype for CSS - if URL==text, link is essentially free
1349                         $linktype = ($text == $url) ? 'free' : 'text';
1350
1351                         # No link text, e.g. [http://domain.tld/some.link]
1352                         if ( $text == '' ) {
1353                                 # Autonumber if allowed. See bug #5918
1354                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1355                                         $text = '[' . ++$this->mAutonumber . ']';
1356                                         $linktype = 'autonumber';
1357                                 } else {
1358                                         # Otherwise just use the URL
1359                                         $text = htmlspecialchars( $url );
1360                                         $linktype = 'free';
1361                                 }
1362                         } else {
1363                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1364                                 # Check for trail
1365                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1366                         }
1367
1368                         $text = $wgContLang->markNoConversion($text);
1369
1370                         $url = Sanitizer::cleanUrl( $url );
1371
1372                         # Process the trail (i.e. everything after this link up until start of the next link),
1373                         # replacing any non-bracketed links
1374                         $trail = $this->replaceFreeExternalLinks( $trail );
1375
1376                         # Use the encoded URL
1377                         # This means that users can paste URLs directly into the text
1378                         # Funny characters like &ouml; aren't valid in URLs anyway
1379                         # This was changed in August 2004
1380                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1381
1382                         # Register link in the output object.
1383                         # Replace unnecessary URL escape codes with the referenced character
1384                         # This prevents spammers from hiding links from the filters
1385                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1386                         $this->mOutput->addExternalLink( $pasteurized );
1387                 }
1388
1389                 wfProfileOut( $fname );
1390                 return $s;
1391         }
1392
1393         /**
1394          * Replace anything that looks like a URL with a link
1395          * @private
1396          */
1397         function replaceFreeExternalLinks( $text ) {
1398                 global $wgContLang;
1399                 $fname = 'Parser::replaceFreeExternalLinks';
1400                 wfProfileIn( $fname );
1401
1402                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1403                 $s = array_shift( $bits );
1404                 $i = 0;
1405
1406                 $sk = $this->mOptions->getSkin();
1407
1408                 while ( $i < count( $bits ) ){
1409                         $protocol = $bits[$i++];
1410                         $remainder = $bits[$i++];
1411
1412                         $m = array();
1413                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1414                                 # Found some characters after the protocol that look promising
1415                                 $url = $protocol . $m[1];
1416                                 $trail = $m[2];
1417
1418                                 # special case: handle urls as url args:
1419                                 # http://www.example.com/foo?=http://www.example.com/bar
1420                                 if(strlen($trail) == 0 &&
1421                                         isset($bits[$i]) &&
1422                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1423                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1424                                 {
1425                                         # add protocol, arg
1426                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1427                                         $i += 2;
1428                                         $trail = $m[2];
1429                                 }
1430
1431                                 # The characters '<' and '>' (which were escaped by
1432                                 # removeHTMLtags()) should not be included in
1433                                 # URLs, per RFC 2396.
1434                                 $m2 = array();
1435                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1436                                         $trail = substr($url, $m2[0][1]) . $trail;
1437                                         $url = substr($url, 0, $m2[0][1]);
1438                                 }
1439
1440                                 # Move trailing punctuation to $trail
1441                                 $sep = ',;\.:!?';
1442                                 # If there is no left bracket, then consider right brackets fair game too
1443                                 if ( strpos( $url, '(' ) === false ) {
1444                                         $sep .= ')';
1445                                 }
1446
1447                                 $numSepChars = strspn( strrev( $url ), $sep );
1448                                 if ( $numSepChars ) {
1449                                         $trail = substr( $url, -$numSepChars ) . $trail;
1450                                         $url = substr( $url, 0, -$numSepChars );
1451                                 }
1452
1453                                 $url = Sanitizer::cleanUrl( $url );
1454
1455                                 # Is this an external image?
1456                                 $text = $this->maybeMakeExternalImage( $url );
1457                                 if ( $text === false ) {
1458                                         # Not an image, make a link
1459                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1460                                         # Register it in the output object...
1461                                         # Replace unnecessary URL escape codes with their equivalent characters
1462                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1463                                         $this->mOutput->addExternalLink( $pasteurized );
1464                                 }
1465                                 $s .= $text . $trail;
1466                         } else {
1467                                 $s .= $protocol . $remainder;
1468                         }
1469                 }
1470                 wfProfileOut( $fname );
1471                 return $s;
1472         }
1473
1474         /**
1475          * Replace unusual URL escape codes with their equivalent characters
1476          * @param string
1477          * @return string
1478          * @static
1479          * @todo  This can merge genuinely required bits in the path or query string,
1480          *        breaking legit URLs. A proper fix would treat the various parts of
1481          *        the URL differently; as a workaround, just use the output for
1482          *        statistical records, not for actual linking/output.
1483          */
1484         static function replaceUnusualEscapes( $url ) {
1485                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1486                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1487         }
1488
1489         /**
1490          * Callback function used in replaceUnusualEscapes().
1491          * Replaces unusual URL escape codes with their equivalent character
1492          * @static
1493          * @private
1494          */
1495         private static function replaceUnusualEscapesCallback( $matches ) {
1496                 $char = urldecode( $matches[0] );
1497                 $ord = ord( $char );
1498                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1499                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1500                         // No, shouldn't be escaped
1501                         return $char;
1502                 } else {
1503                         // Yes, leave it escaped
1504                         return $matches[0];
1505                 }
1506         }
1507
1508         /**
1509          * make an image if it's allowed, either through the global
1510          * option or through the exception
1511          * @private
1512          */
1513         function maybeMakeExternalImage( $url ) {
1514                 $sk = $this->mOptions->getSkin();
1515                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1516                 $imagesexception = !empty($imagesfrom);
1517                 $text = false;
1518                 if ( $this->mOptions->getAllowExternalImages()
1519                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1520                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1521                                 # Image found
1522                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1523                         }
1524                 }
1525                 return $text;
1526         }
1527
1528         /**
1529          * Process [[ ]] wikilinks
1530          *
1531          * @private
1532          */
1533         function replaceInternalLinks( $s ) {
1534                 global $wgContLang;
1535                 static $fname = 'Parser::replaceInternalLinks' ;
1536
1537                 wfProfileIn( $fname );
1538
1539                 wfProfileIn( $fname.'-setup' );
1540                 static $tc = FALSE;
1541                 # the % is needed to support urlencoded titles as well
1542                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1543
1544                 $sk = $this->mOptions->getSkin();
1545
1546                 #split the entire text string on occurences of [[
1547                 $a = explode( '[[', ' ' . $s );
1548                 #get the first element (all text up to first [[), and remove the space we added
1549                 $s = array_shift( $a );
1550                 $s = substr( $s, 1 );
1551
1552                 # Match a link having the form [[namespace:link|alternate]]trail
1553                 static $e1 = FALSE;
1554                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1555                 # Match cases where there is no "]]", which might still be images
1556                 static $e1_img = FALSE;
1557                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1558                 # Match the end of a line for a word that's not followed by whitespace,
1559                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1560                 $e2 = wfMsgForContent( 'linkprefix' );
1561
1562                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1563                 if( is_null( $this->mTitle ) ) {
1564                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1565                 }
1566                 $nottalk = !$this->mTitle->isTalkPage();
1567
1568                 if ( $useLinkPrefixExtension ) {
1569                         $m = array();
1570                         if ( preg_match( $e2, $s, $m ) ) {
1571                                 $first_prefix = $m[2];
1572                         } else {
1573                                 $first_prefix = false;
1574                         }
1575                 } else {
1576                         $prefix = '';
1577                 }
1578
1579                 if($wgContLang->hasVariants()) {
1580                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1581                 } else {
1582                         $selflink = array($this->mTitle->getPrefixedText());
1583                 }
1584                 $useSubpages = $this->areSubpagesAllowed();
1585                 wfProfileOut( $fname.'-setup' );
1586
1587                 # Loop for each link
1588                 for ($k = 0; isset( $a[$k] ); $k++) {
1589                         $line = $a[$k];
1590                         if ( $useLinkPrefixExtension ) {
1591                                 wfProfileIn( $fname.'-prefixhandling' );
1592                                 if ( preg_match( $e2, $s, $m ) ) {
1593                                         $prefix = $m[2];
1594                                         $s = $m[1];
1595                                 } else {
1596                                         $prefix='';
1597                                 }
1598                                 # first link
1599                                 if($first_prefix) {
1600                                         $prefix = $first_prefix;
1601                                         $first_prefix = false;
1602                                 }
1603                                 wfProfileOut( $fname.'-prefixhandling' );
1604                         }
1605
1606                         $might_be_img = false;
1607
1608                         wfProfileIn( "$fname-e1" );
1609                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1610                                 $text = $m[2];
1611                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1612                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1613                                 # the real problem is with the $e1 regex
1614                                 # See bug 1300.
1615                                 #
1616                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1617                                 # and no image is in sight. See bug 2095.
1618                                 #
1619                                 if( $text !== '' &&
1620                                         substr( $m[3], 0, 1 ) === ']' &&
1621                                         strpos($text, '[') !== false
1622                                 )
1623                                 {
1624                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1625                                         $m[3] = substr( $m[3], 1 );
1626                                 }
1627                                 # fix up urlencoded title texts
1628                                 if( strpos( $m[1], '%' ) !== false ) {
1629                                         # Should anchors '#' also be rejected?
1630                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1631                                 }
1632                                 $trail = $m[3];
1633                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1634                                 $might_be_img = true;
1635                                 $text = $m[2];
1636                                 if ( strpos( $m[1], '%' ) !== false ) {
1637                                         $m[1] = urldecode($m[1]);
1638                                 }
1639                                 $trail = "";
1640                         } else { # Invalid form; output directly
1641                                 $s .= $prefix . '[[' . $line ;
1642                                 wfProfileOut( "$fname-e1" );
1643                                 continue;
1644                         }
1645                         wfProfileOut( "$fname-e1" );
1646                         wfProfileIn( "$fname-misc" );
1647
1648                         # Don't allow internal links to pages containing
1649                         # PROTO: where PROTO is a valid URL protocol; these
1650                         # should be external links.
1651                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1652                                 $s .= $prefix . '[[' . $line ;
1653                                 continue;
1654                         }
1655
1656                         # Make subpage if necessary
1657                         if( $useSubpages ) {
1658                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1659                         } else {
1660                                 $link = $m[1];
1661                         }
1662
1663                         $noforce = (substr($m[1], 0, 1) != ':');
1664                         if (!$noforce) {
1665                                 # Strip off leading ':'
1666                                 $link = substr($link, 1);
1667                         }
1668
1669                         wfProfileOut( "$fname-misc" );
1670                         wfProfileIn( "$fname-title" );
1671                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1672                         if( !$nt ) {
1673                                 $s .= $prefix . '[[' . $line;
1674                                 wfProfileOut( "$fname-title" );
1675                                 continue;
1676                         }
1677
1678                         $ns = $nt->getNamespace();
1679                         $iw = $nt->getInterWiki();
1680                         wfProfileOut( "$fname-title" );
1681
1682                         if ($might_be_img) { # if this is actually an invalid link
1683                                 wfProfileIn( "$fname-might_be_img" );
1684                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1685                                         $found = false;
1686                                         while (isset ($a[$k+1]) ) {
1687                                                 #look at the next 'line' to see if we can close it there
1688                                                 $spliced = array_splice( $a, $k + 1, 1 );
1689                                                 $next_line = array_shift( $spliced );
1690                                                 $m = explode( ']]', $next_line, 3 );
1691                                                 if ( count( $m ) == 3 ) {
1692                                                         # the first ]] closes the inner link, the second the image
1693                                                         $found = true;
1694                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1695                                                         $trail = $m[2];
1696                                                         break;
1697                                                 } elseif ( count( $m ) == 2 ) {
1698                                                         #if there's exactly one ]] that's fine, we'll keep looking
1699                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1700                                                 } else {
1701                                                         #if $next_line is invalid too, we need look no further
1702                                                         $text .= '[[' . $next_line;
1703                                                         break;
1704                                                 }
1705                                         }
1706                                         if ( !$found ) {
1707                                                 # we couldn't find the end of this imageLink, so output it raw
1708                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1709                                                 $text = $this->replaceInternalLinks($text);
1710                                                 $s .= "{$prefix}[[$link|$text";
1711                                                 # note: no $trail, because without an end, there *is* no trail
1712                                                 wfProfileOut( "$fname-might_be_img" );
1713                                                 continue;
1714                                         }
1715                                 } else { #it's not an image, so output it raw
1716                                         $s .= "{$prefix}[[$link|$text";
1717                                         # note: no $trail, because without an end, there *is* no trail
1718                                         wfProfileOut( "$fname-might_be_img" );
1719                                         continue;
1720                                 }
1721                                 wfProfileOut( "$fname-might_be_img" );
1722                         }
1723
1724                         $wasblank = ( '' == $text );
1725                         if( $wasblank ) $text = $link;
1726
1727                         # Link not escaped by : , create the various objects
1728                         if( $noforce ) {
1729
1730                                 # Interwikis
1731                                 wfProfileIn( "$fname-interwiki" );
1732                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1733                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1734                                         $s = rtrim($s . $prefix);
1735                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1736                                         wfProfileOut( "$fname-interwiki" );
1737                                         continue;
1738                                 }
1739                                 wfProfileOut( "$fname-interwiki" );
1740
1741                                 if ( $ns == NS_IMAGE ) {
1742                                         wfProfileIn( "$fname-image" );
1743                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1744                                                 # recursively parse links inside the image caption
1745                                                 # actually, this will parse them in any other parameters, too,
1746                                                 # but it might be hard to fix that, and it doesn't matter ATM
1747                                                 $text = $this->replaceExternalLinks($text);
1748                                                 $text = $this->replaceInternalLinks($text);
1749
1750                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1751                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1752                                                 $this->mOutput->addImage( $nt->getDBkey() );
1753
1754                                                 wfProfileOut( "$fname-image" );
1755                                                 continue;
1756                                         } else {
1757                                                 # We still need to record the image's presence on the page
1758                                                 $this->mOutput->addImage( $nt->getDBkey() );
1759                                         }
1760                                         wfProfileOut( "$fname-image" );
1761
1762                                 }
1763
1764                                 if ( $ns == NS_CATEGORY ) {
1765                                         wfProfileIn( "$fname-category" );
1766                                         $s = rtrim($s . "\n"); # bug 87
1767
1768                                         if ( $wasblank ) {
1769                                                 $sortkey = $this->getDefaultSort();
1770                                         } else {
1771                                                 $sortkey = $text;
1772                                         }
1773                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1774                                         $sortkey = str_replace( "\n", '', $sortkey );
1775                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1776                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1777
1778                                         /**
1779                                          * Strip the whitespace Category links produce, see bug 87
1780                                          * @todo We might want to use trim($tmp, "\n") here.
1781                                          */
1782                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1783
1784                                         wfProfileOut( "$fname-category" );
1785                                         continue;
1786                                 }
1787                         }
1788
1789                         # Self-link checking
1790                         if( $nt->getFragment() === '' ) {
1791                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1792                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1793                                         continue;
1794                                 }
1795                         }
1796
1797                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1798                         if( $ns == NS_MEDIA ) {
1799                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1800                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1801                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1802                                 $this->mOutput->addImage( $nt->getDBkey() );
1803                                 continue;
1804                         } elseif( $ns == NS_SPECIAL ) {
1805                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1806                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1807                                 } else {
1808                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1809                                 }
1810                                 continue;
1811                         } elseif( $ns == NS_IMAGE ) {
1812                                 $img = wfFindFile( $nt );
1813                                 if( $img ) {
1814                                         // Force a blue link if the file exists; may be a remote
1815                                         // upload on the shared repository, and we want to see its
1816                                         // auto-generated page.
1817                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1818                                         $this->mOutput->addLink( $nt );
1819                                         continue;
1820                                 }
1821                         }
1822                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1823                 }
1824                 wfProfileOut( $fname );
1825                 return $s;
1826         }
1827
1828         /**
1829          * Make a link placeholder. The text returned can be later resolved to a real link with
1830          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1831          * parsing of interwiki links, and secondly to allow all existence checks and
1832          * article length checks (for stub links) to be bundled into a single query.
1833          *
1834          */
1835         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1836                 wfProfileIn( __METHOD__ );
1837                 if ( ! is_object($nt) ) {
1838                         # Fail gracefully
1839                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1840                 } else {
1841                         # Separate the link trail from the rest of the link
1842                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1843
1844                         if ( $nt->isExternal() ) {
1845                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1846                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1847                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1848                         } else {
1849                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1850                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1851                                 $this->mLinkHolders['queries'][] = $query;
1852                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1853                                 $this->mLinkHolders['titles'][] = $nt;
1854
1855                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1856                         }
1857                 }
1858                 wfProfileOut( __METHOD__ );
1859                 return $retVal;
1860         }
1861
1862         /**
1863          * Render a forced-blue link inline; protect against double expansion of
1864          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1865          * Since this little disaster has to split off the trail text to avoid
1866          * breaking URLs in the following text without breaking trails on the
1867          * wiki links, it's been made into a horrible function.
1868          *
1869          * @param Title $nt
1870          * @param string $text
1871          * @param string $query
1872          * @param string $trail
1873          * @param string $prefix
1874          * @return string HTML-wikitext mix oh yuck
1875          */
1876         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1877                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1878                 $sk = $this->mOptions->getSkin();
1879                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1880                 return $this->armorLinks( $link ) . $trail;
1881         }
1882
1883         /**
1884          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1885          * going to go through further parsing steps before inline URL expansion.
1886          *
1887          * In particular this is important when using action=render, which causes
1888          * full URLs to be included.
1889          *
1890          * Oh man I hate our multi-layer parser!
1891          *
1892          * @param string more-or-less HTML
1893          * @return string less-or-more HTML with NOPARSE bits
1894          */
1895         function armorLinks( $text ) {
1896                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1897                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1898         }
1899
1900         /**
1901          * Return true if subpage links should be expanded on this page.
1902          * @return bool
1903          */
1904         function areSubpagesAllowed() {
1905                 # Some namespaces don't allow subpages
1906                 global $wgNamespacesWithSubpages;
1907                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1908         }
1909
1910         /**
1911          * Handle link to subpage if necessary
1912          * @param string $target the source of the link
1913          * @param string &$text the link text, modified as necessary
1914          * @return string the full name of the link
1915          * @private
1916          */
1917         function maybeDoSubpageLink($target, &$text) {
1918                 # Valid link forms:
1919                 # Foobar -- normal
1920                 # :Foobar -- override special treatment of prefix (images, language links)
1921                 # /Foobar -- convert to CurrentPage/Foobar
1922                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1923                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1924                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1925
1926                 $fname = 'Parser::maybeDoSubpageLink';
1927                 wfProfileIn( $fname );
1928                 $ret = $target; # default return value is no change
1929
1930                 # bug 7425
1931                 $target = trim( $target );
1932
1933                 # Some namespaces don't allow subpages,
1934                 # so only perform processing if subpages are allowed
1935                 if( $this->areSubpagesAllowed() ) {
1936                         # Look at the first character
1937                         if( $target != '' && $target{0} == '/' ) {
1938                                 # / at end means we don't want the slash to be shown
1939                                 $m = array();
1940                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1941                                 if( $trailingSlashes ) {
1942                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1943                                 } else {
1944                                         $noslash = substr( $target, 1 );
1945                                 }
1946
1947                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1948                                 if( '' === $text ) {
1949                                         $text = $target;
1950                                 } # this might be changed for ugliness reasons
1951                         } else {
1952                                 # check for .. subpage backlinks
1953                                 $dotdotcount = 0;
1954                                 $nodotdot = $target;
1955                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1956                                         ++$dotdotcount;
1957                                         $nodotdot = substr( $nodotdot, 3 );
1958                                 }
1959                                 if($dotdotcount > 0) {
1960                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1961                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1962                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1963                                                 # / at the end means don't show full path
1964                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1965                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1966                                                         if( '' === $text ) {
1967                                                                 $text = $nodotdot;
1968                                                         }
1969                                                 }
1970                                                 $nodotdot = trim( $nodotdot );
1971                                                 if( $nodotdot != '' ) {
1972                                                         $ret .= '/' . $nodotdot;
1973                                                 }
1974                                         }
1975                                 }
1976                         }
1977                 }
1978
1979                 wfProfileOut( $fname );
1980                 return $ret;
1981         }
1982
1983         /**#@+
1984          * Used by doBlockLevels()
1985          * @private
1986          */
1987         /* private */ function closeParagraph() {
1988                 $result = '';
1989                 if ( '' != $this->mLastSection ) {
1990                         $result = '</' . $this->mLastSection  . ">\n";
1991                 }
1992                 $this->mInPre = false;
1993                 $this->mLastSection = '';
1994                 return $result;
1995         }
1996         # getCommon() returns the length of the longest common substring
1997         # of both arguments, starting at the beginning of both.
1998         #
1999         /* private */ function getCommon( $st1, $st2 ) {
2000                 $fl = strlen( $st1 );
2001                 $shorter = strlen( $st2 );
2002                 if ( $fl < $shorter ) { $shorter = $fl; }
2003
2004                 for ( $i = 0; $i < $shorter; ++$i ) {
2005                         if ( $st1{$i} != $st2{$i} ) { break; }
2006                 }
2007                 return $i;
2008         }
2009         # These next three functions open, continue, and close the list
2010         # element appropriate to the prefix character passed into them.
2011         #
2012         /* private */ function openList( $char ) {
2013                 $result = $this->closeParagraph();
2014
2015                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2016                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2017                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2018                 else if ( ';' == $char ) {
2019                         $result .= '<dl><dt>';
2020                         $this->mDTopen = true;
2021                 }
2022                 else { $result = '<!-- ERR 1 -->'; }
2023
2024                 return $result;
2025         }
2026
2027         /* private */ function nextItem( $char ) {
2028                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2029                 else if ( ':' == $char || ';' == $char ) {
2030                         $close = '</dd>';
2031                         if ( $this->mDTopen ) { $close = '</dt>'; }
2032                         if ( ';' == $char ) {
2033                                 $this->mDTopen = true;
2034                                 return $close . '<dt>';
2035                         } else {
2036                                 $this->mDTopen = false;
2037                                 return $close . '<dd>';
2038                         }
2039                 }
2040                 return '<!-- ERR 2 -->';
2041         }
2042
2043         /* private */ function closeList( $char ) {
2044                 if ( '*' == $char ) { $text = '</li></ul>'; }
2045                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2046                 else if ( ':' == $char ) {
2047                         if ( $this->mDTopen ) {
2048                                 $this->mDTopen = false;
2049                                 $text = '</dt></dl>';
2050                         } else {
2051                                 $text = '</dd></dl>';
2052                         }
2053                 }
2054                 else {  return '<!-- ERR 3 -->'; }
2055                 return $text."\n";
2056         }
2057         /**#@-*/
2058
2059         /**
2060          * Make lists from lines starting with ':', '*', '#', etc.
2061          *
2062          * @private
2063          * @return string the lists rendered as HTML
2064          */
2065         function doBlockLevels( $text, $linestart ) {
2066                 $fname = 'Parser::doBlockLevels';
2067                 wfProfileIn( $fname );
2068
2069                 # Parsing through the text line by line.  The main thing
2070                 # happening here is handling of block-level elements p, pre,
2071                 # and making lists from lines starting with * # : etc.
2072                 #
2073                 $textLines = explode( "\n", $text );
2074
2075                 $lastPrefix = $output = '';
2076                 $this->mDTopen = $inBlockElem = false;
2077                 $prefixLength = 0;
2078                 $paragraphStack = false;
2079
2080                 if ( !$linestart ) {
2081                         $output .= array_shift( $textLines );
2082                 }
2083                 foreach ( $textLines as $oLine ) {
2084                         $lastPrefixLength = strlen( $lastPrefix );
2085                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2086                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2087                         if ( !$this->mInPre ) {
2088                                 # Multiple prefixes may abut each other for nested lists.
2089                                 $prefixLength = strspn( $oLine, '*#:;' );
2090                                 $pref = substr( $oLine, 0, $prefixLength );
2091
2092                                 # eh?
2093                                 $pref2 = str_replace( ';', ':', $pref );
2094                                 $t = substr( $oLine, $prefixLength );
2095                                 $this->mInPre = !empty($preOpenMatch);
2096                         } else {
2097                                 # Don't interpret any other prefixes in preformatted text
2098                                 $prefixLength = 0;
2099                                 $pref = $pref2 = '';
2100                                 $t = $oLine;
2101                         }
2102
2103                         # List generation
2104                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2105                                 # Same as the last item, so no need to deal with nesting or opening stuff
2106                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2107                                 $paragraphStack = false;
2108
2109                                 if ( substr( $pref, -1 ) == ';') {
2110                                         # The one nasty exception: definition lists work like this:
2111                                         # ; title : definition text
2112                                         # So we check for : in the remainder text to split up the
2113                                         # title and definition, without b0rking links.
2114                                         $term = $t2 = '';
2115                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2116                                                 $t = $t2;
2117                                                 $output .= $term . $this->nextItem( ':' );
2118                                         }
2119                                 }
2120                         } elseif( $prefixLength || $lastPrefixLength ) {
2121                                 # Either open or close a level...
2122                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2123                                 $paragraphStack = false;
2124
2125                                 while( $commonPrefixLength < $lastPrefixLength ) {
2126                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2127                                         --$lastPrefixLength;
2128                                 }
2129                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2130                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2131                                 }
2132                                 while ( $prefixLength > $commonPrefixLength ) {
2133                                         $char = substr( $pref, $commonPrefixLength, 1 );
2134                                         $output .= $this->openList( $char );
2135
2136                                         if ( ';' == $char ) {
2137                                                 # FIXME: This is dupe of code above
2138                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2139                                                         $t = $t2;
2140                                                         $output .= $term . $this->nextItem( ':' );
2141                                                 }
2142                                         }
2143                                         ++$commonPrefixLength;
2144                                 }
2145                                 $lastPrefix = $pref2;
2146                         }
2147                         if( 0 == $prefixLength ) {
2148                                 wfProfileIn( "$fname-paragraph" );
2149                                 # No prefix (not in list)--go to paragraph mode
2150                                 // XXX: use a stack for nestable elements like span, table and div
2151                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2152                                 $closematch = preg_match(
2153                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2154                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2155                                 if ( $openmatch or $closematch ) {
2156                                         $paragraphStack = false;
2157                                         # TODO bug 5718: paragraph closed
2158                                         $output .= $this->closeParagraph();
2159                                         if ( $preOpenMatch and !$preCloseMatch ) {
2160                                                 $this->mInPre = true;
2161                                         }
2162                                         if ( $closematch ) {
2163                                                 $inBlockElem = false;
2164                                         } else {
2165                                                 $inBlockElem = true;
2166                                         }
2167                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2168                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2169                                                 // pre
2170                                                 if ($this->mLastSection != 'pre') {
2171                                                         $paragraphStack = false;
2172                                                         $output .= $this->closeParagraph().'<pre>';
2173                                                         $this->mLastSection = 'pre';
2174                                                 }
2175                                                 $t = substr( $t, 1 );
2176                                         } else {
2177                                                 // paragraph
2178                                                 if ( '' == trim($t) ) {
2179                                                         if ( $paragraphStack ) {
2180                                                                 $output .= $paragraphStack.'<br />';
2181                                                                 $paragraphStack = false;
2182                                                                 $this->mLastSection = 'p';
2183                                                         } else {
2184                                                                 if ($this->mLastSection != 'p' ) {
2185                                                                         $output .= $this->closeParagraph();
2186                                                                         $this->mLastSection = '';
2187                                                                         $paragraphStack = '<p>';
2188                                                                 } else {
2189                                                                         $paragraphStack = '</p><p>';
2190                                                                 }
2191                                                         }
2192                                                 } else {
2193                                                         if ( $paragraphStack ) {
2194                                                                 $output .= $paragraphStack;
2195                                                                 $paragraphStack = false;
2196                                                                 $this->mLastSection = 'p';
2197                                                         } else if ($this->mLastSection != 'p') {
2198                                                                 $output .= $this->closeParagraph().'<p>';
2199                                                                 $this->mLastSection = 'p';
2200                                                         }
2201                                                 }
2202                                         }
2203                                 }
2204                                 wfProfileOut( "$fname-paragraph" );
2205                         }
2206                         // somewhere above we forget to get out of pre block (bug 785)
2207                         if($preCloseMatch && $this->mInPre) {
2208                                 $this->mInPre = false;
2209                         }
2210                         if ($paragraphStack === false) {
2211                                 $output .= $t."\n";
2212                         }
2213                 }
2214                 while ( $prefixLength ) {
2215                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2216                         --$prefixLength;
2217                 }
2218                 if ( '' != $this->mLastSection ) {
2219                         $output .= '</' . $this->mLastSection . '>';
2220                         $this->mLastSection = '';
2221                 }
2222
2223                 wfProfileOut( $fname );
2224                 return $output;
2225         }
2226
2227         /**
2228          * Split up a string on ':', ignoring any occurences inside tags
2229          * to prevent illegal overlapping.
2230          * @param string $str the string to split
2231          * @param string &$before set to everything before the ':'
2232          * @param string &$after set to everything after the ':'
2233          * return string the position of the ':', or false if none found
2234          */
2235         function findColonNoLinks($str, &$before, &$after) {
2236                 $fname = 'Parser::findColonNoLinks';
2237                 wfProfileIn( $fname );
2238
2239                 $pos = strpos( $str, ':' );
2240                 if( $pos === false ) {
2241                         // Nothing to find!
2242                         wfProfileOut( $fname );
2243                         return false;
2244                 }
2245
2246                 $lt = strpos( $str, '<' );
2247                 if( $lt === false || $lt > $pos ) {
2248                         // Easy; no tag nesting to worry about
2249                         $before = substr( $str, 0, $pos );
2250                         $after = substr( $str, $pos+1 );
2251                         wfProfileOut( $fname );
2252                         return $pos;
2253                 }
2254
2255                 // Ugly state machine to walk through avoiding tags.
2256                 $state = MW_COLON_STATE_TEXT;
2257                 $stack = 0;
2258                 $len = strlen( $str );
2259                 for( $i = 0; $i < $len; $i++ ) {
2260                         $c = $str{$i};
2261
2262                         switch( $state ) {
2263                         // (Using the number is a performance hack for common cases)
2264                         case 0: // MW_COLON_STATE_TEXT:
2265                                 switch( $c ) {
2266                                 case "<":
2267                                         // Could be either a <start> tag or an </end> tag
2268                                         $state = MW_COLON_STATE_TAGSTART;
2269                                         break;
2270                                 case ":":
2271                                         if( $stack == 0 ) {
2272                                                 // We found it!
2273                                                 $before = substr( $str, 0, $i );
2274                                                 $after = substr( $str, $i + 1 );
2275                                                 wfProfileOut( $fname );
2276                                                 return $i;
2277                                         }
2278                                         // Embedded in a tag; don't break it.
2279                                         break;
2280                                 default:
2281                                         // Skip ahead looking for something interesting
2282                                         $colon = strpos( $str, ':', $i );
2283                                         if( $colon === false ) {
2284                                                 // Nothing else interesting
2285                                                 wfProfileOut( $fname );
2286                                                 return false;
2287                                         }
2288                                         $lt = strpos( $str, '<', $i );
2289                                         if( $stack === 0 ) {
2290                                                 if( $lt === false || $colon < $lt ) {
2291                                                         // We found it!
2292                                                         $before = substr( $str, 0, $colon );
2293                                                         $after = substr( $str, $colon + 1 );
2294                                                         wfProfileOut( $fname );
2295                                                         return $i;
2296                                                 }
2297                                         }
2298                                         if( $lt === false ) {
2299                                                 // Nothing else interesting to find; abort!
2300                                                 // We're nested, but there's no close tags left. Abort!
2301                                                 break 2;
2302                                         }
2303                                         // Skip ahead to next tag start
2304                                         $i = $lt;
2305                                         $state = MW_COLON_STATE_TAGSTART;
2306                                 }
2307                                 break;
2308                         case 1: // MW_COLON_STATE_TAG:
2309                                 // In a <tag>
2310                                 switch( $c ) {
2311                                 case ">":
2312                                         $stack++;
2313                                         $state = MW_COLON_STATE_TEXT;
2314                                         break;
2315                                 case "/":
2316                                         // Slash may be followed by >?
2317                                         $state = MW_COLON_STATE_TAGSLASH;
2318                                         break;
2319                                 default:
2320                                         // ignore
2321                                 }
2322                                 break;
2323                         case 2: // MW_COLON_STATE_TAGSTART:
2324                                 switch( $c ) {
2325                                 case "/":
2326                                         $state = MW_COLON_STATE_CLOSETAG;
2327                                         break;
2328                                 case "!":
2329                                         $state = MW_COLON_STATE_COMMENT;
2330                                         break;
2331                                 case ">":
2332                                         // Illegal early close? This shouldn't happen D:
2333                                         $state = MW_COLON_STATE_TEXT;
2334                                         break;
2335                                 default:
2336                                         $state = MW_COLON_STATE_TAG;
2337                                 }
2338                                 break;
2339                         case 3: // MW_COLON_STATE_CLOSETAG:
2340                                 // In a </tag>
2341                                 if( $c == ">" ) {
2342                                         $stack--;
2343                                         if( $stack < 0 ) {
2344                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2345                                                 wfProfileOut( $fname );
2346                                                 return false;
2347                                         }
2348                                         $state = MW_COLON_STATE_TEXT;
2349                                 }
2350                                 break;
2351                         case MW_COLON_STATE_TAGSLASH:
2352                                 if( $c == ">" ) {
2353                                         // Yes, a self-closed tag <blah/>
2354                                         $state = MW_COLON_STATE_TEXT;
2355                                 } else {
2356                                         // Probably we're jumping the gun, and this is an attribute
2357                                         $state = MW_COLON_STATE_TAG;
2358                                 }
2359                                 break;
2360                         case 5: // MW_COLON_STATE_COMMENT:
2361                                 if( $c == "-" ) {
2362                                         $state = MW_COLON_STATE_COMMENTDASH;
2363                                 }
2364                                 break;
2365                         case MW_COLON_STATE_COMMENTDASH:
2366                                 if( $c == "-" ) {
2367                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2368                                 } else {
2369                                         $state = MW_COLON_STATE_COMMENT;
2370                                 }
2371                                 break;
2372                         case MW_COLON_STATE_COMMENTDASHDASH:
2373                                 if( $c == ">" ) {
2374                                         $state = MW_COLON_STATE_TEXT;
2375                                 } else {
2376                                         $state = MW_COLON_STATE_COMMENT;
2377                                 }
2378                                 break;
2379                         default:
2380                                 throw new MWException( "State machine error in $fname" );
2381                         }
2382                 }
2383                 if( $stack > 0 ) {
2384                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2385                         return false;
2386                 }
2387                 wfProfileOut( $fname );
2388                 return false;
2389         }
2390
2391         /**
2392          * Return value of a magic variable (like PAGENAME)
2393          *
2394          * @private
2395          */
2396         function getVariableValue( $index ) {
2397                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2398
2399                 /**
2400                  * Some of these require message or data lookups and can be
2401                  * expensive to check many times.
2402                  */
2403                 static $varCache = array();
2404                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2405                         if ( isset( $varCache[$index] ) ) {
2406                                 return $varCache[$index];
2407                         }
2408                 }
2409
2410                 $ts = time();
2411                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2412
2413                 # Use the time zone
2414                 global $wgLocaltimezone;
2415                 if ( isset( $wgLocaltimezone ) ) {
2416                         $oldtz = getenv( 'TZ' );
2417                         putenv( 'TZ='.$wgLocaltimezone );
2418                 }
2419                 $oe_level = error_reporting(E_ALL); #grrrr
2420
2421                 $localTimestamp = date( 'YmdHis', $ts );
2422                 $localMonth = date( 'm', $ts );
2423                 $localMonthName = date( 'n', $ts );
2424                 $localDay = date( 'j', $ts );
2425                 $localDay2 = date( 'd', $ts );
2426                 $localDayOfWeek = date( 'w', $ts );
2427                 $localWeek = date( 'W', $ts );
2428                 $localYear = date( 'Y', $ts );
2429                 $localHour = date( 'H', $ts );
2430                 if ( isset( $wgLocaltimezone ) ) {
2431                         putenv( 'TZ='.$oldtz );
2432                 }
2433
2434                 error_reporting($oe_level); #grrrr
2435
2436                 switch ( $index ) {
2437                         case 'currentmonth':
2438                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2439                         case 'currentmonthname':
2440                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2441                         case 'currentmonthnamegen':
2442                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2443                         case 'currentmonthabbrev':
2444                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2445                         case 'currentday':
2446                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2447                         case 'currentday2':
2448                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2449                         case 'localmonth':
2450                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2451                         case 'localmonthname':
2452                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2453                         case 'localmonthnamegen':
2454                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2455                         case 'localmonthabbrev':
2456                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2457                         case 'localday':
2458                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2459                         case 'localday2':
2460                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2461                         case 'pagename':
2462                                 return wfEscapeWikiText( $this->mTitle->getText() );
2463                         case 'pagenamee':
2464                                 return $this->mTitle->getPartialURL();
2465                         case 'fullpagename':
2466                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2467                         case 'fullpagenamee':
2468                                 return $this->mTitle->getPrefixedURL();
2469                         case 'subpagename':
2470                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2471                         case 'subpagenamee':
2472                                 return $this->mTitle->getSubpageUrlForm();
2473                         case 'basepagename':
2474                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2475                         case 'basepagenamee':
2476                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2477                         case 'talkpagename':
2478                                 if( $this->mTitle->canTalk() ) {
2479                                         $talkPage = $this->mTitle->getTalkPage();
2480                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2481                                 } else {
2482                                         return '';
2483                                 }
2484                         case 'talkpagenamee':
2485                                 if( $this->mTitle->canTalk() ) {
2486                                         $talkPage = $this->mTitle->getTalkPage();
2487                                         return $talkPage->getPrefixedUrl();
2488                                 } else {
2489                                         return '';
2490                                 }
2491                         case 'subjectpagename':
2492                                 $subjPage = $this->mTitle->getSubjectPage();
2493                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2494                         case 'subjectpagenamee':
2495                                 $subjPage = $this->mTitle->getSubjectPage();
2496                                 return $subjPage->getPrefixedUrl();
2497                         case 'revisionid':
2498                                 return $this->mRevisionId;
2499                         case 'revisionday':
2500                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2501                         case 'revisionday2':
2502                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2503                         case 'revisionmonth':
2504                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2505                         case 'revisionyear':
2506                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2507                         case 'revisiontimestamp':
2508                                 return $this->getRevisionTimestamp();
2509                         case 'namespace':
2510                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2511                         case 'namespacee':
2512                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2513                         case 'talkspace':
2514                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2515                         case 'talkspacee':
2516                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2517                         case 'subjectspace':
2518                                 return $this->mTitle->getSubjectNsText();
2519                         case 'subjectspacee':
2520                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2521                         case 'currentdayname':
2522                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2523                         case 'currentyear':
2524                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2525                         case 'currenttime':
2526                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2527                         case 'currenthour':
2528                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2529                         case 'currentweek':
2530                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2531                                 // int to remove the padding
2532                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2533                         case 'currentdow':
2534                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2535                         case 'localdayname':
2536                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2537                         case 'localyear':
2538                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2539                         case 'localtime':
2540                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2541                         case 'localhour':
2542                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2543                         case 'localweek':
2544                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2545                                 // int to remove the padding
2546                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2547                         case 'localdow':
2548                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2549                         case 'numberofarticles':
2550                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2551                         case 'numberoffiles':
2552                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2553                         case 'numberofusers':
2554                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2555                         case 'numberofpages':
2556                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2557                         case 'numberofadmins':
2558                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2559                         case 'numberofedits':
2560                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2561                         case 'currenttimestamp':
2562                                 return $varCache[$index] = wfTimestampNow();
2563                         case 'localtimestamp':
2564                                 return $varCache[$index] = $localTimestamp;
2565                         case 'currentversion':
2566                                 return $varCache[$index] = SpecialVersion::getVersion();
2567                         case 'sitename':
2568                                 return $wgSitename;
2569                         case 'server':
2570                                 return $wgServer;
2571                         case 'servername':
2572                                 return $wgServerName;
2573                         case 'scriptpath':
2574                                 return $wgScriptPath;
2575                         case 'directionmark':
2576                                 return $wgContLang->getDirMark();
2577                         case 'contentlanguage':
2578                                 global $wgContLanguageCode;
2579                                 return $wgContLanguageCode;
2580                         default:
2581                                 $ret = null;
2582                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2583                                         return $ret;
2584                                 else
2585                                         return null;
2586                 }
2587         }
2588
2589         /**
2590          * initialise the magic variables (like CURRENTMONTHNAME)
2591          *
2592          * @private
2593          */
2594         function initialiseVariables() {
2595                 $fname = 'Parser::initialiseVariables';
2596                 wfProfileIn( $fname );
2597                 $variableIDs = MagicWord::getVariableIDs();
2598
2599                 $this->mVariables = array();
2600                 foreach ( $variableIDs as $id ) {
2601                         $mw =& MagicWord::get( $id );
2602                         $mw->addToArray( $this->mVariables, $id );
2603                 }
2604                 wfProfileOut( $fname );
2605         }
2606
2607         /**
2608          * parse any parentheses in format ((title|part|part))
2609          * and call callbacks to get a replacement text for any found piece
2610          *
2611          * @param string $text The text to parse
2612          * @param array $callbacks rules in form:
2613          *     '{' => array(                            # opening parentheses
2614          *                                      'end' => '}',   # closing parentheses
2615          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2616          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2617          *                                                                )
2618          *                                      )
2619          *                                      'min' => 2,     # Minimum parenthesis count in cb
2620          *                                      'max' => 3,     # Maximum parenthesis count in cb
2621          * @private
2622          */
2623         function replace_callback ($text, $callbacks) {
2624                 wfProfileIn( __METHOD__ );
2625                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2626                 $lastOpeningBrace = -1;                 # last not closed parentheses
2627
2628                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2629
2630                 $i = 0;
2631                 while ( $i < strlen( $text ) ) {
2632                         # Find next opening brace, closing brace or pipe
2633                         if ( $lastOpeningBrace == -1 ) {
2634                                 $currentClosing = '';
2635                                 $search = $validOpeningBraces;
2636                         } else {
2637                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2638                                 $search = $validOpeningBraces . '|' . $currentClosing;
2639                         }
2640                         $rule = null;
2641                         $i += strcspn( $text, $search, $i );
2642                         if ( $i < strlen( $text ) ) {
2643                                 if ( $text[$i] == '|' ) {
2644                                         $found = 'pipe';
2645                                 } elseif ( $text[$i] == $currentClosing ) {
2646                                         $found = 'close';
2647                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2648                                         $found = 'open';
2649                                         $rule = $callbacks[$text[$i]];
2650                                 } else {
2651                                         # Some versions of PHP have a strcspn which stops on null characters
2652                                         # Ignore and continue
2653                                         ++$i;
2654                                         continue;
2655                                 }
2656                         } else {
2657                                 # All done
2658                                 break;
2659                         }
2660
2661                         if ( $found == 'open' ) {
2662                                 # found opening brace, let's add it to parentheses stack
2663                                 $piece = array('brace' => $text[$i],
2664                                                            'braceEnd' => $rule['end'],
2665                                                            'title' => '',
2666                                                            'parts' => null);
2667
2668                                 # count opening brace characters
2669                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2670                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2671                                 $i += $piece['count'];
2672
2673                                 # we need to add to stack only if opening brace count is enough for one of the rules
2674                                 if ( $piece['count'] >= $rule['min'] ) {
2675                                         $lastOpeningBrace ++;
2676                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2677                                 }
2678                         } elseif ( $found == 'close' ) {
2679                                 # lets check if it is enough characters for closing brace
2680                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2681                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2682
2683                                 # check for maximum matching characters (if there are 5 closing
2684                                 # characters, we will probably need only 3 - depending on the rules)
2685                                 $matchingCount = 0;
2686                                 $matchingCallback = null;
2687                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2688                                 if ( $count > $cbType['max'] ) {
2689                                         # The specified maximum exists in the callback array, unless the caller
2690                                         # has made an error
2691                                         $matchingCount = $cbType['max'];
2692                                 } else {
2693                                         # Count is less than the maximum
2694                                         # Skip any gaps in the callback array to find the true largest match
2695                                         # Need to use array_key_exists not isset because the callback can be null
2696                                         $matchingCount = $count;
2697                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2698                                                 --$matchingCount;
2699                                         }
2700                                 }
2701
2702                                 if ($matchingCount <= 0) {
2703                                         $i += $count;
2704                                         continue;
2705                                 }
2706                                 $matchingCallback = $cbType['cb'][$matchingCount];
2707
2708                                 # let's set a title or last part (if '|' was found)
2709                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2710                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2711                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2712                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2713                                 } else {
2714                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2715                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2716                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2717                                 }
2718
2719                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2720                                 $pieceEnd = $i + $matchingCount;
2721
2722                                 if( is_callable( $matchingCallback ) ) {
2723                                         $cbArgs = array (
2724                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2725                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2726                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2727                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2728                                                                          );
2729                                         # finally we can call a user callback and replace piece of text
2730                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2731                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2732                                         $i = $pieceStart + strlen($replaceWith);
2733                                 } else {
2734                                         # null value for callback means that parentheses should be parsed, but not replaced
2735                                         $i += $matchingCount;
2736                                 }
2737
2738                                 # reset last opening parentheses, but keep it in case there are unused characters
2739                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2740                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2741                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2742                                                            'title' => '',
2743                                                            'parts' => null,
2744                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2745                                 $openingBraceStack[$lastOpeningBrace--] = null;
2746
2747                                 if ($matchingCount < $piece['count']) {
2748                                         $piece['count'] -= $matchingCount;
2749                                         $piece['startAt'] -= $matchingCount;
2750                                         $piece['partStart'] = $piece['startAt'];
2751                                         # do we still qualify for any callback with remaining count?
2752                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2753                                         while ( $piece['count'] ) {
2754                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2755                                                         $lastOpeningBrace++;
2756                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2757                                                         break;
2758                                                 }
2759                                                 --$piece['count'];
2760                                         }
2761                                 }
2762                         } elseif ( $found == 'pipe' ) {
2763                                 # lets set a title if it is a first separator, or next part otherwise
2764                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2765                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2766                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2767                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2768                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2769                                 } else {
2770                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2771                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2772                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2773                                 }
2774                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2775                         }
2776                 }
2777
2778                 wfProfileOut( __METHOD__ );
2779                 return $text;
2780         }
2781
2782         /**
2783          * Replace magic variables, templates, and template arguments
2784          * with the appropriate text. Templates are substituted recursively,
2785          * taking care to avoid infinite loops.
2786          *
2787          * Note that the substitution depends on value of $mOutputType:
2788          *  OT_WIKI: only {{subst:}} templates
2789          *  OT_MSG: only magic variables
2790          *  OT_HTML: all templates and magic variables
2791          *
2792          * @param string $tex The text to transform
2793          * @param array $args Key-value pairs representing template parameters to substitute
2794          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2795          * @private
2796          */
2797         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2798                 # Prevent too big inclusions
2799                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2800                         return $text;
2801                 }
2802
2803                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2804                 wfProfileIn( $fname );
2805
2806                 # This function is called recursively. To keep track of arguments we need a stack:
2807                 array_push( $this->mArgStack, $args );
2808
2809                 $braceCallbacks = array();
2810                 if ( !$argsOnly ) {
2811                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2812                 }
2813                 if ( $this->mOutputType != OT_MSG ) {
2814                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2815                 }
2816                 if ( $braceCallbacks ) {
2817                         $callbacks = array(
2818                                 '{' => array(
2819                                         'end' => '}',
2820                                         'cb' => $braceCallbacks,
2821                                         'min' => $argsOnly ? 3 : 2,
2822                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2823                                 ),
2824                                 '[' => array(
2825                                         'end' => ']',
2826                                         'cb' => array(2=>null),
2827                                         'min' => 2,
2828                                         'max' => 2,
2829                                 )
2830                         );
2831                         $text = $this->replace_callback ($text, $callbacks);
2832
2833                         array_pop( $this->mArgStack );
2834                 }
2835                 wfProfileOut( $fname );
2836                 return $text;
2837         }
2838
2839         /**
2840          * Replace magic variables
2841          * @private
2842          */
2843         function variableSubstitution( $matches ) {
2844                 global $wgContLang;
2845                 $fname = 'Parser::variableSubstitution';
2846                 $varname = $wgContLang->lc($matches[1]);
2847                 wfProfileIn( $fname );
2848                 $skip = false;
2849                 if ( $this->mOutputType == OT_WIKI ) {
2850                         # Do only magic variables prefixed by SUBST
2851                         $mwSubst =& MagicWord::get( 'subst' );
2852                         if (!$mwSubst->matchStartAndRemove( $varname ))
2853                                 $skip = true;
2854                         # Note that if we don't substitute the variable below,
2855                         # we don't remove the {{subst:}} magic word, in case
2856                         # it is a template rather than a magic variable.
2857                 }
2858                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2859                         $id = $this->mVariables[$varname];
2860                         # Now check if we did really match, case sensitive or not
2861                         $mw =& MagicWord::get( $id );
2862                         if ($mw->match($matches[1])) {
2863                                 $text = $this->getVariableValue( $id );
2864                                 $this->mOutput->mContainsOldMagic = true;
2865                         } else {
2866                                 $text = $matches[0];
2867                         }
2868                 } else {
2869                         $text = $matches[0];
2870                 }
2871                 wfProfileOut( $fname );
2872                 return $text;
2873         }
2874
2875
2876         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2877         static function createAssocArgs( $args ) {
2878                 $assocArgs = array();
2879                 $index = 1;
2880                 foreach( $args as $arg ) {
2881                         $eqpos = strpos( $arg, '=' );
2882                         if ( $eqpos === false ) {
2883                                 $assocArgs[$index++] = $arg;
2884                         } else {
2885                                 $name = trim( substr( $arg, 0, $eqpos ) );
2886                                 $value = trim( substr( $arg, $eqpos+1 ) );
2887                                 if ( $value === false ) {
2888                                         $value = '';
2889                                 }
2890                                 if ( $name !== false ) {
2891                                         $assocArgs[$name] = $value;
2892                                 }
2893                         }
2894                 }
2895
2896                 return $assocArgs;
2897         }
2898
2899         /**
2900          * Return the text of a template, after recursively
2901          * replacing any variables or templates within the template.
2902          *
2903          * @param array $piece The parts of the template
2904          *  $piece['text']: matched text
2905          *  $piece['title']: the title, i.e. the part before the |
2906          *  $piece['parts']: the parameter array
2907          * @return string the text of the template
2908          * @private
2909          */
2910         function braceSubstitution( $piece ) {
2911                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2912                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2913                 wfProfileIn( $fname );
2914                 wfProfileIn( __METHOD__.'-setup' );
2915
2916                 # Flags
2917                 $found = false;             # $text has been filled
2918                 $nowiki = false;            # wiki markup in $text should be escaped
2919                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2920                 $noargs = false;            # Don't replace triple-brace arguments in $text
2921                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2922                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2923                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2924                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2925
2926                 # Title object, where $text came from
2927                 $title = NULL;
2928
2929                 $linestart = '';
2930
2931
2932                 # $part1 is the bit before the first |, and must contain only title characters
2933                 # $args is a list of arguments, starting from index 0, not including $part1
2934
2935                 $titleText = $part1 = $piece['title'];
2936                 # If the third subpattern matched anything, it will start with |
2937
2938                 if (null == $piece['parts']) {
2939                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2940                         if ($replaceWith != $piece['text']) {
2941                                 $text = $replaceWith;
2942                                 $found = true;
2943                                 $noparse = true;
2944                                 $noargs = true;
2945                         }
2946                 }
2947
2948                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2949                 wfProfileOut( __METHOD__.'-setup' );
2950
2951                 # SUBST
2952                 wfProfileIn( __METHOD__.'-modifiers' );
2953                 if ( !$found ) {
2954                         $mwSubst =& MagicWord::get( 'subst' );
2955                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2956                                 # One of two possibilities is true:
2957                                 # 1) Found SUBST but not in the PST phase
2958                                 # 2) Didn't find SUBST and in the PST phase
2959                                 # In either case, return without further processing
2960                                 $text = $piece['text'];
2961                                 $found = true;
2962                                 $noparse = true;
2963                                 $noargs = true;
2964                         }
2965                 }
2966
2967                 # MSG, MSGNW and RAW
2968                 if ( !$found ) {
2969                         # Check for MSGNW:
2970                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2971                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2972                                 $nowiki = true;
2973                         } else {
2974                                 # Remove obsolete MSG:
2975                                 $mwMsg =& MagicWord::get( 'msg' );
2976                                 $mwMsg->matchStartAndRemove( $part1 );
2977                         }
2978
2979                         # Check for RAW:
2980                         $mwRaw =& MagicWord::get( 'raw' );
2981                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2982                                 $forceRawInterwiki = true;
2983                         }
2984                 }
2985                 wfProfileOut( __METHOD__.'-modifiers' );
2986
2987                 //save path level before recursing into functions & templates.
2988                 $lastPathLevel = $this->mTemplatePath;
2989
2990                 # Parser functions
2991                 if ( !$found ) {
2992                         wfProfileIn( __METHOD__ . '-pfunc' );
2993
2994                         $colonPos = strpos( $part1, ':' );
2995                         if ( $colonPos !== false ) {
2996                                 # Case sensitive functions
2997                                 $function = substr( $part1, 0, $colonPos );
2998                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2999                                         $function = $this->mFunctionSynonyms[1][$function];
3000                                 } else {
3001                                         # Case insensitive functions
3002                                         $function = strtolower( $function );
3003                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3004                                                 $function = $this->mFunctionSynonyms[0][$function];
3005                                         } else {
3006                                                 $function = false;
3007                                         }
3008                                 }
3009                                 if ( $function ) {
3010                                         $funcArgs = array_map( 'trim', $args );
3011                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
3012                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
3013                                         $found = true;
3014
3015                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3016                                         //$noargs = true;
3017                                         //$noparse = true;
3018
3019                                         if ( is_array( $result ) ) {
3020                                                 if ( isset( $result[0] ) ) {
3021                                                         $text = $linestart . $result[0];
3022                                                         unset( $result[0] );
3023                                                 }
3024
3025                                                 // Extract flags into the local scope
3026                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3027                                                 extract( $result );
3028                                         } else {
3029                                                 $text = $linestart . $result;
3030                                         }
3031                                 }
3032                         }
3033                         wfProfileOut( __METHOD__ . '-pfunc' );
3034                 }
3035
3036                 # Template table test
3037
3038                 # Did we encounter this template already? If yes, it is in the cache
3039                 # and we need to check for loops.
3040                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3041                         $found = true;
3042
3043                         # Infinite loop test
3044                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3045                                 $noparse = true;
3046                                 $noargs = true;
3047                                 $found = true;
3048                                 $text = $linestart .
3049                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3050                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3051                         } else {
3052                                 # set $text to cached message.
3053                                 $text = $linestart . $this->mTemplates[$piece['title']];
3054                                 #treat title for cached page the same as others
3055                                 $ns = NS_TEMPLATE;
3056                                 $subpage = '';
3057                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3058                                 if ($subpage !== '') {
3059                                   $ns = $this->mTitle->getNamespace();
3060                                 }
3061                                 $title = Title::newFromText( $part1, $ns );
3062                                 //used by include size checking
3063                                 $titleText = $title->getPrefixedText();
3064                                 //used by edit section links
3065                                 $replaceHeadings = true;
3066
3067                         }
3068                 }
3069
3070                 # Load from database
3071                 if ( !$found ) {
3072                         wfProfileIn( __METHOD__ . '-loadtpl' );
3073                         $ns = NS_TEMPLATE;
3074                         # declaring $subpage directly in the function call
3075                         # does not work correctly with references and breaks
3076                         # {{/subpage}}-style inclusions
3077                         $subpage = '';
3078                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3079                         if ($subpage !== '') {
3080                                 $ns = $this->mTitle->getNamespace();
3081                         }
3082                         $title = Title::newFromText( $part1, $ns );
3083
3084
3085                         if ( !is_null( $title ) ) {
3086                                 $titleText = $title->getPrefixedText();
3087                                 # Check for language variants if the template is not found
3088                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3089                                         $wgContLang->findVariantLink($part1, $title);
3090                                 }
3091
3092                                 if ( !$title->isExternal() ) {
3093                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3094                                                 $text = SpecialPage::capturePath( $title );
3095                                                 if ( is_string( $text ) ) {
3096                                                         $found = true;
3097                                                         $noparse = true;
3098                                                         $noargs = true;
3099                                                         $isHTML = true;
3100                                                         $this->disableCache();
3101                                                 }
3102                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3103                                                 $found = false; //access denied
3104                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3105                                         } else {
3106                                                 list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3107                                                 if ( $articleContent !== false ) {
3108                                                         $found = true;
3109                                                         $text = $articleContent;
3110                                                         $replaceHeadings = true;
3111                                                 }
3112                                         }
3113
3114                                         # If the title is valid but undisplayable, make a link to it
3115                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3116                                                 $text = "[[:$titleText]]";
3117                                                 $found = true;
3118                                         }
3119                                 } elseif ( $title->isTrans() ) {
3120                                         // Interwiki transclusion
3121                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3122                                                 $text = $this->interwikiTransclude( $title, 'render' );
3123                                                 $isHTML = true;
3124                                                 $noparse = true;
3125                                         } else {
3126                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3127                                                 $replaceHeadings = true;
3128                                         }
3129                                         $found = true;
3130                                 }
3131
3132                                 # Template cache array insertion
3133                                 # Use the original $piece['title'] not the mangled $part1, so that
3134                                 # modifiers such as RAW: produce separate cache entries
3135                                 if( $found ) {
3136                                         if( $isHTML ) {
3137                                                 // A special page; don't store it in the template cache.
3138                                         } else {
3139                                                 $this->mTemplates[$piece['title']] = $text;
3140                                         }
3141                                         $text = $linestart . $text;
3142                                 }
3143                         }
3144                         wfProfileOut( __METHOD__ . '-loadtpl' );
3145                 }
3146
3147                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3148                         # Error, oversize inclusion
3149                         $text = $linestart .
3150                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3151                         $noparse = true;
3152                         $noargs = true;
3153                 }
3154
3155                 # Recursive parsing, escaping and link table handling
3156                 # Only for HTML output
3157                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3158                         $text = wfEscapeWikiText( $text );
3159                 } elseif ( !$this->ot['msg'] && $found ) {
3160                         if ( $noargs ) {
3161                                 $assocArgs = array();
3162                         } else {
3163                                 # Clean up argument array
3164                                 $assocArgs = self::createAssocArgs($args);
3165                                 # Add a new element to the templace recursion path
3166                                 $this->mTemplatePath[$part1] = 1;
3167                         }
3168
3169                         if ( !$noparse ) {
3170                                 # If there are any <onlyinclude> tags, only include them
3171                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3172                                         $replacer = new OnlyIncludeReplacer;
3173                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3174                                                 array( &$replacer, 'replace' ), $text );
3175                                         $text = $replacer->output;
3176                                 }
3177                                 # Remove <noinclude> sections and <includeonly> tags
3178                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3179                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3180
3181                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3182                                         # Strip <nowiki>, <pre>, etc.
3183                                         $text = $this->strip( $text, $this->mStripState );
3184                                         if ( $this->ot['html'] ) {
3185                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3186                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3187                                                 $text = Sanitizer::removeHTMLcomments( $text );
3188                                         }
3189                                 }
3190                                 $text = $this->replaceVariables( $text, $assocArgs );
3191
3192                                 # If the template begins with a table or block-level
3193                                 # element, it should be treated as beginning a new line.
3194                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3195                                         $text = "\n" . $text;
3196                                 }
3197                         } elseif ( !$noargs ) {
3198                                 # $noparse and !$noargs
3199                                 # Just replace the arguments, not any double-brace items
3200                                 # This is used for rendered interwiki transclusion
3201                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3202                         }
3203                 }
3204                 # Prune lower levels off the recursion check path
3205                 $this->mTemplatePath = $lastPathLevel;
3206
3207                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3208                         # Error, oversize inclusion
3209                         $text = $linestart .
3210                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3211                         $noparse = true;
3212                         $noargs = true;
3213                 }
3214
3215                 if ( !$found ) {
3216                         wfProfileOut( $fname );
3217                         return $piece['text'];
3218                 } else {
3219                         wfProfileIn( __METHOD__ . '-placeholders' );
3220                         if ( $isHTML ) {
3221                                 # Replace raw HTML by a placeholder
3222                                 # Add a blank line preceding, to prevent it from mucking up
3223                                 # immediately preceding headings
3224                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3225                         } else {
3226                                 # replace ==section headers==
3227                                 # XXX this needs to go away once we have a better parser.
3228                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3229                                         if( !is_null( $title ) )
3230                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3231                                         else
3232                                                 $encodedname = base64_encode("");
3233                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3234                                                 PREG_SPLIT_DELIM_CAPTURE);
3235                                         $text = '';
3236                                         $nsec = $headingOffset;
3237
3238                                         for( $i = 0; $i < count($m); $i += 2 ) {
3239                                                 $text .= $m[$i];
3240                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3241                                                 $hl = $m[$i + 1];
3242                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3243                                                         $text .= $hl;
3244                                                         continue;
3245                                                 }
3246                                                 $m2 = array();
3247                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3248                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3249                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3250
3251                                                 $nsec++;
3252                                         }
3253                                 }
3254                         }
3255                         wfProfileOut( __METHOD__ . '-placeholders' );
3256                 }
3257
3258                 # Prune lower levels off the recursion check path
3259                 $this->mTemplatePath = $lastPathLevel;
3260
3261                 if ( !$found ) {
3262                         wfProfileOut( $fname );
3263                         return $piece['text'];
3264                 } else {
3265                         wfProfileOut( $fname );
3266                         return $text;
3267                 }
3268         }
3269
3270         /**
3271          * Fetch the unparsed text of a template and register a reference to it.
3272          */
3273         function fetchTemplateAndtitle( $title ) {
3274                 $text = $skip = false;
3275                 $finalTitle = $title;
3276                 // Loop to fetch the article, with up to 1 redirect
3277                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3278                         # Give extensions a chance to select the revision instead
3279                         $id = false; // Assume current
3280                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( &$this, &$title, &$skip, &$id ) );
3281
3282                         if( $skip ) {
3283                                 $text = false;
3284                                 $this->mOutput->addTemplate( $title, $title->getArticleID(), null );
3285                                 break;
3286                         }
3287                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3288                         $rev_id = $rev ? $rev->getId() : 0;
3289
3290                         $this->mOutput->addTemplate( $title, $title->getArticleID(), $rev_id );
3291
3292                         if( $rev ) {
3293                                 $text = $rev->getText();
3294                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3295                                 global $wgLang;
3296                                 $message = $wgLang->lcfirst( $title->getText() );
3297                                 $text = wfMsgForContentNoTrans( $message );
3298                                 if( wfEmptyMsg( $message, $text ) ) {
3299                                         $text = false;
3300                                         break;
3301                                 }
3302                         } else {
3303                                 break;
3304                         }
3305                         if ( $text === false ) {
3306                                 break;
3307                         }
3308                         // Redirect?
3309                         $finalTitle = $title;
3310                         $title = Title::newFromRedirect( $text );
3311                 }
3312                 return array($text,$finalTitle);
3313         }
3314
3315         function fetchTemplate( $title ) {
3316                 $rv = $this->fetchTemplateAndtitle($title);
3317                 return $rv[0];
3318         }
3319
3320         /**
3321          * Transclude an interwiki link.
3322          */
3323         function interwikiTransclude( $title, $action ) {
3324                 global $wgEnableScaryTranscluding;
3325
3326                 if (!$wgEnableScaryTranscluding)
3327                         return wfMsg('scarytranscludedisabled');
3328
3329                 $url = $title->getFullUrl( "action=$action" );
3330
3331                 if (strlen($url) > 255)
3332                         return wfMsg('scarytranscludetoolong');
3333                 return $this->fetchScaryTemplateMaybeFromCache($url);
3334         }
3335
3336         function fetchScaryTemplateMaybeFromCache($url) {
3337                 global $wgTranscludeCacheExpiry;
3338                 $dbr = wfGetDB(DB_SLAVE);
3339                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3340                                 array('tc_url' => $url));
3341                 if ($obj) {
3342                         $time = $obj->tc_time;
3343                         $text = $obj->tc_contents;
3344                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3345                                 return $text;
3346                         }
3347                 }
3348
3349                 $text = Http::get($url);
3350                 if (!$text)
3351                         return wfMsg('scarytranscludefailed', $url);
3352
3353                 $dbw = wfGetDB(DB_MASTER);
3354                 $dbw->replace('transcache', array('tc_url'), array(
3355                         'tc_url' => $url,
3356                         'tc_time' => time(),
3357                         'tc_contents' => $text));
3358                 return $text;
3359         }
3360
3361
3362         /**
3363          * Triple brace replacement -- used for template arguments
3364          * @private
3365          */
3366         function argSubstitution( $matches ) {
3367                 $arg = trim( $matches['title'] );
3368                 $text = $matches['text'];
3369                 $inputArgs = end( $this->mArgStack );
3370
3371                 if ( array_key_exists( $arg, $inputArgs ) ) {
3372                         $text = $inputArgs[$arg];
3373                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3374                 null != $matches['parts'] && count($matches['parts']) > 0) {
3375                         $text = $matches['parts'][0];
3376                 }
3377                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3378                         $text = $matches['text'] .
3379                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3380                 }
3381
3382                 return $text;
3383         }
3384
3385         /**
3386          * Increment an include size counter
3387          *
3388          * @param string $type The type of expansion
3389          * @param integer $size The size of the text
3390          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3391          */
3392         function incrementIncludeSize( $type, $size ) {
3393                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3394                         return false;
3395                 } else {
3396                         $this->mIncludeSizes[$type] += $size;
3397                         return true;
3398                 }
3399         }
3400
3401         /**
3402          * Detect __NOGALLERY__ magic word and set a placeholder
3403          */
3404         function stripNoGallery( &$text ) {
3405                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3406                 # do not add TOC
3407                 $mw = MagicWord::get( 'nogallery' );
3408                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3409         }
3410
3411         /**
3412          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3413          * placeholder that will then be replaced by the real TOC in
3414          * ->formatHeadings, this works because at this points real
3415          * comments will have already been discarded by the sanitizer.
3416          *
3417          * Any additional __TOC__ magic words left over will be discarded
3418          * as there can only be one TOC on the page.
3419          */
3420         function stripToc( $text ) {
3421                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3422                 # do not add TOC
3423                 $mw = MagicWord::get( 'notoc' );
3424                 if( $mw->matchAndRemove( $text ) ) {
3425                         $this->mShowToc = false;
3426                 }
3427
3428                 $mw = MagicWord::get( 'toc' );
3429                 if( $mw->match( $text ) ) {
3430                         $this->mShowToc = true;
3431                         $this->mForceTocPosition = true;
3432
3433                         // Set a placeholder. At the end we'll fill it in with the TOC.
3434                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3435
3436                         // Only keep the first one.
3437                         $text = $mw->replace( '', $text );
3438                 }
3439                 return $text;
3440         }
3441
3442         /**
3443          * This function accomplishes several tasks:
3444          * 1) Auto-number headings if that option is enabled
3445          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3446          * 3) Add a Table of contents on the top for users who have enabled the option
3447          * 4) Auto-anchor headings
3448          *
3449          * It loops through all headlines, collects the necessary data, then splits up the
3450          * string and re-inserts the newly formatted headlines.
3451          *
3452          * @param string $text
3453          * @param boolean $isMain
3454          * @private
3455          */
3456         function formatHeadings( $text, $isMain=true ) {
3457                 global $wgMaxTocLevel, $wgContLang;
3458
3459                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3460                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3461                         $showEditLink = 0;
3462                 } else {
3463                         $showEditLink = $this->mOptions->getEditSection();
3464                 }
3465
3466                 # Inhibit editsection links if requested in the page
3467                 $esw =& MagicWord::get( 'noeditsection' );
3468                 if( $esw->matchAndRemove( $text ) ) {
3469                         $showEditLink = 0;
3470                 }
3471
3472                 # Get all headlines for numbering them and adding funky stuff like [edit]
3473                 # links - this is for later, but we need the number of headlines right now
3474                 $matches = array();
3475                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3476
3477                 # if there are fewer than 4 headlines in the article, do not show TOC
3478                 # unless it's been explicitly enabled.
3479                 $enoughToc = $this->mShowToc &&
3480                         (($numMatches >= 4) || $this->mForceTocPosition);
3481
3482                 # Allow user to stipulate that a page should have a "new section"
3483                 # link added via __NEWSECTIONLINK__
3484                 $mw =& MagicWord::get( 'newsectionlink' );
3485                 if( $mw->matchAndRemove( $text ) )
3486                         $this->mOutput->setNewSection( true );
3487
3488                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3489                 # override above conditions and always show TOC above first header
3490                 $mw =& MagicWord::get( 'forcetoc' );
3491                 if ($mw->matchAndRemove( $text ) ) {
3492                         $this->mShowToc = true;
3493                         $enoughToc = true;
3494                 }
3495
3496                 # We need this to perform operations on the HTML
3497                 $sk = $this->mOptions->getSkin();
3498
3499                 # headline counter
3500                 $headlineCount = 0;
3501                 $sectionCount = 0; # headlineCount excluding template sections
3502                 $numVisible = 0;
3503
3504                 # Ugh .. the TOC should have neat indentation levels which can be
3505                 # passed to the skin functions. These are determined here
3506                 $toc = '';
3507                 $full = '';
3508                 $head = array();
3509                 $sublevelCount = array();
3510                 $levelCount = array();
3511                 $toclevel = 0;
3512                 $level = 0;
3513                 $prevlevel = 0;
3514                 $toclevel = 0;
3515                 $prevtoclevel = 0;
3516
3517                 foreach( $matches[3] as $headline ) {
3518                         $istemplate = 0;
3519                         $templatetitle = '';
3520                         $templatesection = 0;
3521                         $numbering = '';
3522                         $mat = array();
3523                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3524                                 $istemplate = 1;
3525                                 $templatetitle = base64_decode($mat[1]);
3526                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3527                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3528                         }
3529
3530                         if( $toclevel ) {
3531                                 $prevlevel = $level;
3532                                 $prevtoclevel = $toclevel;
3533                         }
3534                         $level = $matches[1][$headlineCount];
3535
3536                         if( $doNumberHeadings || $enoughToc ) {
3537
3538                                 if ( $level > $prevlevel ) {
3539                                         # Increase TOC level
3540                                         $toclevel++;
3541                                         $sublevelCount[$toclevel] = 0;
3542                                         if( $toclevel<$wgMaxTocLevel ) {
3543                                                 $prevtoclevel = $toclevel;
3544                                                 $toc .= $sk->tocIndent();
3545                                                 $numVisible++;
3546                                         }
3547                                 }
3548                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3549                                         # Decrease TOC level, find level to jump to
3550
3551                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3552                                                 # Can only go down to level 1
3553                                                 $toclevel = 1;
3554                                         } else {
3555                                                 for ($i = $toclevel; $i > 0; $i--) {
3556                                                         if ( $levelCount[$i] == $level ) {
3557                                                                 # Found last matching level
3558                                                                 $toclevel = $i;
3559                                                                 break;
3560                                                         }
3561                                                         elseif ( $levelCount[$i] < $level ) {
3562                                                                 # Found first matching level below current level
3563                                                                 $toclevel = $i + 1;
3564                                                                 break;
3565                                                         }
3566                                                 }
3567                                         }
3568                                         if( $toclevel<$wgMaxTocLevel ) {
3569                                                 if($prevtoclevel < $wgMaxTocLevel) {
3570                                                         # Unindent only if the previous toc level was shown :p
3571                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3572                                                 } else {
3573                                                         $toc .= $sk->tocLineEnd();
3574                                                 }
3575                                         }
3576                                 }
3577                                 else {
3578                                         # No change in level, end TOC line
3579                                         if( $toclevel<$wgMaxTocLevel ) {
3580                                                 $toc .= $sk->tocLineEnd();
3581                                         }
3582                                 }
3583
3584                                 $levelCount[$toclevel] = $level;
3585
3586                                 # count number of headlines for each level
3587                                 @$sublevelCount[$toclevel]++;
3588                                 $dot = 0;
3589                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3590                                         if( !empty( $sublevelCount[$i] ) ) {
3591                                                 if( $dot ) {
3592                                                         $numbering .= '.';
3593                                                 }
3594                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3595                                                 $dot = 1;
3596                                         }
3597                                 }
3598                         }
3599
3600                         # The canonized header is a version of the header text safe to use for links
3601                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3602                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3603
3604                         # Remove link placeholders by the link text.
3605                         #     <!--LINK number-->
3606                         # turns into
3607                         #     link text with suffix
3608                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3609                                                             "\$this->mLinkHolders['texts'][\$1]",
3610                                                             $canonized_headline );
3611                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3612                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3613                                                             $canonized_headline );
3614
3615                         # strip out HTML
3616                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3617                         $tocline = trim( $canonized_headline );
3618                         # Save headline for section edit hint before it's escaped
3619                         $headline_hint = trim( $canonized_headline );
3620                         $canonized_headline = Sanitizer::escapeId( $tocline );
3621                         $refers[$headlineCount] = $canonized_headline;
3622
3623                         # count how many in assoc. array so we can track dupes in anchors
3624                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3625                         $refcount[$headlineCount]=$refers[$canonized_headline];
3626
3627                         # Don't number the heading if it is the only one (looks silly)
3628                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3629                                 # the two are different if the line contains a link
3630                                 $headline=$numbering . ' ' . $headline;
3631                         }
3632
3633                         # Create the anchor for linking from the TOC to the section
3634                         $anchor = $canonized_headline;
3635                         if($refcount[$headlineCount] > 1 ) {
3636                                 $anchor .= '_' . $refcount[$headlineCount];
3637                         }
3638                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3639                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3640                         }
3641                         # give headline the correct <h#> tag
3642                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3643                                 if( $istemplate )
3644                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3645                                 else
3646                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3647                         } else {
3648                                 $editlink = '';
3649                         }
3650                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3651
3652                         $headlineCount++;
3653                         if( !$istemplate )
3654                                 $sectionCount++;
3655                 }
3656
3657                 # Never ever show TOC if no headers
3658                 if( $numVisible < 1 ) {
3659                         $enoughToc = false;
3660                 }
3661
3662                 if( $enoughToc ) {
3663                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3664                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3665                         }
3666                         $toc = $sk->tocList( $toc );
3667                 }
3668
3669                 # split up and insert constructed headlines
3670
3671                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3672                 $i = 0;
3673
3674                 foreach( $blocks as $block ) {
3675                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3676                                 # This is the [edit] link that appears for the top block of text when
3677                                 # section editing is enabled
3678
3679                                 # Disabled because it broke block formatting
3680                                 # For example, a bullet point in the top line
3681                                 # $full .= $sk->editSectionLink(0);
3682                         }
3683                         $full .= $block;
3684                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3685                                 # Top anchor now in skin
3686                                 $full = $full.$toc;
3687                         }
3688
3689                         if( !empty( $head[$i] ) ) {
3690                                 $full .= $head[$i];
3691                         }
3692                         $i++;
3693                 }
3694                 if( $this->mForceTocPosition ) {
3695                         return str_replace( '<!--MWTOC-->', $toc, $full );
3696                 } else {
3697                         return $full;
3698                 }
3699         }
3700
3701         /**
3702          * Transform wiki markup when saving a page by doing \r\n -> \n
3703          * conversion, substitting signatures, {{subst:}} templates, etc.
3704          *
3705          * @param string $text the text to transform
3706          * @param Title &$title the Title object for the current article
3707          * @param User &$user the User object describing the current user
3708          * @param ParserOptions $options parsing options
3709          * @param bool $clearState whether to clear the parser state first
3710          * @return string the altered wiki markup
3711          * @public
3712          */
3713         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3714                 $this->mOptions = $options;
3715                 $this->mTitle =& $title;
3716                 $this->setOutputType( OT_WIKI );
3717
3718                 if ( $clearState ) {
3719                         $this->clearState();
3720                 }
3721
3722                 $stripState = new StripState;
3723                 $pairs = array(
3724                         "\r\n" => "\n",
3725                 );
3726                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3727                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3728                 $text = $this->pstPass2( $text, $stripState, $user );
3729                 $text = $stripState->unstripBoth( $text );
3730                 return $text;
3731         }
3732
3733         /**
3734          * Pre-save transform helper function
3735          * @private
3736          */
3737         function pstPass2( $text, &$stripState, $user ) {
3738                 global $wgContLang, $wgLocaltimezone;
3739
3740                 /* Note: This is the timestamp saved as hardcoded wikitext to
3741                  * the database, we use $wgContLang here in order to give
3742                  * everyone the same signature and use the default one rather
3743                  * than the one selected in each user's preferences.
3744                  */
3745                 if ( isset( $wgLocaltimezone ) ) {
3746                         $oldtz = getenv( 'TZ' );
3747                         putenv( 'TZ='.$wgLocaltimezone );
3748                 }
3749                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3750                   ' (' . date( 'T' ) . ')';
3751                 if ( isset( $wgLocaltimezone ) ) {
3752                         putenv( 'TZ='.$oldtz );
3753                 }
3754
3755                 # Variable replacement
3756                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3757                 $text = $this->replaceVariables( $text );
3758
3759                 # Strip out <nowiki> etc. added via replaceVariables
3760                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3761
3762                 # Signatures
3763                 $sigText = $this->getUserSig( $user );
3764                 $text = strtr( $text, array(
3765                         '~~~~~' => $d,
3766                         '~~~~' => "$sigText $d",
3767                         '~~~' => $sigText
3768                 ) );
3769
3770                 # Context links: [[|name]] and [[name (context)|]]
3771                 #
3772                 global $wgLegalTitleChars;
3773                 $tc = "[$wgLegalTitleChars]";
3774                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3775
3776                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3777                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3778                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3779
3780                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3781                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3782                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3783
3784                 $t = $this->mTitle->getText();
3785                 $m = array();
3786                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3787                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3788                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3789                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3790                 } else {
3791                         # if there's no context, don't bother duplicating the title
3792                         $text = preg_replace( $p2, '[[\\1]]', $text );
3793                 }
3794
3795                 # Trim trailing whitespace
3796                 $text = rtrim( $text );
3797
3798                 return $text;
3799         }
3800
3801         /**
3802          * Fetch the user's signature text, if any, and normalize to
3803          * validated, ready-to-insert wikitext.
3804          *
3805          * @param User $user
3806          * @return string
3807          * @private
3808          */
3809         function getUserSig( &$user ) {
3810                 global $wgMaxSigChars;
3811
3812                 $username = $user->getName();
3813                 $nickname = $user->getOption( 'nickname' );
3814                 $nickname = $nickname === '' ? $username : $nickname;
3815
3816                 if( strlen( $nickname ) > $wgMaxSigChars ) {
3817                         $nickname = $username;
3818                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3819                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3820                         # Sig. might contain markup; validate this
3821                         if( $this->validateSig( $nickname ) !== false ) {
3822                                 # Validated; clean up (if needed) and return it
3823                                 return $this->cleanSig( $nickname, true );
3824                         } else {
3825                                 # Failed to validate; fall back to the default
3826                                 $nickname = $username;
3827                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3828                         }
3829                 }
3830
3831                 // Make sure nickname doesnt get a sig in a sig
3832                 $nickname = $this->cleanSigInSig( $nickname );
3833
3834                 # If we're still here, make it a link to the user page
3835                 $userpage = $user->getUserPage();
3836                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3837         }
3838
3839         /**
3840          * Check that the user's signature contains no bad XML
3841          *
3842          * @param string $text
3843          * @return mixed An expanded string, or false if invalid.
3844          */
3845         function validateSig( $text ) {
3846                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3847         }
3848
3849         /**
3850          * Clean up signature text
3851          *
3852          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3853          * 2) Substitute all transclusions
3854          *
3855          * @param string $text
3856          * @param $parsing Whether we're cleaning (preferences save) or parsing
3857          * @return string Signature text
3858          */
3859         function cleanSig( $text, $parsing = false ) {
3860                 global $wgTitle;
3861                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3862
3863                 $substWord = MagicWord::get( 'subst' );
3864                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3865                 $substText = '{{' . $substWord->getSynonym( 0 );
3866
3867                 $text = preg_replace( $substRegex, $substText, $text );
3868                 $text = $this->cleanSigInSig( $text );
3869                 $text = $this->replaceVariables( $text );
3870
3871                 $this->clearState();
3872                 return $text;
3873         }
3874
3875         /**
3876          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3877          * @param string $text
3878          * @return string Signature text with /~{3,5}/ removed
3879          */
3880         function cleanSigInSig( $text ) {
3881                 $text = preg_replace( '/~{3,5}/', '', $text );
3882                 return $text;
3883         }
3884
3885         /**
3886          * Set up some variables which are usually set up in parse()
3887          * so that an external function can call some class members with confidence
3888          * @public
3889          */
3890         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3891                 $this->mTitle =& $title;
3892                 $this->mOptions = $options;
3893                 $this->setOutputType( $outputType );
3894                 if ( $clearState ) {
3895                         $this->clearState();
3896                 }
3897         }
3898
3899         /**
3900          * Transform a MediaWiki message by replacing magic variables.
3901          *
3902          * @param string $text the text to transform
3903          * @param ParserOptions $options  options
3904          * @return string the text with variables substituted
3905          * @public
3906          */
3907         function transformMsg( $text, $options ) {
3908                 global $wgTitle;
3909                 static $executing = false;
3910
3911                 $fname = "Parser::transformMsg";
3912
3913                 # Guard against infinite recursion
3914                 if ( $executing ) {
3915                         return $text;
3916                 }
3917                 $executing = true;
3918
3919                 wfProfileIn($fname);
3920
3921                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3922                         $this->mTitle = $wgTitle;
3923                 } else {
3924                         $this->mTitle = Title::newFromText('msg');
3925                 }
3926                 $this->mOptions = $options;
3927                 $this->setOutputType( OT_MSG );
3928                 $this->clearState();
3929                 $text = $this->replaceVariables( $text );
3930
3931                 $executing = false;
3932                 wfProfileOut($fname);
3933                 return $text;
3934         }
3935
3936         /**
3937          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3938          * The callback should have the following form:
3939          *    function myParserHook( $text, $params, &$parser ) { ... }
3940          *
3941          * Transform and return $text. Use $parser for any required context, e.g. use
3942          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3943          *
3944          * @public
3945          *
3946          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3947          * @param mixed $callback The callback function (and object) to use for the tag
3948          *
3949          * @return The old value of the mTagHooks array associated with the hook
3950          */
3951         function setHook( $tag, $callback ) {
3952                 $tag = strtolower( $tag );
3953                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3954                 $this->mTagHooks[$tag] = $callback;
3955
3956                 return $oldVal;
3957         }
3958
3959         /**
3960          * Create a function, e.g. {{sum:1|2|3}}
3961          * The callback function should have the form:
3962          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3963          *
3964          * The callback may either return the text result of the function, or an array with the text
3965          * in element 0, and a number of flags in the other elements. The names of the flags are
3966          * specified in the keys. Valid flags are:
3967          *   found                     The text returned is valid, stop processing the template. This
3968          *                             is on by default.
3969          *   nowiki                    Wiki markup in the return value should be escaped
3970          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3971          *   noargs                    Don't replace triple-brace arguments in the return value
3972          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3973          *
3974          * @public
3975          *
3976          * @param string $id The magic word ID
3977          * @param mixed $callback The callback function (and object) to use
3978          * @param integer $flags a combination of the following flags:
3979          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3980          *
3981          * @return The old callback function for this name, if any
3982          */
3983         function setFunctionHook( $id, $callback, $flags = 0 ) {
3984                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3985                 $this->mFunctionHooks[$id] = $callback;
3986
3987                 # Add to function cache
3988                 $mw = MagicWord::get( $id );
3989                 if( !$mw )
3990                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3991
3992                 $synonyms = $mw->getSynonyms();
3993                 $sensitive = intval( $mw->isCaseSensitive() );
3994
3995                 foreach ( $synonyms as $syn ) {
3996                         # Case
3997                         if ( !$sensitive ) {
3998                                 $syn = strtolower( $syn );
3999                         }
4000                         # Add leading hash
4001                         if ( !( $flags & SFH_NO_HASH ) ) {
4002                                 $syn = '#' . $syn;
4003                         }
4004                         # Remove trailing colon
4005                         if ( substr( $syn, -1, 1 ) == ':' ) {
4006                                 $syn = substr( $syn, 0, -1 );
4007                         }
4008                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4009                 }
4010                 return $oldVal;
4011         }
4012
4013         /**
4014          * Get all registered function hook identifiers
4015          *
4016          * @return array
4017          */
4018         function getFunctionHooks() {
4019                 return array_keys( $this->mFunctionHooks );
4020         }
4021
4022         /**
4023          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4024          * Placeholders created in Skin::makeLinkObj()
4025          * Returns an array of links found, indexed by PDBK:
4026          *  0 - broken
4027          *  1 - normal link
4028          *  2 - stub
4029          * $options is a bit field, RLH_FOR_UPDATE to select for update
4030          */
4031         function replaceLinkHolders( &$text, $options = 0 ) {
4032                 global $wgUser;
4033                 global $wgContLang;
4034
4035                 $fname = 'Parser::replaceLinkHolders';
4036                 wfProfileIn( $fname );
4037
4038                 $pdbks = array();
4039                 $colours = array();
4040                 $sk = $this->mOptions->getSkin();
4041                 $linkCache =& LinkCache::singleton();
4042
4043                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4044                         wfProfileIn( $fname.'-check' );
4045                         $dbr = wfGetDB( DB_SLAVE );
4046                         $page = $dbr->tableName( 'page' );
4047                         $threshold = $wgUser->getOption('stubthreshold');
4048
4049                         # Sort by namespace
4050                         asort( $this->mLinkHolders['namespaces'] );
4051
4052                         # Generate query
4053                         $query = false;
4054                         $current = null;
4055                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4056                                 # Make title object
4057                                 $title = $this->mLinkHolders['titles'][$key];
4058
4059                                 # Skip invalid entries.
4060                                 # Result will be ugly, but prevents crash.
4061                                 if ( is_null( $title ) ) {
4062                                         continue;
4063                                 }
4064                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4065
4066                                 # Check if it's a static known link, e.g. interwiki
4067                                 if ( $title->isAlwaysKnown() ) {
4068                                         $colours[$pdbk] = 1;
4069                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4070                                         $colours[$pdbk] = 1;
4071                                         $this->mOutput->addLink( $title, $id );
4072                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4073                                         $colours[$pdbk] = 0;
4074                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4075                                         $colours[$pdbk] = 0;
4076                                 } else {
4077                                         # Not in the link cache, add it to the query
4078                                         if ( !isset( $current ) ) {
4079                                                 $current = $ns;
4080                                                 $query =  "SELECT page_id, page_namespace, page_title";
4081                                                 if ( $threshold > 0 ) {
4082                                                         $query .= ', page_len, page_is_redirect';
4083                                                 }
4084                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4085                                         } elseif ( $current != $ns ) {
4086                                                 $current = $ns;
4087                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4088                                         } else {
4089                                                 $query .= ', ';
4090                                         }
4091
4092                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4093                                 }
4094                         }
4095                         if ( $query ) {
4096                                 $query .= '))';
4097                                 if ( $options & RLH_FOR_UPDATE ) {
4098                                         $query .= ' FOR UPDATE';
4099                                 }
4100
4101                                 $res = $dbr->query( $query, $fname );
4102
4103                                 # Fetch data and form into an associative array
4104                                 # non-existent = broken
4105                                 # 1 = known
4106                                 # 2 = stub
4107                                 while ( $s = $dbr->fetchObject($res) ) {
4108                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4109                                         $pdbk = $title->getPrefixedDBkey();
4110                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4111                                         $this->mOutput->addLink( $title, $s->page_id );
4112
4113                                         $colours[$pdbk] = ( $threshold == 0 || (
4114                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4115                                                                 $s->page_is_redirect ||
4116                                                                 !Namespace::isContent( $s->page_namespace ) )
4117                                                             ? 1 : 2 );
4118                                 }
4119                         }
4120                         wfProfileOut( $fname.'-check' );
4121
4122                         # Do a second query for different language variants of links and categories
4123                         if($wgContLang->hasVariants()){
4124                                 $linkBatch = new LinkBatch();
4125                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4126                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4127                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4128
4129                                 $categories = $this->mOutput->getCategoryLinks();
4130
4131                                 // Add variants of links to link batch
4132                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4133                                         $title = $this->mLinkHolders['titles'][$key];
4134                                         if ( is_null( $title ) )
4135                                                 continue;
4136
4137                                         $pdbk = $title->getPrefixedDBkey();
4138                                         $titleText = $title->getText();
4139
4140                                         // generate all variants of the link title text
4141                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4142
4143                                         // if link was not found (in first query), add all variants to query
4144                                         if ( !isset($colours[$pdbk]) ){
4145                                                 foreach($allTextVariants as $textVariant){
4146                                                         if($textVariant != $titleText){
4147                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4148                                                                 if(is_null($variantTitle)) continue;
4149                                                                 $linkBatch->addObj( $variantTitle );
4150                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4151                                                         }
4152                                                 }
4153                                         }
4154                                 }
4155
4156                                 // process categories, check if a category exists in some variant
4157                                 foreach( $categories as $category ){
4158                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4159                                         foreach($variants as $variant){
4160                                                 if($variant != $category){
4161                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4162                                                         if(is_null($variantTitle)) continue;
4163                                                         $linkBatch->addObj( $variantTitle );
4164                                                         $categoryMap[$variant] = $category;
4165                                                 }
4166                                         }
4167                                 }
4168
4169
4170                                 if(!$linkBatch->isEmpty()){
4171                                         // construct query
4172                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4173
4174                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4175                                         if ( $threshold > 0 ) {
4176                                                 $variantQuery .= ', page_len, page_is_redirect';
4177                                         }
4178
4179                                         $variantQuery .= " FROM $page WHERE $titleClause";
4180                                         if ( $options & RLH_FOR_UPDATE ) {
4181                                                 $variantQuery .= ' FOR UPDATE';
4182                                         }
4183
4184                                         $varRes = $dbr->query( $variantQuery, $fname );
4185
4186                                         // for each found variants, figure out link holders and replace
4187                                         while ( $s = $dbr->fetchObject($varRes) ) {
4188
4189                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4190                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4191                                                 $vardbk = $variantTitle->getDBkey();
4192
4193                                                 $holderKeys = array();
4194                                                 if(isset($variantMap[$varPdbk])){
4195                                                         $holderKeys = $variantMap[$varPdbk];
4196                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4197                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4198                                                 }
4199
4200                                                 // loop over link holders
4201                                                 foreach($holderKeys as $key){
4202                                                         $title = $this->mLinkHolders['titles'][$key];
4203                                                         if ( is_null( $title ) ) continue;
4204
4205                                                         $pdbk = $title->getPrefixedDBkey();
4206
4207                                                         if(!isset($colours[$pdbk])){
4208                                                                 // found link in some of the variants, replace the link holder data
4209                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4210                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4211
4212                                                                 // set pdbk and colour
4213                                                                 $pdbks[$key] = $varPdbk;
4214                                                                 if ( $threshold >  0 ) {
4215                                                                         $size = $s->page_len;
4216                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4217                                                                                 $colours[$varPdbk] = 1;
4218                                                                         } else {
4219                                                                                 $colours[$varPdbk] = 2;
4220                                                                         }
4221                                                                 }
4222                                                                 else {
4223                                                                         $colours[$varPdbk] = 1;
4224                                                                 }
4225                                                         }
4226                                                 }
4227
4228                                                 // check if the object is a variant of a category
4229                                                 if(isset($categoryMap[$vardbk])){
4230                                                         $oldkey = $categoryMap[$vardbk];
4231                                                         if($oldkey != $vardbk)
4232                                                                 $varCategories[$oldkey]=$vardbk;
4233                                                 }
4234                                         }
4235
4236                                         // rebuild the categories in original order (if there are replacements)
4237                                         if(count($varCategories)>0){
4238                                                 $newCats = array();
4239                                                 $originalCats = $this->mOutput->getCategories();
4240                                                 foreach($originalCats as $cat => $sortkey){
4241                                                         // make the replacement
4242                                                         if( array_key_exists($cat,$varCategories) )
4243                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4244                                                         else $newCats[$cat] = $sortkey;
4245                                                 }
4246                                                 $this->mOutput->setCategoryLinks($newCats);
4247                                         }
4248                                 }
4249                         }
4250
4251                         # Construct search and replace arrays
4252                         wfProfileIn( $fname.'-construct' );
4253                         $replacePairs = array();
4254                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4255                                 $pdbk = $pdbks[$key];
4256                                 $searchkey = "<!--LINK $key-->";
4257                                 $title = $this->mLinkHolders['titles'][$key];
4258                                 if ( empty( $colours[$pdbk] ) ) {
4259                                         $linkCache->addBadLinkObj( $title );
4260                                         $colours[$pdbk] = 0;
4261                                         $this->mOutput->addLink( $title, 0 );
4262                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4263                                                                         $this->mLinkHolders['texts'][$key],
4264                                                                         $this->mLinkHolders['queries'][$key] );
4265                                 } elseif ( $colours[$pdbk] == 1 ) {
4266                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4267                                                                         $this->mLinkHolders['texts'][$key],
4268                                                                         $this->mLinkHolders['queries'][$key] );
4269                                 } elseif ( $colours[$pdbk] == 2 ) {
4270                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4271                                                                         $this->mLinkHolders['texts'][$key],
4272                                                                         $this->mLinkHolders['queries'][$key] );
4273                                 }
4274                         }
4275                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4276                         wfProfileOut( $fname.'-construct' );
4277
4278                         # Do the thing
4279                         wfProfileIn( $fname.'-replace' );
4280                         $text = preg_replace_callback(
4281                                 '/(<!--LINK .*?-->)/',
4282                                 $replacer->cb(),
4283                                 $text);
4284
4285                         wfProfileOut( $fname.'-replace' );
4286                 }
4287
4288                 # Now process interwiki link holders
4289                 # This is quite a bit simpler than internal links
4290                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4291                         wfProfileIn( $fname.'-interwiki' );
4292                         # Make interwiki link HTML
4293                         $replacePairs = array();
4294                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4295                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4296                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4297                         }
4298                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4299
4300                         $text = preg_replace_callback(
4301                                 '/<!--IWLINK (.*?)-->/',
4302                                 $replacer->cb(),
4303                                 $text );
4304                         wfProfileOut( $fname.'-interwiki' );
4305                 }
4306
4307                 wfProfileOut( $fname );
4308                 return $colours;
4309         }
4310
4311         /**
4312          * Replace <!--LINK--> link placeholders with plain text of links
4313          * (not HTML-formatted).
4314          * @param string $text
4315          * @return string
4316          */
4317         function replaceLinkHoldersText( $text ) {
4318                 $fname = 'Parser::replaceLinkHoldersText';
4319                 wfProfileIn( $fname );
4320
4321                 $text = preg_replace_callback(
4322                         '/<!--(LINK|IWLINK) (.*?)-->/',
4323                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4324                         $text );
4325
4326                 wfProfileOut( $fname );
4327                 return $text;
4328         }
4329
4330         /**
4331          * @param array $matches
4332          * @return string
4333          * @private
4334          */
4335         function replaceLinkHoldersTextCallback( $matches ) {
4336                 $type = $matches[1];
4337                 $key  = $matches[2];
4338                 if( $type == 'LINK' ) {
4339                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4340                                 return $this->mLinkHolders['texts'][$key];
4341                         }
4342                 } elseif( $type == 'IWLINK' ) {
4343                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4344                                 return $this->mInterwikiLinkHolders['texts'][$key];
4345                         }
4346                 }
4347                 return $matches[0];
4348         }
4349
4350         /**
4351          * Tag hook handler for 'pre'.
4352          */
4353         function renderPreTag( $text, $attribs ) {
4354                 // Backwards-compatibility hack
4355                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4356
4357                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4358                 return wfOpenElement( 'pre', $attribs ) .
4359                         Xml::escapeTagsOnly( $content ) .
4360                         '</pre>';
4361         }
4362
4363         /**
4364          * Renders an image gallery from a text with one line per image.
4365          * text labels may be given by using |-style alternative text. E.g.
4366          *   Image:one.jpg|The number "1"
4367          *   Image:tree.jpg|A tree
4368          * given as text will return the HTML of a gallery with two images,
4369          * labeled 'The number "1"' and
4370          * 'A tree'.
4371          */
4372         function renderImageGallery( $text, $params ) {
4373                 $ig = new ImageGallery();
4374                 $ig->setContextTitle( $this->mTitle );
4375                 $ig->setShowBytes( false );
4376                 $ig->setShowFilename( false );
4377                 $ig->setParsing();
4378                 $ig->useSkin( $this->mOptions->getSkin() );
4379                 $ig->mRevisionId = $this->mRevisionId;
4380
4381                 if( isset( $params['caption'] ) ) {
4382                         $caption = $params['caption'];
4383                         $caption = htmlspecialchars( $caption );
4384                         $caption = $this->replaceInternalLinks( $caption );
4385                         $ig->setCaptionHtml( $caption );
4386                 }
4387                 if( isset( $params['perrow'] ) ) {
4388                         $ig->setPerRow( $params['perrow'] );
4389                 }
4390                 if( isset( $params['widths'] ) ) {
4391                         $ig->setWidths( $params['widths'] );
4392                 }
4393                 if( isset( $params['heights'] ) ) {
4394                         $ig->setHeights( $params['heights'] );
4395                 }
4396
4397                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4398
4399                 $lines = explode( "\n", $text );
4400                 foreach ( $lines as $line ) {
4401                         # match lines like these:
4402                         # Image:someimage.jpg|This is some image
4403                         $matches = array();
4404                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4405                         # Skip empty lines
4406                         if ( count( $matches ) == 0 ) {
4407                                 continue;
4408                         }
4409                         $tp = Title::newFromText( $matches[1] );
4410                         $nt =& $tp;
4411                         if( is_null( $nt ) ) {
4412                                 # Bogus title. Ignore these so we don't bomb out later.
4413                                 continue;
4414                         }
4415                         if ( isset( $matches[3] ) ) {
4416                                 $label = $matches[3];
4417                         } else {
4418                                 $label = '';
4419                         }
4420
4421                         $pout = $this->parse( $label,
4422                                 $this->mTitle,
4423                                 $this->mOptions,
4424                                 false, // Strip whitespace...?
4425                                 false  // Don't clear state!
4426                         );
4427                         $html = $pout->getText();
4428
4429                         $ig->add( $nt, $html );
4430
4431                         # Only add real images (bug #5586)
4432                         if ( $nt->getNamespace() == NS_IMAGE ) {
4433                                 $this->mOutput->addImage( $nt->getDBkey() );
4434                         }
4435                 }
4436                 return $ig->toHTML();
4437         }
4438
4439         /**
4440          * Parse image options text and use it to make an image
4441          */
4442         function makeImage( $nt, $options ) {
4443                 # @TODO: let the MediaHandler specify its transform parameters
4444                 #
4445                 # Check if the options text is of the form "options|alt text"
4446                 # Options are:
4447                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4448                 #  * left               no resizing, just left align. label is used for alt= only
4449                 #  * right              same, but right aligned
4450                 #  * none               same, but not aligned
4451                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4452                 #  * center             center the image
4453                 #  * framed             Keep original image size, no magnify-button.
4454                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4455                 #  * upright            reduce width for upright images, rounded to full __0 px
4456                 #  * border             draw a 1px border around the image
4457                 # vertical-align values (no % or length right now):
4458                 #  * baseline
4459                 #  * sub
4460                 #  * super
4461                 #  * top
4462                 #  * text-top
4463                 #  * middle
4464                 #  * bottom
4465                 #  * text-bottom
4466
4467
4468                 $part = array_map( 'trim', explode( '|', $options) );
4469
4470                 $mwAlign = array();
4471                 $alignments = array( 'left', 'right', 'center', 'none', 'baseline', 'sub', 'super', 'top', 'text-top', 'middle', 'bottom', 'text-bottom' );
4472                 foreach ( $alignments as $alignment ) {
4473                         $mwAlign[$alignment] =& MagicWord::get( 'img_'.$alignment );
4474                 }
4475                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4476                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4477                 $mwWidth  =& MagicWord::get( 'img_width' );
4478                 $mwFramed =& MagicWord::get( 'img_framed' );
4479                 $mwFrameless =& MagicWord::get( 'img_frameless' );
4480                 $mwUpright =& MagicWord::get( 'img_upright' );
4481                 $mwBorder =& MagicWord::get( 'img_border' );
4482                 $mwPage   =& MagicWord::get( 'img_page' );
4483                 $caption = '';
4484
4485                 $params = array();
4486                 $framed = $thumb = false;
4487                 $manual_thumb = '' ;
4488                 $align = $valign = '';
4489                 $sk = $this->mOptions->getSkin();
4490
4491                 foreach( $part as $val ) {
4492                         if ( !is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4493                                 $thumb=true;
4494                         } elseif ( !is_null( $match = $mwUpright->matchVariableStartToEnd( $val ) ) ) {
4495                                 $params['upright'] = true;
4496                                 $params['upright_factor'] = floatval( $match );
4497                         } elseif ( !is_null( $match = $mwFrameless->matchVariableStartToEnd( $val ) ) ) {
4498                                 $params['frameless'] = true;
4499                         } elseif ( !is_null( $mwBorder->matchVariableStartToEnd( $val ) ) ) {
4500                                 $params['border'] = true;
4501                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4502                                 # use manually specified thumbnail
4503                                 $thumb=true;
4504                                 $manual_thumb = $match;
4505                         } else {
4506                                 foreach( $alignments as $alignment ) {
4507                                         if ( ! is_null( $mwAlign[$alignment]->matchVariableStartToEnd($val) ) ) {
4508                                                 switch ( $alignment ) {
4509                                                         case 'left': case 'right': case 'center': case 'none':
4510                                                                 $align = $alignment; break;
4511                                                         default:
4512                                                                 $valign = $alignment;
4513                                                 }
4514                                                 continue 2;
4515                                         }
4516                                 }
4517                                 if ( ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4518                                         # Select a page in a multipage document
4519                                         $params['page'] = $match;
4520                                 } elseif ( !isset( $params['width'] ) && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4521                                         wfDebug( "img_width match: $match\n" );
4522                                         # $match is the image width in pixels
4523                                         $m = array();
4524                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4525                                                  $params['width'] = intval( $m[1] );
4526                                                  $params['height'] = intval( $m[2] );
4527                                         } else {
4528                                                 $params['width'] = intval($match);
4529                                         }
4530                                 } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4531                                         $framed=true;
4532                                 } else {
4533                                         $caption = $val;
4534                                 }
4535                         }
4536                 }
4537                 # Strip bad stuff out of the alt text
4538                 $alt = $this->replaceLinkHoldersText( $caption );
4539
4540                 # make sure there are no placeholders in thumbnail attributes
4541                 # that are later expanded to html- so expand them now and
4542                 # remove the tags
4543                 $alt = $this->mStripState->unstripBoth( $alt );
4544                 $alt = Sanitizer::stripAllTags( $alt );
4545
4546                 # Give extensions a chance to select the file revision for us
4547                 $skip = $time = false;
4548                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
4549
4550                 # Linker does the rest
4551                 if( $skip ) {
4552                         $link = $sk->makeLinkObj( $nt );
4553                 } else {
4554                         $link = $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $params, $framed, $thumb, $manual_thumb, $valign, $time );
4555                 }
4556
4557                 return $link;
4558         }
4559
4560         /**
4561          * Set a flag in the output object indicating that the content is dynamic and
4562          * shouldn't be cached.
4563          */
4564         function disableCache() {
4565                 wfDebug( "Parser output marked as uncacheable.\n" );
4566                 $this->mOutput->mCacheTime = -1;
4567         }
4568
4569         /**#@+
4570          * Callback from the Sanitizer for expanding items found in HTML attribute
4571          * values, so they can be safely tested and escaped.
4572          * @param string $text
4573          * @param array $args
4574          * @return string
4575          * @private
4576          */
4577         function attributeStripCallback( &$text, $args ) {
4578                 $text = $this->replaceVariables( $text, $args );
4579                 $text = $this->mStripState->unstripBoth( $text );
4580                 return $text;
4581         }
4582
4583         /**#@-*/
4584
4585         /**#@+
4586          * Accessor/mutator
4587          */
4588         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4589         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4590         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4591         /**#@-*/
4592
4593         /**#@+
4594          * Accessor
4595          */
4596         function getTags() { return array_keys( $this->mTagHooks ); }
4597         /**#@-*/
4598
4599
4600         /**
4601          * Break wikitext input into sections, and either pull or replace
4602          * some particular section's text.
4603          *
4604          * External callers should use the getSection and replaceSection methods.
4605          *
4606          * @param $text Page wikitext
4607          * @param $section Numbered section. 0 pulls the text before the first
4608          *                 heading; other numbers will pull the given section
4609          *                 along with its lower-level subsections.
4610          * @param $mode One of "get" or "replace"
4611          * @param $newtext Replacement text for section data.
4612          * @return string for "get", the extracted section text.
4613          *                for "replace", the whole page with the section replaced.
4614          */
4615         private function extractSections( $text, $section, $mode, $newtext='' ) {
4616                 # I.... _hope_ this is right.
4617                 # Otherwise, sometimes we don't have things initialized properly.
4618                 $this->clearState();
4619
4620                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4621                 # comments to be stripped as well)
4622                 $stripState = new StripState;
4623
4624                 $oldOutputType = $this->mOutputType;
4625                 $oldOptions = $this->mOptions;
4626                 $this->mOptions = new ParserOptions();
4627                 $this->setOutputType( OT_WIKI );
4628
4629                 $striptext = $this->strip( $text, $stripState, true );
4630
4631                 $this->setOutputType( $oldOutputType );
4632                 $this->mOptions = $oldOptions;
4633
4634                 # now that we can be sure that no pseudo-sections are in the source,
4635                 # split it up by section
4636                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4637                 $comment = "(?:$uniq-!--.*?QINU)";
4638                 $secs = preg_split(
4639                         "/
4640                         (
4641                                 ^
4642                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4643                                 (=+) # Should this be limited to 6?
4644                                 .+?  # Section title...
4645                                 \\2  # Ending = count must match start
4646                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4647                                 $
4648                         |
4649                                 <h([1-6])\b.*?>
4650                                 .*?
4651                                 <\/h\\3\s*>
4652                         )
4653                         /mix",
4654                         $striptext, -1,
4655                         PREG_SPLIT_DELIM_CAPTURE);
4656
4657                 if( $mode == "get" ) {
4658                         if( $section == 0 ) {
4659                                 // "Section 0" returns the content before any other section.
4660                                 $rv = $secs[0];
4661                         } else {
4662                                 //track missing section, will replace if found.
4663                                 $rv = $newtext;
4664                         }
4665                 } elseif( $mode == "replace" ) {
4666                         if( $section == 0 ) {
4667                                 $rv = $newtext . "\n\n";
4668                                 $remainder = true;
4669                         } else {
4670                                 $rv = $secs[0];
4671                                 $remainder = false;
4672                         }
4673                 }
4674                 $count = 0;
4675                 $sectionLevel = 0;
4676                 for( $index = 1; $index < count( $secs ); ) {
4677                         $headerLine = $secs[$index++];
4678                         if( $secs[$index] ) {
4679                                 // A wiki header
4680                                 $headerLevel = strlen( $secs[$index++] );
4681                         } else {
4682                                 // An HTML header
4683                                 $index++;
4684                                 $headerLevel = intval( $secs[$index++] );
4685                         }
4686                         $content = $secs[$index++];
4687
4688                         $count++;
4689                         if( $mode == "get" ) {
4690                                 if( $count == $section ) {
4691                                         $rv = $headerLine . $content;
4692                                         $sectionLevel = $headerLevel;
4693                                 } elseif( $count > $section ) {
4694                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4695                                                 $rv .= $headerLine . $content;
4696                                         } else {
4697                                                 // Broke out to a higher-level section
4698                                                 break;
4699                                         }
4700                                 }
4701                         } elseif( $mode == "replace" ) {
4702                                 if( $count < $section ) {
4703                                         $rv .= $headerLine . $content;
4704                                 } elseif( $count == $section ) {
4705                                         $rv .= $newtext . "\n\n";
4706                                         $sectionLevel = $headerLevel;
4707                                 } elseif( $count > $section ) {
4708                                         if( $headerLevel <= $sectionLevel ) {
4709                                                 // Passed the section's sub-parts.
4710                                                 $remainder = true;
4711                                         }
4712                                         if( $remainder ) {
4713                                                 $rv .= $headerLine . $content;
4714                                         }
4715                                 }
4716                         }
4717                 }
4718                 if (is_string($rv))
4719                         # reinsert stripped tags
4720                         $rv = trim( $stripState->unstripBoth( $rv ) );
4721
4722                 return $rv;
4723         }
4724
4725         /**
4726          * This function returns the text of a section, specified by a number ($section).
4727          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4728          * the first section before any such heading (section 0).
4729          *
4730          * If a section contains subsections, these are also returned.
4731          *
4732          * @param $text String: text to look in
4733          * @param $section Integer: section number
4734          * @param $deftext: default to return if section is not found
4735          * @return string text of the requested section
4736          */
4737         public function getSection( $text, $section, $deftext='' ) {
4738                 return $this->extractSections( $text, $section, "get", $deftext );
4739         }
4740
4741         public function replaceSection( $oldtext, $section, $text ) {
4742                 return $this->extractSections( $oldtext, $section, "replace", $text );
4743         }
4744
4745         /**
4746          * Get the timestamp associated with the current revision, adjusted for
4747          * the default server-local timestamp
4748          */
4749         function getRevisionTimestamp() {
4750                 if ( is_null( $this->mRevisionTimestamp ) ) {
4751                         wfProfileIn( __METHOD__ );
4752                         global $wgContLang;
4753                         $dbr = wfGetDB( DB_SLAVE );
4754                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4755                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4756
4757                         // Normalize timestamp to internal MW format for timezone processing.
4758                         // This has the added side-effect of replacing a null value with
4759                         // the current time, which gives us more sensible behavior for
4760                         // previews.
4761                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4762
4763                         // The cryptic '' timezone parameter tells to use the site-default
4764                         // timezone offset instead of the user settings.
4765                         //
4766                         // Since this value will be saved into the parser cache, served
4767                         // to other users, and potentially even used inside links and such,
4768                         // it needs to be consistent for all visitors.
4769                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4770
4771                         wfProfileOut( __METHOD__ );
4772                 }
4773                 return $this->mRevisionTimestamp;
4774         }
4775
4776         /**
4777          * Mutator for $mDefaultSort
4778          *
4779          * @param $sort New value
4780          */
4781         public function setDefaultSort( $sort ) {
4782                 $this->mDefaultSort = $sort;
4783         }
4784
4785         /**
4786          * Accessor for $mDefaultSort
4787          * Will use the title/prefixed title if none is set
4788          *
4789          * @return string
4790          */
4791         public function getDefaultSort() {
4792                 if( $this->mDefaultSort !== false ) {
4793                         return $this->mDefaultSort;
4794                 } else {
4795                         return $this->mTitle->getNamespace() == NS_CATEGORY
4796                                         ? $this->mTitle->getText()
4797                                         : $this->mTitle->getPrefixedText();
4798                 }
4799         }
4800
4801 }
4802
4803 /**
4804  * @todo document, briefly.
4805  * @addtogroup Parser
4806  */
4807 class OnlyIncludeReplacer {
4808         var $output = '';
4809
4810         function replace( $matches ) {
4811                 if ( substr( $matches[1], -1 ) == "\n" ) {
4812                         $this->output .= substr( $matches[1], 0, -1 );
4813                 } else {
4814                         $this->output .= $matches[1];
4815                 }
4816         }
4817 }
4818
4819 /**
4820  * @todo document, briefly.
4821  * @addtogroup Parser
4822  */
4823 class StripState {
4824         var $general, $nowiki;
4825
4826         function __construct() {
4827                 $this->general = new ReplacementArray;
4828                 $this->nowiki = new ReplacementArray;
4829         }
4830
4831         function unstripGeneral( $text ) {
4832                 wfProfileIn( __METHOD__ );
4833                 $text = $this->general->replace( $text );
4834                 wfProfileOut( __METHOD__ );
4835                 return $text;
4836         }
4837
4838         function unstripNoWiki( $text ) {
4839                 wfProfileIn( __METHOD__ );
4840                 $text = $this->nowiki->replace( $text );
4841                 wfProfileOut( __METHOD__ );
4842                 return $text;
4843         }
4844
4845         function unstripBoth( $text ) {
4846                 wfProfileIn( __METHOD__ );
4847                 $text = $this->general->replace( $text );
4848                 $text = $this->nowiki->replace( $text );
4849                 wfProfileOut( __METHOD__ );
4850                 return $text;
4851         }
4852 }
4853
4854 ?>