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