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', '1.6.2' );
  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, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
 101                 $mImageParams, $mImageParamsMagicArray;
 102
 103         # Cleared with clearState():
 104         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 105         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 106         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 107         var $mIncludeSizes, $mDefaultSort;
 108         var $mTemplates,        // cache of already loaded templates, avoids
 109                                 // multiple SQL queries for the same string
 110             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 111                                 // in this path. Used for loop detection.
 112
 113         # Temporary
 114         # These are variables reset at least once per parse regardless of $clearState
 115         var $mOptions,      // ParserOptions object
 116                 $mTitle,        // Title context, used for self-link rendering and similar things
 117                 $mOutputType,   // Output type, one of the OT_xxx constants
 118                 $ot,            // Shortcut alias, see setOutputType()
 119                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 120                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 121                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 122
 123         /**#@-*/
 124
 125         /**
 126          * Constructor
 127          *
 128          * @public
 129          */
 130         function Parser() {
 131                 $this->mTagHooks = array();
 132                 $this->mTransparentTagHooks = array();
 133                 $this->mFunctionHooks = array();
 134                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 135                 $this->mFirstCall = true;
 136         }
 137
 138         /**
 139          * Do various kinds of initialisation on the first call of the parser
 140          */
 141         function firstCallInit() {
 142                 if ( !$this->mFirstCall ) {
 143                         return;
 144                 }
 145
 146                 wfProfileIn( __METHOD__ );
 147                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 148
 149                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 150
 151                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 170                 $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 171                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 172                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 173                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 174                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 175                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 176                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 177
 178                 if ( $wgAllowDisplayTitle ) {
 179                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 180                 }
 181                 if ( $wgAllowSlowParserFunctions ) {
 182                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 183                 }
 184
 185                 $this->initialiseVariables();
 186                 $this->mFirstCall = false;
 187                 wfProfileOut( __METHOD__ );
 188         }
 189
 190         /**
 191          * Clear Parser state
 192          *
 193          * @private
 194          */
 195         function clearState() {
 196                 wfProfileIn( __METHOD__ );
 197                 if ( $this->mFirstCall ) {
 198                         $this->firstCallInit();
 199                 }
 200                 $this->mOutput = new ParserOutput;
 201                 $this->mAutonumber = 0;
 202                 $this->mLastSection = '';
 203                 $this->mDTopen = false;
 204                 $this->mIncludeCount = array();
 205                 $this->mStripState = new StripState;
 206                 $this->mArgStack = array();
 207                 $this->mInPre = false;
 208                 $this->mInterwikiLinkHolders = array(
 209                         'texts' => array(),
 210                         'titles' => array()
 211                 );
 212                 $this->mLinkHolders = array(
 213                         'namespaces' => array(),
 214                         'dbkeys' => array(),
 215                         'queries' => array(),
 216                         'texts' => array(),
 217                         'titles' => array()
 218                 );
 219                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 220
 221                 /**
 222                  * Prefix for temporary replacement strings for the multipass parser.
 223                  * \x07 should never appear in input as it's disallowed in XML.
 224                  * Using it at the front also gives us a little extra robustness
 225                  * since it shouldn't match when butted up against identifier-like
 226                  * string constructs.
 227                  */
 228                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 229
 230                 # Clear these on every parse, bug 4549
 231                 $this->mTemplates = array();
 232                 $this->mTemplatePath = array();
 233
 234                 $this->mShowToc = true;
 235                 $this->mForceTocPosition = false;
 236                 $this->mIncludeSizes = array(
 237                         'pre-expand' => 0,
 238                         'post-expand' => 0,
 239                         'arg' => 0
 240                 );
 241                 $this->mDefaultSort = false;
 242
 243                 wfRunHooks( 'ParserClearState', array( &$this ) );
 244                 wfProfileOut( __METHOD__ );
 245         }
 246
 247         function setOutputType( $ot ) {
 248                 $this->mOutputType = $ot;
 249                 // Shortcut alias
 250                 $this->ot = array(
 251                         'html' => $ot == OT_HTML,
 252                         'wiki' => $ot == OT_WIKI,
 253                         'msg' => $ot == OT_MSG,
 254                         'pre' => $ot == OT_PREPROCESS,
 255                 );
 256         }
 257
 258         /**
 259          * Accessor for mUniqPrefix.
 260          *
 261          * @public
 262          */
 263         function uniqPrefix() {
 264                 return $this->mUniqPrefix;
 265         }
 266
 267         /**
 268          * Convert wikitext to HTML
 269          * Do not call this function recursively.
 270          *
 271          * @param string $text Text we want to parse
 272          * @param Title &$title A title object
 273          * @param array $options
 274          * @param boolean $linestart
 275          * @param boolean $clearState
 276          * @param int $revid number to pass in {{REVISIONID}}
 277          * @return ParserOutput a ParserOutput
 278          */
 279         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 280                 /**
 281                  * First pass--just handle <nowiki> sections, pass the rest off
 282                  * to internalParse() which does all the real work.
 283                  */
 284
 285                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 286                 $fname = 'Parser::parse-' . wfGetCaller();
 287                 wfProfileIn( __METHOD__ );
 288                 wfProfileIn( $fname );
 289
 290                 if ( $clearState ) {
 291                         $this->clearState();
 292                 }
 293
 294                 $this->mOptions = $options;
 295                 $this->mTitle =& $title;
 296                 $oldRevisionId = $this->mRevisionId;
 297                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 298                 if( $revid !== null ) {
 299                         $this->mRevisionId = $revid;
 300                         $this->mRevisionTimestamp = null;
 301                 }
 302                 $this->setOutputType( OT_HTML );
 303                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 304                 $text = $this->strip( $text, $this->mStripState );
 305                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 306                 $text = $this->internalParse( $text );
 307                 $text = $this->mStripState->unstripGeneral( $text );
 308
 309                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 310                 $fixtags = array(
 311                         # french spaces, last one Guillemet-left
 312                         # only if there is something before the space
 313                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 314                         # french spaces, Guillemet-right
 315                         '/(\\302\\253) /' => '\\1&nbsp;',
 316                 );
 317                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 318
 319                 # only once and last
 320                 $text = $this->doBlockLevels( $text, $linestart );
 321
 322                 $this->replaceLinkHolders( $text );
 323
 324                 # the position of the parserConvert() call should not be changed. it
 325                 # assumes that the links are all replaced and the only thing left
 326                 # is the <nowiki> mark.
 327                 # Side-effects: this calls $this->mOutput->setTitleText()
 328                 $text = $wgContLang->parserConvert( $text, $this );
 329
 330                 $text = $this->mStripState->unstripNoWiki( $text );
 331
 332                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 333
 334 //!JF Move to its own function
 335
 336                 $uniq_prefix = $this->mUniqPrefix;
 337                 $matches = array();
 338                 $elements = array_keys( $this->mTransparentTagHooks );
 339                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 340
 341                 foreach( $matches as $marker => $data ) {
 342                         list( $element, $content, $params, $tag ) = $data;
 343                         $tagName = strtolower( $element );
 344                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 345                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 346                                         array( $content, $params, $this ) );
 347                         } else {
 348                                 $output = $tag;
 349                         }
 350                         $this->mStripState->general->setPair( $marker, $output );
 351                 }
 352                 $text = $this->mStripState->unstripGeneral( $text );
 353
 354                 $text = Sanitizer::normalizeCharReferences( $text );
 355
 356                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 357                         $text = Parser::tidy($text);
 358                 } else {
 359                         # attempt to sanitize at least some nesting problems
 360                         # (bug #2702 and quite a few others)
 361                         $tidyregs = array(
 362                                 # ''Something [http://www.cool.com cool''] -->
 363                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 364                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 365                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 366                                 # fix up an anchor inside another anchor, only
 367                                 # at least for a single single nested link (bug 3695)
 368                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 369                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 370                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 371                                 # contains a div, so fix it up here; replace
 372                                 # div with escaped text
 373                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 374                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 375                                 # remove empty italic or bold tag pairs, some
 376                                 # introduced by rules above
 377                                 '/<([bi])><\/\\1>/' => '',
 378                         );
 379
 380                         $text = preg_replace(
 381                                 array_keys( $tidyregs ),
 382                                 array_values( $tidyregs ),
 383                                 $text );
 384                 }
 385
 386                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 387
 388                 # Information on include size limits, for the benefit of users who try to skirt them
 389                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 390                         $max = $this->mOptions->getMaxIncludeSize();
 391                         $text .= "<!-- \n" .
 392                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 393                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 394                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 395                                 "Maximum: $max bytes\n" .
 396                                 "-->\n";
 397                 }
 398                 $this->mOutput->setText( $text );
 399                 $this->mRevisionId = $oldRevisionId;
 400                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 401                 wfProfileOut( $fname );
 402                 wfProfileOut( __METHOD__ );
 403
 404                 return $this->mOutput;
 405         }
 406
 407         /**
 408          * Recursive parser entry point that can be called from an extension tag
 409          * hook.
 410          */
 411         function recursiveTagParse( $text ) {
 412                 wfProfileIn( __METHOD__ );
 413                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 414                 $text = $this->strip( $text, $this->mStripState );
 415                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 416                 $text = $this->internalParse( $text );
 417                 wfProfileOut( __METHOD__ );
 418                 return $text;
 419         }
 420
 421         /**
 422          * Expand templates and variables in the text, producing valid, static wikitext.
 423          * Also removes comments.
 424          */
 425         function preprocess( $text, $title, $options, $revid = null ) {
 426                 wfProfileIn( __METHOD__ );
 427                 $this->clearState();
 428                 $this->setOutputType( OT_PREPROCESS );
 429                 $this->mOptions = $options;
 430                 $this->mTitle = $title;
 431                 if( $revid !== null ) {
 432                         $this->mRevisionId = $revid;
 433                 }
 434                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 435                 $text = $this->strip( $text, $this->mStripState );
 436                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 437                 if ( $this->mOptions->getRemoveComments() ) {
 438                         $text = Sanitizer::removeHTMLcomments( $text );
 439                 }
 440                 $text = $this->replaceVariables( $text );
 441                 $text = $this->mStripState->unstripBoth( $text );
 442                 wfProfileOut( __METHOD__ );
 443                 return $text;
 444         }
 445
 446         /**
 447          * Get a random string
 448          *
 449          * @private
 450          * @static
 451          */
 452         function getRandomString() {
 453                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 454         }
 455
 456         function &getTitle() { return $this->mTitle; }
 457         function getOptions() { return $this->mOptions; }
 458
 459         function getFunctionLang() {
 460                 global $wgLang, $wgContLang;
 461                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 462         }
 463
 464         /**
 465          * Replaces all occurrences of HTML-style comments and the given tags
 466          * in the text with a random marker and returns teh next text. The output
 467          * parameter $matches will be an associative array filled with data in
 468          * the form:
 469          *   'UNIQ-xxxxx' => array(
 470          *     'element',
 471          *     'tag content',
 472          *     array( 'param' => 'x' ),
 473          *     '<element param="x">tag content</element>' ) )
 474          *
 475          * @param $elements list of element names. Comments are always extracted.
 476          * @param $text Source text string.
 477          * @param $uniq_prefix
 478          *
 479          * @public
 480          * @static
 481          */
 482         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 483                 static $n = 1;
 484                 $stripped = '';
 485                 $matches = array();
 486
 487                 $taglist = implode( '|', $elements );
 488                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 489
 490                 while ( '' != $text ) {
 491                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 492                         $stripped .= $p[0];
 493                         if( count( $p ) < 5 ) {
 494                                 break;
 495                         }
 496                         if( count( $p ) > 5 ) {
 497                                 // comment
 498                                 $element    = $p[4];
 499                                 $attributes = '';
 500                                 $close      = '';
 501                                 $inside     = $p[5];
 502                         } else {
 503                                 // tag
 504                                 $element    = $p[1];
 505                                 $attributes = $p[2];
 506                                 $close      = $p[3];
 507                                 $inside     = $p[4];
 508                         }
 509
 510                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 511                         $stripped .= $marker;
 512
 513                         if ( $close === '/>' ) {
 514                                 // Empty element tag, <tag />
 515                                 $content = null;
 516                                 $text = $inside;
 517                                 $tail = null;
 518                         } else {
 519                                 if( $element == '!--' ) {
 520                                         $end = '/(-->)/';
 521                                 } else {
 522                                         $end = "/(<\\/$element\\s*>)/i";
 523                                 }
 524                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 525                                 $content = $q[0];
 526                                 if( count( $q ) < 3 ) {
 527                                         # No end tag -- let it run out to the end of the text.
 528                                         $tail = '';
 529                                         $text = '';
 530                                 } else {
 531                                         $tail = $q[1];
 532                                         $text = $q[2];
 533                                 }
 534                         }
 535
 536                         $matches[$marker] = array( $element,
 537                                 $content,
 538                                 Sanitizer::decodeTagAttributes( $attributes ),
 539                                 "<$element$attributes$close$content$tail" );
 540                 }
 541                 return $stripped;
 542         }
 543
 544         /**
 545          * Strips and renders nowiki, pre, math, hiero
 546          * If $render is set, performs necessary rendering operations on plugins
 547          * Returns the text, and fills an array with data needed in unstrip()
 548          *
 549          * @param StripState $state
 550          *
 551          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 552          *  will be stripped in addition to other tags. This is important
 553          *  for section editing, where these comments cause confusion when
 554          *  counting the sections in the wikisource
 555          *
 556          * @param array dontstrip contains tags which should not be stripped;
 557          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 558          *
 559          * @private
 560          */
 561         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 562                 global $wgContLang;
 563                 wfProfileIn( __METHOD__ );
 564                 $render = ($this->mOutputType == OT_HTML);
 565
 566                 $uniq_prefix = $this->mUniqPrefix;
 567                 $commentState = new ReplacementArray;
 568                 $nowikiItems = array();
 569                 $generalItems = array();
 570
 571                 $elements = array_merge(
 572                         array( 'nowiki', 'gallery' ),
 573                         array_keys( $this->mTagHooks ) );
 574                 global $wgRawHtml;
 575                 if( $wgRawHtml ) {
 576                         $elements[] = 'html';
 577                 }
 578                 if( $this->mOptions->getUseTeX() ) {
 579                         $elements[] = 'math';
 580                 }
 581
 582                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 583                 foreach ( $elements AS $k => $v ) {
 584                         if ( !in_array ( $v , $dontstrip ) ) continue;
 585                         unset ( $elements[$k] );
 586                 }
 587
 588                 $matches = array();
 589                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 590
 591                 foreach( $matches as $marker => $data ) {
 592                         list( $element, $content, $params, $tag ) = $data;
 593                         if( $render ) {
 594                                 $tagName = strtolower( $element );
 595                                 wfProfileIn( __METHOD__."-render-$tagName" );
 596                                 switch( $tagName ) {
 597                                 case '!--':
 598                                         // Comment
 599                                         if( substr( $tag, -3 ) == '-->' ) {
 600                                                 $output = $tag;
 601                                         } else {
 602                                                 // Unclosed comment in input.
 603                                                 // Close it so later stripping can remove it
 604                                                 $output = "$tag-->";
 605                                         }
 606                                         break;
 607                                 case 'html':
 608                                         if( $wgRawHtml ) {
 609                                                 $output = $content;
 610                                                 break;
 611                                         }
 612                                         // Shouldn't happen otherwise. :)
 613                                 case 'nowiki':
 614                                         $output = Xml::escapeTagsOnly( $content );
 615                                         break;
 616                                 case 'math':
 617                                         $output = $wgContLang->armourMath(
 618                                                 MathRenderer::renderMath( $content, $params ) );
 619                                         break;
 620                                 case 'gallery':
 621                                         $output = $this->renderImageGallery( $content, $params );
 622                                         break;
 623                                 default:
 624                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 625                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 626                                                         array( $content, $params, $this ) );
 627                                         } else {
 628                                                 throw new MWException( "Invalid call hook $element" );
 629                                         }
 630                                 }
 631                                 wfProfileOut( __METHOD__."-render-$tagName" );
 632                         } else {
 633                                 // Just stripping tags; keep the source
 634                                 $output = $tag;
 635                         }
 636
 637                         // Unstrip the output, to support recursive strip() calls
 638                         $output = $state->unstripBoth( $output );
 639
 640                         if( !$stripcomments && $element == '!--' ) {
 641                                 $commentState->setPair( $marker, $output );
 642                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 643                                 $nowikiItems[$marker] = $output;
 644                         } else {
 645                                 $generalItems[$marker] = $output;
 646                         }
 647                 }
 648                 # Add the new items to the state
 649                 # We do this after the loop instead of during it to avoid slowing
 650                 # down the recursive unstrip
 651                 $state->nowiki->mergeArray( $nowikiItems );
 652                 $state->general->mergeArray( $generalItems );
 653
 654                 # Unstrip comments unless explicitly told otherwise.
 655                 # (The comments are always stripped prior to this point, so as to
 656                 # not invoke any extension tags / parser hooks contained within
 657                 # a comment.)
 658                 if ( !$stripcomments ) {
 659                         // Put them all back and forget them
 660                         $text = $commentState->replace( $text );
 661                 }
 662
 663                 wfProfileOut( __METHOD__ );
 664                 return $text;
 665         }
 666
 667         /**
 668          * Restores pre, math, and other extensions removed by strip()
 669          *
 670          * always call unstripNoWiki() after this one
 671          * @private
 672          * @deprecated use $this->mStripState->unstrip()
 673          */
 674         function unstrip( $text, $state ) {
 675                 return $state->unstripGeneral( $text );
 676         }
 677
 678         /**
 679          * Always call this after unstrip() to preserve the order
 680          *
 681          * @private
 682          * @deprecated use $this->mStripState->unstrip()
 683          */
 684         function unstripNoWiki( $text, $state ) {
 685                 return $state->unstripNoWiki( $text );
 686         }
 687
 688         /**
 689          * @deprecated use $this->mStripState->unstripBoth()
 690          */
 691         function unstripForHTML( $text ) {
 692                 return $this->mStripState->unstripBoth( $text );
 693         }
 694
 695         /**
 696          * Add an item to the strip state
 697          * Returns the unique tag which must be inserted into the stripped text
 698          * The tag will be replaced with the original text in unstrip()
 699          *
 700          * @private
 701          */
 702         function insertStripItem( $text, &$state ) {
 703                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 704                 $state->general->setPair( $rnd, $text );
 705                 return $rnd;
 706         }
 707
 708         /**
 709          * Interface with html tidy, used if $wgUseTidy = true.
 710          * If tidy isn't able to correct the markup, the original will be
 711          * returned in all its glory with a warning comment appended.
 712          *
 713          * Either the external tidy program or the in-process tidy extension
 714          * will be used depending on availability. Override the default
 715          * $wgTidyInternal setting to disable the internal if it's not working.
 716          *
 717          * @param string $text Hideous HTML input
 718          * @return string Corrected HTML output
 719          * @public
 720          * @static
 721          */
 722         function tidy( $text ) {
 723                 global $wgTidyInternal;
 724                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 725 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 726 '<head><title>test</title></head><body>'.$text.'</body></html>';
 727                 if( $wgTidyInternal ) {
 728                         $correctedtext = Parser::internalTidy( $wrappedtext );
 729                 } else {
 730                         $correctedtext = Parser::externalTidy( $wrappedtext );
 731                 }
 732                 if( is_null( $correctedtext ) ) {
 733                         wfDebug( "Tidy error detected!\n" );
 734                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 735                 }
 736                 return $correctedtext;
 737         }
 738
 739         /**
 740          * Spawn an external HTML tidy process and get corrected markup back from it.
 741          *
 742          * @private
 743          * @static
 744          */
 745         function externalTidy( $text ) {
 746                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 747                 $fname = 'Parser::externalTidy';
 748                 wfProfileIn( $fname );
 749
 750                 $cleansource = '';
 751                 $opts = ' -utf8';
 752
 753                 $descriptorspec = array(
 754                         0 => array('pipe', 'r'),
 755                         1 => array('pipe', 'w'),
 756                         2 => array('file', wfGetNull(), 'a')
 757                 );
 758                 $pipes = array();
 759                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 760                 if (is_resource($process)) {
 761                         // Theoretically, this style of communication could cause a deadlock
 762                         // here. If the stdout buffer fills up, then writes to stdin could
 763                         // block. This doesn't appear to happen with tidy, because tidy only
 764                         // writes to stdout after it's finished reading from stdin. Search
 765                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 766                         fwrite($pipes[0], $text);
 767                         fclose($pipes[0]);
 768                         while (!feof($pipes[1])) {
 769                                 $cleansource .= fgets($pipes[1], 1024);
 770                         }
 771                         fclose($pipes[1]);
 772                         proc_close($process);
 773                 }
 774
 775                 wfProfileOut( $fname );
 776
 777                 if( $cleansource == '' && $text != '') {
 778                         // Some kind of error happened, so we couldn't get the corrected text.
 779                         // Just give up; we'll use the source text and append a warning.
 780                         return null;
 781                 } else {
 782                         return $cleansource;
 783                 }
 784         }
 785
 786         /**
 787          * Use the HTML tidy PECL extension to use the tidy library in-process,
 788          * saving the overhead of spawning a new process. Currently written to
 789          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 790          *
 791          * 'pear install tidy' should be able to compile the extension module.
 792          *
 793          * @private
 794          * @static
 795          */
 796         function internalTidy( $text ) {
 797                 global $wgTidyConf;
 798                 $fname = 'Parser::internalTidy';
 799                 wfProfileIn( $fname );
 800
 801                 tidy_load_config( $wgTidyConf );
 802                 tidy_set_encoding( 'utf8' );
 803                 tidy_parse_string( $text );
 804                 tidy_clean_repair();
 805                 if( tidy_get_status() == 2 ) {
 806                         // 2 is magic number for fatal error
 807                         // http://www.php.net/manual/en/function.tidy-get-status.php
 808                         $cleansource = null;
 809                 } else {
 810                         $cleansource = tidy_get_output();
 811                 }
 812                 wfProfileOut( $fname );
 813                 return $cleansource;
 814         }
 815
 816         /**
 817          * parse the wiki syntax used to render tables
 818          *
 819          * @private
 820          */
 821         function doTableStuff ( $text ) {
 822                 $fname = 'Parser::doTableStuff';
 823                 wfProfileIn( $fname );
 824
 825                 $lines = explode ( "\n" , $text );
 826                 $td_history = array (); // Is currently a td tag open?
 827                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 828                 $tr_history = array (); // Is currently a tr tag open?
 829                 $tr_attributes = array (); // history of tr attributes
 830                 $has_opened_tr = array(); // Did this table open a <tr> element?
 831                 $indent_level = 0; // indent level of the table
 832                 foreach ( $lines as $key => $line )
 833                 {
 834                         $line = trim ( $line );
 835
 836                         if( $line == '' ) { // empty line, go to next line
 837                                 continue;
 838                         }
 839                         $first_character = $line{0};
 840                         $matches = array();
 841
 842                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 843                                 // First check if we are starting a new table
 844                                 $indent_level = strlen( $matches[1] );
 845
 846                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 847                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 848
 849                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 850                                 array_push ( $td_history , false );
 851                                 array_push ( $last_tag_history , '' );
 852                                 array_push ( $tr_history , false );
 853                                 array_push ( $tr_attributes , '' );
 854                                 array_push ( $has_opened_tr , false );
 855                         } else if ( count ( $td_history ) == 0 ) {
 856                                 // Don't do any of the following
 857                                 continue;
 858                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 859                                 // We are ending a table
 860                                 $line = '</table>' . substr ( $line , 2 );
 861                                 $last_tag = array_pop ( $last_tag_history );
 862
 863                                 if ( !array_pop ( $has_opened_tr ) ) {
 864                                         $line = "<tr><td></td></tr>{$line}";
 865                                 }
 866
 867                                 if ( array_pop ( $tr_history ) ) {
 868                                         $line = "</tr>{$line}";
 869                                 }
 870
 871                                 if ( array_pop ( $td_history ) ) {
 872                                         $line = "</{$last_tag}>{$line}";
 873                                 }
 874                                 array_pop ( $tr_attributes );
 875                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 876                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 877                                 // Now we have a table row
 878                                 $line = preg_replace( '#^\|-+#', '', $line );
 879
 880                                 // Whats after the tag is now only attributes
 881                                 $attributes = $this->mStripState->unstripBoth( $line );
 882                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 883                                 array_pop ( $tr_attributes );
 884                                 array_push ( $tr_attributes , $attributes );
 885
 886                                 $line = '';
 887                                 $last_tag = array_pop ( $last_tag_history );
 888                                 array_pop ( $has_opened_tr );
 889                                 array_push ( $has_opened_tr , true );
 890
 891                                 if ( array_pop ( $tr_history ) ) {
 892                                         $line = '</tr>';
 893                                 }
 894
 895                                 if ( array_pop ( $td_history ) ) {
 896                                         $line = "</{$last_tag}>{$line}";
 897                                 }
 898
 899                                 $lines[$key] = $line;
 900                                 array_push ( $tr_history , false );
 901                                 array_push ( $td_history , false );
 902                                 array_push ( $last_tag_history , '' );
 903                         }
 904                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 905                                 // This might be cell elements, td, th or captions
 906                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 907                                         $first_character = '+';
 908                                         $line = substr ( $line , 1 );
 909                                 }
 910
 911                                 $line = substr ( $line , 1 );
 912
 913                                 if ( $first_character == '!' ) {
 914                                         $line = str_replace ( '!!' , '||' , $line );
 915                                 }
 916
 917                                 // Split up multiple cells on the same line.
 918                                 // FIXME : This can result in improper nesting of tags processed
 919                                 // by earlier parser steps, but should avoid splitting up eg
 920                                 // attribute values containing literal "||".
 921                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 922
 923                                 $lines[$key] = '';
 924
 925                                 // Loop through each table cell
 926                                 foreach ( $cells as $cell )
 927                                 {
 928                                         $previous = '';
 929                                         if ( $first_character != '+' )
 930                                         {
 931                                                 $tr_after = array_pop ( $tr_attributes );
 932                                                 if ( !array_pop ( $tr_history ) ) {
 933                                                         $previous = "<tr{$tr_after}>\n";
 934                                                 }
 935                                                 array_push ( $tr_history , true );
 936                                                 array_push ( $tr_attributes , '' );
 937                                                 array_pop ( $has_opened_tr );
 938                                                 array_push ( $has_opened_tr , true );
 939                                         }
 940
 941                                         $last_tag = array_pop ( $last_tag_history );
 942
 943                                         if ( array_pop ( $td_history ) ) {
 944                                                 $previous = "</{$last_tag}>{$previous}";
 945                                         }
 946
 947                                         if ( $first_character == '|' ) {
 948                                                 $last_tag = 'td';
 949                                         } else if ( $first_character == '!' ) {
 950                                                 $last_tag = 'th';
 951                                         } else if ( $first_character == '+' ) {
 952                                                 $last_tag = 'caption';
 953                                         } else {
 954                                                 $last_tag = '';
 955                                         }
 956
 957                                         array_push ( $last_tag_history , $last_tag );
 958
 959                                         // A cell could contain both parameters and data
 960                                         $cell_data = explode ( '|' , $cell , 2 );
 961
 962                                         // Bug 553: Note that a '|' inside an invalid link should not
 963                                         // be mistaken as delimiting cell parameters
 964                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 965                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 966                                         } else if ( count ( $cell_data ) == 1 )
 967                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 968                                         else {
 969                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 970                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 971                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 972                                         }
 973
 974                                         $lines[$key] .= $cell;
 975                                         array_push ( $td_history , true );
 976                                 }
 977                         }
 978                 }
 979
 980                 // Closing open td, tr && table
 981                 while ( count ( $td_history ) > 0 )
 982                 {
 983                         if ( array_pop ( $td_history ) ) {
 984                                 $lines[] = '</td>' ;
 985                         }
 986                         if ( array_pop ( $tr_history ) ) {
 987                                 $lines[] = '</tr>' ;
 988                         }
 989                         if ( !array_pop ( $has_opened_tr ) ) {
 990                                 $lines[] = "<tr><td></td></tr>" ;
 991                         }
 992
 993                         $lines[] = '</table>' ;
 994                 }
 995
 996                 $output = implode ( "\n" , $lines ) ;
 997
 998                 // special case: don't return empty table
 999                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
1000                         $output = '';
1001                 }
1002
1003                 wfProfileOut( $fname );
1004
1005                 return $output;
1006         }
1007
1008         /**
1009          * Helper function for parse() that transforms wiki markup into
1010          * HTML. Only called for $mOutputType == OT_HTML.
1011          *
1012          * @private
1013          */
1014         function internalParse( $text ) {
1015                 $args = array();
1016                 $isMain = true;
1017                 $fname = 'Parser::internalParse';
1018                 wfProfileIn( $fname );
1019
1020                 # Hook to suspend the parser in this state
1021                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1022                         wfProfileOut( $fname );
1023                         return $text ;
1024                 }
1025
1026                 # Remove <noinclude> tags and <includeonly> sections
1027                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
1028                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
1029                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
1030
1031                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
1032
1033                 $text = $this->replaceVariables( $text, $args );
1034                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1035
1036                 // Tables need to come after variable replacement for things to work
1037                 // properly; putting them before other transformations should keep
1038                 // exciting things like link expansions from showing up in surprising
1039                 // places.
1040                 $text = $this->doTableStuff( $text );
1041
1042                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1043
1044                 $text = $this->stripToc( $text );
1045                 $this->stripNoGallery( $text );
1046                 $text = $this->doHeadings( $text );
1047                 if($this->mOptions->getUseDynamicDates()) {
1048                         $df =& DateFormatter::getInstance();
1049                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1050                 }
1051                 $text = $this->doAllQuotes( $text );
1052                 $text = $this->replaceInternalLinks( $text );
1053                 $text = $this->replaceExternalLinks( $text );
1054
1055                 # replaceInternalLinks may sometimes leave behind
1056                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1057                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1058
1059                 $text = $this->doMagicLinks( $text );
1060                 $text = $this->formatHeadings( $text, $isMain );
1061
1062                 wfProfileOut( $fname );
1063                 return $text;
1064         }
1065
1066         /**
1067          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1068          * magic external links.
1069          *
1070          * @private
1071          */
1072         function &doMagicLinks( &$text ) {
1073                 wfProfileIn( __METHOD__ );
1074                 $text = preg_replace_callback(
1075                         '!(?:                           # Start cases
1076                             <a.*?</a> |                 # Skip link text
1077                             <.*?> |                     # Skip stuff inside HTML elements
1078                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1079                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1080                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1081                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1082                                      [0-9Xx]                 # check digit
1083                                    \b)
1084                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1085                 wfProfileOut( __METHOD__ );
1086                 return $text;
1087         }
1088
1089         function magicLinkCallback( $m ) {
1090                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1091                         # Skip HTML element
1092                         return $m[0];
1093                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1094                         $isbn = $m[2];
1095                         $num = strtr( $isbn, array(
1096                                 '-' => '',
1097                                 ' ' => '',
1098                                 'x' => 'X',
1099                         ));
1100                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1101                         $text = '<a href="' .
1102                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1103                                 "\" class=\"internal\">ISBN $isbn</a>";
1104                 } else {
1105                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1106                                 $keyword = 'RFC';
1107                                 $urlmsg = 'rfcurl';
1108                                 $id = $m[1];
1109                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1110                                 $keyword = 'PMID';
1111                                 $urlmsg = 'pubmedurl';
1112                                 $id = $m[1];
1113                         } else {
1114                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1115                                         substr($m[0], 0, 20 ) . '"' );
1116                         }
1117
1118                         $url = wfMsg( $urlmsg, $id);
1119                         $sk = $this->mOptions->getSkin();
1120                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1121                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1122                 }
1123                 return $text;
1124         }
1125
1126         /**
1127          * Parse headers and return html
1128          *
1129          * @private
1130          */
1131         function doHeadings( $text ) {
1132                 $fname = 'Parser::doHeadings';
1133                 wfProfileIn( $fname );
1134                 for ( $i = 6; $i >= 1; --$i ) {
1135                         $h = str_repeat( '=', $i );
1136                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1137                           "<h{$i}>\\1</h{$i}>\\2", $text );
1138                 }
1139                 wfProfileOut( $fname );
1140                 return $text;
1141         }
1142
1143         /**
1144          * Replace single quotes with HTML markup
1145          * @private
1146          * @return string the altered text
1147          */
1148         function doAllQuotes( $text ) {
1149                 $fname = 'Parser::doAllQuotes';
1150                 wfProfileIn( $fname );
1151                 $outtext = '';
1152                 $lines = explode( "\n", $text );
1153                 foreach ( $lines as $line ) {
1154                         $outtext .= $this->doQuotes ( $line ) . "\n";
1155                 }
1156                 $outtext = substr($outtext, 0,-1);
1157                 wfProfileOut( $fname );
1158                 return $outtext;
1159         }
1160
1161         /**
1162          * Helper function for doAllQuotes()
1163          */
1164         public function doQuotes( $text ) {
1165                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1166                 if ( count( $arr ) == 1 )
1167                         return $text;
1168                 else
1169                 {
1170                         # First, do some preliminary work. This may shift some apostrophes from
1171                         # being mark-up to being text. It also counts the number of occurrences
1172                         # of bold and italics mark-ups.
1173                         $i = 0;
1174                         $numbold = 0;
1175                         $numitalics = 0;
1176                         foreach ( $arr as $r )
1177                         {
1178                                 if ( ( $i % 2 ) == 1 )
1179                                 {
1180                                         # If there are ever four apostrophes, assume the first is supposed to
1181                                         # be text, and the remaining three constitute mark-up for bold text.
1182                                         if ( strlen( $arr[$i] ) == 4 )
1183                                         {
1184                                                 $arr[$i-1] .= "'";
1185                                                 $arr[$i] = "'''";
1186                                         }
1187                                         # If there are more than 5 apostrophes in a row, assume they're all
1188                                         # text except for the last 5.
1189                                         else if ( strlen( $arr[$i] ) > 5 )
1190                                         {
1191                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1192                                                 $arr[$i] = "'''''";
1193                                         }
1194                                         # Count the number of occurrences of bold and italics mark-ups.
1195                                         # We are not counting sequences of five apostrophes.
1196                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1197                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1198                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1199                                 }
1200                                 $i++;
1201                         }
1202
1203                         # If there is an odd number of both bold and italics, it is likely
1204                         # that one of the bold ones was meant to be an apostrophe followed
1205                         # by italics. Which one we cannot know for certain, but it is more
1206                         # likely to be one that has a single-letter word before it.
1207                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1208                         {
1209                                 $i = 0;
1210                                 $firstsingleletterword = -1;
1211                                 $firstmultiletterword = -1;
1212                                 $firstspace = -1;
1213                                 foreach ( $arr as $r )
1214                                 {
1215                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1216                                         {
1217                                                 $x1 = substr ($arr[$i-1], -1);
1218                                                 $x2 = substr ($arr[$i-1], -2, 1);
1219                                                 if ($x1 == ' ') {
1220                                                         if ($firstspace == -1) $firstspace = $i;
1221                                                 } else if ($x2 == ' ') {
1222                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1223                                                 } else {
1224                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1225                                                 }
1226                                         }
1227                                         $i++;
1228                                 }
1229
1230                                 # If there is a single-letter word, use it!
1231                                 if ($firstsingleletterword > -1)
1232                                 {
1233                                         $arr [ $firstsingleletterword ] = "''";
1234                                         $arr [ $firstsingleletterword-1 ] .= "'";
1235                                 }
1236                                 # If not, but there's a multi-letter word, use that one.
1237                                 else if ($firstmultiletterword > -1)
1238                                 {
1239                                         $arr [ $firstmultiletterword ] = "''";
1240                                         $arr [ $firstmultiletterword-1 ] .= "'";
1241                                 }
1242                                 # ... otherwise use the first one that has neither.
1243                                 # (notice that it is possible for all three to be -1 if, for example,
1244                                 # there is only one pentuple-apostrophe in the line)
1245                                 else if ($firstspace > -1)
1246                                 {
1247                                         $arr [ $firstspace ] = "''";
1248                                         $arr [ $firstspace-1 ] .= "'";
1249                                 }
1250                         }
1251
1252                         # Now let's actually convert our apostrophic mush to HTML!
1253                         $output = '';
1254                         $buffer = '';
1255                         $state = '';
1256                         $i = 0;
1257                         foreach ($arr as $r)
1258                         {
1259                                 if (($i % 2) == 0)
1260                                 {
1261                                         if ($state == 'both')
1262                                                 $buffer .= $r;
1263                                         else
1264                                                 $output .= $r;
1265                                 }
1266                                 else
1267                                 {
1268                                         if (strlen ($r) == 2)
1269                                         {
1270                                                 if ($state == 'i')
1271                                                 { $output .= '</i>'; $state = ''; }
1272                                                 else if ($state == 'bi')
1273                                                 { $output .= '</i>'; $state = 'b'; }
1274                                                 else if ($state == 'ib')
1275                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1276                                                 else if ($state == 'both')
1277                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1278                                                 else # $state can be 'b' or ''
1279                                                 { $output .= '<i>'; $state .= 'i'; }
1280                                         }
1281                                         else if (strlen ($r) == 3)
1282                                         {
1283                                                 if ($state == 'b')
1284                                                 { $output .= '</b>'; $state = ''; }
1285                                                 else if ($state == 'bi')
1286                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1287                                                 else if ($state == 'ib')
1288                                                 { $output .= '</b>'; $state = 'i'; }
1289                                                 else if ($state == 'both')
1290                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1291                                                 else # $state can be 'i' or ''
1292                                                 { $output .= '<b>'; $state .= 'b'; }
1293                                         }
1294                                         else if (strlen ($r) == 5)
1295                                         {
1296                                                 if ($state == 'b')
1297                                                 { $output .= '</b><i>'; $state = 'i'; }
1298                                                 else if ($state == 'i')
1299                                                 { $output .= '</i><b>'; $state = 'b'; }
1300                                                 else if ($state == 'bi')
1301                                                 { $output .= '</i></b>'; $state = ''; }
1302                                                 else if ($state == 'ib')
1303                                                 { $output .= '</b></i>'; $state = ''; }
1304                                                 else if ($state == 'both')
1305                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1306                                                 else # ($state == '')
1307                                                 { $buffer = ''; $state = 'both'; }
1308                                         }
1309                                 }
1310                                 $i++;
1311                         }
1312                         # Now close all remaining tags.  Notice that the order is important.
1313                         if ($state == 'b' || $state == 'ib')
1314                                 $output .= '</b>';
1315                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1316                                 $output .= '</i>';
1317                         if ($state == 'bi')
1318                                 $output .= '</b>';
1319                         # There might be lonely ''''', so make sure we have a buffer
1320                         if ($state == 'both' && $buffer)
1321                                 $output .= '<b><i>'.$buffer.'</i></b>';
1322                         return $output;
1323                 }
1324         }
1325
1326         /**
1327          * Replace external links
1328          *
1329          * Note: this is all very hackish and the order of execution matters a lot.
1330          * Make sure to run maintenance/parserTests.php if you change this code.
1331          *
1332          * @private
1333          */
1334         function replaceExternalLinks( $text ) {
1335                 global $wgContLang;
1336                 $fname = 'Parser::replaceExternalLinks';
1337                 wfProfileIn( $fname );
1338
1339                 $sk = $this->mOptions->getSkin();
1340
1341                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1342
1343                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1344
1345                 $i = 0;
1346                 while ( $i<count( $bits ) ) {
1347                         $url = $bits[$i++];
1348                         $protocol = $bits[$i++];
1349                         $text = $bits[$i++];
1350                         $trail = $bits[$i++];
1351
1352                         # The characters '<' and '>' (which were escaped by
1353                         # removeHTMLtags()) should not be included in
1354                         # URLs, per RFC 2396.
1355                         $m2 = array();
1356                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1357                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1358                                 $url = substr($url, 0, $m2[0][1]);
1359                         }
1360
1361                         # If the link text is an image URL, replace it with an <img> tag
1362                         # This happened by accident in the original parser, but some people used it extensively
1363                         $img = $this->maybeMakeExternalImage( $text );
1364                         if ( $img !== false ) {
1365                                 $text = $img;
1366                         }
1367
1368                         $dtrail = '';
1369
1370                         # Set linktype for CSS - if URL==text, link is essentially free
1371                         $linktype = ($text == $url) ? 'free' : 'text';
1372
1373                         # No link text, e.g. [http://domain.tld/some.link]
1374                         if ( $text == '' ) {
1375                                 # Autonumber if allowed. See bug #5918
1376                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1377                                         $text = '[' . ++$this->mAutonumber . ']';
1378                                         $linktype = 'autonumber';
1379                                 } else {
1380                                         # Otherwise just use the URL
1381                                         $text = htmlspecialchars( $url );
1382                                         $linktype = 'free';
1383                                 }
1384                         } else {
1385                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1386                                 # Check for trail
1387                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1388                         }
1389
1390                         $text = $wgContLang->markNoConversion($text);
1391
1392                         $url = Sanitizer::cleanUrl( $url );
1393
1394                         # Process the trail (i.e. everything after this link up until start of the next link),
1395                         # replacing any non-bracketed links
1396                         $trail = $this->replaceFreeExternalLinks( $trail );
1397
1398                         # Use the encoded URL
1399                         # This means that users can paste URLs directly into the text
1400                         # Funny characters like &ouml; aren't valid in URLs anyway
1401                         # This was changed in August 2004
1402                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1403
1404                         # Register link in the output object.
1405                         # Replace unnecessary URL escape codes with the referenced character
1406                         # This prevents spammers from hiding links from the filters
1407                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1408                         $this->mOutput->addExternalLink( $pasteurized );
1409                 }
1410
1411                 wfProfileOut( $fname );
1412                 return $s;
1413         }
1414
1415         /**
1416          * Replace anything that looks like a URL with a link
1417          * @private
1418          */
1419         function replaceFreeExternalLinks( $text ) {
1420                 global $wgContLang;
1421                 $fname = 'Parser::replaceFreeExternalLinks';
1422                 wfProfileIn( $fname );
1423
1424                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1425                 $s = array_shift( $bits );
1426                 $i = 0;
1427
1428                 $sk = $this->mOptions->getSkin();
1429
1430                 while ( $i < count( $bits ) ){
1431                         $protocol = $bits[$i++];
1432                         $remainder = $bits[$i++];
1433
1434                         $m = array();
1435                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1436                                 # Found some characters after the protocol that look promising
1437                                 $url = $protocol . $m[1];
1438                                 $trail = $m[2];
1439
1440                                 # special case: handle urls as url args:
1441                                 # http://www.example.com/foo?=http://www.example.com/bar
1442                                 if(strlen($trail) == 0 &&
1443                                         isset($bits[$i]) &&
1444                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1445                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1446                                 {
1447                                         # add protocol, arg
1448                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1449                                         $i += 2;
1450                                         $trail = $m[2];
1451                                 }
1452
1453                                 # The characters '<' and '>' (which were escaped by
1454                                 # removeHTMLtags()) should not be included in
1455                                 # URLs, per RFC 2396.
1456                                 $m2 = array();
1457                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1458                                         $trail = substr($url, $m2[0][1]) . $trail;
1459                                         $url = substr($url, 0, $m2[0][1]);
1460                                 }
1461
1462                                 # Move trailing punctuation to $trail
1463                                 $sep = ',;\.:!?';
1464                                 # If there is no left bracket, then consider right brackets fair game too
1465                                 if ( strpos( $url, '(' ) === false ) {
1466                                         $sep .= ')';
1467                                 }
1468
1469                                 $numSepChars = strspn( strrev( $url ), $sep );
1470                                 if ( $numSepChars ) {
1471                                         $trail = substr( $url, -$numSepChars ) . $trail;
1472                                         $url = substr( $url, 0, -$numSepChars );
1473                                 }
1474
1475                                 $url = Sanitizer::cleanUrl( $url );
1476
1477                                 # Is this an external image?
1478                                 $text = $this->maybeMakeExternalImage( $url );
1479                                 if ( $text === false ) {
1480                                         # Not an image, make a link
1481                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1482                                         # Register it in the output object...
1483                                         # Replace unnecessary URL escape codes with their equivalent characters
1484                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1485                                         $this->mOutput->addExternalLink( $pasteurized );
1486                                 }
1487                                 $s .= $text . $trail;
1488                         } else {
1489                                 $s .= $protocol . $remainder;
1490                         }
1491                 }
1492                 wfProfileOut( $fname );
1493                 return $s;
1494         }
1495
1496         /**
1497          * Replace unusual URL escape codes with their equivalent characters
1498          * @param string
1499          * @return string
1500          * @static
1501          * @todo  This can merge genuinely required bits in the path or query string,
1502          *        breaking legit URLs. A proper fix would treat the various parts of
1503          *        the URL differently; as a workaround, just use the output for
1504          *        statistical records, not for actual linking/output.
1505          */
1506         static function replaceUnusualEscapes( $url ) {
1507                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1508                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1509         }
1510
1511         /**
1512          * Callback function used in replaceUnusualEscapes().
1513          * Replaces unusual URL escape codes with their equivalent character
1514          * @static
1515          * @private
1516          */
1517         private static function replaceUnusualEscapesCallback( $matches ) {
1518                 $char = urldecode( $matches[0] );
1519                 $ord = ord( $char );
1520                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1521                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1522                         // No, shouldn't be escaped
1523                         return $char;
1524                 } else {
1525                         // Yes, leave it escaped
1526                         return $matches[0];
1527                 }
1528         }
1529
1530         /**
1531          * make an image if it's allowed, either through the global
1532          * option or through the exception
1533          * @private
1534          */
1535         function maybeMakeExternalImage( $url ) {
1536                 $sk = $this->mOptions->getSkin();
1537                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1538                 $imagesexception = !empty($imagesfrom);
1539                 $text = false;
1540                 if ( $this->mOptions->getAllowExternalImages()
1541                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1542                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1543                                 # Image found
1544                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1545                         }
1546                 }
1547                 return $text;
1548         }
1549
1550         /**
1551          * Process [[ ]] wikilinks
1552          *
1553          * @private
1554          */
1555         function replaceInternalLinks( $s ) {
1556                 global $wgContLang;
1557                 static $fname = 'Parser::replaceInternalLinks' ;
1558
1559                 wfProfileIn( $fname );
1560
1561                 wfProfileIn( $fname.'-setup' );
1562                 static $tc = FALSE;
1563                 # the % is needed to support urlencoded titles as well
1564                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1565
1566                 $sk = $this->mOptions->getSkin();
1567
1568                 #split the entire text string on occurences of [[
1569                 $a = explode( '[[', ' ' . $s );
1570                 #get the first element (all text up to first [[), and remove the space we added
1571                 $s = array_shift( $a );
1572                 $s = substr( $s, 1 );
1573
1574                 # Match a link having the form [[namespace:link|alternate]]trail
1575                 static $e1 = FALSE;
1576                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1577                 # Match cases where there is no "]]", which might still be images
1578                 static $e1_img = FALSE;
1579                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1580                 # Match the end of a line for a word that's not followed by whitespace,
1581                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1582                 $e2 = wfMsgForContent( 'linkprefix' );
1583
1584                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1585                 if( is_null( $this->mTitle ) ) {
1586                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1587                 }
1588                 $nottalk = !$this->mTitle->isTalkPage();
1589
1590                 if ( $useLinkPrefixExtension ) {
1591                         $m = array();
1592                         if ( preg_match( $e2, $s, $m ) ) {
1593                                 $first_prefix = $m[2];
1594                         } else {
1595                                 $first_prefix = false;
1596                         }
1597                 } else {
1598                         $prefix = '';
1599                 }
1600
1601                 if($wgContLang->hasVariants()) {
1602                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1603                 } else {
1604                         $selflink = array($this->mTitle->getPrefixedText());
1605                 }
1606                 $useSubpages = $this->areSubpagesAllowed();
1607                 wfProfileOut( $fname.'-setup' );
1608
1609                 # Loop for each link
1610                 for ($k = 0; isset( $a[$k] ); $k++) {
1611                         $line = $a[$k];
1612                         if ( $useLinkPrefixExtension ) {
1613                                 wfProfileIn( $fname.'-prefixhandling' );
1614                                 if ( preg_match( $e2, $s, $m ) ) {
1615                                         $prefix = $m[2];
1616                                         $s = $m[1];
1617                                 } else {
1618                                         $prefix='';
1619                                 }
1620                                 # first link
1621                                 if($first_prefix) {
1622                                         $prefix = $first_prefix;
1623                                         $first_prefix = false;
1624                                 }
1625                                 wfProfileOut( $fname.'-prefixhandling' );
1626                         }
1627
1628                         $might_be_img = false;
1629
1630                         wfProfileIn( "$fname-e1" );
1631                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1632                                 $text = $m[2];
1633                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1634                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1635                                 # the real problem is with the $e1 regex
1636                                 # See bug 1300.
1637                                 #
1638                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1639                                 # and no image is in sight. See bug 2095.
1640                                 #
1641                                 if( $text !== '' &&
1642                                         substr( $m[3], 0, 1 ) === ']' &&
1643                                         strpos($text, '[') !== false
1644                                 )
1645                                 {
1646                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1647                                         $m[3] = substr( $m[3], 1 );
1648                                 }
1649                                 # fix up urlencoded title texts
1650                                 if( strpos( $m[1], '%' ) !== false ) {
1651                                         # Should anchors '#' also be rejected?
1652                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1653                                 }
1654                                 $trail = $m[3];
1655                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1656                                 $might_be_img = true;
1657                                 $text = $m[2];
1658                                 if ( strpos( $m[1], '%' ) !== false ) {
1659                                         $m[1] = urldecode($m[1]);
1660                                 }
1661                                 $trail = "";
1662                         } else { # Invalid form; output directly
1663                                 $s .= $prefix . '[[' . $line ;
1664                                 wfProfileOut( "$fname-e1" );
1665                                 continue;
1666                         }
1667                         wfProfileOut( "$fname-e1" );
1668                         wfProfileIn( "$fname-misc" );
1669
1670                         # Don't allow internal links to pages containing
1671                         # PROTO: where PROTO is a valid URL protocol; these
1672                         # should be external links.
1673                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1674                                 $s .= $prefix . '[[' . $line ;
1675                                 continue;
1676                         }
1677
1678                         # Make subpage if necessary
1679                         if( $useSubpages ) {
1680                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1681                         } else {
1682                                 $link = $m[1];
1683                         }
1684
1685                         $noforce = (substr($m[1], 0, 1) != ':');
1686                         if (!$noforce) {
1687                                 # Strip off leading ':'
1688                                 $link = substr($link, 1);
1689                         }
1690
1691                         wfProfileOut( "$fname-misc" );
1692                         wfProfileIn( "$fname-title" );
1693                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1694                         if( !$nt ) {
1695                                 $s .= $prefix . '[[' . $line;
1696                                 wfProfileOut( "$fname-title" );
1697                                 continue;
1698                         }
1699
1700                         $ns = $nt->getNamespace();
1701                         $iw = $nt->getInterWiki();
1702                         wfProfileOut( "$fname-title" );
1703
1704                         if ($might_be_img) { # if this is actually an invalid link
1705                                 wfProfileIn( "$fname-might_be_img" );
1706                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1707                                         $found = false;
1708                                         while (isset ($a[$k+1]) ) {
1709                                                 #look at the next 'line' to see if we can close it there
1710                                                 $spliced = array_splice( $a, $k + 1, 1 );
1711                                                 $next_line = array_shift( $spliced );
1712                                                 $m = explode( ']]', $next_line, 3 );
1713                                                 if ( count( $m ) == 3 ) {
1714                                                         # the first ]] closes the inner link, the second the image
1715                                                         $found = true;
1716                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1717                                                         $trail = $m[2];
1718                                                         break;
1719                                                 } elseif ( count( $m ) == 2 ) {
1720                                                         #if there's exactly one ]] that's fine, we'll keep looking
1721                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1722                                                 } else {
1723                                                         #if $next_line is invalid too, we need look no further
1724                                                         $text .= '[[' . $next_line;
1725                                                         break;
1726                                                 }
1727                                         }
1728                                         if ( !$found ) {
1729                                                 # we couldn't find the end of this imageLink, so output it raw
1730                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1731                                                 $text = $this->replaceInternalLinks($text);
1732                                                 $s .= "{$prefix}[[$link|$text";
1733                                                 # note: no $trail, because without an end, there *is* no trail
1734                                                 wfProfileOut( "$fname-might_be_img" );
1735                                                 continue;
1736                                         }
1737                                 } else { #it's not an image, so output it raw
1738                                         $s .= "{$prefix}[[$link|$text";
1739                                         # note: no $trail, because without an end, there *is* no trail
1740                                         wfProfileOut( "$fname-might_be_img" );
1741                                         continue;
1742                                 }
1743                                 wfProfileOut( "$fname-might_be_img" );
1744                         }
1745
1746                         $wasblank = ( '' == $text );
1747                         if( $wasblank ) $text = $link;
1748
1749                         # Link not escaped by : , create the various objects
1750                         if( $noforce ) {
1751
1752                                 # Interwikis
1753                                 wfProfileIn( "$fname-interwiki" );
1754                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1755                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1756                                         $s = rtrim($s . $prefix);
1757                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1758                                         wfProfileOut( "$fname-interwiki" );
1759                                         continue;
1760                                 }
1761                                 wfProfileOut( "$fname-interwiki" );
1762
1763                                 if ( $ns == NS_IMAGE ) {
1764                                         wfProfileIn( "$fname-image" );
1765                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1766                                                 # recursively parse links inside the image caption
1767                                                 # actually, this will parse them in any other parameters, too,
1768                                                 # but it might be hard to fix that, and it doesn't matter ATM
1769                                                 $text = $this->replaceExternalLinks($text);
1770                                                 $text = $this->replaceInternalLinks($text);
1771
1772                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1773                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1774                                                 $this->mOutput->addImage( $nt->getDBkey() );
1775
1776                                                 wfProfileOut( "$fname-image" );
1777                                                 continue;
1778                                         } else {
1779                                                 # We still need to record the image's presence on the page
1780                                                 $this->mOutput->addImage( $nt->getDBkey() );
1781                                         }
1782                                         wfProfileOut( "$fname-image" );
1783
1784                                 }
1785
1786                                 if ( $ns == NS_CATEGORY ) {
1787                                         wfProfileIn( "$fname-category" );
1788                                         $s = rtrim($s . "\n"); # bug 87
1789
1790                                         if ( $wasblank ) {
1791                                                 $sortkey = $this->getDefaultSort();
1792                                         } else {
1793                                                 $sortkey = $text;
1794                                         }
1795                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1796                                         $sortkey = str_replace( "\n", '', $sortkey );
1797                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1798                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1799
1800                                         /**
1801                                          * Strip the whitespace Category links produce, see bug 87
1802                                          * @todo We might want to use trim($tmp, "\n") here.
1803                                          */
1804                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1805
1806                                         wfProfileOut( "$fname-category" );
1807                                         continue;
1808                                 }
1809                         }
1810
1811                         # Self-link checking
1812                         if( $nt->getFragment() === '' ) {
1813                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1814                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1815                                         continue;
1816                                 }
1817                         }
1818
1819                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1820                         if( $ns == NS_MEDIA ) {
1821                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1822                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1823                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1824                                 $this->mOutput->addImage( $nt->getDBkey() );
1825                                 continue;
1826                         } elseif( $ns == NS_SPECIAL ) {
1827                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1828                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1829                                 } else {
1830                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1831                                 }
1832                                 continue;
1833                         } elseif( $ns == NS_IMAGE ) {
1834                                 $img = wfFindFile( $nt );
1835                                 if( $img ) {
1836                                         // Force a blue link if the file exists; may be a remote
1837                                         // upload on the shared repository, and we want to see its
1838                                         // auto-generated page.
1839                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1840                                         $this->mOutput->addLink( $nt );
1841                                         continue;
1842                                 }
1843                         }
1844                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1845                 }
1846                 wfProfileOut( $fname );
1847                 return $s;
1848         }
1849
1850         /**
1851          * Make a link placeholder. The text returned can be later resolved to a real link with
1852          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1853          * parsing of interwiki links, and secondly to allow all existence checks and
1854          * article length checks (for stub links) to be bundled into a single query.
1855          *
1856          */
1857         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1858                 wfProfileIn( __METHOD__ );
1859                 if ( ! is_object($nt) ) {
1860                         # Fail gracefully
1861                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1862                 } else {
1863                         # Separate the link trail from the rest of the link
1864                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1865
1866                         if ( $nt->isExternal() ) {
1867                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1868                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1869                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1870                         } else {
1871                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1872                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1873                                 $this->mLinkHolders['queries'][] = $query;
1874                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1875                                 $this->mLinkHolders['titles'][] = $nt;
1876
1877                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1878                         }
1879                 }
1880                 wfProfileOut( __METHOD__ );
1881                 return $retVal;
1882         }
1883
1884         /**
1885          * Render a forced-blue link inline; protect against double expansion of
1886          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1887          * Since this little disaster has to split off the trail text to avoid
1888          * breaking URLs in the following text without breaking trails on the
1889          * wiki links, it's been made into a horrible function.
1890          *
1891          * @param Title $nt
1892          * @param string $text
1893          * @param string $query
1894          * @param string $trail
1895          * @param string $prefix
1896          * @return string HTML-wikitext mix oh yuck
1897          */
1898         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1899                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1900                 $sk = $this->mOptions->getSkin();
1901                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1902                 return $this->armorLinks( $link ) . $trail;
1903         }
1904
1905         /**
1906          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1907          * going to go through further parsing steps before inline URL expansion.
1908          *
1909          * In particular this is important when using action=render, which causes
1910          * full URLs to be included.
1911          *
1912          * Oh man I hate our multi-layer parser!
1913          *
1914          * @param string more-or-less HTML
1915          * @return string less-or-more HTML with NOPARSE bits
1916          */
1917         function armorLinks( $text ) {
1918                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1919                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1920         }
1921
1922         /**
1923          * Return true if subpage links should be expanded on this page.
1924          * @return bool
1925          */
1926         function areSubpagesAllowed() {
1927                 # Some namespaces don't allow subpages
1928                 global $wgNamespacesWithSubpages;
1929                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1930         }
1931
1932         /**
1933          * Handle link to subpage if necessary
1934          * @param string $target the source of the link
1935          * @param string &$text the link text, modified as necessary
1936          * @return string the full name of the link
1937          * @private
1938          */
1939         function maybeDoSubpageLink($target, &$text) {
1940                 # Valid link forms:
1941                 # Foobar -- normal
1942                 # :Foobar -- override special treatment of prefix (images, language links)
1943                 # /Foobar -- convert to CurrentPage/Foobar
1944                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1945                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1946                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1947
1948                 $fname = 'Parser::maybeDoSubpageLink';
1949                 wfProfileIn( $fname );
1950                 $ret = $target; # default return value is no change
1951
1952                 # Some namespaces don't allow subpages,
1953                 # so only perform processing if subpages are allowed
1954                 if( $this->areSubpagesAllowed() ) {
1955                         $hash = strpos( $target, '#' );
1956                         if( $hash !== false ) {
1957                                 $suffix = substr( $target, $hash );
1958                                 $target = substr( $target, 0, $hash );
1959                         } else {
1960                                 $suffix = '';
1961                         }
1962                         # bug 7425
1963                         $target = trim( $target );
1964                         # Look at the first character
1965                         if( $target != '' && $target{0} == '/' ) {
1966                                 # / at end means we don't want the slash to be shown
1967                                 $m = array();
1968                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1969                                 if( $trailingSlashes ) {
1970                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1971                                 } else {
1972                                         $noslash = substr( $target, 1 );
1973                                 }
1974
1975                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1976                                 if( '' === $text ) {
1977                                         $text = $target . $suffix;
1978                                 } # this might be changed for ugliness reasons
1979                         } else {
1980                                 # check for .. subpage backlinks
1981                                 $dotdotcount = 0;
1982                                 $nodotdot = $target;
1983                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1984                                         ++$dotdotcount;
1985                                         $nodotdot = substr( $nodotdot, 3 );
1986                                 }
1987                                 if($dotdotcount > 0) {
1988                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1989                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1990                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1991                                                 # / at the end means don't show full path
1992                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1993                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1994                                                         if( '' === $text ) {
1995                                                                 $text = $nodotdot . $suffix;
1996                                                         }
1997                                                 }
1998                                                 $nodotdot = trim( $nodotdot );
1999                                                 if( $nodotdot != '' ) {
2000                                                         $ret .= '/' . $nodotdot;
2001                                                 }
2002                                                 $ret .= $suffix;
2003                                         }
2004                                 }
2005                         }
2006                 }
2007
2008                 wfProfileOut( $fname );
2009                 return $ret;
2010         }
2011
2012         /**#@+
2013          * Used by doBlockLevels()
2014          * @private
2015          */
2016         /* private */ function closeParagraph() {
2017                 $result = '';
2018                 if ( '' != $this->mLastSection ) {
2019                         $result = '</' . $this->mLastSection  . ">\n";
2020                 }
2021                 $this->mInPre = false;
2022                 $this->mLastSection = '';
2023                 return $result;
2024         }
2025         # getCommon() returns the length of the longest common substring
2026         # of both arguments, starting at the beginning of both.
2027         #
2028         /* private */ function getCommon( $st1, $st2 ) {
2029                 $fl = strlen( $st1 );
2030                 $shorter = strlen( $st2 );
2031                 if ( $fl < $shorter ) { $shorter = $fl; }
2032
2033                 for ( $i = 0; $i < $shorter; ++$i ) {
2034                         if ( $st1{$i} != $st2{$i} ) { break; }
2035                 }
2036                 return $i;
2037         }
2038         # These next three functions open, continue, and close the list
2039         # element appropriate to the prefix character passed into them.
2040         #
2041         /* private */ function openList( $char ) {
2042                 $result = $this->closeParagraph();
2043
2044                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2045                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2046                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2047                 else if ( ';' == $char ) {
2048                         $result .= '<dl><dt>';
2049                         $this->mDTopen = true;
2050                 }
2051                 else { $result = '<!-- ERR 1 -->'; }
2052
2053                 return $result;
2054         }
2055
2056         /* private */ function nextItem( $char ) {
2057                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2058                 else if ( ':' == $char || ';' == $char ) {
2059                         $close = '</dd>';
2060                         if ( $this->mDTopen ) { $close = '</dt>'; }
2061                         if ( ';' == $char ) {
2062                                 $this->mDTopen = true;
2063                                 return $close . '<dt>';
2064                         } else {
2065                                 $this->mDTopen = false;
2066                                 return $close . '<dd>';
2067                         }
2068                 }
2069                 return '<!-- ERR 2 -->';
2070         }
2071
2072         /* private */ function closeList( $char ) {
2073                 if ( '*' == $char ) { $text = '</li></ul>'; }
2074                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2075                 else if ( ':' == $char ) {
2076                         if ( $this->mDTopen ) {
2077                                 $this->mDTopen = false;
2078                                 $text = '</dt></dl>';
2079                         } else {
2080                                 $text = '</dd></dl>';
2081                         }
2082                 }
2083                 else {  return '<!-- ERR 3 -->'; }
2084                 return $text."\n";
2085         }
2086         /**#@-*/
2087
2088         /**
2089          * Make lists from lines starting with ':', '*', '#', etc.
2090          *
2091          * @private
2092          * @return string the lists rendered as HTML
2093          */
2094         function doBlockLevels( $text, $linestart ) {
2095                 $fname = 'Parser::doBlockLevels';
2096                 wfProfileIn( $fname );
2097
2098                 # Parsing through the text line by line.  The main thing
2099                 # happening here is handling of block-level elements p, pre,
2100                 # and making lists from lines starting with * # : etc.
2101                 #
2102                 $textLines = explode( "\n", $text );
2103
2104                 $lastPrefix = $output = '';
2105                 $this->mDTopen = $inBlockElem = false;
2106                 $prefixLength = 0;
2107                 $paragraphStack = false;
2108
2109                 if ( !$linestart ) {
2110                         $output .= array_shift( $textLines );
2111                 }
2112                 foreach ( $textLines as $oLine ) {
2113                         $lastPrefixLength = strlen( $lastPrefix );
2114                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2115                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2116                         if ( !$this->mInPre ) {
2117                                 # Multiple prefixes may abut each other for nested lists.
2118                                 $prefixLength = strspn( $oLine, '*#:;' );
2119                                 $pref = substr( $oLine, 0, $prefixLength );
2120
2121                                 # eh?
2122                                 $pref2 = str_replace( ';', ':', $pref );
2123                                 $t = substr( $oLine, $prefixLength );
2124                                 $this->mInPre = !empty($preOpenMatch);
2125                         } else {
2126                                 # Don't interpret any other prefixes in preformatted text
2127                                 $prefixLength = 0;
2128                                 $pref = $pref2 = '';
2129                                 $t = $oLine;
2130                         }
2131
2132                         # List generation
2133                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2134                                 # Same as the last item, so no need to deal with nesting or opening stuff
2135                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2136                                 $paragraphStack = false;
2137
2138                                 if ( substr( $pref, -1 ) == ';') {
2139                                         # The one nasty exception: definition lists work like this:
2140                                         # ; title : definition text
2141                                         # So we check for : in the remainder text to split up the
2142                                         # title and definition, without b0rking links.
2143                                         $term = $t2 = '';
2144                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2145                                                 $t = $t2;
2146                                                 $output .= $term . $this->nextItem( ':' );
2147                                         }
2148                                 }
2149                         } elseif( $prefixLength || $lastPrefixLength ) {
2150                                 # Either open or close a level...
2151                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2152                                 $paragraphStack = false;
2153
2154                                 while( $commonPrefixLength < $lastPrefixLength ) {
2155                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2156                                         --$lastPrefixLength;
2157                                 }
2158                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2159                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2160                                 }
2161                                 while ( $prefixLength > $commonPrefixLength ) {
2162                                         $char = substr( $pref, $commonPrefixLength, 1 );
2163                                         $output .= $this->openList( $char );
2164
2165                                         if ( ';' == $char ) {
2166                                                 # FIXME: This is dupe of code above
2167                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2168                                                         $t = $t2;
2169                                                         $output .= $term . $this->nextItem( ':' );
2170                                                 }
2171                                         }
2172                                         ++$commonPrefixLength;
2173                                 }
2174                                 $lastPrefix = $pref2;
2175                         }
2176                         if( 0 == $prefixLength ) {
2177                                 wfProfileIn( "$fname-paragraph" );
2178                                 # No prefix (not in list)--go to paragraph mode
2179                                 // XXX: use a stack for nestable elements like span, table and div
2180                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2181                                 $closematch = preg_match(
2182                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2183                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2184                                 if ( $openmatch or $closematch ) {
2185                                         $paragraphStack = false;
2186                                         # TODO bug 5718: paragraph closed
2187                                         $output .= $this->closeParagraph();
2188                                         if ( $preOpenMatch and !$preCloseMatch ) {
2189                                                 $this->mInPre = true;
2190                                         }
2191                                         if ( $closematch ) {
2192                                                 $inBlockElem = false;
2193                                         } else {
2194                                                 $inBlockElem = true;
2195                                         }
2196                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2197                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2198                                                 // pre
2199                                                 if ($this->mLastSection != 'pre') {
2200                                                         $paragraphStack = false;
2201                                                         $output .= $this->closeParagraph().'<pre>';
2202                                                         $this->mLastSection = 'pre';
2203                                                 }
2204                                                 $t = substr( $t, 1 );
2205                                         } else {
2206                                                 // paragraph
2207                                                 if ( '' == trim($t) ) {
2208                                                         if ( $paragraphStack ) {
2209                                                                 $output .= $paragraphStack.'<br />';
2210                                                                 $paragraphStack = false;
2211                                                                 $this->mLastSection = 'p';
2212                                                         } else {
2213                                                                 if ($this->mLastSection != 'p' ) {
2214                                                                         $output .= $this->closeParagraph();
2215                                                                         $this->mLastSection = '';
2216                                                                         $paragraphStack = '<p>';
2217                                                                 } else {
2218                                                                         $paragraphStack = '</p><p>';
2219                                                                 }
2220                                                         }
2221                                                 } else {
2222                                                         if ( $paragraphStack ) {
2223                                                                 $output .= $paragraphStack;
2224                                                                 $paragraphStack = false;
2225                                                                 $this->mLastSection = 'p';
2226                                                         } else if ($this->mLastSection != 'p') {
2227                                                                 $output .= $this->closeParagraph().'<p>';
2228                                                                 $this->mLastSection = 'p';
2229                                                         }
2230                                                 }
2231                                         }
2232                                 }
2233                                 wfProfileOut( "$fname-paragraph" );
2234                         }
2235                         // somewhere above we forget to get out of pre block (bug 785)
2236                         if($preCloseMatch && $this->mInPre) {
2237                                 $this->mInPre = false;
2238                         }
2239                         if ($paragraphStack === false) {
2240                                 $output .= $t."\n";
2241                         }
2242                 }
2243                 while ( $prefixLength ) {
2244                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2245                         --$prefixLength;
2246                 }
2247                 if ( '' != $this->mLastSection ) {
2248                         $output .= '</' . $this->mLastSection . '>';
2249                         $this->mLastSection = '';
2250                 }
2251
2252                 wfProfileOut( $fname );
2253                 return $output;
2254         }
2255
2256         /**
2257          * Split up a string on ':', ignoring any occurences inside tags
2258          * to prevent illegal overlapping.
2259          * @param string $str the string to split
2260          * @param string &$before set to everything before the ':'
2261          * @param string &$after set to everything after the ':'
2262          * return string the position of the ':', or false if none found
2263          */
2264         function findColonNoLinks($str, &$before, &$after) {
2265                 $fname = 'Parser::findColonNoLinks';
2266                 wfProfileIn( $fname );
2267
2268                 $pos = strpos( $str, ':' );
2269                 if( $pos === false ) {
2270                         // Nothing to find!
2271                         wfProfileOut( $fname );
2272                         return false;
2273                 }
2274
2275                 $lt = strpos( $str, '<' );
2276                 if( $lt === false || $lt > $pos ) {
2277                         // Easy; no tag nesting to worry about
2278                         $before = substr( $str, 0, $pos );
2279                         $after = substr( $str, $pos+1 );
2280                         wfProfileOut( $fname );
2281                         return $pos;
2282                 }
2283
2284                 // Ugly state machine to walk through avoiding tags.
2285                 $state = MW_COLON_STATE_TEXT;
2286                 $stack = 0;
2287                 $len = strlen( $str );
2288                 for( $i = 0; $i < $len; $i++ ) {
2289                         $c = $str{$i};
2290
2291                         switch( $state ) {
2292                         // (Using the number is a performance hack for common cases)
2293                         case 0: // MW_COLON_STATE_TEXT:
2294                                 switch( $c ) {
2295                                 case "<":
2296                                         // Could be either a <start> tag or an </end> tag
2297                                         $state = MW_COLON_STATE_TAGSTART;
2298                                         break;
2299                                 case ":":
2300                                         if( $stack == 0 ) {
2301                                                 // We found it!
2302                                                 $before = substr( $str, 0, $i );
2303                                                 $after = substr( $str, $i + 1 );
2304                                                 wfProfileOut( $fname );
2305                                                 return $i;
2306                                         }
2307                                         // Embedded in a tag; don't break it.
2308                                         break;
2309                                 default:
2310                                         // Skip ahead looking for something interesting
2311                                         $colon = strpos( $str, ':', $i );
2312                                         if( $colon === false ) {
2313                                                 // Nothing else interesting
2314                                                 wfProfileOut( $fname );
2315                                                 return false;
2316                                         }
2317                                         $lt = strpos( $str, '<', $i );
2318                                         if( $stack === 0 ) {
2319                                                 if( $lt === false || $colon < $lt ) {
2320                                                         // We found it!
2321                                                         $before = substr( $str, 0, $colon );
2322                                                         $after = substr( $str, $colon + 1 );
2323                                                         wfProfileOut( $fname );
2324                                                         return $i;
2325                                                 }
2326                                         }
2327                                         if( $lt === false ) {
2328                                                 // Nothing else interesting to find; abort!
2329                                                 // We're nested, but there's no close tags left. Abort!
2330                                                 break 2;
2331                                         }
2332                                         // Skip ahead to next tag start
2333                                         $i = $lt;
2334                                         $state = MW_COLON_STATE_TAGSTART;
2335                                 }
2336                                 break;
2337                         case 1: // MW_COLON_STATE_TAG:
2338                                 // In a <tag>
2339                                 switch( $c ) {
2340                                 case ">":
2341                                         $stack++;
2342                                         $state = MW_COLON_STATE_TEXT;
2343                                         break;
2344                                 case "/":
2345                                         // Slash may be followed by >?
2346                                         $state = MW_COLON_STATE_TAGSLASH;
2347                                         break;
2348                                 default:
2349                                         // ignore
2350                                 }
2351                                 break;
2352                         case 2: // MW_COLON_STATE_TAGSTART:
2353                                 switch( $c ) {
2354                                 case "/":
2355                                         $state = MW_COLON_STATE_CLOSETAG;
2356                                         break;
2357                                 case "!":
2358                                         $state = MW_COLON_STATE_COMMENT;
2359                                         break;
2360                                 case ">":
2361                                         // Illegal early close? This shouldn't happen D:
2362                                         $state = MW_COLON_STATE_TEXT;
2363                                         break;
2364                                 default:
2365                                         $state = MW_COLON_STATE_TAG;
2366                                 }
2367                                 break;
2368                         case 3: // MW_COLON_STATE_CLOSETAG:
2369                                 // In a </tag>
2370                                 if( $c == ">" ) {
2371                                         $stack--;
2372                                         if( $stack < 0 ) {
2373                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2374                                                 wfProfileOut( $fname );
2375                                                 return false;
2376                                         }
2377                                         $state = MW_COLON_STATE_TEXT;
2378                                 }
2379                                 break;
2380                         case MW_COLON_STATE_TAGSLASH:
2381                                 if( $c == ">" ) {
2382                                         // Yes, a self-closed tag <blah/>
2383                                         $state = MW_COLON_STATE_TEXT;
2384                                 } else {
2385                                         // Probably we're jumping the gun, and this is an attribute
2386                                         $state = MW_COLON_STATE_TAG;
2387                                 }
2388                                 break;
2389                         case 5: // MW_COLON_STATE_COMMENT:
2390                                 if( $c == "-" ) {
2391                                         $state = MW_COLON_STATE_COMMENTDASH;
2392                                 }
2393                                 break;
2394                         case MW_COLON_STATE_COMMENTDASH:
2395                                 if( $c == "-" ) {
2396                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2397                                 } else {
2398                                         $state = MW_COLON_STATE_COMMENT;
2399                                 }
2400                                 break;
2401                         case MW_COLON_STATE_COMMENTDASHDASH:
2402                                 if( $c == ">" ) {
2403                                         $state = MW_COLON_STATE_TEXT;
2404                                 } else {
2405                                         $state = MW_COLON_STATE_COMMENT;
2406                                 }
2407                                 break;
2408                         default:
2409                                 throw new MWException( "State machine error in $fname" );
2410                         }
2411                 }
2412                 if( $stack > 0 ) {
2413                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2414                         return false;
2415                 }
2416                 wfProfileOut( $fname );
2417                 return false;
2418         }
2419
2420         /**
2421          * Return value of a magic variable (like PAGENAME)
2422          *
2423          * @private
2424          */
2425         function getVariableValue( $index ) {
2426                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2427
2428                 /**
2429                  * Some of these require message or data lookups and can be
2430                  * expensive to check many times.
2431                  */
2432                 static $varCache = array();
2433                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2434                         if ( isset( $varCache[$index] ) ) {
2435                                 return $varCache[$index];
2436                         }
2437                 }
2438
2439                 $ts = time();
2440                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2441
2442                 # Use the time zone
2443                 global $wgLocaltimezone;
2444                 if ( isset( $wgLocaltimezone ) ) {
2445                         $oldtz = getenv( 'TZ' );
2446                         putenv( 'TZ='.$wgLocaltimezone );
2447                 }
2448
2449                 wfSuppressWarnings(); // E_STRICT system time bitching
2450                 $localTimestamp = date( 'YmdHis', $ts );
2451                 $localMonth = date( 'm', $ts );
2452                 $localMonthName = date( 'n', $ts );
2453                 $localDay = date( 'j', $ts );
2454                 $localDay2 = date( 'd', $ts );
2455                 $localDayOfWeek = date( 'w', $ts );
2456                 $localWeek = date( 'W', $ts );
2457                 $localYear = date( 'Y', $ts );
2458                 $localHour = date( 'H', $ts );
2459                 if ( isset( $wgLocaltimezone ) ) {
2460                         putenv( 'TZ='.$oldtz );
2461                 }
2462                 wfRestoreWarnings();
2463
2464                 switch ( $index ) {
2465                         case 'currentmonth':
2466                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2467                         case 'currentmonthname':
2468                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2469                         case 'currentmonthnamegen':
2470                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2471                         case 'currentmonthabbrev':
2472                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2473                         case 'currentday':
2474                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2475                         case 'currentday2':
2476                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2477                         case 'localmonth':
2478                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2479                         case 'localmonthname':
2480                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2481                         case 'localmonthnamegen':
2482                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2483                         case 'localmonthabbrev':
2484                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2485                         case 'localday':
2486                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2487                         case 'localday2':
2488                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2489                         case 'pagename':
2490                                 return wfEscapeWikiText( $this->mTitle->getText() );
2491                         case 'pagenamee':
2492                                 return $this->mTitle->getPartialURL();
2493                         case 'fullpagename':
2494                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2495                         case 'fullpagenamee':
2496                                 return $this->mTitle->getPrefixedURL();
2497                         case 'subpagename':
2498                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2499                         case 'subpagenamee':
2500                                 return $this->mTitle->getSubpageUrlForm();
2501                         case 'basepagename':
2502                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2503                         case 'basepagenamee':
2504                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2505                         case 'talkpagename':
2506                                 if( $this->mTitle->canTalk() ) {
2507                                         $talkPage = $this->mTitle->getTalkPage();
2508                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2509                                 } else {
2510                                         return '';
2511                                 }
2512                         case 'talkpagenamee':
2513                                 if( $this->mTitle->canTalk() ) {
2514                                         $talkPage = $this->mTitle->getTalkPage();
2515                                         return $talkPage->getPrefixedUrl();
2516                                 } else {
2517                                         return '';
2518                                 }
2519                         case 'subjectpagename':
2520                                 $subjPage = $this->mTitle->getSubjectPage();
2521                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2522                         case 'subjectpagenamee':
2523                                 $subjPage = $this->mTitle->getSubjectPage();
2524                                 return $subjPage->getPrefixedUrl();
2525                         case 'revisionid':
2526                                 return $this->mRevisionId;
2527                         case 'revisionday':
2528                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2529                         case 'revisionday2':
2530                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2531                         case 'revisionmonth':
2532                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2533                         case 'revisionyear':
2534                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2535                         case 'revisiontimestamp':
2536                                 return $this->getRevisionTimestamp();
2537                         case 'namespace':
2538                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2539                         case 'namespacee':
2540                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2541                         case 'talkspace':
2542                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2543                         case 'talkspacee':
2544                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2545                         case 'subjectspace':
2546                                 return $this->mTitle->getSubjectNsText();
2547                         case 'subjectspacee':
2548                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2549                         case 'currentdayname':
2550                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2551                         case 'currentyear':
2552                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2553                         case 'currenttime':
2554                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2555                         case 'currenthour':
2556                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2557                         case 'currentweek':
2558                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2559                                 // int to remove the padding
2560                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2561                         case 'currentdow':
2562                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2563                         case 'localdayname':
2564                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2565                         case 'localyear':
2566                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2567                         case 'localtime':
2568                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2569                         case 'localhour':
2570                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2571                         case 'localweek':
2572                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2573                                 // int to remove the padding
2574                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2575                         case 'localdow':
2576                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2577                         case 'numberofarticles':
2578                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2579                         case 'numberoffiles':
2580                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2581                         case 'numberofusers':
2582                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2583                         case 'numberofpages':
2584                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2585                         case 'numberofadmins':
2586                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2587                         case 'numberofedits':
2588                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2589                         case 'currenttimestamp':
2590                                 return $varCache[$index] = wfTimestampNow();
2591                         case 'localtimestamp':
2592                                 return $varCache[$index] = $localTimestamp;
2593                         case 'currentversion':
2594                                 return $varCache[$index] = SpecialVersion::getVersion();
2595                         case 'sitename':
2596                                 return $wgSitename;
2597                         case 'server':
2598                                 return $wgServer;
2599                         case 'servername':
2600                                 return $wgServerName;
2601                         case 'scriptpath':
2602                                 return $wgScriptPath;
2603                         case 'directionmark':
2604                                 return $wgContLang->getDirMark();
2605                         case 'contentlanguage':
2606                                 global $wgContLanguageCode;
2607                                 return $wgContLanguageCode;
2608                         default:
2609                                 $ret = null;
2610                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2611                                         return $ret;
2612                                 else
2613                                         return null;
2614                 }
2615         }
2616
2617         /**
2618          * initialise the magic variables (like CURRENTMONTHNAME)
2619          *
2620          * @private
2621          */
2622         function initialiseVariables() {
2623                 $fname = 'Parser::initialiseVariables';
2624                 wfProfileIn( $fname );
2625                 $variableIDs = MagicWord::getVariableIDs();
2626
2627                 $this->mVariables = array();
2628                 foreach ( $variableIDs as $id ) {
2629                         $mw =& MagicWord::get( $id );
2630                         $mw->addToArray( $this->mVariables, $id );
2631                 }
2632                 wfProfileOut( $fname );
2633         }
2634
2635         /**
2636          * parse any parentheses in format ((title|part|part))
2637          * and call callbacks to get a replacement text for any found piece
2638          *
2639          * @param string $text The text to parse
2640          * @param array $callbacks rules in form:
2641          *     '{' => array(                            # opening parentheses
2642          *                                      'end' => '}',   # closing parentheses
2643          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2644          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2645          *                                                                )
2646          *                                      )
2647          *                                      'min' => 2,     # Minimum parenthesis count in cb
2648          *                                      'max' => 3,     # Maximum parenthesis count in cb
2649          * @private
2650          */
2651         function replace_callback ($text, $callbacks) {
2652                 wfProfileIn( __METHOD__ );
2653                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2654                 $lastOpeningBrace = -1;                 # last not closed parentheses
2655
2656                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2657
2658                 $i = 0;
2659                 while ( $i < strlen( $text ) ) {
2660                         # Find next opening brace, closing brace or pipe
2661                         if ( $lastOpeningBrace == -1 ) {
2662                                 $currentClosing = '';
2663                                 $search = $validOpeningBraces;
2664                         } else {
2665                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2666                                 $search = $validOpeningBraces . '|' . $currentClosing;
2667                         }
2668                         $rule = null;
2669                         $i += strcspn( $text, $search, $i );
2670                         if ( $i < strlen( $text ) ) {
2671                                 if ( $text[$i] == '|' ) {
2672                                         $found = 'pipe';
2673                                 } elseif ( $text[$i] == $currentClosing ) {
2674                                         $found = 'close';
2675                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2676                                         $found = 'open';
2677                                         $rule = $callbacks[$text[$i]];
2678                                 } else {
2679                                         # Some versions of PHP have a strcspn which stops on null characters
2680                                         # Ignore and continue
2681                                         ++$i;
2682                                         continue;
2683                                 }
2684                         } else {
2685                                 # All done
2686                                 break;
2687                         }
2688
2689                         if ( $found == 'open' ) {
2690                                 # found opening brace, let's add it to parentheses stack
2691                                 $piece = array('brace' => $text[$i],
2692                                                            'braceEnd' => $rule['end'],
2693                                                            'title' => '',
2694                                                            'parts' => null);
2695
2696                                 # count opening brace characters
2697                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2698                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2699                                 $i += $piece['count'];
2700
2701                                 # we need to add to stack only if opening brace count is enough for one of the rules
2702                                 if ( $piece['count'] >= $rule['min'] ) {
2703                                         $lastOpeningBrace ++;
2704                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2705                                 }
2706                         } elseif ( $found == 'close' ) {
2707                                 # lets check if it is enough characters for closing brace
2708                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2709                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2710
2711                                 # check for maximum matching characters (if there are 5 closing
2712                                 # characters, we will probably need only 3 - depending on the rules)
2713                                 $matchingCount = 0;
2714                                 $matchingCallback = null;
2715                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2716                                 if ( $count > $cbType['max'] ) {
2717                                         # The specified maximum exists in the callback array, unless the caller
2718                                         # has made an error
2719                                         $matchingCount = $cbType['max'];
2720                                 } else {
2721                                         # Count is less than the maximum
2722                                         # Skip any gaps in the callback array to find the true largest match
2723                                         # Need to use array_key_exists not isset because the callback can be null
2724                                         $matchingCount = $count;
2725                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2726                                                 --$matchingCount;
2727                                         }
2728                                 }
2729
2730                                 if ($matchingCount <= 0) {
2731                                         $i += $count;
2732                                         continue;
2733                                 }
2734                                 $matchingCallback = $cbType['cb'][$matchingCount];
2735
2736                                 # let's set a title or last part (if '|' was found)
2737                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2738                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2739                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2740                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2741                                 } else {
2742                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2743                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2744                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2745                                 }
2746
2747                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2748                                 $pieceEnd = $i + $matchingCount;
2749
2750                                 if( is_callable( $matchingCallback ) ) {
2751                                         $cbArgs = array (
2752                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2753                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2754                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2755                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2756                                                                          );
2757                                         # finally we can call a user callback and replace piece of text
2758                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2759                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2760                                         $i = $pieceStart + strlen($replaceWith);
2761                                 } else {
2762                                         # null value for callback means that parentheses should be parsed, but not replaced
2763                                         $i += $matchingCount;
2764                                 }
2765
2766                                 # reset last opening parentheses, but keep it in case there are unused characters
2767                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2768                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2769                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2770                                                            'title' => '',
2771                                                            'parts' => null,
2772                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2773                                 $openingBraceStack[$lastOpeningBrace--] = null;
2774
2775                                 if ($matchingCount < $piece['count']) {
2776                                         $piece['count'] -= $matchingCount;
2777                                         $piece['startAt'] -= $matchingCount;
2778                                         $piece['partStart'] = $piece['startAt'];
2779                                         # do we still qualify for any callback with remaining count?
2780                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2781                                         while ( $piece['count'] ) {
2782                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2783                                                         $lastOpeningBrace++;
2784                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2785                                                         break;
2786                                                 }
2787                                                 --$piece['count'];
2788                                         }
2789                                 }
2790                         } elseif ( $found == 'pipe' ) {
2791                                 # lets set a title if it is a first separator, or next part otherwise
2792                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2793                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2794                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2795                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2796                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2797                                 } else {
2798                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2799                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2800                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2801                                 }
2802                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2803                         }
2804                 }
2805
2806                 wfProfileOut( __METHOD__ );
2807                 return $text;
2808         }
2809
2810         /**
2811          * Replace magic variables, templates, and template arguments
2812          * with the appropriate text. Templates are substituted recursively,
2813          * taking care to avoid infinite loops.
2814          *
2815          * Note that the substitution depends on value of $mOutputType:
2816          *  OT_WIKI: only {{subst:}} templates
2817          *  OT_MSG: only magic variables
2818          *  OT_HTML: all templates and magic variables
2819          *
2820          * @param string $tex The text to transform
2821          * @param array $args Key-value pairs representing template parameters to substitute
2822          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2823          * @private
2824          */
2825         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2826                 # Prevent too big inclusions
2827                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2828                         return $text;
2829                 }
2830
2831                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2832                 wfProfileIn( $fname );
2833
2834                 # This function is called recursively. To keep track of arguments we need a stack:
2835                 array_push( $this->mArgStack, $args );
2836
2837                 $braceCallbacks = array();
2838                 if ( !$argsOnly ) {
2839                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2840                 }
2841                 if ( $this->mOutputType != OT_MSG ) {
2842                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2843                 }
2844                 if ( $braceCallbacks ) {
2845                         $callbacks = array(
2846                                 '{' => array(
2847                                         'end' => '}',
2848                                         'cb' => $braceCallbacks,
2849                                         'min' => $argsOnly ? 3 : 2,
2850                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2851                                 ),
2852                                 '[' => array(
2853                                         'end' => ']',
2854                                         'cb' => array(2=>null),
2855                                         'min' => 2,
2856                                         'max' => 2,
2857                                 )
2858                         );
2859                         $text = $this->replace_callback ($text, $callbacks);
2860
2861                         array_pop( $this->mArgStack );
2862                 }
2863                 wfProfileOut( $fname );
2864                 return $text;
2865         }
2866
2867         /**
2868          * Replace magic variables
2869          * @private
2870          */
2871         function variableSubstitution( $matches ) {
2872                 global $wgContLang;
2873                 $fname = 'Parser::variableSubstitution';
2874                 $varname = $wgContLang->lc($matches[1]);
2875                 wfProfileIn( $fname );
2876                 $skip = false;
2877                 if ( $this->mOutputType == OT_WIKI ) {
2878                         # Do only magic variables prefixed by SUBST
2879                         $mwSubst =& MagicWord::get( 'subst' );
2880                         if (!$mwSubst->matchStartAndRemove( $varname ))
2881                                 $skip = true;
2882                         # Note that if we don't substitute the variable below,
2883                         # we don't remove the {{subst:}} magic word, in case
2884                         # it is a template rather than a magic variable.
2885                 }
2886                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2887                         $id = $this->mVariables[$varname];
2888                         # Now check if we did really match, case sensitive or not
2889                         $mw =& MagicWord::get( $id );
2890                         if ($mw->match($matches[1])) {
2891                                 $text = $this->getVariableValue( $id );
2892                                 $this->mOutput->mContainsOldMagic = true;
2893                         } else {
2894                                 $text = $matches[0];
2895                         }
2896                 } else {
2897                         $text = $matches[0];
2898                 }
2899                 wfProfileOut( $fname );
2900                 return $text;
2901         }
2902
2903
2904         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2905         static function createAssocArgs( $args ) {
2906                 $assocArgs = array();
2907                 $index = 1;
2908                 foreach( $args as $arg ) {
2909                         $eqpos = strpos( $arg, '=' );
2910                         if ( $eqpos === false ) {
2911                                 $assocArgs[$index++] = $arg;
2912                         } else {
2913                                 $name = trim( substr( $arg, 0, $eqpos ) );
2914                                 $value = trim( substr( $arg, $eqpos+1 ) );
2915                                 if ( $value === false ) {
2916                                         $value = '';
2917                                 }
2918                                 if ( $name !== false ) {
2919                                         $assocArgs[$name] = $value;
2920                                 }
2921                         }
2922                 }
2923
2924                 return $assocArgs;
2925         }
2926
2927         /**
2928          * Return the text of a template, after recursively
2929          * replacing any variables or templates within the template.
2930          *
2931          * @param array $piece The parts of the template
2932          *  $piece['text']: matched text
2933          *  $piece['title']: the title, i.e. the part before the |
2934          *  $piece['parts']: the parameter array
2935          * @return string the text of the template
2936          * @private
2937          */
2938         function braceSubstitution( $piece ) {
2939                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2940                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2941                 wfProfileIn( $fname );
2942                 wfProfileIn( __METHOD__.'-setup' );
2943
2944                 # Flags
2945                 $found = false;             # $text has been filled
2946                 $nowiki = false;            # wiki markup in $text should be escaped
2947                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2948                 $noargs = false;            # Don't replace triple-brace arguments in $text
2949                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2950                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2951                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2952                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2953
2954                 # Title object, where $text came from
2955                 $title = NULL;
2956
2957                 $linestart = '';
2958
2959
2960                 # $part1 is the bit before the first |, and must contain only title characters
2961                 # $args is a list of arguments, starting from index 0, not including $part1
2962
2963                 $titleText = $part1 = $piece['title'];
2964                 # If the third subpattern matched anything, it will start with |
2965
2966                 if (null == $piece['parts']) {
2967                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2968                         if ($replaceWith != $piece['text']) {
2969                                 $text = $replaceWith;
2970                                 $found = true;
2971                                 $noparse = true;
2972                                 $noargs = true;
2973                         }
2974                 }
2975
2976                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2977                 wfProfileOut( __METHOD__.'-setup' );
2978
2979                 # SUBST
2980                 wfProfileIn( __METHOD__.'-modifiers' );
2981                 if ( !$found ) {
2982                         $mwSubst =& MagicWord::get( 'subst' );
2983                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2984                                 # One of two possibilities is true:
2985                                 # 1) Found SUBST but not in the PST phase
2986                                 # 2) Didn't find SUBST and in the PST phase
2987                                 # In either case, return without further processing
2988                                 $text = $piece['text'];
2989                                 $found = true;
2990                                 $noparse = true;
2991                                 $noargs = true;
2992                         }
2993                 }
2994
2995                 # MSG, MSGNW and RAW
2996                 if ( !$found ) {
2997                         # Check for MSGNW:
2998                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2999                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3000                                 $nowiki = true;
3001                         } else {
3002                                 # Remove obsolete MSG:
3003                                 $mwMsg =& MagicWord::get( 'msg' );
3004                                 $mwMsg->matchStartAndRemove( $part1 );
3005                         }
3006
3007                         # Check for RAW:
3008                         $mwRaw =& MagicWord::get( 'raw' );
3009                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3010                                 $forceRawInterwiki = true;
3011                         }
3012                 }
3013                 wfProfileOut( __METHOD__.'-modifiers' );
3014
3015                 //save path level before recursing into functions & templates.
3016                 $lastPathLevel = $this->mTemplatePath;
3017
3018                 # Parser functions
3019                 if ( !$found ) {
3020                         wfProfileIn( __METHOD__ . '-pfunc' );
3021
3022                         $colonPos = strpos( $part1, ':' );
3023                         if ( $colonPos !== false ) {
3024                                 # Case sensitive functions
3025                                 $function = substr( $part1, 0, $colonPos );
3026                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3027                                         $function = $this->mFunctionSynonyms[1][$function];
3028                                 } else {
3029                                         # Case insensitive functions
3030                                         $function = strtolower( $function );
3031                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3032                                                 $function = $this->mFunctionSynonyms[0][$function];
3033                                         } else {
3034                                                 $function = false;
3035                                         }
3036                                 }
3037                                 if ( $function ) {
3038                                         $funcArgs = array_map( 'trim', $args );
3039                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
3040                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
3041                                         $found = true;
3042
3043                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3044                                         //$noargs = true;
3045                                         //$noparse = true;
3046
3047                                         if ( is_array( $result ) ) {
3048                                                 if ( isset( $result[0] ) ) {
3049                                                         $text = $linestart . $result[0];
3050                                                         unset( $result[0] );
3051                                                 }
3052
3053                                                 // Extract flags into the local scope
3054                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3055                                                 extract( $result );
3056                                         } else {
3057                                                 $text = $linestart . $result;
3058                                         }
3059                                 }
3060                         }
3061                         wfProfileOut( __METHOD__ . '-pfunc' );
3062                 }
3063
3064                 # Template table test
3065
3066                 # Did we encounter this template already? If yes, it is in the cache
3067                 # and we need to check for loops.
3068                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3069                         $found = true;
3070
3071                         # Infinite loop test
3072                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3073                                 $noparse = true;
3074                                 $noargs = true;
3075                                 $found = true;
3076                                 $text = $linestart .
3077                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3078                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3079                         } else {
3080                                 # set $text to cached message.
3081                                 $text = $linestart . $this->mTemplates[$piece['title']];
3082                                 #treat title for cached page the same as others
3083                                 $ns = NS_TEMPLATE;
3084                                 $subpage = '';
3085                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3086                                 if ($subpage !== '') {
3087                                   $ns = $this->mTitle->getNamespace();
3088                                 }
3089                                 $title = Title::newFromText( $part1, $ns );
3090                                 //used by include size checking
3091                                 $titleText = $title->getPrefixedText();
3092                                 //used by edit section links
3093                                 $replaceHeadings = true;
3094
3095                         }
3096                 }
3097
3098                 # Load from database
3099                 if ( !$found ) {
3100                         wfProfileIn( __METHOD__ . '-loadtpl' );
3101                         $ns = NS_TEMPLATE;
3102                         # declaring $subpage directly in the function call
3103                         # does not work correctly with references and breaks
3104                         # {{/subpage}}-style inclusions
3105                         $subpage = '';
3106                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3107                         if ($subpage !== '') {
3108                                 $ns = $this->mTitle->getNamespace();
3109                         }
3110                         $title = Title::newFromText( $part1, $ns );
3111
3112
3113                         if ( !is_null( $title ) ) {
3114                                 $titleText = $title->getPrefixedText();
3115                                 # Check for language variants if the template is not found
3116                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3117                                         $wgContLang->findVariantLink($part1, $title);
3118                                 }
3119
3120                                 if ( !$title->isExternal() ) {
3121                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3122                                                 $text = SpecialPage::capturePath( $title );
3123                                                 if ( is_string( $text ) ) {
3124                                                         $found = true;
3125                                                         $noparse = true;
3126                                                         $noargs = true;
3127                                                         $isHTML = true;
3128                                                         $this->disableCache();
3129                                                 }
3130                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3131                                                 $found = false; //access denied
3132                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3133                                         } else {
3134                                                 list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3135                                                 if ( $articleContent !== false ) {
3136                                                         $found = true;
3137                                                         $text = $articleContent;
3138                                                         $replaceHeadings = true;
3139                                                 }
3140                                         }
3141
3142                                         # If the title is valid but undisplayable, make a link to it
3143                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3144                                                 $text = "[[:$titleText]]";
3145                                                 $found = true;
3146                                         }
3147                                 } elseif ( $title->isTrans() ) {
3148                                         // Interwiki transclusion
3149                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3150                                                 $text = $this->interwikiTransclude( $title, 'render' );
3151                                                 $isHTML = true;
3152                                                 $noparse = true;
3153                                         } else {
3154                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3155                                                 $replaceHeadings = true;
3156                                         }
3157                                         $found = true;
3158                                 }
3159
3160                                 # Template cache array insertion
3161                                 # Use the original $piece['title'] not the mangled $part1, so that
3162                                 # modifiers such as RAW: produce separate cache entries
3163                                 if( $found ) {
3164                                         if( $isHTML ) {
3165                                                 // A special page; don't store it in the template cache.
3166                                         } else {
3167                                                 $this->mTemplates[$piece['title']] = $text;
3168                                         }
3169                                         $text = $linestart . $text;
3170                                 }
3171                         }
3172                         wfProfileOut( __METHOD__ . '-loadtpl' );
3173                 }
3174
3175                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3176                         # Error, oversize inclusion
3177                         $text = $linestart .
3178                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3179                         $noparse = true;
3180                         $noargs = true;
3181                 }
3182
3183                 # Recursive parsing, escaping and link table handling
3184                 # Only for HTML output
3185                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3186                         $text = wfEscapeWikiText( $text );
3187                 } elseif ( !$this->ot['msg'] && $found ) {
3188                         if ( $noargs ) {
3189                                 $assocArgs = array();
3190                         } else {
3191                                 # Clean up argument array
3192                                 $assocArgs = self::createAssocArgs($args);
3193                                 # Add a new element to the templace recursion path
3194                                 $this->mTemplatePath[$part1] = 1;
3195                         }
3196
3197                         if ( !$noparse ) {
3198                                 # If there are any <onlyinclude> tags, only include them
3199                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3200                                         $replacer = new OnlyIncludeReplacer;
3201                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3202                                                 array( &$replacer, 'replace' ), $text );
3203                                         $text = $replacer->output;
3204                                 }
3205                                 # Remove <noinclude> sections and <includeonly> tags
3206                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3207                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3208
3209                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3210                                         # Strip <nowiki>, <pre>, etc.
3211                                         $text = $this->strip( $text, $this->mStripState );
3212                                         if ( $this->ot['html'] ) {
3213                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3214                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3215                                                 $text = Sanitizer::removeHTMLcomments( $text );
3216                                         }
3217                                 }
3218                                 $text = $this->replaceVariables( $text, $assocArgs );
3219
3220                                 # If the template begins with a table or block-level
3221                                 # element, it should be treated as beginning a new line.
3222                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3223                                         $text = "\n" . $text;
3224                                 }
3225                         } elseif ( !$noargs ) {
3226                                 # $noparse and !$noargs
3227                                 # Just replace the arguments, not any double-brace items
3228                                 # This is used for rendered interwiki transclusion
3229                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3230                         }
3231                 }
3232                 # Prune lower levels off the recursion check path
3233                 $this->mTemplatePath = $lastPathLevel;
3234
3235                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3236                         # Error, oversize inclusion
3237                         $text = $linestart .
3238                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3239                         $noparse = true;
3240                         $noargs = true;
3241                 }
3242
3243                 if ( !$found ) {
3244                         wfProfileOut( $fname );
3245                         return $piece['text'];
3246                 } else {
3247                         wfProfileIn( __METHOD__ . '-placeholders' );
3248                         if ( $isHTML ) {
3249                                 # Replace raw HTML by a placeholder
3250                                 # Add a blank line preceding, to prevent it from mucking up
3251                                 # immediately preceding headings
3252                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3253                         } else {
3254                                 # replace ==section headers==
3255                                 # XXX this needs to go away once we have a better parser.
3256                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3257                                         if( !is_null( $title ) )
3258                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3259                                         else
3260                                                 $encodedname = base64_encode("");
3261                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3262                                                 PREG_SPLIT_DELIM_CAPTURE);
3263                                         $text = '';
3264                                         $nsec = $headingOffset;
3265
3266                                         for( $i = 0; $i < count($m); $i += 2 ) {
3267                                                 $text .= $m[$i];
3268                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3269                                                 $hl = $m[$i + 1];
3270                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3271                                                         $text .= $hl;
3272                                                         continue;
3273                                                 }
3274                                                 $m2 = array();
3275                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3276                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3277                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3278
3279                                                 $nsec++;
3280                                         }
3281                                 }
3282                         }
3283                         wfProfileOut( __METHOD__ . '-placeholders' );
3284                 }
3285
3286                 # Prune lower levels off the recursion check path
3287                 $this->mTemplatePath = $lastPathLevel;
3288
3289                 if ( !$found ) {
3290                         wfProfileOut( $fname );
3291                         return $piece['text'];
3292                 } else {
3293                         wfProfileOut( $fname );
3294                         return $text;
3295                 }
3296         }
3297
3298         /**
3299          * Fetch the unparsed text of a template and register a reference to it.
3300          */
3301         function fetchTemplateAndtitle( $title ) {
3302                 $text = $skip = false;
3303                 $finalTitle = $title;
3304                 // Loop to fetch the article, with up to 1 redirect
3305                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3306                         # Give extensions a chance to select the revision instead
3307                         $id = false; // Assume current
3308                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( &$this, &$title, &$skip, &$id ) );
3309
3310                         if( $skip ) {
3311                                 $text = false;
3312                                 $this->mOutput->addTemplate( $title, $title->getArticleID(), null );
3313                                 break;
3314                         }
3315                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3316                         $rev_id = $rev ? $rev->getId() : 0;
3317
3318                         $this->mOutput->addTemplate( $title, $title->getArticleID(), $rev_id );
3319
3320                         if( $rev ) {
3321                                 $text = $rev->getText();
3322                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3323                                 global $wgLang;
3324                                 $message = $wgLang->lcfirst( $title->getText() );
3325                                 $text = wfMsgForContentNoTrans( $message );
3326                                 if( wfEmptyMsg( $message, $text ) ) {
3327                                         $text = false;
3328                                         break;
3329                                 }
3330                         } else {
3331                                 break;
3332                         }
3333                         if ( $text === false ) {
3334                                 break;
3335                         }
3336                         // Redirect?
3337                         $finalTitle = $title;
3338                         $title = Title::newFromRedirect( $text );
3339                 }
3340                 return array($text,$finalTitle);
3341         }
3342
3343         function fetchTemplate( $title ) {
3344                 $rv = $this->fetchTemplateAndtitle($title);
3345                 return $rv[0];
3346         }
3347
3348         /**
3349          * Transclude an interwiki link.
3350          */
3351         function interwikiTransclude( $title, $action ) {
3352                 global $wgEnableScaryTranscluding;
3353
3354                 if (!$wgEnableScaryTranscluding)
3355                         return wfMsg('scarytranscludedisabled');
3356
3357                 $url = $title->getFullUrl( "action=$action" );
3358
3359                 if (strlen($url) > 255)
3360                         return wfMsg('scarytranscludetoolong');
3361                 return $this->fetchScaryTemplateMaybeFromCache($url);
3362         }
3363
3364         function fetchScaryTemplateMaybeFromCache($url) {
3365                 global $wgTranscludeCacheExpiry;
3366                 $dbr = wfGetDB(DB_SLAVE);
3367                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3368                                 array('tc_url' => $url));
3369                 if ($obj) {
3370                         $time = $obj->tc_time;
3371                         $text = $obj->tc_contents;
3372                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3373                                 return $text;
3374                         }
3375                 }
3376
3377                 $text = Http::get($url);
3378                 if (!$text)
3379                         return wfMsg('scarytranscludefailed', $url);
3380
3381                 $dbw = wfGetDB(DB_MASTER);
3382                 $dbw->replace('transcache', array('tc_url'), array(
3383                         'tc_url' => $url,
3384                         'tc_time' => time(),
3385                         'tc_contents' => $text));
3386                 return $text;
3387         }
3388
3389
3390         /**
3391          * Triple brace replacement -- used for template arguments
3392          * @private
3393          */
3394         function argSubstitution( $matches ) {
3395                 $arg = trim( $matches['title'] );
3396                 $text = $matches['text'];
3397                 $inputArgs = end( $this->mArgStack );
3398
3399                 if ( array_key_exists( $arg, $inputArgs ) ) {
3400                         $text = $inputArgs[$arg];
3401                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3402                 null != $matches['parts'] && count($matches['parts']) > 0) {
3403                         $text = $matches['parts'][0];
3404                 }
3405                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3406                         $text = $matches['text'] .
3407                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3408                 }
3409
3410                 return $text;
3411         }
3412
3413         /**
3414          * Increment an include size counter
3415          *
3416          * @param string $type The type of expansion
3417          * @param integer $size The size of the text
3418          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3419          */
3420         function incrementIncludeSize( $type, $size ) {
3421                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3422                         return false;
3423                 } else {
3424                         $this->mIncludeSizes[$type] += $size;
3425                         return true;
3426                 }
3427         }
3428
3429         /**
3430          * Detect __NOGALLERY__ magic word and set a placeholder
3431          */
3432         function stripNoGallery( &$text ) {
3433                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3434                 # do not add TOC
3435                 $mw = MagicWord::get( 'nogallery' );
3436                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3437         }
3438
3439         /**
3440          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3441          * placeholder that will then be replaced by the real TOC in
3442          * ->formatHeadings, this works because at this points real
3443          * comments will have already been discarded by the sanitizer.
3444          *
3445          * Any additional __TOC__ magic words left over will be discarded
3446          * as there can only be one TOC on the page.
3447          */
3448         function stripToc( $text ) {
3449                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3450                 # do not add TOC
3451                 $mw = MagicWord::get( 'notoc' );
3452                 if( $mw->matchAndRemove( $text ) ) {
3453                         $this->mShowToc = false;
3454                 }
3455
3456                 $mw = MagicWord::get( 'toc' );
3457                 if( $mw->match( $text ) ) {
3458                         $this->mShowToc = true;
3459                         $this->mForceTocPosition = true;
3460
3461                         // Set a placeholder. At the end we'll fill it in with the TOC.
3462                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3463
3464                         // Only keep the first one.
3465                         $text = $mw->replace( '', $text );
3466                 }
3467                 return $text;
3468         }
3469
3470         /**
3471          * This function accomplishes several tasks:
3472          * 1) Auto-number headings if that option is enabled
3473          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3474          * 3) Add a Table of contents on the top for users who have enabled the option
3475          * 4) Auto-anchor headings
3476          *
3477          * It loops through all headlines, collects the necessary data, then splits up the
3478          * string and re-inserts the newly formatted headlines.
3479          *
3480          * @param string $text
3481          * @param boolean $isMain
3482          * @private
3483          */
3484         function formatHeadings( $text, $isMain=true ) {
3485                 global $wgMaxTocLevel, $wgContLang;
3486
3487                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3488                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3489                         $showEditLink = 0;
3490                 } else {
3491                         $showEditLink = $this->mOptions->getEditSection();
3492                 }
3493
3494                 # Inhibit editsection links if requested in the page
3495                 $esw =& MagicWord::get( 'noeditsection' );
3496                 if( $esw->matchAndRemove( $text ) ) {
3497                         $showEditLink = 0;
3498                 }
3499
3500                 # Get all headlines for numbering them and adding funky stuff like [edit]
3501                 # links - this is for later, but we need the number of headlines right now
3502                 $matches = array();
3503                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3504
3505                 # if there are fewer than 4 headlines in the article, do not show TOC
3506                 # unless it's been explicitly enabled.
3507                 $enoughToc = $this->mShowToc &&
3508                         (($numMatches >= 4) || $this->mForceTocPosition);
3509
3510                 # Allow user to stipulate that a page should have a "new section"
3511                 # link added via __NEWSECTIONLINK__
3512                 $mw =& MagicWord::get( 'newsectionlink' );
3513                 if( $mw->matchAndRemove( $text ) )
3514                         $this->mOutput->setNewSection( true );
3515
3516                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3517                 # override above conditions and always show TOC above first header
3518                 $mw =& MagicWord::get( 'forcetoc' );
3519                 if ($mw->matchAndRemove( $text ) ) {
3520                         $this->mShowToc = true;
3521                         $enoughToc = true;
3522                 }
3523
3524                 # We need this to perform operations on the HTML
3525                 $sk = $this->mOptions->getSkin();
3526
3527                 # headline counter
3528                 $headlineCount = 0;
3529                 $sectionCount = 0; # headlineCount excluding template sections
3530                 $numVisible = 0;
3531
3532                 # Ugh .. the TOC should have neat indentation levels which can be
3533                 # passed to the skin functions. These are determined here
3534                 $toc = '';
3535                 $full = '';
3536                 $head = array();
3537                 $sublevelCount = array();
3538                 $levelCount = array();
3539                 $toclevel = 0;
3540                 $level = 0;
3541                 $prevlevel = 0;
3542                 $toclevel = 0;
3543                 $prevtoclevel = 0;
3544
3545                 foreach( $matches[3] as $headline ) {
3546                         $istemplate = 0;
3547                         $templatetitle = '';
3548                         $templatesection = 0;
3549                         $numbering = '';
3550                         $mat = array();
3551                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3552                                 $istemplate = 1;
3553                                 $templatetitle = base64_decode($mat[1]);
3554                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3555                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3556                         }
3557
3558                         if( $toclevel ) {
3559                                 $prevlevel = $level;
3560                                 $prevtoclevel = $toclevel;
3561                         }
3562                         $level = $matches[1][$headlineCount];
3563
3564                         if( $doNumberHeadings || $enoughToc ) {
3565
3566                                 if ( $level > $prevlevel ) {
3567                                         # Increase TOC level
3568                                         $toclevel++;
3569                                         $sublevelCount[$toclevel] = 0;
3570                                         if( $toclevel<$wgMaxTocLevel ) {
3571                                                 $prevtoclevel = $toclevel;
3572                                                 $toc .= $sk->tocIndent();
3573                                                 $numVisible++;
3574                                         }
3575                                 }
3576                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3577                                         # Decrease TOC level, find level to jump to
3578
3579                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3580                                                 # Can only go down to level 1
3581                                                 $toclevel = 1;
3582                                         } else {
3583                                                 for ($i = $toclevel; $i > 0; $i--) {
3584                                                         if ( $levelCount[$i] == $level ) {
3585                                                                 # Found last matching level
3586                                                                 $toclevel = $i;
3587                                                                 break;
3588                                                         }
3589                                                         elseif ( $levelCount[$i] < $level ) {
3590                                                                 # Found first matching level below current level
3591                                                                 $toclevel = $i + 1;
3592                                                                 break;
3593                                                         }
3594                                                 }
3595                                         }
3596                                         if( $toclevel<$wgMaxTocLevel ) {
3597                                                 if($prevtoclevel < $wgMaxTocLevel) {
3598                                                         # Unindent only if the previous toc level was shown :p
3599                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3600                                                 } else {
3601                                                         $toc .= $sk->tocLineEnd();
3602                                                 }
3603                                         }
3604                                 }
3605                                 else {
3606                                         # No change in level, end TOC line
3607                                         if( $toclevel<$wgMaxTocLevel ) {
3608                                                 $toc .= $sk->tocLineEnd();
3609                                         }
3610                                 }
3611
3612                                 $levelCount[$toclevel] = $level;
3613
3614                                 # count number of headlines for each level
3615                                 @$sublevelCount[$toclevel]++;
3616                                 $dot = 0;
3617                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3618                                         if( !empty( $sublevelCount[$i] ) ) {
3619                                                 if( $dot ) {
3620                                                         $numbering .= '.';
3621                                                 }
3622                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3623                                                 $dot = 1;
3624                                         }
3625                                 }
3626                         }
3627
3628                         # The canonized header is a version of the header text safe to use for links
3629                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3630                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3631
3632                         # Remove link placeholders by the link text.
3633                         #     <!--LINK number-->
3634                         # turns into
3635                         #     link text with suffix
3636                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3637                                                             "\$this->mLinkHolders['texts'][\$1]",
3638                                                             $canonized_headline );
3639                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3640                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3641                                                             $canonized_headline );
3642
3643                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3644                         $tocline = preg_replace(
3645                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3646                                 array( '',                          '<$1>'),
3647                                 $canonized_headline
3648                         );
3649                         $tocline = trim( $tocline );
3650
3651                         # For the anchor, strip out HTML-y stuff period
3652                         $canonized_headline = preg_replace( '/<.*?'.'>/', '', $canonized_headline );
3653                         $canonized_headline = trim( $canonized_headline );
3654
3655                         # Save headline for section edit hint before it's escaped
3656                         $headline_hint = $canonized_headline;
3657                         $canonized_headline = Sanitizer::escapeId( $canonized_headline );
3658                         $refers[$headlineCount] = $canonized_headline;
3659
3660                         # count how many in assoc. array so we can track dupes in anchors
3661                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3662                         $refcount[$headlineCount]=$refers[$canonized_headline];
3663
3664                         # Don't number the heading if it is the only one (looks silly)
3665                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3666                                 # the two are different if the line contains a link
3667                                 $headline=$numbering . ' ' . $headline;
3668                         }
3669
3670                         # Create the anchor for linking from the TOC to the section
3671                         $anchor = $canonized_headline;
3672                         if($refcount[$headlineCount] > 1 ) {
3673                                 $anchor .= '_' . $refcount[$headlineCount];
3674                         }
3675                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3676                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3677                         }
3678                         # give headline the correct <h#> tag
3679                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3680                                 if( $istemplate )
3681                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3682                                 else
3683                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3684                         } else {
3685                                 $editlink = '';
3686                         }
3687                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3688
3689                         $headlineCount++;
3690                         if( !$istemplate )
3691                                 $sectionCount++;
3692                 }
3693
3694                 # Never ever show TOC if no headers
3695                 if( $numVisible < 1 ) {
3696                         $enoughToc = false;
3697                 }
3698
3699                 if( $enoughToc ) {
3700                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3701                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3702                         }
3703                         $toc = $sk->tocList( $toc );
3704                 }
3705
3706                 # split up and insert constructed headlines
3707
3708                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3709                 $i = 0;
3710
3711                 foreach( $blocks as $block ) {
3712                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3713                                 # This is the [edit] link that appears for the top block of text when
3714                                 # section editing is enabled
3715
3716                                 # Disabled because it broke block formatting
3717                                 # For example, a bullet point in the top line
3718                                 # $full .= $sk->editSectionLink(0);
3719                         }
3720                         $full .= $block;
3721                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3722                                 # Top anchor now in skin
3723                                 $full = $full.$toc;
3724                         }
3725
3726                         if( !empty( $head[$i] ) ) {
3727                                 $full .= $head[$i];
3728                         }
3729                         $i++;
3730                 }
3731                 if( $this->mForceTocPosition ) {
3732                         return str_replace( '<!--MWTOC-->', $toc, $full );
3733                 } else {
3734                         return $full;
3735                 }
3736         }
3737
3738         /**
3739          * Transform wiki markup when saving a page by doing \r\n -> \n
3740          * conversion, substitting signatures, {{subst:}} templates, etc.
3741          *
3742          * @param string $text the text to transform
3743          * @param Title &$title the Title object for the current article
3744          * @param User &$user the User object describing the current user
3745          * @param ParserOptions $options parsing options
3746          * @param bool $clearState whether to clear the parser state first
3747          * @return string the altered wiki markup
3748          * @public
3749          */
3750         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3751                 $this->mOptions = $options;
3752                 $this->mTitle =& $title;
3753                 $this->setOutputType( OT_WIKI );
3754
3755                 if ( $clearState ) {
3756                         $this->clearState();
3757                 }
3758
3759                 $stripState = new StripState;
3760                 $pairs = array(
3761                         "\r\n" => "\n",
3762                 );
3763                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3764                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3765                 $text = $this->pstPass2( $text, $stripState, $user );
3766                 $text = $stripState->unstripBoth( $text );
3767                 return $text;
3768         }
3769
3770         /**
3771          * Pre-save transform helper function
3772          * @private
3773          */
3774         function pstPass2( $text, &$stripState, $user ) {
3775                 global $wgContLang, $wgLocaltimezone;
3776
3777                 /* Note: This is the timestamp saved as hardcoded wikitext to
3778                  * the database, we use $wgContLang here in order to give
3779                  * everyone the same signature and use the default one rather
3780                  * than the one selected in each user's preferences.
3781                  */
3782                 if ( isset( $wgLocaltimezone ) ) {
3783                         $oldtz = getenv( 'TZ' );
3784                         putenv( 'TZ='.$wgLocaltimezone );
3785                 }
3786                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3787                   ' (' . date( 'T' ) . ')';
3788                 if ( isset( $wgLocaltimezone ) ) {
3789                         putenv( 'TZ='.$oldtz );
3790                 }
3791
3792                 # Variable replacement
3793                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3794                 $text = $this->replaceVariables( $text );
3795
3796                 # Strip out <nowiki> etc. added via replaceVariables
3797                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3798
3799                 # Signatures
3800                 $sigText = $this->getUserSig( $user );
3801                 $text = strtr( $text, array(
3802                         '~~~~~' => $d,
3803                         '~~~~' => "$sigText $d",
3804                         '~~~' => $sigText
3805                 ) );
3806
3807                 # Context links: [[|name]] and [[name (context)|]]
3808                 #
3809                 global $wgLegalTitleChars;
3810                 $tc = "[$wgLegalTitleChars]";
3811                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3812
3813                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3814                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3815                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3816
3817                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3818                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3819                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3820
3821                 $t = $this->mTitle->getText();
3822                 $m = array();
3823                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3824                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3825                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3826                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3827                 } else {
3828                         # if there's no context, don't bother duplicating the title
3829                         $text = preg_replace( $p2, '[[\\1]]', $text );
3830                 }
3831
3832                 # Trim trailing whitespace
3833                 $text = rtrim( $text );
3834
3835                 return $text;
3836         }
3837
3838         /**
3839          * Fetch the user's signature text, if any, and normalize to
3840          * validated, ready-to-insert wikitext.
3841          *
3842          * @param User $user
3843          * @return string
3844          * @private
3845          */
3846         function getUserSig( &$user ) {
3847                 global $wgMaxSigChars;
3848
3849                 $username = $user->getName();
3850                 $nickname = $user->getOption( 'nickname' );
3851                 $nickname = $nickname === '' ? $username : $nickname;
3852
3853                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3854                         $nickname = $username;
3855                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3856                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3857                         # Sig. might contain markup; validate this
3858                         if( $this->validateSig( $nickname ) !== false ) {
3859                                 # Validated; clean up (if needed) and return it
3860                                 return $this->cleanSig( $nickname, true );
3861                         } else {
3862                                 # Failed to validate; fall back to the default
3863                                 $nickname = $username;
3864                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3865                         }
3866                 }
3867
3868                 // Make sure nickname doesnt get a sig in a sig
3869                 $nickname = $this->cleanSigInSig( $nickname );
3870
3871                 # If we're still here, make it a link to the user page
3872                 $userpage = $user->getUserPage();
3873                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3874         }
3875
3876         /**
3877          * Check that the user's signature contains no bad XML
3878          *
3879          * @param string $text
3880          * @return mixed An expanded string, or false if invalid.
3881          */
3882         function validateSig( $text ) {
3883                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3884         }
3885
3886         /**
3887          * Clean up signature text
3888          *
3889          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3890          * 2) Substitute all transclusions
3891          *
3892          * @param string $text
3893          * @param $parsing Whether we're cleaning (preferences save) or parsing
3894          * @return string Signature text
3895          */
3896         function cleanSig( $text, $parsing = false ) {
3897                 global $wgTitle;
3898                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3899
3900                 $substWord = MagicWord::get( 'subst' );
3901                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3902                 $substText = '{{' . $substWord->getSynonym( 0 );
3903
3904                 $text = preg_replace( $substRegex, $substText, $text );
3905                 $text = $this->cleanSigInSig( $text );
3906                 $text = $this->replaceVariables( $text );
3907
3908                 $this->clearState();
3909                 return $text;
3910         }
3911
3912         /**
3913          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3914          * @param string $text
3915          * @return string Signature text with /~{3,5}/ removed
3916          */
3917         function cleanSigInSig( $text ) {
3918                 $text = preg_replace( '/~{3,5}/', '', $text );
3919                 return $text;
3920         }
3921
3922         /**
3923          * Set up some variables which are usually set up in parse()
3924          * so that an external function can call some class members with confidence
3925          * @public
3926          */
3927         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3928                 $this->mTitle =& $title;
3929                 $this->mOptions = $options;
3930                 $this->setOutputType( $outputType );
3931                 if ( $clearState ) {
3932                         $this->clearState();
3933                 }
3934         }
3935
3936         /**
3937          * Transform a MediaWiki message by replacing magic variables.
3938          *
3939          * @param string $text the text to transform
3940          * @param ParserOptions $options  options
3941          * @return string the text with variables substituted
3942          * @public
3943          */
3944         function transformMsg( $text, $options ) {
3945                 global $wgTitle;
3946                 static $executing = false;
3947
3948                 $fname = "Parser::transformMsg";
3949
3950                 # Guard against infinite recursion
3951                 if ( $executing ) {
3952                         return $text;
3953                 }
3954                 $executing = true;
3955
3956                 wfProfileIn($fname);
3957
3958                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3959                         $this->mTitle = $wgTitle;
3960                 } else {
3961                         $this->mTitle = Title::newFromText('msg');
3962                 }
3963                 $this->mOptions = $options;
3964                 $this->setOutputType( OT_MSG );
3965                 $this->clearState();
3966                 $text = $this->replaceVariables( $text );
3967
3968                 $executing = false;
3969                 wfProfileOut($fname);
3970                 return $text;
3971         }
3972
3973         /**
3974          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3975          * The callback should have the following form:
3976          *    function myParserHook( $text, $params, &$parser ) { ... }
3977          *
3978          * Transform and return $text. Use $parser for any required context, e.g. use
3979          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3980          *
3981          * @public
3982          *
3983          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3984          * @param mixed $callback The callback function (and object) to use for the tag
3985          *
3986          * @return The old value of the mTagHooks array associated with the hook
3987          */
3988         function setHook( $tag, $callback ) {
3989                 $tag = strtolower( $tag );
3990                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3991                 $this->mTagHooks[$tag] = $callback;
3992
3993                 return $oldVal;
3994         }
3995
3996         function setTransparentTagHook( $tag, $callback ) {
3997                 $tag = strtolower( $tag );
3998                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3999                 $this->mTransparentTagHooks[$tag] = $callback;
4000
4001                 return $oldVal;
4002         }
4003
4004         /**
4005          * Create a function, e.g. {{sum:1|2|3}}
4006          * The callback function should have the form:
4007          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4008          *
4009          * The callback may either return the text result of the function, or an array with the text
4010          * in element 0, and a number of flags in the other elements. The names of the flags are
4011          * specified in the keys. Valid flags are:
4012          *   found                     The text returned is valid, stop processing the template. This
4013          *                             is on by default.
4014          *   nowiki                    Wiki markup in the return value should be escaped
4015          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4016          *   noargs                    Don't replace triple-brace arguments in the return value
4017          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4018          *
4019          * @public
4020          *
4021          * @param string $id The magic word ID
4022          * @param mixed $callback The callback function (and object) to use
4023          * @param integer $flags a combination of the following flags:
4024          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4025          *
4026          * @return The old callback function for this name, if any
4027          */
4028         function setFunctionHook( $id, $callback, $flags = 0 ) {
4029                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
4030                 $this->mFunctionHooks[$id] = $callback;
4031
4032                 # Add to function cache
4033                 $mw = MagicWord::get( $id );
4034                 if( !$mw )
4035                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4036
4037                 $synonyms = $mw->getSynonyms();
4038                 $sensitive = intval( $mw->isCaseSensitive() );
4039
4040                 foreach ( $synonyms as $syn ) {
4041                         # Case
4042                         if ( !$sensitive ) {
4043                                 $syn = strtolower( $syn );
4044                         }
4045                         # Add leading hash
4046                         if ( !( $flags & SFH_NO_HASH ) ) {
4047                                 $syn = '#' . $syn;
4048                         }
4049                         # Remove trailing colon
4050                         if ( substr( $syn, -1, 1 ) == ':' ) {
4051                                 $syn = substr( $syn, 0, -1 );
4052                         }
4053                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4054                 }
4055                 return $oldVal;
4056         }
4057
4058         /**
4059          * Get all registered function hook identifiers
4060          *
4061          * @return array
4062          */
4063         function getFunctionHooks() {
4064                 return array_keys( $this->mFunctionHooks );
4065         }
4066
4067         /**
4068          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4069          * Placeholders created in Skin::makeLinkObj()
4070          * Returns an array of links found, indexed by PDBK:
4071          *  0 - broken
4072          *  1 - normal link
4073          *  2 - stub
4074          * $options is a bit field, RLH_FOR_UPDATE to select for update
4075          */
4076         function replaceLinkHolders( &$text, $options = 0 ) {
4077                 global $wgUser;
4078                 global $wgContLang;
4079
4080                 $fname = 'Parser::replaceLinkHolders';
4081                 wfProfileIn( $fname );
4082
4083                 $pdbks = array();
4084                 $colours = array();
4085                 $sk = $this->mOptions->getSkin();
4086                 $linkCache =& LinkCache::singleton();
4087
4088                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4089                         wfProfileIn( $fname.'-check' );
4090                         $dbr = wfGetDB( DB_SLAVE );
4091                         $page = $dbr->tableName( 'page' );
4092                         $threshold = $wgUser->getOption('stubthreshold');
4093
4094                         # Sort by namespace
4095                         asort( $this->mLinkHolders['namespaces'] );
4096
4097                         # Generate query
4098                         $query = false;
4099                         $current = null;
4100                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4101                                 # Make title object
4102                                 $title = $this->mLinkHolders['titles'][$key];
4103
4104                                 # Skip invalid entries.
4105                                 # Result will be ugly, but prevents crash.
4106                                 if ( is_null( $title ) ) {
4107                                         continue;
4108                                 }
4109                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4110
4111                                 # Check if it's a static known link, e.g. interwiki
4112                                 if ( $title->isAlwaysKnown() ) {
4113                                         $colours[$pdbk] = 1;
4114                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4115                                         $colours[$pdbk] = 1;
4116                                         $this->mOutput->addLink( $title, $id );
4117                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4118                                         $colours[$pdbk] = 0;
4119                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4120                                         $colours[$pdbk] = 0;
4121                                 } else {
4122                                         # Not in the link cache, add it to the query
4123                                         if ( !isset( $current ) ) {
4124                                                 $current = $ns;
4125                                                 $query =  "SELECT page_id, page_namespace, page_title";
4126                                                 if ( $threshold > 0 ) {
4127                                                         $query .= ', page_len, page_is_redirect';
4128                                                 }
4129                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4130                                         } elseif ( $current != $ns ) {
4131                                                 $current = $ns;
4132                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4133                                         } else {
4134                                                 $query .= ', ';
4135                                         }
4136
4137                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4138                                 }
4139                         }
4140                         if ( $query ) {
4141                                 $query .= '))';
4142                                 if ( $options & RLH_FOR_UPDATE ) {
4143                                         $query .= ' FOR UPDATE';
4144                                 }
4145
4146                                 $res = $dbr->query( $query, $fname );
4147
4148                                 # Fetch data and form into an associative array
4149                                 # non-existent = broken
4150                                 # 1 = known
4151                                 # 2 = stub
4152                                 while ( $s = $dbr->fetchObject($res) ) {
4153                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4154                                         $pdbk = $title->getPrefixedDBkey();
4155                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4156                                         $this->mOutput->addLink( $title, $s->page_id );
4157
4158                                         $colours[$pdbk] = ( $threshold == 0 || (
4159                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4160                                                                 $s->page_is_redirect ||
4161                                                                 !Namespace::isContent( $s->page_namespace ) )
4162                                                             ? 1 : 2 );
4163                                 }
4164                         }
4165                         wfProfileOut( $fname.'-check' );
4166
4167                         # Do a second query for different language variants of links and categories
4168                         if($wgContLang->hasVariants()){
4169                                 $linkBatch = new LinkBatch();
4170                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4171                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4172                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4173
4174                                 $categories = $this->mOutput->getCategoryLinks();
4175
4176                                 // Add variants of links to link batch
4177                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4178                                         $title = $this->mLinkHolders['titles'][$key];
4179                                         if ( is_null( $title ) )
4180                                                 continue;
4181
4182                                         $pdbk = $title->getPrefixedDBkey();
4183                                         $titleText = $title->getText();
4184
4185                                         // generate all variants of the link title text
4186                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4187
4188                                         // if link was not found (in first query), add all variants to query
4189                                         if ( !isset($colours[$pdbk]) ){
4190                                                 foreach($allTextVariants as $textVariant){
4191                                                         if($textVariant != $titleText){
4192                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4193                                                                 if(is_null($variantTitle)) continue;
4194                                                                 $linkBatch->addObj( $variantTitle );
4195                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4196                                                         }
4197                                                 }
4198                                         }
4199                                 }
4200
4201                                 // process categories, check if a category exists in some variant
4202                                 foreach( $categories as $category ){
4203                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4204                                         foreach($variants as $variant){
4205                                                 if($variant != $category){
4206                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4207                                                         if(is_null($variantTitle)) continue;
4208                                                         $linkBatch->addObj( $variantTitle );
4209                                                         $categoryMap[$variant] = $category;
4210                                                 }
4211                                         }
4212                                 }
4213
4214
4215                                 if(!$linkBatch->isEmpty()){
4216                                         // construct query
4217                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4218
4219                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4220                                         if ( $threshold > 0 ) {
4221                                                 $variantQuery .= ', page_len, page_is_redirect';
4222                                         }
4223
4224                                         $variantQuery .= " FROM $page WHERE $titleClause";
4225                                         if ( $options & RLH_FOR_UPDATE ) {
4226                                                 $variantQuery .= ' FOR UPDATE';
4227                                         }
4228
4229                                         $varRes = $dbr->query( $variantQuery, $fname );
4230
4231                                         // for each found variants, figure out link holders and replace
4232                                         while ( $s = $dbr->fetchObject($varRes) ) {
4233
4234                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4235                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4236                                                 $vardbk = $variantTitle->getDBkey();
4237
4238                                                 $holderKeys = array();
4239                                                 if(isset($variantMap[$varPdbk])){
4240                                                         $holderKeys = $variantMap[$varPdbk];
4241                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4242                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4243                                                 }
4244
4245                                                 // loop over link holders
4246                                                 foreach($holderKeys as $key){
4247                                                         $title = $this->mLinkHolders['titles'][$key];
4248                                                         if ( is_null( $title ) ) continue;
4249
4250                                                         $pdbk = $title->getPrefixedDBkey();
4251
4252                                                         if(!isset($colours[$pdbk])){
4253                                                                 // found link in some of the variants, replace the link holder data
4254                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4255                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4256
4257                                                                 // set pdbk and colour
4258                                                                 $pdbks[$key] = $varPdbk;
4259                                                                 if ( $threshold >  0 ) {
4260                                                                         $size = $s->page_len;
4261                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4262                                                                                 $colours[$varPdbk] = 1;
4263                                                                         } else {
4264                                                                                 $colours[$varPdbk] = 2;
4265                                                                         }
4266                                                                 }
4267                                                                 else {
4268                                                                         $colours[$varPdbk] = 1;
4269                                                                 }
4270                                                         }
4271                                                 }
4272
4273                                                 // check if the object is a variant of a category
4274                                                 if(isset($categoryMap[$vardbk])){
4275                                                         $oldkey = $categoryMap[$vardbk];
4276                                                         if($oldkey != $vardbk)
4277                                                                 $varCategories[$oldkey]=$vardbk;
4278                                                 }
4279                                         }
4280
4281                                         // rebuild the categories in original order (if there are replacements)
4282                                         if(count($varCategories)>0){
4283                                                 $newCats = array();
4284                                                 $originalCats = $this->mOutput->getCategories();
4285                                                 foreach($originalCats as $cat => $sortkey){
4286                                                         // make the replacement
4287                                                         if( array_key_exists($cat,$varCategories) )
4288                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4289                                                         else $newCats[$cat] = $sortkey;
4290                                                 }
4291                                                 $this->mOutput->setCategoryLinks($newCats);
4292                                         }
4293                                 }
4294                         }
4295
4296                         # Construct search and replace arrays
4297                         wfProfileIn( $fname.'-construct' );
4298                         $replacePairs = array();
4299                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4300                                 $pdbk = $pdbks[$key];
4301                                 $searchkey = "<!--LINK $key-->";
4302                                 $title = $this->mLinkHolders['titles'][$key];
4303                                 if ( empty( $colours[$pdbk] ) ) {
4304                                         $linkCache->addBadLinkObj( $title );
4305                                         $colours[$pdbk] = 0;
4306                                         $this->mOutput->addLink( $title, 0 );
4307                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4308                                                                         $this->mLinkHolders['texts'][$key],
4309                                                                         $this->mLinkHolders['queries'][$key] );
4310                                 } elseif ( $colours[$pdbk] == 1 ) {
4311                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4312                                                                         $this->mLinkHolders['texts'][$key],
4313                                                                         $this->mLinkHolders['queries'][$key] );
4314                                 } elseif ( $colours[$pdbk] == 2 ) {
4315                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4316                                                                         $this->mLinkHolders['texts'][$key],
4317                                                                         $this->mLinkHolders['queries'][$key] );
4318                                 }
4319                         }
4320                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4321                         wfProfileOut( $fname.'-construct' );
4322
4323                         # Do the thing
4324                         wfProfileIn( $fname.'-replace' );
4325                         $text = preg_replace_callback(
4326                                 '/(<!--LINK .*?-->)/',
4327                                 $replacer->cb(),
4328                                 $text);
4329
4330                         wfProfileOut( $fname.'-replace' );
4331                 }
4332
4333                 # Now process interwiki link holders
4334                 # This is quite a bit simpler than internal links
4335                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4336                         wfProfileIn( $fname.'-interwiki' );
4337                         # Make interwiki link HTML
4338                         $replacePairs = array();
4339                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4340                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4341                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4342                         }
4343                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4344
4345                         $text = preg_replace_callback(
4346                                 '/<!--IWLINK (.*?)-->/',
4347                                 $replacer->cb(),
4348                                 $text );
4349                         wfProfileOut( $fname.'-interwiki' );
4350                 }
4351
4352                 wfProfileOut( $fname );
4353                 return $colours;
4354         }
4355
4356         /**
4357          * Replace <!--LINK--> link placeholders with plain text of links
4358          * (not HTML-formatted).
4359          * @param string $text
4360          * @return string
4361          */
4362         function replaceLinkHoldersText( $text ) {
4363                 $fname = 'Parser::replaceLinkHoldersText';
4364                 wfProfileIn( $fname );
4365
4366                 $text = preg_replace_callback(
4367                         '/<!--(LINK|IWLINK) (.*?)-->/',
4368                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4369                         $text );
4370
4371                 wfProfileOut( $fname );
4372                 return $text;
4373         }
4374
4375         /**
4376          * @param array $matches
4377          * @return string
4378          * @private
4379          */
4380         function replaceLinkHoldersTextCallback( $matches ) {
4381                 $type = $matches[1];
4382                 $key  = $matches[2];
4383                 if( $type == 'LINK' ) {
4384                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4385                                 return $this->mLinkHolders['texts'][$key];
4386                         }
4387                 } elseif( $type == 'IWLINK' ) {
4388                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4389                                 return $this->mInterwikiLinkHolders['texts'][$key];
4390                         }
4391                 }
4392                 return $matches[0];
4393         }
4394
4395         /**
4396          * Tag hook handler for 'pre'.
4397          */
4398         function renderPreTag( $text, $attribs ) {
4399                 // Backwards-compatibility hack
4400                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4401
4402                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4403                 return wfOpenElement( 'pre', $attribs ) .
4404                         Xml::escapeTagsOnly( $content ) .
4405                         '</pre>';
4406         }
4407
4408         /**
4409          * Renders an image gallery from a text with one line per image.
4410          * text labels may be given by using |-style alternative text. E.g.
4411          *   Image:one.jpg|The number "1"
4412          *   Image:tree.jpg|A tree
4413          * given as text will return the HTML of a gallery with two images,
4414          * labeled 'The number "1"' and
4415          * 'A tree'.
4416          */
4417         function renderImageGallery( $text, $params ) {
4418                 $ig = new ImageGallery();
4419                 $ig->setContextTitle( $this->mTitle );
4420                 $ig->setShowBytes( false );
4421                 $ig->setShowFilename( false );
4422                 $ig->setParser( $this );
4423                 $ig->setHideBadImages();
4424                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4425                 $ig->useSkin( $this->mOptions->getSkin() );
4426                 $ig->mRevisionId = $this->mRevisionId;
4427
4428                 if( isset( $params['caption'] ) ) {
4429                         $caption = $params['caption'];
4430                         $caption = htmlspecialchars( $caption );
4431                         $caption = $this->replaceInternalLinks( $caption );
4432                         $ig->setCaptionHtml( $caption );
4433                 }
4434                 if( isset( $params['perrow'] ) ) {
4435                         $ig->setPerRow( $params['perrow'] );
4436                 }
4437                 if( isset( $params['widths'] ) ) {
4438                         $ig->setWidths( $params['widths'] );
4439                 }
4440                 if( isset( $params['heights'] ) ) {
4441                         $ig->setHeights( $params['heights'] );
4442                 }
4443
4444                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4445
4446                 $lines = explode( "\n", $text );
4447                 foreach ( $lines as $line ) {
4448                         # match lines like these:
4449                         # Image:someimage.jpg|This is some image
4450                         $matches = array();
4451                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4452                         # Skip empty lines
4453                         if ( count( $matches ) == 0 ) {
4454                                 continue;
4455                         }
4456                         $tp = Title::newFromText( $matches[1] );
4457                         $nt =& $tp;
4458                         if( is_null( $nt ) ) {
4459                                 # Bogus title. Ignore these so we don't bomb out later.
4460                                 continue;
4461                         }
4462                         if ( isset( $matches[3] ) ) {
4463                                 $label = $matches[3];
4464                         } else {
4465                                 $label = '';
4466                         }
4467
4468                         $pout = $this->parse( $label,
4469                                 $this->mTitle,
4470                                 $this->mOptions,
4471                                 false, // Strip whitespace...?
4472                                 false  // Don't clear state!
4473                         );
4474                         $html = $pout->getText();
4475
4476                         $ig->add( $nt, $html );
4477
4478                         # Only add real images (bug #5586)
4479                         if ( $nt->getNamespace() == NS_IMAGE ) {
4480                                 $this->mOutput->addImage( $nt->getDBkey() );
4481                         }
4482                 }
4483                 return $ig->toHTML();
4484         }
4485
4486         function getImageParams( $handler ) {
4487                 if ( $handler ) {
4488                         $handlerClass = get_class( $handler );
4489                 } else {
4490                         $handlerClass = '';
4491                 }
4492                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4493                         // Initialise static lists
4494                         static $internalParamNames = array(
4495                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4496                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4497                                         'bottom', 'text-bottom' ),
4498                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4499                                         'upright', 'border' ),
4500                         );
4501                         static $internalParamMap;
4502                         if ( !$internalParamMap ) {
4503                                 $internalParamMap = array();
4504                                 foreach ( $internalParamNames as $type => $names ) {
4505                                         foreach ( $names as $name ) {
4506                                                 $magicName = str_replace( '-', '_', "img_$name" );
4507                                                 $internalParamMap[$magicName] = array( $type, $name );
4508                                         }
4509                                 }
4510                         }
4511
4512                         // Add handler params
4513                         $paramMap = $internalParamMap;
4514                         if ( $handler ) {
4515                                 $handlerParamMap = $handler->getParamMap();
4516                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4517                                         $paramMap[$magic] = array( 'handler', $paramName );
4518                                 }
4519                         }
4520                         $this->mImageParams[$handlerClass] = $paramMap;
4521                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4522                 }
4523                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4524         }
4525
4526         /**
4527          * Parse image options text and use it to make an image
4528          */
4529         function makeImage( $title, $options ) {
4530                 # @TODO: let the MediaHandler specify its transform parameters
4531                 #
4532                 # Check if the options text is of the form "options|alt text"
4533                 # Options are:
4534                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4535                 #  * left               no resizing, just left align. label is used for alt= only
4536                 #  * right              same, but right aligned
4537                 #  * none               same, but not aligned
4538                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4539                 #  * center             center the image
4540                 #  * framed             Keep original image size, no magnify-button.
4541                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4542                 #  * upright            reduce width for upright images, rounded to full __0 px
4543                 #  * border             draw a 1px border around the image
4544                 # vertical-align values (no % or length right now):
4545                 #  * baseline
4546                 #  * sub
4547                 #  * super
4548                 #  * top
4549                 #  * text-top
4550                 #  * middle
4551                 #  * bottom
4552                 #  * text-bottom
4553
4554                 $parts = array_map( 'trim', explode( '|', $options) );
4555                 $sk = $this->mOptions->getSkin();
4556
4557                 # Give extensions a chance to select the file revision for us
4558                 $skip = $time = false;
4559                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4560
4561                 if ( $skip ) {
4562                         return $sk->makeLinkObj( $title );
4563                 }
4564
4565                 # Get parameter map
4566                 $file = wfFindFile( $title, $time );
4567                 $handler = $file ? $file->getHandler() : false;
4568
4569                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4570
4571                 # Process the input parameters
4572                 $caption = '';
4573                 $params = array( 'frame' => array(), 'handler' => array(),
4574                         'horizAlign' => array(), 'vertAlign' => array() );
4575                 foreach( $parts as $part ) {
4576                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4577                         if ( isset( $paramMap[$magicName] ) ) {
4578                                 list( $type, $paramName ) = $paramMap[$magicName];
4579                                 $params[$type][$paramName] = $value;
4580
4581                                 // Special case; width and height come in one variable together
4582                                 if( $type == 'handler' && $paramName == 'width' ) {
4583                                         $m = array();
4584                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4585                                                 $params[$type]['width'] = intval( $m[1] );
4586                                                 $params[$type]['height'] = intval( $m[2] );
4587                                         } else {
4588                                                 $params[$type]['width'] = intval( $value );
4589                                         }
4590                                 }
4591                         } else {
4592                                 $caption = $part;
4593                         }
4594                 }
4595
4596                 # Process alignment parameters
4597                 if ( $params['horizAlign'] ) {
4598                         $params['frame']['align'] = key( $params['horizAlign'] );
4599                 }
4600                 if ( $params['vertAlign'] ) {
4601                         $params['frame']['valign'] = key( $params['vertAlign'] );
4602                 }
4603
4604                 # Validate the handler parameters
4605                 if ( $handler ) {
4606                         foreach ( $params['handler'] as $name => $value ) {
4607                                 if ( !$handler->validateParam( $name, $value ) ) {
4608                                         unset( $params['handler'][$name] );
4609                                 }
4610                         }
4611                 }
4612
4613                 # Strip bad stuff out of the alt text
4614                 $alt = $this->replaceLinkHoldersText( $caption );
4615
4616                 # make sure there are no placeholders in thumbnail attributes
4617                 # that are later expanded to html- so expand them now and
4618                 # remove the tags
4619                 $alt = $this->mStripState->unstripBoth( $alt );
4620                 $alt = Sanitizer::stripAllTags( $alt );
4621
4622                 $params['frame']['alt'] = $alt;
4623                 $params['frame']['caption'] = $caption;
4624
4625                 # Linker does the rest
4626                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4627
4628                 # Give the handler a chance to modify the parser object
4629                 if ( $handler ) {
4630                         $handler->parserTransformHook( $this, $file );
4631                 }
4632
4633                 return $ret;
4634         }
4635
4636         /**
4637          * Set a flag in the output object indicating that the content is dynamic and
4638          * shouldn't be cached.
4639          */
4640         function disableCache() {
4641                 wfDebug( "Parser output marked as uncacheable.\n" );
4642                 $this->mOutput->mCacheTime = -1;
4643         }
4644
4645         /**#@+
4646          * Callback from the Sanitizer for expanding items found in HTML attribute
4647          * values, so they can be safely tested and escaped.
4648          * @param string $text
4649          * @param array $args
4650          * @return string
4651          * @private
4652          */
4653         function attributeStripCallback( &$text, $args ) {
4654                 $text = $this->replaceVariables( $text, $args );
4655                 $text = $this->mStripState->unstripBoth( $text );
4656                 return $text;
4657         }
4658
4659         /**#@-*/
4660
4661         /**#@+
4662          * Accessor/mutator
4663          */
4664         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4665         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4666         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4667         /**#@-*/
4668
4669         /**#@+
4670          * Accessor
4671          */
4672         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4673         /**#@-*/
4674
4675
4676         /**
4677          * Break wikitext input into sections, and either pull or replace
4678          * some particular section's text.
4679          *
4680          * External callers should use the getSection and replaceSection methods.
4681          *
4682          * @param $text Page wikitext
4683          * @param $section Numbered section. 0 pulls the text before the first
4684          *                 heading; other numbers will pull the given section
4685          *                 along with its lower-level subsections.
4686          * @param $mode One of "get" or "replace"
4687          * @param $newtext Replacement text for section data.
4688          * @return string for "get", the extracted section text.
4689          *                for "replace", the whole page with the section replaced.
4690          */
4691         private function extractSections( $text, $section, $mode, $newtext='' ) {
4692                 # I.... _hope_ this is right.
4693                 # Otherwise, sometimes we don't have things initialized properly.
4694                 $this->clearState();
4695
4696                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4697                 # comments to be stripped as well)
4698                 $stripState = new StripState;
4699
4700                 $oldOutputType = $this->mOutputType;
4701                 $oldOptions = $this->mOptions;
4702                 $this->mOptions = new ParserOptions();
4703                 $this->setOutputType( OT_WIKI );
4704
4705                 $striptext = $this->strip( $text, $stripState, true );
4706
4707                 $this->setOutputType( $oldOutputType );
4708                 $this->mOptions = $oldOptions;
4709
4710                 # now that we can be sure that no pseudo-sections are in the source,
4711                 # split it up by section
4712                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4713                 $comment = "(?:$uniq-!--.*?QINU\x07)";
4714                 $secs = preg_split(
4715                         "/
4716                         (
4717                                 ^
4718                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4719                                 (=+) # Should this be limited to 6?
4720                                 .+?  # Section title...
4721                                 \\2  # Ending = count must match start
4722                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4723                                 $
4724                         |
4725                                 <h([1-6])\b.*?>
4726                                 .*?
4727                                 <\/h\\3\s*>
4728                         )
4729                         /mix",
4730                         $striptext, -1,
4731                         PREG_SPLIT_DELIM_CAPTURE);
4732
4733                 if( $mode == "get" ) {
4734                         if( $section == 0 ) {
4735                                 // "Section 0" returns the content before any other section.
4736                                 $rv = $secs[0];
4737                         } else {
4738                                 //track missing section, will replace if found.
4739                                 $rv = $newtext;
4740                         }
4741                 } elseif( $mode == "replace" ) {
4742                         if( $section == 0 ) {
4743                                 $rv = $newtext . "\n\n";
4744                                 $remainder = true;
4745                         } else {
4746                                 $rv = $secs[0];
4747                                 $remainder = false;
4748                         }
4749                 }
4750                 $count = 0;
4751                 $sectionLevel = 0;
4752                 for( $index = 1; $index < count( $secs ); ) {
4753                         $headerLine = $secs[$index++];
4754                         if( $secs[$index] ) {
4755                                 // A wiki header
4756                                 $headerLevel = strlen( $secs[$index++] );
4757                         } else {
4758                                 // An HTML header
4759                                 $index++;
4760                                 $headerLevel = intval( $secs[$index++] );
4761                         }
4762                         $content = $secs[$index++];
4763
4764                         $count++;
4765                         if( $mode == "get" ) {
4766                                 if( $count == $section ) {
4767                                         $rv = $headerLine . $content;
4768                                         $sectionLevel = $headerLevel;
4769                                 } elseif( $count > $section ) {
4770                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4771                                                 $rv .= $headerLine . $content;
4772                                         } else {
4773                                                 // Broke out to a higher-level section
4774                                                 break;
4775                                         }
4776                                 }
4777                         } elseif( $mode == "replace" ) {
4778                                 if( $count < $section ) {
4779                                         $rv .= $headerLine . $content;
4780                                 } elseif( $count == $section ) {
4781                                         $rv .= $newtext . "\n\n";
4782                                         $sectionLevel = $headerLevel;
4783                                 } elseif( $count > $section ) {
4784                                         if( $headerLevel <= $sectionLevel ) {
4785                                                 // Passed the section's sub-parts.
4786                                                 $remainder = true;
4787                                         }
4788                                         if( $remainder ) {
4789                                                 $rv .= $headerLine . $content;
4790                                         }
4791                                 }
4792                         }
4793                 }
4794                 if (is_string($rv))
4795                         # reinsert stripped tags
4796                         $rv = trim( $stripState->unstripBoth( $rv ) );
4797
4798                 return $rv;
4799         }
4800
4801         /**
4802          * This function returns the text of a section, specified by a number ($section).
4803          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4804          * the first section before any such heading (section 0).
4805          *
4806          * If a section contains subsections, these are also returned.
4807          *
4808          * @param $text String: text to look in
4809          * @param $section Integer: section number
4810          * @param $deftext: default to return if section is not found
4811          * @return string text of the requested section
4812          */
4813         public function getSection( $text, $section, $deftext='' ) {
4814                 return $this->extractSections( $text, $section, "get", $deftext );
4815         }
4816
4817         public function replaceSection( $oldtext, $section, $text ) {
4818                 return $this->extractSections( $oldtext, $section, "replace", $text );
4819         }
4820
4821         /**
4822          * Get the timestamp associated with the current revision, adjusted for
4823          * the default server-local timestamp
4824          */
4825         function getRevisionTimestamp() {
4826                 if ( is_null( $this->mRevisionTimestamp ) ) {
4827                         wfProfileIn( __METHOD__ );
4828                         global $wgContLang;
4829                         $dbr = wfGetDB( DB_SLAVE );
4830                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4831                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4832
4833                         // Normalize timestamp to internal MW format for timezone processing.
4834                         // This has the added side-effect of replacing a null value with
4835                         // the current time, which gives us more sensible behavior for
4836                         // previews.
4837                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4838
4839                         // The cryptic '' timezone parameter tells to use the site-default
4840                         // timezone offset instead of the user settings.
4841                         //
4842                         // Since this value will be saved into the parser cache, served
4843                         // to other users, and potentially even used inside links and such,
4844                         // it needs to be consistent for all visitors.
4845                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4846
4847                         wfProfileOut( __METHOD__ );
4848                 }
4849                 return $this->mRevisionTimestamp;
4850         }
4851
4852         /**
4853          * Mutator for $mDefaultSort
4854          *
4855          * @param $sort New value
4856          */
4857         public function setDefaultSort( $sort ) {
4858                 $this->mDefaultSort = $sort;
4859         }
4860
4861         /**
4862          * Accessor for $mDefaultSort
4863          * Will use the title/prefixed title if none is set
4864          *
4865          * @return string
4866          */
4867         public function getDefaultSort() {
4868                 if( $this->mDefaultSort !== false ) {
4869                         return $this->mDefaultSort;
4870                 } else {
4871                         return $this->mTitle->getNamespace() == NS_CATEGORY
4872                                         ? $this->mTitle->getText()
4873                                         : $this->mTitle->getPrefixedText();
4874                 }
4875         }
4876
4877         /**
4878          * Try to guess the section anchor name based on a wikitext fragment
4879          * presumably extracted from a heading, for example "Header" from
4880          * "== Header ==".
4881          */
4882         public function guessSectionNameFromWikiText( $text ) {
4883                 # Strip out wikitext links(they break the anchor)
4884                 $text = $this->stripSectionName( $text );
4885                 $headline = Sanitizer::decodeCharReferences( $text );
4886                 # strip out HTML
4887                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4888                 $headline = trim( $headline );
4889                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4890                 $replacearray = array(
4891                         '%3A' => ':',
4892                         '%' => '.'
4893                 );
4894                 return str_replace(
4895                         array_keys( $replacearray ),
4896                         array_values( $replacearray ),
4897                         $sectionanchor );
4898         }
4899
4900         /**
4901          * Strips a text string of wikitext for use in a section anchor
4902          *
4903          * Accepts a text string and then removes all wikitext from the
4904          * string and leaves only the resultant text (i.e. the result of
4905          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4906          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4907          * to create valid section anchors by mimicing the output of the
4908          * parser when headings are parsed.
4909          *
4910          * @param $text string Text string to be stripped of wikitext
4911          * for use in a Section anchor
4912          * @return Filtered text string
4913          */
4914         public function stripSectionName( $text ) {
4915                 # Strip internal link markup
4916                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4917                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4918
4919                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4920                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4921                 # on how many empty links there are on the page - need to figure that out.
4922                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4923
4924                 # Parse wikitext quotes (italics & bold)
4925                 $text = $this->doQuotes($text);
4926
4927                 # Strip HTML tags
4928                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4929                 return $text;
4930         }
4931 }
4932
4933 /**
4934  * @todo document, briefly.
4935  * @addtogroup Parser
4936  */
4937 class OnlyIncludeReplacer {
4938         var $output = '';
4939
4940         function replace( $matches ) {
4941                 if ( substr( $matches[1], -1 ) == "\n" ) {
4942                         $this->output .= substr( $matches[1], 0, -1 );
4943                 } else {
4944                         $this->output .= $matches[1];
4945                 }
4946         }
4947 }
4948
4949 /**
4950  * @todo document, briefly.
4951  * @addtogroup Parser
4952  */
4953 class StripState {
4954         var $general, $nowiki;
4955
4956         function __construct() {
4957                 $this->general = new ReplacementArray;
4958                 $this->nowiki = new ReplacementArray;
4959         }
4960
4961         function unstripGeneral( $text ) {
4962                 wfProfileIn( __METHOD__ );
4963                 $text = $this->general->replace( $text );
4964                 wfProfileOut( __METHOD__ );
4965                 return $text;
4966         }
4967
4968         function unstripNoWiki( $text ) {
4969                 wfProfileIn( __METHOD__ );
4970                 $text = $this->nowiki->replace( $text );
4971                 wfProfileOut( __METHOD__ );
4972                 return $text;
4973         }
4974
4975         function unstripBoth( $text ) {
4976                 wfProfileIn( __METHOD__ );
4977                 $text = $this->general->replace( $text );
4978                 $text = $this->nowiki->replace( $text );
4979                 wfProfileOut( __METHOD__ );
4980                 return $text;
4981         }
4982 }