includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are four main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * transformMsg()
  24  *   performs brace substitution on MediaWiki messages
  25  * preprocess()
  26  *   removes HTML comments and expands templates
  27  *
  28  * Globals used:
  29  *    objects:   $wgLang, $wgContLang
  30  *
  31  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  32  *
  33  * settings:
  34  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  35  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  36  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  37  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.2';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  62                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  63
  64         // State constants for the definition list colon extraction
  65         const COLON_STATE_TEXT = 0;
  66         const COLON_STATE_TAG = 1;
  67         const COLON_STATE_TAGSTART = 2;
  68         const COLON_STATE_CLOSETAG = 3;
  69         const COLON_STATE_TAGSLASH = 4;
  70         const COLON_STATE_COMMENT = 5;
  71         const COLON_STATE_COMMENTDASH = 6;
  72         const COLON_STATE_COMMENTDASHDASH = 7;
  73
  74         /**#@+
  75          * @private
  76          */
  77         # Persistent:
  78         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  79                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  80                 $mExtLinkBracketedRegex;
  81
  82         # Cleared with clearState():
  83         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  84         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  85         var $mInterwikiLinkHolders, $mLinkHolders;
  86         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  87         var $mTplExpandCache,// empty-frame expansion cache
  88             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  89                                 // in this path. Used for loop detection.
  90         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  91
  92         # Temporary
  93         # These are variables reset at least once per parse regardless of $clearState
  94         var $mOptions,      // ParserOptions object
  95                 $mTitle,        // Title context, used for self-link rendering and similar things
  96                 $mOutputType,   // Output type, one of the OT_xxx constants
  97                 $ot,            // Shortcut alias, see setOutputType()
  98                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  99                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 100                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 101
 102         /**#@-*/
 103
 104         /**
 105          * Constructor
 106          *
 107          * @public
 108          */
 109         function __construct( $conf = array() ) {
 110                 $this->mTagHooks = array();
 111                 $this->mTransparentTagHooks = array();
 112                 $this->mFunctionHooks = array();
 113                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 114                 $this->mStripList = array( 'nowiki', 'gallery' );
 115                 $this->mMarkerSuffix = "-QINU\x7f";
 116                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 117                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 118                 $this->mFirstCall = true;
 119         }
 120
 121         /**
 122          * Do various kinds of initialisation on the first call of the parser
 123          */
 124         function firstCallInit() {
 125                 if ( !$this->mFirstCall ) {
 126                         return;
 127                 }
 128
 129                 wfProfileIn( __METHOD__ );
 130                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 131
 132                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 133
 134                 # Syntax for arguments (see self::setFunctionHook):
 135                 #  "name for lookup in localized magic words array",
 136                 #  function callback,
 137                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 138                 #    instead of {{#int:...}})
 139                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 140                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 141                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 142                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 164                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 166
 167                 if ( $wgAllowDisplayTitle ) {
 168                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 169                 }
 170                 if ( $wgAllowSlowParserFunctions ) {
 171                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 172                 }
 173
 174                 $this->initialiseVariables();
 175                 $this->mFirstCall = false;
 176                 wfProfileOut( __METHOD__ );
 177         }
 178
 179         /**
 180          * Clear Parser state
 181          *
 182          * @private
 183          */
 184         function clearState() {
 185                 wfProfileIn( __METHOD__ );
 186                 if ( $this->mFirstCall ) {
 187                         $this->firstCallInit();
 188                 }
 189                 $this->mOutput = new ParserOutput;
 190                 $this->mAutonumber = 0;
 191                 $this->mLastSection = '';
 192                 $this->mDTopen = false;
 193                 $this->mIncludeCount = array();
 194                 $this->mStripState = new StripState;
 195                 $this->mArgStack = false;
 196                 $this->mInPre = false;
 197                 $this->mInterwikiLinkHolders = array(
 198                         'texts' => array(),
 199                         'titles' => array()
 200                 );
 201                 $this->mLinkHolders = array(
 202                         'namespaces' => array(),
 203                         'dbkeys' => array(),
 204                         'queries' => array(),
 205                         'texts' => array(),
 206                         'titles' => array()
 207                 );
 208                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 209
 210                 /**
 211                  * Prefix for temporary replacement strings for the multipass parser.
 212                  * \x07 should never appear in input as it's disallowed in XML.
 213                  * Using it at the front also gives us a little extra robustness
 214                  * since it shouldn't match when butted up against identifier-like
 215                  * string constructs.
 216                  *
 217                  * Must not consist of all title characters, or else it will change
 218                  * the behaviour of <nowiki> in a link.
 219                  */
 220                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 221                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTemplatePath = array();
 225                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'post-expand' => 0,
 231                         'arg' => 0,
 232                 );
 233                 $this->mPPNodeCount = 0;
 234                 $this->mDefaultSort = false;
 235                 $this->mHeadings = array();
 236
 237                 wfRunHooks( 'ParserClearState', array( &$this ) );
 238                 wfProfileOut( __METHOD__ );
 239         }
 240
 241         function setOutputType( $ot ) {
 242                 $this->mOutputType = $ot;
 243                 // Shortcut alias
 244                 $this->ot = array(
 245                         'html' => $ot == OT_HTML,
 246                         'wiki' => $ot == OT_WIKI,
 247                         'msg' => $ot == OT_MSG,
 248                         'pre' => $ot == OT_PREPROCESS,
 249                 );
 250         }
 251
 252         /**
 253          * Accessor for mUniqPrefix.
 254          *
 255          * @public
 256          */
 257         function uniqPrefix() {
 258                 return $this->mUniqPrefix;
 259         }
 260
 261         /**
 262          * Convert wikitext to HTML
 263          * Do not call this function recursively.
 264          *
 265          * @param string $text Text we want to parse
 266          * @param Title &$title A title object
 267          * @param array $options
 268          * @param boolean $linestart
 269          * @param boolean $clearState
 270          * @param int $revid number to pass in {{REVISIONID}}
 271          * @return ParserOutput a ParserOutput
 272          */
 273         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 274                 /**
 275                  * First pass--just handle <nowiki> sections, pass the rest off
 276                  * to internalParse() which does all the real work.
 277                  */
 278
 279                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 280                 $fname = 'Parser::parse-' . wfGetCaller();
 281                 wfProfileIn( __METHOD__ );
 282                 wfProfileIn( $fname );
 283
 284                 if ( $clearState ) {
 285                         $this->clearState();
 286                 }
 287
 288                 $this->mOptions = $options;
 289                 $this->mTitle =& $title;
 290                 $oldRevisionId = $this->mRevisionId;
 291                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 292                 if( $revid !== null ) {
 293                         $this->mRevisionId = $revid;
 294                         $this->mRevisionTimestamp = null;
 295                 }
 296                 $this->setOutputType( OT_HTML );
 297                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 298                 # No more strip!
 299                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 300                 $text = $this->internalParse( $text );
 301                 $text = $this->mStripState->unstripGeneral( $text );
 302
 303                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 304                 $fixtags = array(
 305                         # french spaces, last one Guillemet-left
 306                         # only if there is something before the space
 307                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 308                         # french spaces, Guillemet-right
 309                         '/(\\302\\253) /' => '\\1&nbsp;',
 310                 );
 311                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 312
 313                 # only once and last
 314                 $text = $this->doBlockLevels( $text, $linestart );
 315
 316                 $this->replaceLinkHolders( $text );
 317
 318                 # the position of the parserConvert() call should not be changed. it
 319                 # assumes that the links are all replaced and the only thing left
 320                 # is the <nowiki> mark.
 321                 # Side-effects: this calls $this->mOutput->setTitleText()
 322                 $text = $wgContLang->parserConvert( $text, $this );
 323
 324                 $text = $this->mStripState->unstripNoWiki( $text );
 325
 326                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 327
 328 //!JF Move to its own function
 329
 330                 $uniq_prefix = $this->mUniqPrefix;
 331                 $matches = array();
 332                 $elements = array_keys( $this->mTransparentTagHooks );
 333                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 334
 335                 foreach( $matches as $marker => $data ) {
 336                         list( $element, $content, $params, $tag ) = $data;
 337                         $tagName = strtolower( $element );
 338                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 339                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 340                                         array( $content, $params, $this ) );
 341                         } else {
 342                                 $output = $tag;
 343                         }
 344                         $this->mStripState->general->setPair( $marker, $output );
 345                 }
 346                 $text = $this->mStripState->unstripGeneral( $text );
 347
 348                 $text = Sanitizer::normalizeCharReferences( $text );
 349
 350                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 351                         $text = Parser::tidy($text);
 352                 } else {
 353                         # attempt to sanitize at least some nesting problems
 354                         # (bug #2702 and quite a few others)
 355                         $tidyregs = array(
 356                                 # ''Something [http://www.cool.com cool''] -->
 357                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 358                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 359                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 360                                 # fix up an anchor inside another anchor, only
 361                                 # at least for a single single nested link (bug 3695)
 362                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 363                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 364                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 365                                 # contains a div, so fix it up here; replace
 366                                 # div with escaped text
 367                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 368                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 369                                 # remove empty italic or bold tag pairs, some
 370                                 # introduced by rules above
 371                                 '/<([bi])><\/\\1>/' => '',
 372                         );
 373
 374                         $text = preg_replace(
 375                                 array_keys( $tidyregs ),
 376                                 array_values( $tidyregs ),
 377                                 $text );
 378                 }
 379
 380                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 381
 382                 # Information on include size limits, for the benefit of users who try to skirt them
 383                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 384                         $max = $this->mOptions->getMaxIncludeSize();
 385                         $text .= "<!-- \n" .
 386                                 "Preprocessor node count: {$this->mPPNodeCount}\n" .
 387                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 388                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 389                                 "Maximum: $max bytes\n" .
 390                                 "-->\n";
 391                 }
 392                 $this->mOutput->setText( $text );
 393                 $this->mRevisionId = $oldRevisionId;
 394                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 395                 wfProfileOut( $fname );
 396                 wfProfileOut( __METHOD__ );
 397
 398                 return $this->mOutput;
 399         }
 400
 401         /**
 402          * Recursive parser entry point that can be called from an extension tag
 403          * hook.
 404          */
 405         function recursiveTagParse( $text ) {
 406                 wfProfileIn( __METHOD__ );
 407                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 408                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 409                 $text = $this->internalParse( $text );
 410                 wfProfileOut( __METHOD__ );
 411                 return $text;
 412         }
 413
 414         /**
 415          * Expand templates and variables in the text, producing valid, static wikitext.
 416          * Also removes comments.
 417          */
 418         function preprocess( $text, $title, $options, $revid = null ) {
 419                 wfProfileIn( __METHOD__ );
 420                 $this->clearState();
 421                 $this->setOutputType( OT_PREPROCESS );
 422                 $this->mOptions = $options;
 423                 $this->mTitle = $title;
 424                 if( $revid !== null ) {
 425                         $this->mRevisionId = $revid;
 426                 }
 427                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 428                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 429                 $text = $this->replaceVariables( $text );
 430                 if ( $this->mOptions->getRemoveComments() ) {
 431                         $text = Sanitizer::removeHTMLcomments( $text );
 432                 }
 433                 $text = $this->mStripState->unstripBoth( $text );
 434                 wfProfileOut( __METHOD__ );
 435                 return $text;
 436         }
 437
 438         /**
 439          * Get a random string
 440          *
 441          * @private
 442          * @static
 443          */
 444         function getRandomString() {
 445                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 446         }
 447
 448         function &getTitle() { return $this->mTitle; }
 449         function getOptions() { return $this->mOptions; }
 450
 451         function getFunctionLang() {
 452                 global $wgLang, $wgContLang;
 453                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 454         }
 455
 456         /**
 457          * Replaces all occurrences of HTML-style comments and the given tags
 458          * in the text with a random marker and returns teh next text. The output
 459          * parameter $matches will be an associative array filled with data in
 460          * the form:
 461          *   'UNIQ-xxxxx' => array(
 462          *     'element',
 463          *     'tag content',
 464          *     array( 'param' => 'x' ),
 465          *     '<element param="x">tag content</element>' ) )
 466          *
 467          * @param $elements list of element names. Comments are always extracted.
 468          * @param $text Source text string.
 469          * @param $uniq_prefix
 470          *
 471          * @public
 472          * @static
 473          */
 474         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 475                 static $n = 1;
 476                 $stripped = '';
 477                 $matches = array();
 478
 479                 $taglist = implode( '|', $elements );
 480                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 481
 482                 while ( '' != $text ) {
 483                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 484                         $stripped .= $p[0];
 485                         if( count( $p ) < 5 ) {
 486                                 break;
 487                         }
 488                         if( count( $p ) > 5 ) {
 489                                 // comment
 490                                 $element    = $p[4];
 491                                 $attributes = '';
 492                                 $close      = '';
 493                                 $inside     = $p[5];
 494                         } else {
 495                                 // tag
 496                                 $element    = $p[1];
 497                                 $attributes = $p[2];
 498                                 $close      = $p[3];
 499                                 $inside     = $p[4];
 500                         }
 501
 502                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 503                         $stripped .= $marker;
 504
 505                         if ( $close === '/>' ) {
 506                                 // Empty element tag, <tag />
 507                                 $content = null;
 508                                 $text = $inside;
 509                                 $tail = null;
 510                         } else {
 511                                 if( $element == '!--' ) {
 512                                         $end = '/(-->)/';
 513                                 } else {
 514                                         $end = "/(<\\/$element\\s*>)/i";
 515                                 }
 516                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 517                                 $content = $q[0];
 518                                 if( count( $q ) < 3 ) {
 519                                         # No end tag -- let it run out to the end of the text.
 520                                         $tail = '';
 521                                         $text = '';
 522                                 } else {
 523                                         $tail = $q[1];
 524                                         $text = $q[2];
 525                                 }
 526                         }
 527
 528                         $matches[$marker] = array( $element,
 529                                 $content,
 530                                 Sanitizer::decodeTagAttributes( $attributes ),
 531                                 "<$element$attributes$close$content$tail" );
 532                 }
 533                 return $stripped;
 534         }
 535
 536         /**
 537          * Get a list of strippable XML-like elements
 538          */
 539         function getStripList() {
 540                 global $wgRawHtml;
 541                 $elements = $this->mStripList;
 542                 if( $wgRawHtml ) {
 543                         $elements[] = 'html';
 544                 }
 545                 if( $this->mOptions->getUseTeX() ) {
 546                         $elements[] = 'math';
 547                 }
 548                 return $elements;
 549         }
 550
 551         /**
 552          * @deprecated use replaceVariables
 553          */
 554         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 555                 return $text;
 556         }
 557
 558         /**
 559          * Restores pre, math, and other extensions removed by strip()
 560          *
 561          * always call unstripNoWiki() after this one
 562          * @private
 563          * @deprecated use $this->mStripState->unstrip()
 564          */
 565         function unstrip( $text, $state ) {
 566                 return $state->unstripGeneral( $text );
 567         }
 568
 569         /**
 570          * Always call this after unstrip() to preserve the order
 571          *
 572          * @private
 573          * @deprecated use $this->mStripState->unstrip()
 574          */
 575         function unstripNoWiki( $text, $state ) {
 576                 return $state->unstripNoWiki( $text );
 577         }
 578
 579         /**
 580          * @deprecated use $this->mStripState->unstripBoth()
 581          */
 582         function unstripForHTML( $text ) {
 583                 return $this->mStripState->unstripBoth( $text );
 584         }
 585
 586         /**
 587          * Add an item to the strip state
 588          * Returns the unique tag which must be inserted into the stripped text
 589          * The tag will be replaced with the original text in unstrip()
 590          *
 591          * @private
 592          */
 593         function insertStripItem( $text ) {
 594                 static $n = 0;
 595                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 596                 $this->mStripState->general->setPair( $rnd, $text );
 597                 return $rnd;
 598         }
 599
 600         /**
 601          * Interface with html tidy, used if $wgUseTidy = true.
 602          * If tidy isn't able to correct the markup, the original will be
 603          * returned in all its glory with a warning comment appended.
 604          *
 605          * Either the external tidy program or the in-process tidy extension
 606          * will be used depending on availability. Override the default
 607          * $wgTidyInternal setting to disable the internal if it's not working.
 608          *
 609          * @param string $text Hideous HTML input
 610          * @return string Corrected HTML output
 611          * @public
 612          * @static
 613          */
 614         function tidy( $text ) {
 615                 global $wgTidyInternal;
 616                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 617 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 618 '<head><title>test</title></head><body>'.$text.'</body></html>';
 619                 if( $wgTidyInternal ) {
 620                         $correctedtext = Parser::internalTidy( $wrappedtext );
 621                 } else {
 622                         $correctedtext = Parser::externalTidy( $wrappedtext );
 623                 }
 624                 if( is_null( $correctedtext ) ) {
 625                         wfDebug( "Tidy error detected!\n" );
 626                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 627                 }
 628                 return $correctedtext;
 629         }
 630
 631         /**
 632          * Spawn an external HTML tidy process and get corrected markup back from it.
 633          *
 634          * @private
 635          * @static
 636          */
 637         function externalTidy( $text ) {
 638                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 639                 $fname = 'Parser::externalTidy';
 640                 wfProfileIn( $fname );
 641
 642                 $cleansource = '';
 643                 $opts = ' -utf8';
 644
 645                 $descriptorspec = array(
 646                         0 => array('pipe', 'r'),
 647                         1 => array('pipe', 'w'),
 648                         2 => array('file', wfGetNull(), 'a')
 649                 );
 650                 $pipes = array();
 651                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 652                 if (is_resource($process)) {
 653                         // Theoretically, this style of communication could cause a deadlock
 654                         // here. If the stdout buffer fills up, then writes to stdin could
 655                         // block. This doesn't appear to happen with tidy, because tidy only
 656                         // writes to stdout after it's finished reading from stdin. Search
 657                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 658                         fwrite($pipes[0], $text);
 659                         fclose($pipes[0]);
 660                         while (!feof($pipes[1])) {
 661                                 $cleansource .= fgets($pipes[1], 1024);
 662                         }
 663                         fclose($pipes[1]);
 664                         proc_close($process);
 665                 }
 666
 667                 wfProfileOut( $fname );
 668
 669                 if( $cleansource == '' && $text != '') {
 670                         // Some kind of error happened, so we couldn't get the corrected text.
 671                         // Just give up; we'll use the source text and append a warning.
 672                         return null;
 673                 } else {
 674                         return $cleansource;
 675                 }
 676         }
 677
 678         /**
 679          * Use the HTML tidy PECL extension to use the tidy library in-process,
 680          * saving the overhead of spawning a new process.
 681          *
 682          * 'pear install tidy' should be able to compile the extension module.
 683          *
 684          * @private
 685          * @static
 686          */
 687         function internalTidy( $text ) {
 688                 global $wgTidyConf, $IP;
 689                 $fname = 'Parser::internalTidy';
 690                 wfProfileIn( $fname );
 691
 692                 $tidy = new tidy;
 693                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 694                 $tidy->cleanRepair();
 695                 if( $tidy->getStatus() == 2 ) {
 696                         // 2 is magic number for fatal error
 697                         // http://www.php.net/manual/en/function.tidy-get-status.php
 698                         $cleansource = null;
 699                 } else {
 700                         $cleansource = tidy_get_output( $tidy );
 701                 }
 702                 wfProfileOut( $fname );
 703                 return $cleansource;
 704         }
 705
 706         /**
 707          * parse the wiki syntax used to render tables
 708          *
 709          * @private
 710          */
 711         function doTableStuff ( $text ) {
 712                 $fname = 'Parser::doTableStuff';
 713                 wfProfileIn( $fname );
 714
 715                 $lines = explode ( "\n" , $text );
 716                 $td_history = array (); // Is currently a td tag open?
 717                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 718                 $tr_history = array (); // Is currently a tr tag open?
 719                 $tr_attributes = array (); // history of tr attributes
 720                 $has_opened_tr = array(); // Did this table open a <tr> element?
 721                 $indent_level = 0; // indent level of the table
 722                 foreach ( $lines as $key => $line )
 723                 {
 724                         $line = trim ( $line );
 725
 726                         if( $line == '' ) { // empty line, go to next line
 727                                 continue;
 728                         }
 729                         $first_character = $line{0};
 730                         $matches = array();
 731
 732                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 733                                 // First check if we are starting a new table
 734                                 $indent_level = strlen( $matches[1] );
 735
 736                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 737                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 738
 739                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 740                                 array_push ( $td_history , false );
 741                                 array_push ( $last_tag_history , '' );
 742                                 array_push ( $tr_history , false );
 743                                 array_push ( $tr_attributes , '' );
 744                                 array_push ( $has_opened_tr , false );
 745                         } else if ( count ( $td_history ) == 0 ) {
 746                                 // Don't do any of the following
 747                                 continue;
 748                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 749                                 // We are ending a table
 750                                 $line = '</table>' . substr ( $line , 2 );
 751                                 $last_tag = array_pop ( $last_tag_history );
 752
 753                                 if ( !array_pop ( $has_opened_tr ) ) {
 754                                         $line = "<tr><td></td></tr>{$line}";
 755                                 }
 756
 757                                 if ( array_pop ( $tr_history ) ) {
 758                                         $line = "</tr>{$line}";
 759                                 }
 760
 761                                 if ( array_pop ( $td_history ) ) {
 762                                         $line = "</{$last_tag}>{$line}";
 763                                 }
 764                                 array_pop ( $tr_attributes );
 765                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 766                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 767                                 // Now we have a table row
 768                                 $line = preg_replace( '#^\|-+#', '', $line );
 769
 770                                 // Whats after the tag is now only attributes
 771                                 $attributes = $this->mStripState->unstripBoth( $line );
 772                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 773                                 array_pop ( $tr_attributes );
 774                                 array_push ( $tr_attributes , $attributes );
 775
 776                                 $line = '';
 777                                 $last_tag = array_pop ( $last_tag_history );
 778                                 array_pop ( $has_opened_tr );
 779                                 array_push ( $has_opened_tr , true );
 780
 781                                 if ( array_pop ( $tr_history ) ) {
 782                                         $line = '</tr>';
 783                                 }
 784
 785                                 if ( array_pop ( $td_history ) ) {
 786                                         $line = "</{$last_tag}>{$line}";
 787                                 }
 788
 789                                 $lines[$key] = $line;
 790                                 array_push ( $tr_history , false );
 791                                 array_push ( $td_history , false );
 792                                 array_push ( $last_tag_history , '' );
 793                         }
 794                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 795                                 // This might be cell elements, td, th or captions
 796                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 797                                         $first_character = '+';
 798                                         $line = substr ( $line , 1 );
 799                                 }
 800
 801                                 $line = substr ( $line , 1 );
 802
 803                                 if ( $first_character == '!' ) {
 804                                         $line = str_replace ( '!!' , '||' , $line );
 805                                 }
 806
 807                                 // Split up multiple cells on the same line.
 808                                 // FIXME : This can result in improper nesting of tags processed
 809                                 // by earlier parser steps, but should avoid splitting up eg
 810                                 // attribute values containing literal "||".
 811                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 812
 813                                 $lines[$key] = '';
 814
 815                                 // Loop through each table cell
 816                                 foreach ( $cells as $cell )
 817                                 {
 818                                         $previous = '';
 819                                         if ( $first_character != '+' )
 820                                         {
 821                                                 $tr_after = array_pop ( $tr_attributes );
 822                                                 if ( !array_pop ( $tr_history ) ) {
 823                                                         $previous = "<tr{$tr_after}>\n";
 824                                                 }
 825                                                 array_push ( $tr_history , true );
 826                                                 array_push ( $tr_attributes , '' );
 827                                                 array_pop ( $has_opened_tr );
 828                                                 array_push ( $has_opened_tr , true );
 829                                         }
 830
 831                                         $last_tag = array_pop ( $last_tag_history );
 832
 833                                         if ( array_pop ( $td_history ) ) {
 834                                                 $previous = "</{$last_tag}>{$previous}";
 835                                         }
 836
 837                                         if ( $first_character == '|' ) {
 838                                                 $last_tag = 'td';
 839                                         } else if ( $first_character == '!' ) {
 840                                                 $last_tag = 'th';
 841                                         } else if ( $first_character == '+' ) {
 842                                                 $last_tag = 'caption';
 843                                         } else {
 844                                                 $last_tag = '';
 845                                         }
 846
 847                                         array_push ( $last_tag_history , $last_tag );
 848
 849                                         // A cell could contain both parameters and data
 850                                         $cell_data = explode ( '|' , $cell , 2 );
 851
 852                                         // Bug 553: Note that a '|' inside an invalid link should not
 853                                         // be mistaken as delimiting cell parameters
 854                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 855                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 856                                         } else if ( count ( $cell_data ) == 1 )
 857                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 858                                         else {
 859                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 860                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 861                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 862                                         }
 863
 864                                         $lines[$key] .= $cell;
 865                                         array_push ( $td_history , true );
 866                                 }
 867                         }
 868                 }
 869
 870                 // Closing open td, tr && table
 871                 while ( count ( $td_history ) > 0 )
 872                 {
 873                         if ( array_pop ( $td_history ) ) {
 874                                 $lines[] = '</td>' ;
 875                         }
 876                         if ( array_pop ( $tr_history ) ) {
 877                                 $lines[] = '</tr>' ;
 878                         }
 879                         if ( !array_pop ( $has_opened_tr ) ) {
 880                                 $lines[] = "<tr><td></td></tr>" ;
 881                         }
 882
 883                         $lines[] = '</table>' ;
 884                 }
 885
 886                 $output = implode ( "\n" , $lines ) ;
 887
 888                 // special case: don't return empty table
 889                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 890                         $output = '';
 891                 }
 892
 893                 wfProfileOut( $fname );
 894
 895                 return $output;
 896         }
 897
 898         /**
 899          * Helper function for parse() that transforms wiki markup into
 900          * HTML. Only called for $mOutputType == OT_HTML.
 901          *
 902          * @private
 903          */
 904         function internalParse( $text ) {
 905                 $isMain = true;
 906                 $fname = 'Parser::internalParse';
 907                 wfProfileIn( $fname );
 908
 909                 # Hook to suspend the parser in this state
 910                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 911                         wfProfileOut( $fname );
 912                         return $text ;
 913                 }
 914
 915                 # Remove <noinclude> tags and <includeonly> sections
 916                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 917                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 918                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 919
 920                 $text = $this->replaceVariables( $text );
 921                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 922                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 923
 924                 // Tables need to come after variable replacement for things to work
 925                 // properly; putting them before other transformations should keep
 926                 // exciting things like link expansions from showing up in surprising
 927                 // places.
 928                 $text = $this->doTableStuff( $text );
 929
 930                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 931
 932                 $text = $this->stripToc( $text );
 933                 $this->stripNoGallery( $text );
 934                 $text = $this->doHeadings( $text );
 935                 if($this->mOptions->getUseDynamicDates()) {
 936                         $df =& DateFormatter::getInstance();
 937                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 938                 }
 939                 $text = $this->doAllQuotes( $text );
 940                 $text = $this->replaceInternalLinks( $text );
 941                 $text = $this->replaceExternalLinks( $text );
 942
 943                 # replaceInternalLinks may sometimes leave behind
 944                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 945                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 946
 947                 $text = $this->doMagicLinks( $text );
 948                 $text = $this->formatHeadings( $text, $isMain );
 949
 950                 wfProfileOut( $fname );
 951                 return $text;
 952         }
 953
 954         /**
 955          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 956          * magic external links.
 957          *
 958          * @private
 959          */
 960         function doMagicLinks( $text ) {
 961                 wfProfileIn( __METHOD__ );
 962                 $text = preg_replace_callback(
 963                         '!(?:                           # Start cases
 964                             <a.*?</a> |                 # Skip link text
 965                             <.*?> |                     # Skip stuff inside HTML elements
 966                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 967                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 968                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 969                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 970                                      [0-9Xx]                 # check digit
 971                                    \b)
 972                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 973                 wfProfileOut( __METHOD__ );
 974                 return $text;
 975         }
 976
 977         function magicLinkCallback( $m ) {
 978                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 979                         # Skip HTML element
 980                         return $m[0];
 981                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 982                         $isbn = $m[2];
 983                         $num = strtr( $isbn, array(
 984                                 '-' => '',
 985                                 ' ' => '',
 986                                 'x' => 'X',
 987                         ));
 988                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
 989                         $text = '<a href="' .
 990                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
 991                                 "\" class=\"internal\">ISBN $isbn</a>";
 992                 } else {
 993                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
 994                                 $keyword = 'RFC';
 995                                 $urlmsg = 'rfcurl';
 996                                 $id = $m[1];
 997                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
 998                                 $keyword = 'PMID';
 999                                 $urlmsg = 'pubmedurl';
1000                                 $id = $m[1];
1001                         } else {
1002                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1003                                         substr($m[0], 0, 20 ) . '"' );
1004                         }
1005
1006                         $url = wfMsg( $urlmsg, $id);
1007                         $sk = $this->mOptions->getSkin();
1008                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1009                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1010                 }
1011                 return $text;
1012         }
1013
1014         /**
1015          * Parse headers and return html
1016          *
1017          * @private
1018          */
1019         function doHeadings( $text ) {
1020                 $fname = 'Parser::doHeadings';
1021                 wfProfileIn( $fname );
1022                 for ( $i = 6; $i >= 1; --$i ) {
1023                         $h = str_repeat( '=', $i );
1024                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1025                           "<h$i>\\1</h$i>", $text );
1026                 }
1027                 wfProfileOut( $fname );
1028                 return $text;
1029         }
1030
1031         /**
1032          * Replace single quotes with HTML markup
1033          * @private
1034          * @return string the altered text
1035          */
1036         function doAllQuotes( $text ) {
1037                 $fname = 'Parser::doAllQuotes';
1038                 wfProfileIn( $fname );
1039                 $outtext = '';
1040                 $lines = explode( "\n", $text );
1041                 foreach ( $lines as $line ) {
1042                         $outtext .= $this->doQuotes ( $line ) . "\n";
1043                 }
1044                 $outtext = substr($outtext, 0,-1);
1045                 wfProfileOut( $fname );
1046                 return $outtext;
1047         }
1048
1049         /**
1050          * Helper function for doAllQuotes()
1051          */
1052         public function doQuotes( $text ) {
1053                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1054                 if ( count( $arr ) == 1 )
1055                         return $text;
1056                 else
1057                 {
1058                         # First, do some preliminary work. This may shift some apostrophes from
1059                         # being mark-up to being text. It also counts the number of occurrences
1060                         # of bold and italics mark-ups.
1061                         $i = 0;
1062                         $numbold = 0;
1063                         $numitalics = 0;
1064                         foreach ( $arr as $r )
1065                         {
1066                                 if ( ( $i % 2 ) == 1 )
1067                                 {
1068                                         # If there are ever four apostrophes, assume the first is supposed to
1069                                         # be text, and the remaining three constitute mark-up for bold text.
1070                                         if ( strlen( $arr[$i] ) == 4 )
1071                                         {
1072                                                 $arr[$i-1] .= "'";
1073                                                 $arr[$i] = "'''";
1074                                         }
1075                                         # If there are more than 5 apostrophes in a row, assume they're all
1076                                         # text except for the last 5.
1077                                         else if ( strlen( $arr[$i] ) > 5 )
1078                                         {
1079                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1080                                                 $arr[$i] = "'''''";
1081                                         }
1082                                         # Count the number of occurrences of bold and italics mark-ups.
1083                                         # We are not counting sequences of five apostrophes.
1084                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1085                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1086                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1087                                 }
1088                                 $i++;
1089                         }
1090
1091                         # If there is an odd number of both bold and italics, it is likely
1092                         # that one of the bold ones was meant to be an apostrophe followed
1093                         # by italics. Which one we cannot know for certain, but it is more
1094                         # likely to be one that has a single-letter word before it.
1095                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1096                         {
1097                                 $i = 0;
1098                                 $firstsingleletterword = -1;
1099                                 $firstmultiletterword = -1;
1100                                 $firstspace = -1;
1101                                 foreach ( $arr as $r )
1102                                 {
1103                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1104                                         {
1105                                                 $x1 = substr ($arr[$i-1], -1);
1106                                                 $x2 = substr ($arr[$i-1], -2, 1);
1107                                                 if ($x1 == ' ') {
1108                                                         if ($firstspace == -1) $firstspace = $i;
1109                                                 } else if ($x2 == ' ') {
1110                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1111                                                 } else {
1112                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1113                                                 }
1114                                         }
1115                                         $i++;
1116                                 }
1117
1118                                 # If there is a single-letter word, use it!
1119                                 if ($firstsingleletterword > -1)
1120                                 {
1121                                         $arr [ $firstsingleletterword ] = "''";
1122                                         $arr [ $firstsingleletterword-1 ] .= "'";
1123                                 }
1124                                 # If not, but there's a multi-letter word, use that one.
1125                                 else if ($firstmultiletterword > -1)
1126                                 {
1127                                         $arr [ $firstmultiletterword ] = "''";
1128                                         $arr [ $firstmultiletterword-1 ] .= "'";
1129                                 }
1130                                 # ... otherwise use the first one that has neither.
1131                                 # (notice that it is possible for all three to be -1 if, for example,
1132                                 # there is only one pentuple-apostrophe in the line)
1133                                 else if ($firstspace > -1)
1134                                 {
1135                                         $arr [ $firstspace ] = "''";
1136                                         $arr [ $firstspace-1 ] .= "'";
1137                                 }
1138                         }
1139
1140                         # Now let's actually convert our apostrophic mush to HTML!
1141                         $output = '';
1142                         $buffer = '';
1143                         $state = '';
1144                         $i = 0;
1145                         foreach ($arr as $r)
1146                         {
1147                                 if (($i % 2) == 0)
1148                                 {
1149                                         if ($state == 'both')
1150                                                 $buffer .= $r;
1151                                         else
1152                                                 $output .= $r;
1153                                 }
1154                                 else
1155                                 {
1156                                         if (strlen ($r) == 2)
1157                                         {
1158                                                 if ($state == 'i')
1159                                                 { $output .= '</i>'; $state = ''; }
1160                                                 else if ($state == 'bi')
1161                                                 { $output .= '</i>'; $state = 'b'; }
1162                                                 else if ($state == 'ib')
1163                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1164                                                 else if ($state == 'both')
1165                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1166                                                 else # $state can be 'b' or ''
1167                                                 { $output .= '<i>'; $state .= 'i'; }
1168                                         }
1169                                         else if (strlen ($r) == 3)
1170                                         {
1171                                                 if ($state == 'b')
1172                                                 { $output .= '</b>'; $state = ''; }
1173                                                 else if ($state == 'bi')
1174                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1175                                                 else if ($state == 'ib')
1176                                                 { $output .= '</b>'; $state = 'i'; }
1177                                                 else if ($state == 'both')
1178                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1179                                                 else # $state can be 'i' or ''
1180                                                 { $output .= '<b>'; $state .= 'b'; }
1181                                         }
1182                                         else if (strlen ($r) == 5)
1183                                         {
1184                                                 if ($state == 'b')
1185                                                 { $output .= '</b><i>'; $state = 'i'; }
1186                                                 else if ($state == 'i')
1187                                                 { $output .= '</i><b>'; $state = 'b'; }
1188                                                 else if ($state == 'bi')
1189                                                 { $output .= '</i></b>'; $state = ''; }
1190                                                 else if ($state == 'ib')
1191                                                 { $output .= '</b></i>'; $state = ''; }
1192                                                 else if ($state == 'both')
1193                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1194                                                 else # ($state == '')
1195                                                 { $buffer = ''; $state = 'both'; }
1196                                         }
1197                                 }
1198                                 $i++;
1199                         }
1200                         # Now close all remaining tags.  Notice that the order is important.
1201                         if ($state == 'b' || $state == 'ib')
1202                                 $output .= '</b>';
1203                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1204                                 $output .= '</i>';
1205                         if ($state == 'bi')
1206                                 $output .= '</b>';
1207                         # There might be lonely ''''', so make sure we have a buffer
1208                         if ($state == 'both' && $buffer)
1209                                 $output .= '<b><i>'.$buffer.'</i></b>';
1210                         return $output;
1211                 }
1212         }
1213
1214         /**
1215          * Replace external links
1216          *
1217          * Note: this is all very hackish and the order of execution matters a lot.
1218          * Make sure to run maintenance/parserTests.php if you change this code.
1219          *
1220          * @private
1221          */
1222         function replaceExternalLinks( $text ) {
1223                 global $wgContLang;
1224                 $fname = 'Parser::replaceExternalLinks';
1225                 wfProfileIn( $fname );
1226
1227                 $sk = $this->mOptions->getSkin();
1228
1229                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1230
1231                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1232
1233                 $i = 0;
1234                 while ( $i<count( $bits ) ) {
1235                         $url = $bits[$i++];
1236                         $protocol = $bits[$i++];
1237                         $text = $bits[$i++];
1238                         $trail = $bits[$i++];
1239
1240                         # The characters '<' and '>' (which were escaped by
1241                         # removeHTMLtags()) should not be included in
1242                         # URLs, per RFC 2396.
1243                         $m2 = array();
1244                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1245                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1246                                 $url = substr($url, 0, $m2[0][1]);
1247                         }
1248
1249                         # If the link text is an image URL, replace it with an <img> tag
1250                         # This happened by accident in the original parser, but some people used it extensively
1251                         $img = $this->maybeMakeExternalImage( $text );
1252                         if ( $img !== false ) {
1253                                 $text = $img;
1254                         }
1255
1256                         $dtrail = '';
1257
1258                         # Set linktype for CSS - if URL==text, link is essentially free
1259                         $linktype = ($text == $url) ? 'free' : 'text';
1260
1261                         # No link text, e.g. [http://domain.tld/some.link]
1262                         if ( $text == '' ) {
1263                                 # Autonumber if allowed. See bug #5918
1264                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1265                                         $text = '[' . ++$this->mAutonumber . ']';
1266                                         $linktype = 'autonumber';
1267                                 } else {
1268                                         # Otherwise just use the URL
1269                                         $text = htmlspecialchars( $url );
1270                                         $linktype = 'free';
1271                                 }
1272                         } else {
1273                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1274                                 # Check for trail
1275                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1276                         }
1277
1278                         $text = $wgContLang->markNoConversion($text);
1279
1280                         $url = Sanitizer::cleanUrl( $url );
1281
1282                         # Process the trail (i.e. everything after this link up until start of the next link),
1283                         # replacing any non-bracketed links
1284                         $trail = $this->replaceFreeExternalLinks( $trail );
1285
1286                         # Use the encoded URL
1287                         # This means that users can paste URLs directly into the text
1288                         # Funny characters like &ouml; aren't valid in URLs anyway
1289                         # This was changed in August 2004
1290                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1291
1292                         # Register link in the output object.
1293                         # Replace unnecessary URL escape codes with the referenced character
1294                         # This prevents spammers from hiding links from the filters
1295                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1296                         $this->mOutput->addExternalLink( $pasteurized );
1297                 }
1298
1299                 wfProfileOut( $fname );
1300                 return $s;
1301         }
1302
1303         /**
1304          * Replace anything that looks like a URL with a link
1305          * @private
1306          */
1307         function replaceFreeExternalLinks( $text ) {
1308                 global $wgContLang;
1309                 $fname = 'Parser::replaceFreeExternalLinks';
1310                 wfProfileIn( $fname );
1311
1312                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1313                 $s = array_shift( $bits );
1314                 $i = 0;
1315
1316                 $sk = $this->mOptions->getSkin();
1317
1318                 while ( $i < count( $bits ) ){
1319                         $protocol = $bits[$i++];
1320                         $remainder = $bits[$i++];
1321
1322                         $m = array();
1323                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1324                                 # Found some characters after the protocol that look promising
1325                                 $url = $protocol . $m[1];
1326                                 $trail = $m[2];
1327
1328                                 # special case: handle urls as url args:
1329                                 # http://www.example.com/foo?=http://www.example.com/bar
1330                                 if(strlen($trail) == 0 &&
1331                                         isset($bits[$i]) &&
1332                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1333                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1334                                 {
1335                                         # add protocol, arg
1336                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1337                                         $i += 2;
1338                                         $trail = $m[2];
1339                                 }
1340
1341                                 # The characters '<' and '>' (which were escaped by
1342                                 # removeHTMLtags()) should not be included in
1343                                 # URLs, per RFC 2396.
1344                                 $m2 = array();
1345                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1346                                         $trail = substr($url, $m2[0][1]) . $trail;
1347                                         $url = substr($url, 0, $m2[0][1]);
1348                                 }
1349
1350                                 # Move trailing punctuation to $trail
1351                                 $sep = ',;\.:!?';
1352                                 # If there is no left bracket, then consider right brackets fair game too
1353                                 if ( strpos( $url, '(' ) === false ) {
1354                                         $sep .= ')';
1355                                 }
1356
1357                                 $numSepChars = strspn( strrev( $url ), $sep );
1358                                 if ( $numSepChars ) {
1359                                         $trail = substr( $url, -$numSepChars ) . $trail;
1360                                         $url = substr( $url, 0, -$numSepChars );
1361                                 }
1362
1363                                 $url = Sanitizer::cleanUrl( $url );
1364
1365                                 # Is this an external image?
1366                                 $text = $this->maybeMakeExternalImage( $url );
1367                                 if ( $text === false ) {
1368                                         # Not an image, make a link
1369                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1370                                         # Register it in the output object...
1371                                         # Replace unnecessary URL escape codes with their equivalent characters
1372                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1373                                         $this->mOutput->addExternalLink( $pasteurized );
1374                                 }
1375                                 $s .= $text . $trail;
1376                         } else {
1377                                 $s .= $protocol . $remainder;
1378                         }
1379                 }
1380                 wfProfileOut( $fname );
1381                 return $s;
1382         }
1383
1384         /**
1385          * Replace unusual URL escape codes with their equivalent characters
1386          * @param string
1387          * @return string
1388          * @static
1389          * @todo  This can merge genuinely required bits in the path or query string,
1390          *        breaking legit URLs. A proper fix would treat the various parts of
1391          *        the URL differently; as a workaround, just use the output for
1392          *        statistical records, not for actual linking/output.
1393          */
1394         static function replaceUnusualEscapes( $url ) {
1395                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1396                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1397         }
1398
1399         /**
1400          * Callback function used in replaceUnusualEscapes().
1401          * Replaces unusual URL escape codes with their equivalent character
1402          * @static
1403          * @private
1404          */
1405         private static function replaceUnusualEscapesCallback( $matches ) {
1406                 $char = urldecode( $matches[0] );
1407                 $ord = ord( $char );
1408                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1409                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1410                         // No, shouldn't be escaped
1411                         return $char;
1412                 } else {
1413                         // Yes, leave it escaped
1414                         return $matches[0];
1415                 }
1416         }
1417
1418         /**
1419          * make an image if it's allowed, either through the global
1420          * option or through the exception
1421          * @private
1422          */
1423         function maybeMakeExternalImage( $url ) {
1424                 $sk = $this->mOptions->getSkin();
1425                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1426                 $imagesexception = !empty($imagesfrom);
1427                 $text = false;
1428                 if ( $this->mOptions->getAllowExternalImages()
1429                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1430                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1431                                 # Image found
1432                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1433                         }
1434                 }
1435                 return $text;
1436         }
1437
1438         /**
1439          * Process [[ ]] wikilinks
1440          *
1441          * @private
1442          */
1443         function replaceInternalLinks( $s ) {
1444                 global $wgContLang;
1445                 static $fname = 'Parser::replaceInternalLinks' ;
1446
1447                 wfProfileIn( $fname );
1448
1449                 wfProfileIn( $fname.'-setup' );
1450                 static $tc = FALSE;
1451                 # the % is needed to support urlencoded titles as well
1452                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1453
1454                 $sk = $this->mOptions->getSkin();
1455
1456                 #split the entire text string on occurences of [[
1457                 $a = explode( '[[', ' ' . $s );
1458                 #get the first element (all text up to first [[), and remove the space we added
1459                 $s = array_shift( $a );
1460                 $s = substr( $s, 1 );
1461
1462                 # Match a link having the form [[namespace:link|alternate]]trail
1463                 static $e1 = FALSE;
1464                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1465                 # Match cases where there is no "]]", which might still be images
1466                 static $e1_img = FALSE;
1467                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1468
1469                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1470                 $e2 = null;
1471                 if ( $useLinkPrefixExtension ) {
1472                         # Match the end of a line for a word that's not followed by whitespace,
1473                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1474                         $e2 = wfMsgForContent( 'linkprefix' );
1475                 }
1476
1477                 if( is_null( $this->mTitle ) ) {
1478                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1479                 }
1480                 $nottalk = !$this->mTitle->isTalkPage();
1481
1482                 if ( $useLinkPrefixExtension ) {
1483                         $m = array();
1484                         if ( preg_match( $e2, $s, $m ) ) {
1485                                 $first_prefix = $m[2];
1486                         } else {
1487                                 $first_prefix = false;
1488                         }
1489                 } else {
1490                         $prefix = '';
1491                 }
1492
1493                 if($wgContLang->hasVariants()) {
1494                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1495                 } else {
1496                         $selflink = array($this->mTitle->getPrefixedText());
1497                 }
1498                 $useSubpages = $this->areSubpagesAllowed();
1499                 wfProfileOut( $fname.'-setup' );
1500
1501                 # Loop for each link
1502                 for ($k = 0; isset( $a[$k] ); $k++) {
1503                         $line = $a[$k];
1504                         if ( $useLinkPrefixExtension ) {
1505                                 wfProfileIn( $fname.'-prefixhandling' );
1506                                 if ( preg_match( $e2, $s, $m ) ) {
1507                                         $prefix = $m[2];
1508                                         $s = $m[1];
1509                                 } else {
1510                                         $prefix='';
1511                                 }
1512                                 # first link
1513                                 if($first_prefix) {
1514                                         $prefix = $first_prefix;
1515                                         $first_prefix = false;
1516                                 }
1517                                 wfProfileOut( $fname.'-prefixhandling' );
1518                         }
1519
1520                         $might_be_img = false;
1521
1522                         wfProfileIn( "$fname-e1" );
1523                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1524                                 $text = $m[2];
1525                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1526                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1527                                 # the real problem is with the $e1 regex
1528                                 # See bug 1300.
1529                                 #
1530                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1531                                 # and no image is in sight. See bug 2095.
1532                                 #
1533                                 if( $text !== '' &&
1534                                         substr( $m[3], 0, 1 ) === ']' &&
1535                                         strpos($text, '[') !== false
1536                                 )
1537                                 {
1538                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1539                                         $m[3] = substr( $m[3], 1 );
1540                                 }
1541                                 # fix up urlencoded title texts
1542                                 if( strpos( $m[1], '%' ) !== false ) {
1543                                         # Should anchors '#' also be rejected?
1544                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1545                                 }
1546                                 $trail = $m[3];
1547                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1548                                 $might_be_img = true;
1549                                 $text = $m[2];
1550                                 if ( strpos( $m[1], '%' ) !== false ) {
1551                                         $m[1] = urldecode($m[1]);
1552                                 }
1553                                 $trail = "";
1554                         } else { # Invalid form; output directly
1555                                 $s .= $prefix . '[[' . $line ;
1556                                 wfProfileOut( "$fname-e1" );
1557                                 continue;
1558                         }
1559                         wfProfileOut( "$fname-e1" );
1560                         wfProfileIn( "$fname-misc" );
1561
1562                         # Don't allow internal links to pages containing
1563                         # PROTO: where PROTO is a valid URL protocol; these
1564                         # should be external links.
1565                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1566                                 $s .= $prefix . '[[' . $line ;
1567                                 continue;
1568                         }
1569
1570                         # Make subpage if necessary
1571                         if( $useSubpages ) {
1572                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1573                         } else {
1574                                 $link = $m[1];
1575                         }
1576
1577                         $noforce = (substr($m[1], 0, 1) != ':');
1578                         if (!$noforce) {
1579                                 # Strip off leading ':'
1580                                 $link = substr($link, 1);
1581                         }
1582
1583                         wfProfileOut( "$fname-misc" );
1584                         wfProfileIn( "$fname-title" );
1585                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1586                         if( !$nt ) {
1587                                 $s .= $prefix . '[[' . $line;
1588                                 wfProfileOut( "$fname-title" );
1589                                 continue;
1590                         }
1591
1592                         $ns = $nt->getNamespace();
1593                         $iw = $nt->getInterWiki();
1594                         wfProfileOut( "$fname-title" );
1595
1596                         if ($might_be_img) { # if this is actually an invalid link
1597                                 wfProfileIn( "$fname-might_be_img" );
1598                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1599                                         $found = false;
1600                                         while (isset ($a[$k+1]) ) {
1601                                                 #look at the next 'line' to see if we can close it there
1602                                                 $spliced = array_splice( $a, $k + 1, 1 );
1603                                                 $next_line = array_shift( $spliced );
1604                                                 $m = explode( ']]', $next_line, 3 );
1605                                                 if ( count( $m ) == 3 ) {
1606                                                         # the first ]] closes the inner link, the second the image
1607                                                         $found = true;
1608                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1609                                                         $trail = $m[2];
1610                                                         break;
1611                                                 } elseif ( count( $m ) == 2 ) {
1612                                                         #if there's exactly one ]] that's fine, we'll keep looking
1613                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1614                                                 } else {
1615                                                         #if $next_line is invalid too, we need look no further
1616                                                         $text .= '[[' . $next_line;
1617                                                         break;
1618                                                 }
1619                                         }
1620                                         if ( !$found ) {
1621                                                 # we couldn't find the end of this imageLink, so output it raw
1622                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1623                                                 $text = $this->replaceInternalLinks($text);
1624                                                 $s .= "{$prefix}[[$link|$text";
1625                                                 # note: no $trail, because without an end, there *is* no trail
1626                                                 wfProfileOut( "$fname-might_be_img" );
1627                                                 continue;
1628                                         }
1629                                 } else { #it's not an image, so output it raw
1630                                         $s .= "{$prefix}[[$link|$text";
1631                                         # note: no $trail, because without an end, there *is* no trail
1632                                         wfProfileOut( "$fname-might_be_img" );
1633                                         continue;
1634                                 }
1635                                 wfProfileOut( "$fname-might_be_img" );
1636                         }
1637
1638                         $wasblank = ( '' == $text );
1639                         if( $wasblank ) $text = $link;
1640
1641                         # Link not escaped by : , create the various objects
1642                         if( $noforce ) {
1643
1644                                 # Interwikis
1645                                 wfProfileIn( "$fname-interwiki" );
1646                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1647                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1648                                         $s = rtrim($s . $prefix);
1649                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1650                                         wfProfileOut( "$fname-interwiki" );
1651                                         continue;
1652                                 }
1653                                 wfProfileOut( "$fname-interwiki" );
1654
1655                                 if ( $ns == NS_IMAGE ) {
1656                                         wfProfileIn( "$fname-image" );
1657                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1658                                                 # recursively parse links inside the image caption
1659                                                 # actually, this will parse them in any other parameters, too,
1660                                                 # but it might be hard to fix that, and it doesn't matter ATM
1661                                                 $text = $this->replaceExternalLinks($text);
1662                                                 $text = $this->replaceInternalLinks($text);
1663
1664                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1665                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1666                                                 $this->mOutput->addImage( $nt->getDBkey() );
1667
1668                                                 wfProfileOut( "$fname-image" );
1669                                                 continue;
1670                                         } else {
1671                                                 # We still need to record the image's presence on the page
1672                                                 $this->mOutput->addImage( $nt->getDBkey() );
1673                                         }
1674                                         wfProfileOut( "$fname-image" );
1675
1676                                 }
1677
1678                                 if ( $ns == NS_CATEGORY ) {
1679                                         wfProfileIn( "$fname-category" );
1680                                         $s = rtrim($s . "\n"); # bug 87
1681
1682                                         if ( $wasblank ) {
1683                                                 $sortkey = $this->getDefaultSort();
1684                                         } else {
1685                                                 $sortkey = $text;
1686                                         }
1687                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1688                                         $sortkey = str_replace( "\n", '', $sortkey );
1689                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1690                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1691
1692                                         /**
1693                                          * Strip the whitespace Category links produce, see bug 87
1694                                          * @todo We might want to use trim($tmp, "\n") here.
1695                                          */
1696                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1697
1698                                         wfProfileOut( "$fname-category" );
1699                                         continue;
1700                                 }
1701                         }
1702
1703                         # Self-link checking
1704                         if( $nt->getFragment() === '' ) {
1705                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1706                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1707                                         continue;
1708                                 }
1709                         }
1710
1711                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1712                         if( $ns == NS_MEDIA ) {
1713                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1714                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1715                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1716                                 $this->mOutput->addImage( $nt->getDBkey() );
1717                                 continue;
1718                         } elseif( $ns == NS_SPECIAL ) {
1719                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1720                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1721                                 } else {
1722                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1723                                 }
1724                                 continue;
1725                         } elseif( $ns == NS_IMAGE ) {
1726                                 $img = wfFindFile( $nt );
1727                                 if( $img ) {
1728                                         // Force a blue link if the file exists; may be a remote
1729                                         // upload on the shared repository, and we want to see its
1730                                         // auto-generated page.
1731                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1732                                         $this->mOutput->addLink( $nt );
1733                                         continue;
1734                                 }
1735                         }
1736                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1737                 }
1738                 wfProfileOut( $fname );
1739                 return $s;
1740         }
1741
1742         /**
1743          * Make a link placeholder. The text returned can be later resolved to a real link with
1744          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1745          * parsing of interwiki links, and secondly to allow all existence checks and
1746          * article length checks (for stub links) to be bundled into a single query.
1747          *
1748          */
1749         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1750                 wfProfileIn( __METHOD__ );
1751                 if ( ! is_object($nt) ) {
1752                         # Fail gracefully
1753                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1754                 } else {
1755                         # Separate the link trail from the rest of the link
1756                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1757
1758                         if ( $nt->isExternal() ) {
1759                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1760                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1761                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1762                         } else {
1763                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1764                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1765                                 $this->mLinkHolders['queries'][] = $query;
1766                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1767                                 $this->mLinkHolders['titles'][] = $nt;
1768
1769                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1770                         }
1771                 }
1772                 wfProfileOut( __METHOD__ );
1773                 return $retVal;
1774         }
1775
1776         /**
1777          * Render a forced-blue link inline; protect against double expansion of
1778          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1779          * Since this little disaster has to split off the trail text to avoid
1780          * breaking URLs in the following text without breaking trails on the
1781          * wiki links, it's been made into a horrible function.
1782          *
1783          * @param Title $nt
1784          * @param string $text
1785          * @param string $query
1786          * @param string $trail
1787          * @param string $prefix
1788          * @return string HTML-wikitext mix oh yuck
1789          */
1790         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1791                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1792                 $sk = $this->mOptions->getSkin();
1793                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1794                 return $this->armorLinks( $link ) . $trail;
1795         }
1796
1797         /**
1798          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1799          * going to go through further parsing steps before inline URL expansion.
1800          *
1801          * In particular this is important when using action=render, which causes
1802          * full URLs to be included.
1803          *
1804          * Oh man I hate our multi-layer parser!
1805          *
1806          * @param string more-or-less HTML
1807          * @return string less-or-more HTML with NOPARSE bits
1808          */
1809         function armorLinks( $text ) {
1810                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1811                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1812         }
1813
1814         /**
1815          * Return true if subpage links should be expanded on this page.
1816          * @return bool
1817          */
1818         function areSubpagesAllowed() {
1819                 # Some namespaces don't allow subpages
1820                 global $wgNamespacesWithSubpages;
1821                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1822         }
1823
1824         /**
1825          * Handle link to subpage if necessary
1826          * @param string $target the source of the link
1827          * @param string &$text the link text, modified as necessary
1828          * @return string the full name of the link
1829          * @private
1830          */
1831         function maybeDoSubpageLink($target, &$text) {
1832                 # Valid link forms:
1833                 # Foobar -- normal
1834                 # :Foobar -- override special treatment of prefix (images, language links)
1835                 # /Foobar -- convert to CurrentPage/Foobar
1836                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1837                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1838                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1839
1840                 $fname = 'Parser::maybeDoSubpageLink';
1841                 wfProfileIn( $fname );
1842                 $ret = $target; # default return value is no change
1843
1844                 # Some namespaces don't allow subpages,
1845                 # so only perform processing if subpages are allowed
1846                 if( $this->areSubpagesAllowed() ) {
1847                         $hash = strpos( $target, '#' );
1848                         if( $hash !== false ) {
1849                                 $suffix = substr( $target, $hash );
1850                                 $target = substr( $target, 0, $hash );
1851                         } else {
1852                                 $suffix = '';
1853                         }
1854                         # bug 7425
1855                         $target = trim( $target );
1856                         # Look at the first character
1857                         if( $target != '' && $target{0} == '/' ) {
1858                                 # / at end means we don't want the slash to be shown
1859                                 $m = array();
1860                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1861                                 if( $trailingSlashes ) {
1862                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1863                                 } else {
1864                                         $noslash = substr( $target, 1 );
1865                                 }
1866
1867                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1868                                 if( '' === $text ) {
1869                                         $text = $target . $suffix;
1870                                 } # this might be changed for ugliness reasons
1871                         } else {
1872                                 # check for .. subpage backlinks
1873                                 $dotdotcount = 0;
1874                                 $nodotdot = $target;
1875                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1876                                         ++$dotdotcount;
1877                                         $nodotdot = substr( $nodotdot, 3 );
1878                                 }
1879                                 if($dotdotcount > 0) {
1880                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1881                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1882                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1883                                                 # / at the end means don't show full path
1884                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1885                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1886                                                         if( '' === $text ) {
1887                                                                 $text = $nodotdot . $suffix;
1888                                                         }
1889                                                 }
1890                                                 $nodotdot = trim( $nodotdot );
1891                                                 if( $nodotdot != '' ) {
1892                                                         $ret .= '/' . $nodotdot;
1893                                                 }
1894                                                 $ret .= $suffix;
1895                                         }
1896                                 }
1897                         }
1898                 }
1899
1900                 wfProfileOut( $fname );
1901                 return $ret;
1902         }
1903
1904         /**#@+
1905          * Used by doBlockLevels()
1906          * @private
1907          */
1908         /* private */ function closeParagraph() {
1909                 $result = '';
1910                 if ( '' != $this->mLastSection ) {
1911                         $result = '</' . $this->mLastSection  . ">\n";
1912                 }
1913                 $this->mInPre = false;
1914                 $this->mLastSection = '';
1915                 return $result;
1916         }
1917         # getCommon() returns the length of the longest common substring
1918         # of both arguments, starting at the beginning of both.
1919         #
1920         /* private */ function getCommon( $st1, $st2 ) {
1921                 $fl = strlen( $st1 );
1922                 $shorter = strlen( $st2 );
1923                 if ( $fl < $shorter ) { $shorter = $fl; }
1924
1925                 for ( $i = 0; $i < $shorter; ++$i ) {
1926                         if ( $st1{$i} != $st2{$i} ) { break; }
1927                 }
1928                 return $i;
1929         }
1930         # These next three functions open, continue, and close the list
1931         # element appropriate to the prefix character passed into them.
1932         #
1933         /* private */ function openList( $char ) {
1934                 $result = $this->closeParagraph();
1935
1936                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1937                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1938                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1939                 else if ( ';' == $char ) {
1940                         $result .= '<dl><dt>';
1941                         $this->mDTopen = true;
1942                 }
1943                 else { $result = '<!-- ERR 1 -->'; }
1944
1945                 return $result;
1946         }
1947
1948         /* private */ function nextItem( $char ) {
1949                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1950                 else if ( ':' == $char || ';' == $char ) {
1951                         $close = '</dd>';
1952                         if ( $this->mDTopen ) { $close = '</dt>'; }
1953                         if ( ';' == $char ) {
1954                                 $this->mDTopen = true;
1955                                 return $close . '<dt>';
1956                         } else {
1957                                 $this->mDTopen = false;
1958                                 return $close . '<dd>';
1959                         }
1960                 }
1961                 return '<!-- ERR 2 -->';
1962         }
1963
1964         /* private */ function closeList( $char ) {
1965                 if ( '*' == $char ) { $text = '</li></ul>'; }
1966                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1967                 else if ( ':' == $char ) {
1968                         if ( $this->mDTopen ) {
1969                                 $this->mDTopen = false;
1970                                 $text = '</dt></dl>';
1971                         } else {
1972                                 $text = '</dd></dl>';
1973                         }
1974                 }
1975                 else {  return '<!-- ERR 3 -->'; }
1976                 return $text."\n";
1977         }
1978         /**#@-*/
1979
1980         /**
1981          * Make lists from lines starting with ':', '*', '#', etc.
1982          *
1983          * @private
1984          * @return string the lists rendered as HTML
1985          */
1986         function doBlockLevels( $text, $linestart ) {
1987                 $fname = 'Parser::doBlockLevels';
1988                 wfProfileIn( $fname );
1989
1990                 # Parsing through the text line by line.  The main thing
1991                 # happening here is handling of block-level elements p, pre,
1992                 # and making lists from lines starting with * # : etc.
1993                 #
1994                 $textLines = explode( "\n", $text );
1995
1996                 $lastPrefix = $output = '';
1997                 $this->mDTopen = $inBlockElem = false;
1998                 $prefixLength = 0;
1999                 $paragraphStack = false;
2000
2001                 if ( !$linestart ) {
2002                         $output .= array_shift( $textLines );
2003                 }
2004                 foreach ( $textLines as $oLine ) {
2005                         $lastPrefixLength = strlen( $lastPrefix );
2006                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2007                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2008                         if ( !$this->mInPre ) {
2009                                 # Multiple prefixes may abut each other for nested lists.
2010                                 $prefixLength = strspn( $oLine, '*#:;' );
2011                                 $pref = substr( $oLine, 0, $prefixLength );
2012
2013                                 # eh?
2014                                 $pref2 = str_replace( ';', ':', $pref );
2015                                 $t = substr( $oLine, $prefixLength );
2016                                 $this->mInPre = !empty($preOpenMatch);
2017                         } else {
2018                                 # Don't interpret any other prefixes in preformatted text
2019                                 $prefixLength = 0;
2020                                 $pref = $pref2 = '';
2021                                 $t = $oLine;
2022                         }
2023
2024                         # List generation
2025                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2026                                 # Same as the last item, so no need to deal with nesting or opening stuff
2027                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2028                                 $paragraphStack = false;
2029
2030                                 if ( substr( $pref, -1 ) == ';') {
2031                                         # The one nasty exception: definition lists work like this:
2032                                         # ; title : definition text
2033                                         # So we check for : in the remainder text to split up the
2034                                         # title and definition, without b0rking links.
2035                                         $term = $t2 = '';
2036                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2037                                                 $t = $t2;
2038                                                 $output .= $term . $this->nextItem( ':' );
2039                                         }
2040                                 }
2041                         } elseif( $prefixLength || $lastPrefixLength ) {
2042                                 # Either open or close a level...
2043                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2044                                 $paragraphStack = false;
2045
2046                                 while( $commonPrefixLength < $lastPrefixLength ) {
2047                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2048                                         --$lastPrefixLength;
2049                                 }
2050                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2051                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2052                                 }
2053                                 while ( $prefixLength > $commonPrefixLength ) {
2054                                         $char = substr( $pref, $commonPrefixLength, 1 );
2055                                         $output .= $this->openList( $char );
2056
2057                                         if ( ';' == $char ) {
2058                                                 # FIXME: This is dupe of code above
2059                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2060                                                         $t = $t2;
2061                                                         $output .= $term . $this->nextItem( ':' );
2062                                                 }
2063                                         }
2064                                         ++$commonPrefixLength;
2065                                 }
2066                                 $lastPrefix = $pref2;
2067                         }
2068                         if( 0 == $prefixLength ) {
2069                                 wfProfileIn( "$fname-paragraph" );
2070                                 # No prefix (not in list)--go to paragraph mode
2071                                 // XXX: use a stack for nestable elements like span, table and div
2072                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2073                                 $closematch = preg_match(
2074                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2075                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2076                                 if ( $openmatch or $closematch ) {
2077                                         $paragraphStack = false;
2078                                         # TODO bug 5718: paragraph closed
2079                                         $output .= $this->closeParagraph();
2080                                         if ( $preOpenMatch and !$preCloseMatch ) {
2081                                                 $this->mInPre = true;
2082                                         }
2083                                         if ( $closematch ) {
2084                                                 $inBlockElem = false;
2085                                         } else {
2086                                                 $inBlockElem = true;
2087                                         }
2088                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2089                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2090                                                 // pre
2091                                                 if ($this->mLastSection != 'pre') {
2092                                                         $paragraphStack = false;
2093                                                         $output .= $this->closeParagraph().'<pre>';
2094                                                         $this->mLastSection = 'pre';
2095                                                 }
2096                                                 $t = substr( $t, 1 );
2097                                         } else {
2098                                                 // paragraph
2099                                                 if ( '' == trim($t) ) {
2100                                                         if ( $paragraphStack ) {
2101                                                                 $output .= $paragraphStack.'<br />';
2102                                                                 $paragraphStack = false;
2103                                                                 $this->mLastSection = 'p';
2104                                                         } else {
2105                                                                 if ($this->mLastSection != 'p' ) {
2106                                                                         $output .= $this->closeParagraph();
2107                                                                         $this->mLastSection = '';
2108                                                                         $paragraphStack = '<p>';
2109                                                                 } else {
2110                                                                         $paragraphStack = '</p><p>';
2111                                                                 }
2112                                                         }
2113                                                 } else {
2114                                                         if ( $paragraphStack ) {
2115                                                                 $output .= $paragraphStack;
2116                                                                 $paragraphStack = false;
2117                                                                 $this->mLastSection = 'p';
2118                                                         } else if ($this->mLastSection != 'p') {
2119                                                                 $output .= $this->closeParagraph().'<p>';
2120                                                                 $this->mLastSection = 'p';
2121                                                         }
2122                                                 }
2123                                         }
2124                                 }
2125                                 wfProfileOut( "$fname-paragraph" );
2126                         }
2127                         // somewhere above we forget to get out of pre block (bug 785)
2128                         if($preCloseMatch && $this->mInPre) {
2129                                 $this->mInPre = false;
2130                         }
2131                         if ($paragraphStack === false) {
2132                                 $output .= $t."\n";
2133                         }
2134                 }
2135                 while ( $prefixLength ) {
2136                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2137                         --$prefixLength;
2138                 }
2139                 if ( '' != $this->mLastSection ) {
2140                         $output .= '</' . $this->mLastSection . '>';
2141                         $this->mLastSection = '';
2142                 }
2143
2144                 wfProfileOut( $fname );
2145                 return $output;
2146         }
2147
2148         /**
2149          * Split up a string on ':', ignoring any occurences inside tags
2150          * to prevent illegal overlapping.
2151          * @param string $str the string to split
2152          * @param string &$before set to everything before the ':'
2153          * @param string &$after set to everything after the ':'
2154          * return string the position of the ':', or false if none found
2155          */
2156         function findColonNoLinks($str, &$before, &$after) {
2157                 $fname = 'Parser::findColonNoLinks';
2158                 wfProfileIn( $fname );
2159
2160                 $pos = strpos( $str, ':' );
2161                 if( $pos === false ) {
2162                         // Nothing to find!
2163                         wfProfileOut( $fname );
2164                         return false;
2165                 }
2166
2167                 $lt = strpos( $str, '<' );
2168                 if( $lt === false || $lt > $pos ) {
2169                         // Easy; no tag nesting to worry about
2170                         $before = substr( $str, 0, $pos );
2171                         $after = substr( $str, $pos+1 );
2172                         wfProfileOut( $fname );
2173                         return $pos;
2174                 }
2175
2176                 // Ugly state machine to walk through avoiding tags.
2177                 $state = self::COLON_STATE_TEXT;
2178                 $stack = 0;
2179                 $len = strlen( $str );
2180                 for( $i = 0; $i < $len; $i++ ) {
2181                         $c = $str{$i};
2182
2183                         switch( $state ) {
2184                         // (Using the number is a performance hack for common cases)
2185                         case 0: // self::COLON_STATE_TEXT:
2186                                 switch( $c ) {
2187                                 case "<":
2188                                         // Could be either a <start> tag or an </end> tag
2189                                         $state = self::COLON_STATE_TAGSTART;
2190                                         break;
2191                                 case ":":
2192                                         if( $stack == 0 ) {
2193                                                 // We found it!
2194                                                 $before = substr( $str, 0, $i );
2195                                                 $after = substr( $str, $i + 1 );
2196                                                 wfProfileOut( $fname );
2197                                                 return $i;
2198                                         }
2199                                         // Embedded in a tag; don't break it.
2200                                         break;
2201                                 default:
2202                                         // Skip ahead looking for something interesting
2203                                         $colon = strpos( $str, ':', $i );
2204                                         if( $colon === false ) {
2205                                                 // Nothing else interesting
2206                                                 wfProfileOut( $fname );
2207                                                 return false;
2208                                         }
2209                                         $lt = strpos( $str, '<', $i );
2210                                         if( $stack === 0 ) {
2211                                                 if( $lt === false || $colon < $lt ) {
2212                                                         // We found it!
2213                                                         $before = substr( $str, 0, $colon );
2214                                                         $after = substr( $str, $colon + 1 );
2215                                                         wfProfileOut( $fname );
2216                                                         return $i;
2217                                                 }
2218                                         }
2219                                         if( $lt === false ) {
2220                                                 // Nothing else interesting to find; abort!
2221                                                 // We're nested, but there's no close tags left. Abort!
2222                                                 break 2;
2223                                         }
2224                                         // Skip ahead to next tag start
2225                                         $i = $lt;
2226                                         $state = self::COLON_STATE_TAGSTART;
2227                                 }
2228                                 break;
2229                         case 1: // self::COLON_STATE_TAG:
2230                                 // In a <tag>
2231                                 switch( $c ) {
2232                                 case ">":
2233                                         $stack++;
2234                                         $state = self::COLON_STATE_TEXT;
2235                                         break;
2236                                 case "/":
2237                                         // Slash may be followed by >?
2238                                         $state = self::COLON_STATE_TAGSLASH;
2239                                         break;
2240                                 default:
2241                                         // ignore
2242                                 }
2243                                 break;
2244                         case 2: // self::COLON_STATE_TAGSTART:
2245                                 switch( $c ) {
2246                                 case "/":
2247                                         $state = self::COLON_STATE_CLOSETAG;
2248                                         break;
2249                                 case "!":
2250                                         $state = self::COLON_STATE_COMMENT;
2251                                         break;
2252                                 case ">":
2253                                         // Illegal early close? This shouldn't happen D:
2254                                         $state = self::COLON_STATE_TEXT;
2255                                         break;
2256                                 default:
2257                                         $state = self::COLON_STATE_TAG;
2258                                 }
2259                                 break;
2260                         case 3: // self::COLON_STATE_CLOSETAG:
2261                                 // In a </tag>
2262                                 if( $c == ">" ) {
2263                                         $stack--;
2264                                         if( $stack < 0 ) {
2265                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2266                                                 wfProfileOut( $fname );
2267                                                 return false;
2268                                         }
2269                                         $state = self::COLON_STATE_TEXT;
2270                                 }
2271                                 break;
2272                         case self::COLON_STATE_TAGSLASH:
2273                                 if( $c == ">" ) {
2274                                         // Yes, a self-closed tag <blah/>
2275                                         $state = self::COLON_STATE_TEXT;
2276                                 } else {
2277                                         // Probably we're jumping the gun, and this is an attribute
2278                                         $state = self::COLON_STATE_TAG;
2279                                 }
2280                                 break;
2281                         case 5: // self::COLON_STATE_COMMENT:
2282                                 if( $c == "-" ) {
2283                                         $state = self::COLON_STATE_COMMENTDASH;
2284                                 }
2285                                 break;
2286                         case self::COLON_STATE_COMMENTDASH:
2287                                 if( $c == "-" ) {
2288                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2289                                 } else {
2290                                         $state = self::COLON_STATE_COMMENT;
2291                                 }
2292                                 break;
2293                         case self::COLON_STATE_COMMENTDASHDASH:
2294                                 if( $c == ">" ) {
2295                                         $state = self::COLON_STATE_TEXT;
2296                                 } else {
2297                                         $state = self::COLON_STATE_COMMENT;
2298                                 }
2299                                 break;
2300                         default:
2301                                 throw new MWException( "State machine error in $fname" );
2302                         }
2303                 }
2304                 if( $stack > 0 ) {
2305                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2306                         return false;
2307                 }
2308                 wfProfileOut( $fname );
2309                 return false;
2310         }
2311
2312         /**
2313          * Return value of a magic variable (like PAGENAME)
2314          *
2315          * @private
2316          */
2317         function getVariableValue( $index ) {
2318                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2319
2320                 /**
2321                  * Some of these require message or data lookups and can be
2322                  * expensive to check many times.
2323                  */
2324                 static $varCache = array();
2325                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2326                         if ( isset( $varCache[$index] ) ) {
2327                                 return $varCache[$index];
2328                         }
2329                 }
2330
2331                 $ts = time();
2332                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2333
2334                 # Use the time zone
2335                 global $wgLocaltimezone;
2336                 if ( isset( $wgLocaltimezone ) ) {
2337                         $oldtz = getenv( 'TZ' );
2338                         putenv( 'TZ='.$wgLocaltimezone );
2339                 }
2340
2341                 wfSuppressWarnings(); // E_STRICT system time bitching
2342                 $localTimestamp = date( 'YmdHis', $ts );
2343                 $localMonth = date( 'm', $ts );
2344                 $localMonthName = date( 'n', $ts );
2345                 $localDay = date( 'j', $ts );
2346                 $localDay2 = date( 'd', $ts );
2347                 $localDayOfWeek = date( 'w', $ts );
2348                 $localWeek = date( 'W', $ts );
2349                 $localYear = date( 'Y', $ts );
2350                 $localHour = date( 'H', $ts );
2351                 if ( isset( $wgLocaltimezone ) ) {
2352                         putenv( 'TZ='.$oldtz );
2353                 }
2354                 wfRestoreWarnings();
2355
2356                 switch ( $index ) {
2357                         case 'currentmonth':
2358                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2359                         case 'currentmonthname':
2360                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2361                         case 'currentmonthnamegen':
2362                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2363                         case 'currentmonthabbrev':
2364                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2365                         case 'currentday':
2366                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2367                         case 'currentday2':
2368                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2369                         case 'localmonth':
2370                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2371                         case 'localmonthname':
2372                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2373                         case 'localmonthnamegen':
2374                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2375                         case 'localmonthabbrev':
2376                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2377                         case 'localday':
2378                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2379                         case 'localday2':
2380                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2381                         case 'pagename':
2382                                 return wfEscapeWikiText( $this->mTitle->getText() );
2383                         case 'pagenamee':
2384                                 return $this->mTitle->getPartialURL();
2385                         case 'fullpagename':
2386                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2387                         case 'fullpagenamee':
2388                                 return $this->mTitle->getPrefixedURL();
2389                         case 'subpagename':
2390                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2391                         case 'subpagenamee':
2392                                 return $this->mTitle->getSubpageUrlForm();
2393                         case 'basepagename':
2394                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2395                         case 'basepagenamee':
2396                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2397                         case 'talkpagename':
2398                                 if( $this->mTitle->canTalk() ) {
2399                                         $talkPage = $this->mTitle->getTalkPage();
2400                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2401                                 } else {
2402                                         return '';
2403                                 }
2404                         case 'talkpagenamee':
2405                                 if( $this->mTitle->canTalk() ) {
2406                                         $talkPage = $this->mTitle->getTalkPage();
2407                                         return $talkPage->getPrefixedUrl();
2408                                 } else {
2409                                         return '';
2410                                 }
2411                         case 'subjectpagename':
2412                                 $subjPage = $this->mTitle->getSubjectPage();
2413                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2414                         case 'subjectpagenamee':
2415                                 $subjPage = $this->mTitle->getSubjectPage();
2416                                 return $subjPage->getPrefixedUrl();
2417                         case 'revisionid':
2418                                 // Let the edit saving system know we should parse the page
2419                                 // *after* a revision ID has been assigned.
2420                                 $this->mOutput->setFlag( 'vary-revision' );
2421                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2422                                 return $this->mRevisionId;
2423                         case 'revisionday':
2424                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2425                         case 'revisionday2':
2426                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2427                         case 'revisionmonth':
2428                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2429                         case 'revisionyear':
2430                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2431                         case 'revisiontimestamp':
2432                                 return $this->getRevisionTimestamp();
2433                         case 'namespace':
2434                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2435                         case 'namespacee':
2436                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2437                         case 'talkspace':
2438                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2439                         case 'talkspacee':
2440                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2441                         case 'subjectspace':
2442                                 return $this->mTitle->getSubjectNsText();
2443                         case 'subjectspacee':
2444                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2445                         case 'currentdayname':
2446                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2447                         case 'currentyear':
2448                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2449                         case 'currenttime':
2450                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2451                         case 'currenthour':
2452                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2453                         case 'currentweek':
2454                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2455                                 // int to remove the padding
2456                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2457                         case 'currentdow':
2458                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2459                         case 'localdayname':
2460                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2461                         case 'localyear':
2462                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2463                         case 'localtime':
2464                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2465                         case 'localhour':
2466                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2467                         case 'localweek':
2468                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2469                                 // int to remove the padding
2470                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2471                         case 'localdow':
2472                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2473                         case 'numberofarticles':
2474                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2475                         case 'numberoffiles':
2476                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2477                         case 'numberofusers':
2478                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2479                         case 'numberofpages':
2480                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2481                         case 'numberofadmins':
2482                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2483                         case 'numberofedits':
2484                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2485                         case 'currenttimestamp':
2486                                 return $varCache[$index] = wfTimestampNow();
2487                         case 'localtimestamp':
2488                                 return $varCache[$index] = $localTimestamp;
2489                         case 'currentversion':
2490                                 return $varCache[$index] = SpecialVersion::getVersion();
2491                         case 'sitename':
2492                                 return $wgSitename;
2493                         case 'server':
2494                                 return $wgServer;
2495                         case 'servername':
2496                                 return $wgServerName;
2497                         case 'scriptpath':
2498                                 return $wgScriptPath;
2499                         case 'directionmark':
2500                                 return $wgContLang->getDirMark();
2501                         case 'contentlanguage':
2502                                 global $wgContLanguageCode;
2503                                 return $wgContLanguageCode;
2504                         default:
2505                                 $ret = null;
2506                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2507                                         return $ret;
2508                                 else
2509                                         return null;
2510                 }
2511         }
2512
2513         /**
2514          * initialise the magic variables (like CURRENTMONTHNAME)
2515          *
2516          * @private
2517          */
2518         function initialiseVariables() {
2519                 $fname = 'Parser::initialiseVariables';
2520                 wfProfileIn( $fname );
2521                 $variableIDs = MagicWord::getVariableIDs();
2522
2523                 $this->mVariables = new MagicWordArray( $variableIDs );
2524                 wfProfileOut( $fname );
2525         }
2526
2527         /**
2528          * Parse any parentheses in format ((title|part|part)} and return the document tree
2529          * This is the ghost of replace_variables().
2530          *
2531          * @param string $text The text to parse
2532          * @private
2533          */
2534         function preprocessToDom ( $text ) {
2535                 wfProfileIn( __METHOD__ );
2536                 wfProfileIn( __METHOD__.'-makexml' );
2537
2538                 static $msgRules, $normalRules;
2539                 if ( !$msgRules ) {
2540                         $msgRules = array(
2541                                 '{' => array(
2542                                         'end' => '}',
2543                                         'names' => array(
2544                                                 2 => 'template',
2545                                         ),
2546                                         'min' => 2,
2547                                         'max' => 2,
2548                                 ),
2549                                 '[' => array(
2550                                         'end' => ']',
2551                                         'names' => array( 2 => null ),
2552                                         'min' => 2,
2553                                         'max' => 2,
2554                                 )
2555                         );
2556                         $normalRules = array(
2557                                 '{' => array(
2558                                         'end' => '}',
2559                                         'names' => array(
2560                                                 2 => 'template',
2561                                                 3 => 'tplarg',
2562                                         ),
2563                                         'min' => 2,
2564                                         'max' => 3,
2565                                 ),
2566                                 '[' => array(
2567                                         'end' => ']',
2568                                         'names' => array( 2 => null ),
2569                                         'min' => 2,
2570                                         'max' => 2,
2571                                 )
2572                         );
2573                 }
2574                 if ( $this->ot['msg'] ) {
2575                         $rules = $msgRules;
2576                 } else {
2577                         $rules = $normalRules;
2578                 }
2579
2580                 $extElements = implode( '|', $this->getStripList() );
2581                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2582                 $extElementsRegex = "/($extElements)(?:\s|\/>|>)|(!--)/iA";
2583
2584                 $stack = array();      # Stack of unclosed parentheses
2585                 $stackIndex = -1;      # Stack read pointer
2586
2587                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2588
2589                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2590                 $topAccum = '<root>';      # Top level text accumulator
2591                 $accum =& $topAccum;            # Current text accumulator
2592                 $findEquals = false;            # True to find equals signs in arguments
2593                 $findHeading = false;           # True to look at LF characters for possible headings
2594                 $findPipe = false;              # True to take notice of pipe characters
2595                 $headingIndex = 1;
2596                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2597
2598                 while ( $i < strlen( $text ) ) {
2599                         if ( $i == -1 ) {
2600                                 $found = 'line-start';
2601                                 $curChar = '';
2602                         } else {
2603                                 # Find next opening brace, closing brace or pipe
2604                                 $search = $searchBase;
2605                                 if ( $stackIndex == -1 ) {
2606                                         $currentClosing = '';
2607                                         // Look for headings only at the top stack level
2608                                         // Among other things, this resolves the ambiguity between =
2609                                         // for headings and = for template arguments
2610                                         $search .= "\n";
2611                                 } else {
2612                                         $currentClosing = $stack[$stackIndex]['close'];
2613                                         $search .= $currentClosing;
2614                                 }
2615                                 if ( $findPipe ) {
2616                                         $search .= '|';
2617                                 }
2618                                 if ( $findEquals ) {
2619                                         $search .= '=';
2620                                 }
2621                                 $rule = null;
2622                                 # Output literal section, advance input counter
2623                                 $literalLength = strcspn( $text, $search, $i );
2624                                 if ( $literalLength > 0 ) {
2625                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2626                                         $i += $literalLength;
2627                                 }
2628                                 if ( $i >= strlen( $text ) ) {
2629                                         if ( $currentClosing == "\n" ) {
2630                                                 // Do a past-the-end run to finish off the heading
2631                                                 $curChar = '';
2632                                                 $found = 'line-end';
2633                                         } else {
2634                                                 # All done
2635                                                 break;
2636                                         }
2637                                 } else {
2638                                         $curChar = $text[$i];
2639                                         if ( $curChar == '|' ) {
2640                                                 $found = 'pipe';
2641                                         } elseif ( $curChar == '=' ) {
2642                                                 $found = 'equals';
2643                                         } elseif ( $curChar == '<' ) {
2644                                                 $found = 'angle';
2645                                         } elseif ( $curChar == "\n" ) {
2646                                                 if ( $stackIndex == -1 ) {
2647                                                         $found = 'line-start';
2648                                                 } else {
2649                                                         $found = 'line-end';
2650                                                 }
2651                                         } elseif ( $curChar == $currentClosing ) {
2652                                                 $found = 'close';
2653                                         } elseif ( isset( $rules[$curChar] ) ) {
2654                                                 $found = 'open';
2655                                                 $rule = $rules[$curChar];
2656                                         } else {
2657                                                 # Some versions of PHP have a strcspn which stops on null characters
2658                                                 # Ignore and continue
2659                                                 ++$i;
2660                                                 continue;
2661                                         }
2662                                 }
2663                         }
2664
2665                         if ( $found == 'angle' ) {
2666                                 $matches = false;
2667                                 // Determine element name
2668                                 if ( !preg_match( $extElementsRegex, $text, $matches, 0, $i + 1 ) ) {
2669                                         // Element name missing or not listed
2670                                         $accum .= '&lt;';
2671                                         ++$i;
2672                                         continue;
2673                                 }
2674                                 // Handle comments
2675                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2676                                         // HTML comment, scan to end
2677                                         $endpos = strpos( $text, '-->', $i + 4 );
2678                                         if ( $endpos === false ) {
2679                                                 // Unclosed comment in input, runs to end
2680                                                 $accum .= htmlspecialchars( substr( $text, $i ) );
2681                                                 if ( $this->ot['html'] ) {
2682                                                         // Close it so later stripping can remove it
2683                                                         $accum .= htmlspecialchars( '-->' );
2684                                                 }
2685                                                 $i = strlen( $text );
2686                                                 continue;
2687                                         }
2688                                         $accum .= htmlspecialchars( substr( $text, $i, $endpos - $i + 3 ) );
2689                                         #$inner = substr( $text, $i + 4, $endpos - $i - 4 );
2690                                         #$accum .= '<ext><name>!--</name><inner>' . htmlspecialchars( $inner ) . '</inner></ext>';
2691                                         $i = $endpos + 3;
2692                                         continue;
2693                                 }
2694                                 $name = $matches[1];
2695                                 $attrStart = $i + strlen( $name ) + 1;
2696
2697                                 // Find end of tag
2698                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2699                                 if ( $tagEndPos === false ) {
2700                                         // Infinite backtrack
2701                                         // Disable tag search to prevent worst-case O(N^2) performance
2702                                         $noMoreGT = true;
2703                                         $accum .= '&lt;';
2704                                         ++$i;
2705                                         continue;
2706                                 }
2707                                 if ( $text[$tagEndPos-1] == '/' ) {
2708                                         $attrEnd = $tagEndPos - 1;
2709                                         $inner = null;
2710                                         $i = $tagEndPos + 1;
2711                                         $close = '';
2712                                 } else {
2713                                         $attrEnd = $tagEndPos;
2714                                         // Find closing tag
2715                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2716                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2717                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2718                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2719                                         } else {
2720                                                 // No end tag -- let it run out to the end of the text.
2721                                                 $inner = substr( $text, $tagEndPos + 1 );
2722                                                 $i = strlen( $text );
2723                                                 $close = '';
2724                                         }
2725                                 }
2726                                 $accum .= '<ext>';
2727                                 if ( $attrEnd <= $attrStart ) {
2728                                         $attr = '';
2729                                 } else {
2730                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2731                                 }
2732                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2733                                         // Note that the attr element contains the whitespace between name and attribute,
2734                                         // this is necessary for precise reconstruction during pre-save transform.
2735                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2736                                 if ( $inner !== null ) {
2737                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2738                                 }
2739                                 $accum .= $close . '</ext>';
2740                         }
2741
2742                         elseif ( $found == 'line-start' ) {
2743                                 // Is this the start of a heading?
2744                                 // Line break belongs before the heading element in any case
2745                                 $accum .= $curChar;
2746                                 $i++;
2747
2748                                 $count = strspn( $text, '=', $i, 6 );
2749                                 if ( $count > 0 ) {
2750                                         $piece = array(
2751                                                 'open' => "\n",
2752                                                 'close' => "\n",
2753                                                 'parts' => array( str_repeat( '=', $count ) ),
2754                                                 'count' => $count );
2755                                         $stack[++$stackIndex] = $piece;
2756                                         $i += $count;
2757                                         $accum =& $stack[$stackIndex]['parts'][0];
2758                                         $findPipe = false;
2759                                 }
2760                         }
2761
2762                         elseif ( $found == 'line-end' ) {
2763                                 $piece = $stack[$stackIndex];
2764                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2765                                 assert( $piece['open'] == "\n" );
2766                                 assert( $stackIndex == 0 );
2767                                 // Search back through the accumulator to see if it has a proper close
2768                                 // No efficient way to do this in PHP AFAICT: strrev, PCRE search with $ anchor
2769                                 // and rtrim are all O(N) in total size. Optimal would be O(N) in trailing
2770                                 // whitespace size only.
2771                                 $m = false;
2772                                 $count = $piece['count'];
2773                                 if ( preg_match( "/(={{$count}})\s*$/", $accum, $m, 0, $count ) ) {
2774                                         // Found match, output <h>
2775                                         $count = min( strlen( $m[1] ), $count );
2776                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2777                                         $headingIndex++;
2778                                 } else {
2779                                         // No match, no <h>, just pass down the inner text
2780                                         $element = $accum;
2781                                 }
2782                                 // Unwind the stack
2783                                 // Headings can only occur on the top level, so this is a bit simpler than the
2784                                 // generic stack unwind operation in the close case
2785                                 unset( $stack[$stackIndex--] );
2786                                 $accum =& $topAccum;
2787                                 $findEquals = false;
2788                                 $findPipe = false;
2789
2790                                 // Append the result to the enclosing accumulator
2791                                 $accum .= $element;
2792                                 // Note that we do NOT increment the input pointer.
2793                                 // This is because the closing linebreak could be the opening linebreak of
2794                                 // another heading. Infinite loops are avoided because the next iteration MUST
2795                                 // hit the heading open case above, which unconditionally increments the
2796                                 // input pointer.
2797                         }
2798
2799                         elseif ( $found == 'open' ) {
2800                                 # count opening brace characters
2801                                 $count = strspn( $text, $curChar, $i );
2802
2803                                 # we need to add to stack only if opening brace count is enough for one of the rules
2804                                 if ( $count >= $rule['min'] ) {
2805                                         # Add it to the stack
2806                                         $piece = array(
2807                                                 'open' => $curChar,
2808                                                 'close' => $rule['end'],
2809                                                 'count' => $count,
2810                                                 'parts' => array( '' ),
2811                                                 'eqpos' => array(),
2812                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2813                                         );
2814
2815                                         $stackIndex ++;
2816                                         $stack[$stackIndex] = $piece;
2817                                         $accum =& $stack[$stackIndex]['parts'][0];
2818                                         $findEquals = false;
2819                                         $findPipe = true;
2820                                 } else {
2821                                         # Add literal brace(s)
2822                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2823                                 }
2824                                 $i += $count;
2825                         }
2826
2827                         elseif ( $found == 'close' ) {
2828                                 $piece = $stack[$stackIndex];
2829                                 # lets check if there are enough characters for closing brace
2830                                 $maxCount = $piece['count'];
2831                                 $count = strspn( $text, $curChar, $i, $maxCount );
2832
2833                                 # check for maximum matching characters (if there are 5 closing
2834                                 # characters, we will probably need only 3 - depending on the rules)
2835                                 $matchingCount = 0;
2836                                 $rule = $rules[$piece['open']];
2837                                 if ( $count > $rule['max'] ) {
2838                                         # The specified maximum exists in the callback array, unless the caller
2839                                         # has made an error
2840                                         $matchingCount = $rule['max'];
2841                                 } else {
2842                                         # Count is less than the maximum
2843                                         # Skip any gaps in the callback array to find the true largest match
2844                                         # Need to use array_key_exists not isset because the callback can be null
2845                                         $matchingCount = $count;
2846                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2847                                                 --$matchingCount;
2848                                         }
2849                                 }
2850
2851                                 if ($matchingCount <= 0) {
2852                                         # No matching element found in callback array
2853                                         # Output a literal closing brace and continue
2854                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2855                                         $i += $count;
2856                                         continue;
2857                                 }
2858                                 $name = $rule['names'][$matchingCount];
2859                                 if ( $name === null ) {
2860                                         // No element, just literal text
2861                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2862                                                 implode( '|', $piece['parts'] ) .
2863                                                 str_repeat( $rule['end'], $matchingCount );
2864                                 } else {
2865                                         # Create XML element
2866                                         # Note: $parts is already XML, does not need to be encoded further
2867                                         $parts = $piece['parts'];
2868                                         $title = $parts[0];
2869                                         unset( $parts[0] );
2870
2871                                         # The invocation is at the start of the line if lineStart is set in
2872                                         # the stack, and all opening brackets are used up.
2873                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
2874                                                 $attr = ' lineStart="1"';
2875                                         } else {
2876                                                 $attr = '';
2877                                         }
2878
2879                                         $element = "<$name$attr>";
2880                                         $element .= "<title>$title</title>";
2881                                         $argIndex = 1;
2882                                         foreach ( $parts as $partIndex => $part ) {
2883                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
2884                                                         $eqpos = $piece['eqpos'][$partIndex];
2885                                                         list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
2886                                                         list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
2887                                                         $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
2888                                                 } else {
2889                                                         list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
2890                                                         $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
2891                                                         $argIndex++;
2892                                                 }
2893                                         }
2894                                         $element .= "</$name>";
2895                                 }
2896
2897                                 # Advance input pointer
2898                                 $i += $matchingCount;
2899
2900                                 # Unwind the stack
2901                                 unset( $stack[$stackIndex--] );
2902                                 if ( $stackIndex == -1 ) {
2903                                         $accum =& $topAccum;
2904                                         $findEquals = false;
2905                                         $findPipe = false;
2906                                 } else {
2907                                         $partCount = count( $stack[$stackIndex]['parts'] );
2908                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
2909                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
2910                                         $findEquals = $findPipe && $partCount > 1
2911                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
2912                                 }
2913
2914                                 # Re-add the old stack element if it still has unmatched opening characters remaining
2915                                 if ($matchingCount < $piece['count']) {
2916                                         $piece['parts'] = array( '' );
2917                                         $piece['count'] -= $matchingCount;
2918                                         $piece['eqpos'] = array();
2919                                         # do we still qualify for any callback with remaining count?
2920                                         $names = $rules[$piece['open']]['names'];
2921                                         $skippedBraces = 0;
2922                                         $enclosingAccum =& $accum;
2923                                         while ( $piece['count'] ) {
2924                                                 if ( array_key_exists( $piece['count'], $names ) ) {
2925                                                         $stackIndex++;
2926                                                         $stack[$stackIndex] = $piece;
2927                                                         $accum =& $stack[$stackIndex]['parts'][0];
2928                                                         $findEquals = true;
2929                                                         $findPipe = true;
2930                                                         break;
2931                                                 }
2932                                                 --$piece['count'];
2933                                                 $skippedBraces ++;
2934                                         }
2935                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
2936                                 }
2937
2938                                 # Add XML element to the enclosing accumulator
2939                                 $accum .= $element;
2940                         }
2941
2942                         elseif ( $found == 'pipe' ) {
2943                                 $stack[$stackIndex]['parts'][] = '';
2944                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2945                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
2946                                 $findEquals = true;
2947                                 ++$i;
2948                         }
2949
2950                         elseif ( $found == 'equals' ) {
2951                                 $findEquals = false;
2952                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2953                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
2954                                 $accum .= '=';
2955                                 ++$i;
2956                         }
2957                 }
2958
2959                 # Output any remaining unclosed brackets
2960                 foreach ( $stack as $piece ) {
2961                         if ( $piece['open'] == "\n" ) {
2962                                 $topAccum .= $piece['parts'][0];
2963                         } else {
2964                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
2965                         }
2966                 }
2967                 $topAccum .= '</root>';
2968
2969                 wfProfileOut( __METHOD__.'-makexml' );
2970                 wfProfileIn( __METHOD__.'-loadXML' );
2971                 $dom = new DOMDocument;
2972                 if ( !$dom->loadXML( $topAccum ) ) {
2973                         throw new MWException( __METHOD__.' generated invalid XML' );
2974                 }
2975                 wfProfileOut( __METHOD__.'-loadXML' );
2976                 wfProfileOut( __METHOD__ );
2977                 return $dom;
2978         }
2979
2980         /*
2981          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2982          */
2983         public static function splitWhitespace( $s ) {
2984                 $ltrimmed = ltrim( $s );
2985                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2986                 $trimmed = rtrim( $ltrimmed );
2987                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2988                 if ( $diff > 0 ) {
2989                         $w2 = substr( $ltrimmed, -$diff );
2990                 } else {
2991                         $w2 = '';
2992                 }
2993                 return array( $w1, $trimmed, $w2 );
2994         }
2995
2996         /**
2997          * Replace magic variables, templates, and template arguments
2998          * with the appropriate text. Templates are substituted recursively,
2999          * taking care to avoid infinite loops.
3000          *
3001          * Note that the substitution depends on value of $mOutputType:
3002          *  OT_WIKI: only {{subst:}} templates
3003          *  OT_MSG: only magic variables
3004          *  OT_HTML: all templates and magic variables
3005          *
3006          * @param string $tex The text to transform
3007          * @param PPFrame $frame Object describing the arguments passed to the template
3008          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3009          * @private
3010          */
3011         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3012                 # Prevent too big inclusions
3013                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3014                         return $text;
3015                 }
3016
3017                 $fname = __METHOD__;
3018                 wfProfileIn( $fname );
3019
3020                 if ( $frame === false ) {
3021                         $frame = new PPFrame( $this );
3022                 } elseif ( !( $frame instanceof PPFrame ) ) {
3023                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3024                 }
3025
3026                 $dom = $this->preprocessToDom( $text );
3027                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3028                 $text = $frame->expand( $dom, 0, $flags );
3029
3030                 wfProfileOut( $fname );
3031                 return $text;
3032         }
3033
3034         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3035         static function createAssocArgs( $args ) {
3036                 $assocArgs = array();
3037                 $index = 1;
3038                 foreach( $args as $arg ) {
3039                         $eqpos = strpos( $arg, '=' );
3040                         if ( $eqpos === false ) {
3041                                 $assocArgs[$index++] = $arg;
3042                         } else {
3043                                 $name = trim( substr( $arg, 0, $eqpos ) );
3044                                 $value = trim( substr( $arg, $eqpos+1 ) );
3045                                 if ( $value === false ) {
3046                                         $value = '';
3047                                 }
3048                                 if ( $name !== false ) {
3049                                         $assocArgs[$name] = $value;
3050                                 }
3051                         }
3052                 }
3053
3054                 return $assocArgs;
3055         }
3056
3057         /**
3058          * Return the text of a template, after recursively
3059          * replacing any variables or templates within the template.
3060          *
3061          * @param array $piece The parts of the template
3062          *  $piece['text']: matched text
3063          *  $piece['title']: the title, i.e. the part before the |
3064          *  $piece['parts']: the parameter array
3065          * @param PPFrame The current frame, contains template arguments
3066          * @return string the text of the template
3067          * @private
3068          */
3069         function braceSubstitution( $piece, $frame ) {
3070                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3071                 $fname = __METHOD__;
3072                 wfProfileIn( $fname );
3073                 wfProfileIn( __METHOD__.'-setup' );
3074
3075                 # Flags
3076                 $found = false;             # $text has been filled
3077                 $nowiki = false;            # wiki markup in $text should be escaped
3078                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
3079                 $noargs = false;            # Don't replace triple-brace arguments in $text
3080                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3081                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3082                 $isDOM = false;             # $text is a DOM node needing expansion
3083
3084                 # Title object, where $text came from
3085                 $title = NULL;
3086
3087                 # $part1 is the bit before the first |, and must contain only title characters.
3088                 # Various prefixes will be stripped from it later.
3089                 $titleWithSpaces = $frame->expand( $piece['title'] );
3090                 $part1 = trim( $titleWithSpaces );
3091                 $titleText = false;
3092
3093                 # Original title text preserved for various purposes
3094                 $originalTitle = $part1;
3095
3096                 # $args is a list of argument nodes, starting from index 0, not including $part1
3097                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3098                 wfProfileOut( __METHOD__.'-setup' );
3099
3100                 # SUBST
3101                 wfProfileIn( __METHOD__.'-modifiers' );
3102                 if ( !$found ) {
3103                         $mwSubst =& MagicWord::get( 'subst' );
3104                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3105                                 # One of two possibilities is true:
3106                                 # 1) Found SUBST but not in the PST phase
3107                                 # 2) Didn't find SUBST and in the PST phase
3108                                 # In either case, return without further processing
3109                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3110                                 $found = true;
3111                                 $noparse = true;
3112                                 $noargs = true;
3113                         }
3114                 }
3115
3116                 # Variables
3117                 if ( !$found && $args->length == 0 ) {
3118                         $id = $this->mVariables->matchStartToEnd( $part1 );
3119                         if ( $id !== false ) {
3120                                 $text = $this->getVariableValue( $id );
3121                                 $this->mOutput->mContainsOldMagic = true;
3122                                 $found = true;
3123                                 $noparse = true;
3124                                 $noargs = true;
3125                         }
3126                 }
3127
3128                 # MSG, MSGNW and RAW
3129                 if ( !$found ) {
3130                         # Check for MSGNW:
3131                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3132                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3133                                 $nowiki = true;
3134                         } else {
3135                                 # Remove obsolete MSG:
3136                                 $mwMsg =& MagicWord::get( 'msg' );
3137                                 $mwMsg->matchStartAndRemove( $part1 );
3138                         }
3139
3140                         # Check for RAW:
3141                         $mwRaw =& MagicWord::get( 'raw' );
3142                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3143                                 $forceRawInterwiki = true;
3144                         }
3145                 }
3146                 wfProfileOut( __METHOD__.'-modifiers' );
3147
3148                 # Save path level before recursing into functions & templates.
3149                 $lastPathLevel = $this->mTemplatePath;
3150
3151                 # Parser functions
3152                 if ( !$found ) {
3153                         wfProfileIn( __METHOD__ . '-pfunc' );
3154
3155                         $colonPos = strpos( $part1, ':' );
3156                         if ( $colonPos !== false ) {
3157                                 # Case sensitive functions
3158                                 $function = substr( $part1, 0, $colonPos );
3159                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3160                                         $function = $this->mFunctionSynonyms[1][$function];
3161                                 } else {
3162                                         # Case insensitive functions
3163                                         $function = strtolower( $function );
3164                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3165                                                 $function = $this->mFunctionSynonyms[0][$function];
3166                                         } else {
3167                                                 $function = false;
3168                                         }
3169                                 }
3170                                 if ( $function ) {
3171                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3172                                         $initialArgs = array( &$this );
3173                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3174                                         if ( $flags & SFH_OBJECT_ARGS ) {
3175                                                 # Add a frame parameter, and pass the arguments as an array
3176                                                 $allArgs = $initialArgs;
3177                                                 $allArgs[] = $frame;
3178                                                 foreach ( $args as $arg ) {
3179                                                         $funcArgs[] = $arg;
3180                                                 }
3181                                                 $allArgs[] = $funcArgs;
3182                                         } else {
3183                                                 # Convert arguments to plain text
3184                                                 foreach ( $args as $arg ) {
3185                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3186                                                 }
3187                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3188                                         }
3189
3190                                         $result = call_user_func_array( $callback, $allArgs );
3191                                         $found = true;
3192
3193                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3194                                         $noargs = true;
3195                                         $noparse = true;
3196
3197                                         if ( is_array( $result ) ) {
3198                                                 if ( isset( $result[0] ) ) {
3199                                                         $text = $result[0];
3200                                                         unset( $result[0] );
3201                                                 }
3202
3203                                                 // Extract flags into the local scope
3204                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3205                                                 extract( $result );
3206                                         } else {
3207                                                 $text = $result;
3208                                         }
3209                                 }
3210                         }
3211                         wfProfileOut( __METHOD__ . '-pfunc' );
3212                 }
3213
3214                 # Finish mangling title and then check for loops.
3215                 # Set $title to a Title object and $titleText to the PDBK
3216                 if ( !$found ) {
3217                         $ns = NS_TEMPLATE;
3218                         # Split the title into page and subpage
3219                         $subpage = '';
3220                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3221                         if ($subpage !== '') {
3222                                 $ns = $this->mTitle->getNamespace();
3223                         }
3224                         $title = Title::newFromText( $part1, $ns );
3225                         if ( $title ) {
3226                                 $titleText = $title->getPrefixedText();
3227                                 # Check for language variants if the template is not found
3228                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3229                                         $wgContLang->findVariantLink($part1, $title);
3230                                 }
3231                                 # Do infinite loop check
3232                                 if ( isset( $this->mTemplatePath[$titleText] ) ) {
3233                                         $noparse = true;
3234                                         $noargs = true;
3235                                         $found = true;
3236                                         $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
3237                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3238                                 }
3239                         }
3240                 }
3241
3242                 # Load from database
3243                 if ( !$found && $title ) {
3244                         wfProfileIn( __METHOD__ . '-loadtpl' );
3245                         if ( !$title->isExternal() ) {
3246                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3247                                         $text = SpecialPage::capturePath( $title );
3248                                         if ( is_string( $text ) ) {
3249                                                 $found = true;
3250                                                 $noparse = true;
3251                                                 $noargs = true;
3252                                                 $isHTML = true;
3253                                                 $this->disableCache();
3254                                         }
3255                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3256                                         $found = false; //access denied
3257                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3258                                 } else {
3259                                         list( $text, $title ) = $this->getTemplateDom( $title );
3260                                         if ( $text !== false ) {
3261                                                 $found = true;
3262                                                 $isDOM = true;
3263                                         }
3264                                 }
3265
3266                                 # If the title is valid but undisplayable, make a link to it
3267                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3268                                         $text = "[[:$titleText]]";
3269                                         $found = true;
3270                                 }
3271                         } elseif ( $title->isTrans() ) {
3272                                 // Interwiki transclusion
3273                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3274                                         $text = $this->interwikiTransclude( $title, 'render' );
3275                                         $isHTML = true;
3276                                         $noparse = true;
3277                                 } else {
3278                                         $text = $this->interwikiTransclude( $title, 'raw' );
3279                                 }
3280                                 $found = true;
3281                         }
3282                         wfProfileOut( __METHOD__ . '-loadtpl' );
3283                 }
3284
3285                 # Recursive parsing, escaping and link table handling
3286                 # Only for HTML output
3287                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3288                         if ( $isDOM ) {
3289                                 $text = $frame->expand( $text );
3290                         }
3291                         $text = wfEscapeWikiText( $text );
3292                 } elseif ( !$this->ot['msg'] && $found ) {
3293                         if ( $noargs ) {
3294                                 $newFrame = $frame->newChild();
3295                         } else {
3296                                 # Clean up argument array
3297                                 $newFrame = $frame->newChild( $args, $title );
3298                                 # Add a new element to the templace recursion path
3299                                 $this->mTemplatePath[$titleText] = 1;
3300                         }
3301
3302                         if ( !$noparse ) {
3303                                 if ( $isDOM ) {
3304                                         if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
3305                                                 # Expansion is eligible for the empty-frame cache
3306                                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3307                                                         $text = $this->mTplExpandCache[$titleText];
3308                                                 } else {
3309                                                         $text = $newFrame->expand( $text );
3310                                                         $this->mTplExpandCache[$titleText] = $text;
3311                                                 }
3312                                         } else {
3313                                                 $text = $newFrame->expand( $text );
3314                                         }
3315                                 } else {
3316                                         $text = $this->replaceVariables( $text, $newFrame );
3317                                 }
3318
3319                                 # strip woz 'ere 2004-07
3320
3321                                 # Bug 529: if the template begins with a table or block-level
3322                                 # element, it should be treated as beginning a new line.
3323                                 # This behaviour is somewhat controversial.
3324                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3325                                         $text = "\n" . $text;
3326                                 }
3327                         } elseif ( !$noargs ) {
3328                                 # $noparse and !$noargs
3329                                 # Just replace the arguments, not any double-brace items
3330                                 # This is used for rendered interwiki transclusion
3331                                 if ( $isDOM ) {
3332                                         $text = $newFrame->expand( $text, 0, PPFrame::NO_TEMPLATES );
3333                                 } else {
3334                                         $text = $this->replaceVariables( $text, $newFrame, true );
3335                                 }
3336                         } elseif ( $isDOM ) {
3337                                 $text = $frame->expand( $text );
3338                         }
3339                 } elseif ( $isDOM ) {
3340                         $text = $frame->expand( $text, 0, PPFrame::NO_TEMPLATES | PPFrame::NO_ARGS );
3341                 }
3342
3343                 # Prune lower levels off the recursion check path
3344                 $this->mTemplatePath = $lastPathLevel;
3345
3346                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3347                         # Error, oversize inclusion
3348                         $text = "[[$originalTitle]]" .
3349                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3350                         $noparse = true;
3351                         $noargs = true;
3352                 }
3353
3354                 if ( !$found ) {
3355                         wfProfileOut( $fname );
3356                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3357                 } else {
3358                         wfProfileIn( __METHOD__ . '-placeholders' );
3359                         if ( $isHTML ) {
3360                                 # Replace raw HTML by a placeholder
3361                                 # Add a blank line preceding, to prevent it from mucking up
3362                                 # immediately preceding headings
3363                                 $text = "\n\n" . $this->insertStripItem( $text );
3364                         }
3365                         wfProfileOut( __METHOD__ . '-placeholders' );
3366                 }
3367
3368                 # Prune lower levels off the recursion check path
3369                 $this->mTemplatePath = $lastPathLevel;
3370
3371                 if ( !$found ) {
3372                         wfProfileOut( $fname );
3373                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3374                 } else {
3375                         wfProfileOut( $fname );
3376                         return $text;
3377                 }
3378         }
3379
3380         /**
3381          * Get the semi-parsed DOM representation of a template with a given title,
3382          * and its redirect destination title. Cached.
3383          */
3384         function getTemplateDom( $title ) {
3385                 $cacheTitle = $title;
3386                 $titleText = $title->getPrefixedDBkey();
3387
3388                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3389                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3390                         $title = Title::makeTitle( $ns, $dbk );
3391                         $titleText = $title->getPrefixedDBkey();
3392                 }
3393                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3394                         return array( $this->mTplDomCache[$titleText], $title );
3395                 }
3396
3397                 // Cache miss, go to the database
3398                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3399
3400                 if ( $text === false ) {
3401                         $this->mTplDomCache[$titleText] = false;
3402                         return array( false, $title );
3403                 }
3404
3405                 # If there are any <onlyinclude> tags, only include them
3406                 if ( !$this->ot['msg'] ) {
3407                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3408                                 $replacer = new OnlyIncludeReplacer;
3409                                 StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3410                                         array( &$replacer, 'replace' ), $text );
3411                                 $text = $replacer->output;
3412                         }
3413                         # Remove <noinclude> sections and <includeonly> tags
3414                         $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3415                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3416                 }
3417
3418                 $dom = $this->preprocessToDom( $text );
3419
3420                 $this->mTplDomCache[ $titleText ] = $dom;
3421
3422                 if (! $title->equals($cacheTitle)) {
3423                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3424                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3425                 }
3426
3427                 return array( $dom, $title );
3428         }
3429
3430         /**
3431          * Fetch the unparsed text of a template and register a reference to it.
3432          */
3433         function fetchTemplateAndTitle( $title ) {
3434                 $templateCb = $this->mOptions->getTemplateCallback();
3435                 $stuff = call_user_func( $templateCb, $title );
3436                 $text = $stuff['text'];
3437                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3438                 if ( isset( $stuff['deps'] ) ) {
3439                         foreach ( $stuff['deps'] as $dep ) {
3440                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3441                         }
3442                 }
3443                 return array($text,$finalTitle);
3444         }
3445
3446         function fetchTemplate( $title ) {
3447                 $rv = $this->fetchTemplateAndTitle($title);
3448                 return $rv[0];
3449         }
3450
3451         /**
3452          * Static function to get a template
3453          * Can be overridden via ParserOptions::setTemplateCallback().
3454          */
3455         static function statelessFetchTemplate( $title ) {
3456                 $text = $skip = false;
3457                 $finalTitle = $title;
3458                 $deps = array();
3459
3460                 // Loop to fetch the article, with up to 1 redirect
3461                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3462                         # Give extensions a chance to select the revision instead
3463                         $id = false; // Assume current
3464                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3465
3466                         if( $skip ) {
3467                                 $text = false;
3468                                 $deps[] = array(
3469                                         'title' => $title,
3470                                         'page_id' => $title->getArticleID(),
3471                                         'rev_id' => null );
3472                                 break;
3473                         }
3474                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3475                         $rev_id = $rev ? $rev->getId() : 0;
3476
3477                         $deps[] = array(
3478                                 'title' => $title,
3479                                 'page_id' => $title->getArticleID(),
3480                                 'rev_id' => $rev_id );
3481
3482                         if( $rev ) {
3483                                 $text = $rev->getText();
3484                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3485                                 global $wgLang;
3486                                 $message = $wgLang->lcfirst( $title->getText() );
3487                                 $text = wfMsgForContentNoTrans( $message );
3488                                 if( wfEmptyMsg( $message, $text ) ) {
3489                                         $text = false;
3490                                         break;
3491                                 }
3492                         } else {
3493                                 break;
3494                         }
3495                         if ( $text === false ) {
3496                                 break;
3497                         }
3498                         // Redirect?
3499                         $finalTitle = $title;
3500                         $title = Title::newFromRedirect( $text );
3501                 }
3502                 return array(
3503                         'text' => $text,
3504                         'finalTitle' => $finalTitle,
3505                         'deps' => $deps );
3506         }
3507
3508         /**
3509          * Transclude an interwiki link.
3510          */
3511         function interwikiTransclude( $title, $action ) {
3512                 global $wgEnableScaryTranscluding;
3513
3514                 if (!$wgEnableScaryTranscluding)
3515                         return wfMsg('scarytranscludedisabled');
3516
3517                 $url = $title->getFullUrl( "action=$action" );
3518
3519                 if (strlen($url) > 255)
3520                         return wfMsg('scarytranscludetoolong');
3521                 return $this->fetchScaryTemplateMaybeFromCache($url);
3522         }
3523
3524         function fetchScaryTemplateMaybeFromCache($url) {
3525                 global $wgTranscludeCacheExpiry;
3526                 $dbr = wfGetDB(DB_SLAVE);
3527                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3528                                 array('tc_url' => $url));
3529                 if ($obj) {
3530                         $time = $obj->tc_time;
3531                         $text = $obj->tc_contents;
3532                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3533                                 return $text;
3534                         }
3535                 }
3536
3537                 $text = Http::get($url);
3538                 if (!$text)
3539                         return wfMsg('scarytranscludefailed', $url);
3540
3541                 $dbw = wfGetDB(DB_MASTER);
3542                 $dbw->replace('transcache', array('tc_url'), array(
3543                         'tc_url' => $url,
3544                         'tc_time' => time(),
3545                         'tc_contents' => $text));
3546                 return $text;
3547         }
3548
3549
3550         /**
3551          * Triple brace replacement -- used for template arguments
3552          * @private
3553          */
3554         function argSubstitution( $piece, $frame ) {
3555                 wfProfileIn( __METHOD__ );
3556
3557                 $text = false;
3558                 $error = false;
3559                 $parts = $piece['parts'];
3560                 $argWithSpaces = $frame->expand( $piece['title'] );
3561                 $arg = trim( $argWithSpaces );
3562
3563                 if ( isset( $frame->args[$arg] ) ) {
3564                         $text = $frame->parent->expand( $frame->args[$arg] );
3565                 } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3566                         $text = $frame->expand( $parts->item( 0 ) );
3567                 }
3568                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3569                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3570                 }
3571
3572                 if ( $text === false ) {
3573                         $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
3574                 }
3575                 if ( $error !== false ) {
3576                         $text .= $error;
3577                 }
3578
3579                 wfProfileOut( __METHOD__ );
3580                 return $text;
3581         }
3582
3583         /**
3584          * Return the text to be used for a given extension tag.
3585          * This is the ghost of strip().
3586          *
3587          * @param array $params Associative array of parameters:
3588          *     name       DOMNode for the tag name
3589          *     attrText   DOMNode for unparsed text where tag attributes are thought to be
3590          *     inner      Contents of extension element
3591          *     noClose    Original text did not have a close tag
3592          * @param PPFrame $frame
3593          */
3594         function extensionSubstitution( $params, $frame ) {
3595                 global $wgRawHtml, $wgContLang;
3596                 static $n = 1;
3597
3598                 $name = $frame->expand( $params['name'] );
3599                 $attrText = is_null( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3600                 $content = is_null( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3601
3602                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3603
3604                 if ( $this->ot['html'] ) {
3605                         if ( $name == '!--' ) {
3606                                 return '';
3607                         }
3608                         $name = strtolower( $name );
3609
3610                         $params = Sanitizer::decodeTagAttributes( $attrText );
3611                         switch ( $name ) {
3612                                 case 'html':
3613                                         if( $wgRawHtml ) {
3614                                                 $output = $content;
3615                                                 break;
3616                                         } else {
3617                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3618                                         }
3619                                 case 'nowiki':
3620                                         $output = Xml::escapeTagsOnly( $content );
3621                                         break;
3622                                 case 'math':
3623                                         $output = $wgContLang->armourMath(
3624                                                 MathRenderer::renderMath( $content, $params ) );
3625                                         break;
3626                                 case 'gallery':
3627                                         $output = $this->renderImageGallery( $content, $params );
3628                                         break;
3629                                 default:
3630                                         if( isset( $this->mTagHooks[$name] ) ) {
3631                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3632                                                         array( $content, $params, $this ) );
3633                                         } else {
3634                                                 throw new MWException( "Invalid call hook $name" );
3635                                         }
3636                         }
3637                 } else {
3638                         if ( $name == '!--' ) {
3639                                 $output = '<!--' . $content . '-->';
3640                         } else {
3641                                 if ( $content === null ) {
3642                                         $output = "<$name$attrText/>";
3643                                 } else {
3644                                         $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3645                                         $output = "<$name$attrText>$content$close";
3646                                 }
3647                         }
3648                 }
3649
3650                 if ( $name == 'html' || $name == 'nowiki' ) {
3651                         $this->mStripState->nowiki->setPair( $marker, $output );
3652                 } else {
3653                         $this->mStripState->general->setPair( $marker, $output );
3654                 }
3655                 return $marker;
3656         }
3657
3658         /**
3659          * Increment an include size counter
3660          *
3661          * @param string $type The type of expansion
3662          * @param integer $size The size of the text
3663          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3664          */
3665         function incrementIncludeSize( $type, $size ) {
3666                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3667                         return false;
3668                 } else {
3669                         $this->mIncludeSizes[$type] += $size;
3670                         return true;
3671                 }
3672         }
3673
3674         /**
3675          * Detect __NOGALLERY__ magic word and set a placeholder
3676          */
3677         function stripNoGallery( &$text ) {
3678                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3679                 # do not add TOC
3680                 $mw = MagicWord::get( 'nogallery' );
3681                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3682         }
3683
3684         /**
3685          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3686          * placeholder that will then be replaced by the real TOC in
3687          * ->formatHeadings, this works because at this points real
3688          * comments will have already been discarded by the sanitizer.
3689          *
3690          * Any additional __TOC__ magic words left over will be discarded
3691          * as there can only be one TOC on the page.
3692          */
3693         function stripToc( $text ) {
3694                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3695                 # do not add TOC
3696                 $mw = MagicWord::get( 'notoc' );
3697                 if( $mw->matchAndRemove( $text ) ) {
3698                         $this->mShowToc = false;
3699                 }
3700
3701                 $mw = MagicWord::get( 'toc' );
3702                 if( $mw->match( $text ) ) {
3703                         $this->mShowToc = true;
3704                         $this->mForceTocPosition = true;
3705
3706                         // Set a placeholder. At the end we'll fill it in with the TOC.
3707                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3708
3709                         // Only keep the first one.
3710                         $text = $mw->replace( '', $text );
3711                 }
3712                 return $text;
3713         }
3714
3715         /**
3716          * This function accomplishes several tasks:
3717          * 1) Auto-number headings if that option is enabled
3718          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3719          * 3) Add a Table of contents on the top for users who have enabled the option
3720          * 4) Auto-anchor headings
3721          *
3722          * It loops through all headlines, collects the necessary data, then splits up the
3723          * string and re-inserts the newly formatted headlines.
3724          *
3725          * @param string $text
3726          * @param boolean $isMain
3727          * @private
3728          */
3729         function formatHeadings( $text, $isMain=true ) {
3730                 global $wgMaxTocLevel, $wgContLang;
3731
3732                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3733                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3734                         $showEditLink = 0;
3735                 } else {
3736                         $showEditLink = $this->mOptions->getEditSection();
3737                 }
3738
3739                 # Inhibit editsection links if requested in the page
3740                 $esw =& MagicWord::get( 'noeditsection' );
3741                 if( $esw->matchAndRemove( $text ) ) {
3742                         $showEditLink = 0;
3743                 }
3744
3745                 # Get all headlines for numbering them and adding funky stuff like [edit]
3746                 # links - this is for later, but we need the number of headlines right now
3747                 $matches = array();
3748                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3749
3750                 # if there are fewer than 4 headlines in the article, do not show TOC
3751                 # unless it's been explicitly enabled.
3752                 $enoughToc = $this->mShowToc &&
3753                         (($numMatches >= 4) || $this->mForceTocPosition);
3754
3755                 # Allow user to stipulate that a page should have a "new section"
3756                 # link added via __NEWSECTIONLINK__
3757                 $mw =& MagicWord::get( 'newsectionlink' );
3758                 if( $mw->matchAndRemove( $text ) )
3759                         $this->mOutput->setNewSection( true );
3760
3761                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3762                 # override above conditions and always show TOC above first header
3763                 $mw =& MagicWord::get( 'forcetoc' );
3764                 if ($mw->matchAndRemove( $text ) ) {
3765                         $this->mShowToc = true;
3766                         $enoughToc = true;
3767                 }
3768
3769                 # We need this to perform operations on the HTML
3770                 $sk = $this->mOptions->getSkin();
3771
3772                 # headline counter
3773                 $headlineCount = 0;
3774                 $numVisible = 0;
3775
3776                 # Ugh .. the TOC should have neat indentation levels which can be
3777                 # passed to the skin functions. These are determined here
3778                 $toc = '';
3779                 $full = '';
3780                 $head = array();
3781                 $sublevelCount = array();
3782                 $levelCount = array();
3783                 $toclevel = 0;
3784                 $level = 0;
3785                 $prevlevel = 0;
3786                 $toclevel = 0;
3787                 $prevtoclevel = 0;
3788                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3789                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3790
3791                 foreach( $matches[3] as $headline ) {
3792                         $isTemplate = false;
3793                         $titleText = false;
3794                         $sectionIndex = false;
3795                         $numbering = '';
3796                         $markerMatches = array();
3797                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3798                                 $serial = $markerMatches[1];
3799                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3800                                 $isTemplate = ($titleText != $baseTitleText);
3801                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3802                         }
3803
3804                         if( $toclevel ) {
3805                                 $prevlevel = $level;
3806                                 $prevtoclevel = $toclevel;
3807                         }
3808                         $level = $matches[1][$headlineCount];
3809
3810                         if( $doNumberHeadings || $enoughToc ) {
3811
3812                                 if ( $level > $prevlevel ) {
3813                                         # Increase TOC level
3814                                         $toclevel++;
3815                                         $sublevelCount[$toclevel] = 0;
3816                                         if( $toclevel<$wgMaxTocLevel ) {
3817                                                 $prevtoclevel = $toclevel;
3818                                                 $toc .= $sk->tocIndent();
3819                                                 $numVisible++;
3820                                         }
3821                                 }
3822                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3823                                         # Decrease TOC level, find level to jump to
3824
3825                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3826                                                 # Can only go down to level 1
3827                                                 $toclevel = 1;
3828                                         } else {
3829                                                 for ($i = $toclevel; $i > 0; $i--) {
3830                                                         if ( $levelCount[$i] == $level ) {
3831                                                                 # Found last matching level
3832                                                                 $toclevel = $i;
3833                                                                 break;
3834                                                         }
3835                                                         elseif ( $levelCount[$i] < $level ) {
3836                                                                 # Found first matching level below current level
3837                                                                 $toclevel = $i + 1;
3838                                                                 break;
3839                                                         }
3840                                                 }
3841                                         }
3842                                         if( $toclevel<$wgMaxTocLevel ) {
3843                                                 if($prevtoclevel < $wgMaxTocLevel) {
3844                                                         # Unindent only if the previous toc level was shown :p
3845                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3846                                                 } else {
3847                                                         $toc .= $sk->tocLineEnd();
3848                                                 }
3849                                         }
3850                                 }
3851                                 else {
3852                                         # No change in level, end TOC line
3853                                         if( $toclevel<$wgMaxTocLevel ) {
3854                                                 $toc .= $sk->tocLineEnd();
3855                                         }
3856                                 }
3857
3858                                 $levelCount[$toclevel] = $level;
3859
3860                                 # count number of headlines for each level
3861                                 @$sublevelCount[$toclevel]++;
3862                                 $dot = 0;
3863                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3864                                         if( !empty( $sublevelCount[$i] ) ) {
3865                                                 if( $dot ) {
3866                                                         $numbering .= '.';
3867                                                 }
3868                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3869                                                 $dot = 1;
3870                                         }
3871                                 }
3872                         }
3873
3874                         # The safe header is a version of the header text safe to use for links
3875                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3876                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3877
3878                         # Remove link placeholders by the link text.
3879                         #     <!--LINK number-->
3880                         # turns into
3881                         #     link text with suffix
3882                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3883                                                             "\$this->mLinkHolders['texts'][\$1]",
3884                                                             $safeHeadline );
3885                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3886                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3887                                                             $safeHeadline );
3888
3889                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3890                         $tocline = preg_replace(
3891                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3892                                 array( '',                          '<$1>'),
3893                                 $safeHeadline
3894                         );
3895                         $tocline = trim( $tocline );
3896
3897                         # For the anchor, strip out HTML-y stuff period
3898                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3899                         $safeHeadline = trim( $safeHeadline );
3900
3901                         # Save headline for section edit hint before it's escaped
3902                         $headlineHint = $safeHeadline;
3903                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3904                         $refers[$headlineCount] = $safeHeadline;
3905
3906                         # count how many in assoc. array so we can track dupes in anchors
3907                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3908                         $refcount[$headlineCount] = $refers[$safeHeadline];
3909
3910                         # Don't number the heading if it is the only one (looks silly)
3911                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3912                                 # the two are different if the line contains a link
3913                                 $headline=$numbering . ' ' . $headline;
3914                         }
3915
3916                         # Create the anchor for linking from the TOC to the section
3917                         $anchor = $safeHeadline;
3918                         if($refcount[$headlineCount] > 1 ) {
3919                                 $anchor .= '_' . $refcount[$headlineCount];
3920                         }
3921                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3922                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3923                         }
3924                         # give headline the correct <h#> tag
3925                         if( $showEditLink && $sectionIndex !== false ) {
3926                                 if( $isTemplate )
3927                                         $editlink = $sk->editSectionLinkForOther($titleText, $sectionIndex);
3928                                 else
3929                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3930                         } else {
3931                                 $editlink = '';
3932                         }
3933                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3934
3935                         $headlineCount++;
3936                 }
3937
3938                 # Never ever show TOC if no headers
3939                 if( $numVisible < 1 ) {
3940                         $enoughToc = false;
3941                 }
3942
3943                 if( $enoughToc ) {
3944                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3945                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3946                         }
3947                         $toc = $sk->tocList( $toc );
3948                 }
3949
3950                 # split up and insert constructed headlines
3951
3952                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3953                 $i = 0;
3954
3955                 foreach( $blocks as $block ) {
3956                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3957                                 # This is the [edit] link that appears for the top block of text when
3958                                 # section editing is enabled
3959
3960                                 # Disabled because it broke block formatting
3961                                 # For example, a bullet point in the top line
3962                                 # $full .= $sk->editSectionLink(0);
3963                         }
3964                         $full .= $block;
3965                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3966                                 # Top anchor now in skin
3967                                 $full = $full.$toc;
3968                         }
3969
3970                         if( !empty( $head[$i] ) ) {
3971                                 $full .= $head[$i];
3972                         }
3973                         $i++;
3974                 }
3975                 if( $this->mForceTocPosition ) {
3976                         return str_replace( '<!--MWTOC-->', $toc, $full );
3977                 } else {
3978                         return $full;
3979                 }
3980         }
3981
3982         /**
3983          * Transform wiki markup when saving a page by doing \r\n -> \n
3984          * conversion, substitting signatures, {{subst:}} templates, etc.
3985          *
3986          * @param string $text the text to transform
3987          * @param Title &$title the Title object for the current article
3988          * @param User &$user the User object describing the current user
3989          * @param ParserOptions $options parsing options
3990          * @param bool $clearState whether to clear the parser state first
3991          * @return string the altered wiki markup
3992          * @public
3993          */
3994         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3995                 $this->mOptions = $options;
3996                 $this->mTitle =& $title;
3997                 $this->setOutputType( OT_WIKI );
3998
3999                 if ( $clearState ) {
4000                         $this->clearState();
4001                 }
4002
4003                 $pairs = array(
4004                         "\r\n" => "\n",
4005                 );
4006                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4007                 $text = $this->pstPass2( $text, $user );
4008                 $text = $this->mStripState->unstripBoth( $text );
4009                 return $text;
4010         }
4011
4012         /**
4013          * Pre-save transform helper function
4014          * @private
4015          */
4016         function pstPass2( $text, $user ) {
4017                 global $wgContLang, $wgLocaltimezone;
4018
4019                 /* Note: This is the timestamp saved as hardcoded wikitext to
4020                  * the database, we use $wgContLang here in order to give
4021                  * everyone the same signature and use the default one rather
4022                  * than the one selected in each user's preferences.
4023                  */
4024                 if ( isset( $wgLocaltimezone ) ) {
4025                         $oldtz = getenv( 'TZ' );
4026                         putenv( 'TZ='.$wgLocaltimezone );
4027                 }
4028                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4029                   ' (' . date( 'T' ) . ')';
4030                 if ( isset( $wgLocaltimezone ) ) {
4031                         putenv( 'TZ='.$oldtz );
4032                 }
4033
4034                 # Variable replacement
4035                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4036                 $text = $this->replaceVariables( $text );
4037
4038                 # Strip out <nowiki> etc. added via replaceVariables
4039                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4040
4041                 # Signatures
4042                 $sigText = $this->getUserSig( $user );
4043                 $text = strtr( $text, array(
4044                         '~~~~~' => $d,
4045                         '~~~~' => "$sigText $d",
4046                         '~~~' => $sigText
4047                 ) );
4048
4049                 # Context links: [[|name]] and [[name (context)|]]
4050                 #
4051                 global $wgLegalTitleChars;
4052                 $tc = "[$wgLegalTitleChars]";
4053                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4054
4055                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4056                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4057                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4058
4059                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4060                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4061                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4062
4063                 $t = $this->mTitle->getText();
4064                 $m = array();
4065                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4066                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4067                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4068                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4069                 } else {
4070                         # if there's no context, don't bother duplicating the title
4071                         $text = preg_replace( $p2, '[[\\1]]', $text );
4072                 }
4073
4074                 # Trim trailing whitespace
4075                 $text = rtrim( $text );
4076
4077                 return $text;
4078         }
4079
4080         /**
4081          * Fetch the user's signature text, if any, and normalize to
4082          * validated, ready-to-insert wikitext.
4083          *
4084          * @param User $user
4085          * @return string
4086          * @private
4087          */
4088         function getUserSig( &$user ) {
4089                 global $wgMaxSigChars;
4090
4091                 $username = $user->getName();
4092                 $nickname = $user->getOption( 'nickname' );
4093                 $nickname = $nickname === '' ? $username : $nickname;
4094
4095                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4096                         $nickname = $username;
4097                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4098                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4099                         # Sig. might contain markup; validate this
4100                         if( $this->validateSig( $nickname ) !== false ) {
4101                                 # Validated; clean up (if needed) and return it
4102                                 return $this->cleanSig( $nickname, true );
4103                         } else {
4104                                 # Failed to validate; fall back to the default
4105                                 $nickname = $username;
4106                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4107                         }
4108                 }
4109
4110                 // Make sure nickname doesnt get a sig in a sig
4111                 $nickname = $this->cleanSigInSig( $nickname );
4112
4113                 # If we're still here, make it a link to the user page
4114                 $userText = wfEscapeWikiText( $username );
4115                 $nickText = wfEscapeWikiText( $nickname );
4116                 if ( $user->isAnon() )  {
4117                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4118                 } else {
4119                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4120                 }
4121         }
4122
4123         /**
4124          * Check that the user's signature contains no bad XML
4125          *
4126          * @param string $text
4127          * @return mixed An expanded string, or false if invalid.
4128          */
4129         function validateSig( $text ) {
4130                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4131         }
4132
4133         /**
4134          * Clean up signature text
4135          *
4136          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4137          * 2) Substitute all transclusions
4138          *
4139          * @param string $text
4140          * @param $parsing Whether we're cleaning (preferences save) or parsing
4141          * @return string Signature text
4142          */
4143         function cleanSig( $text, $parsing = false ) {
4144                 global $wgTitle;
4145                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
4146
4147                 $substWord = MagicWord::get( 'subst' );
4148                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4149                 $substText = '{{' . $substWord->getSynonym( 0 );
4150
4151                 $text = preg_replace( $substRegex, $substText, $text );
4152                 $text = $this->cleanSigInSig( $text );
4153                 $text = $this->replaceVariables( $text );
4154
4155                 $this->clearState();
4156                 return $text;
4157         }
4158
4159         /**
4160          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4161          * @param string $text
4162          * @return string Signature text with /~{3,5}/ removed
4163          */
4164         function cleanSigInSig( $text ) {
4165                 $text = preg_replace( '/~{3,5}/', '', $text );
4166                 return $text;
4167         }
4168
4169         /**
4170          * Set up some variables which are usually set up in parse()
4171          * so that an external function can call some class members with confidence
4172          * @public
4173          */
4174         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4175                 $this->mTitle =& $title;
4176                 $this->mOptions = $options;
4177                 $this->setOutputType( $outputType );
4178                 if ( $clearState ) {
4179                         $this->clearState();
4180                 }
4181         }
4182
4183         /**
4184          * Transform a MediaWiki message by replacing magic variables.
4185          *
4186          * For some unknown reason, it also expands templates, but only to the
4187          * first recursion level. This is wrong and broken, probably introduced
4188          * accidentally during refactoring, but probably relied upon by thousands
4189          * of users.
4190          *
4191          * @param string $text the text to transform
4192          * @param ParserOptions $options  options
4193          * @return string the text with variables substituted
4194          * @public
4195          */
4196         function transformMsg( $text, $options ) {
4197                 global $wgTitle;
4198                 static $executing = false;
4199
4200                 $fname = "Parser::transformMsg";
4201
4202                 # Guard against infinite recursion
4203                 if ( $executing ) {
4204                         return $text;
4205                 }
4206                 $executing = true;
4207
4208                 wfProfileIn($fname);
4209
4210                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4211                         $this->mTitle = $wgTitle;
4212                 } else {
4213                         $this->mTitle = Title::newFromText('msg');
4214                 }
4215                 $this->mOptions = $options;
4216                 $this->setOutputType( OT_MSG );
4217                 $this->clearState();
4218                 $text = $this->replaceVariables( $text );
4219                 $text = $this->mStripState->unstripBoth( $text );
4220
4221                 $executing = false;
4222                 wfProfileOut($fname);
4223                 return $text;
4224         }
4225
4226         /**
4227          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4228          * The callback should have the following form:
4229          *    function myParserHook( $text, $params, &$parser ) { ... }
4230          *
4231          * Transform and return $text. Use $parser for any required context, e.g. use
4232          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4233          *
4234          * @public
4235          *
4236          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4237          * @param mixed $callback The callback function (and object) to use for the tag
4238          *
4239          * @return The old value of the mTagHooks array associated with the hook
4240          */
4241         function setHook( $tag, $callback ) {
4242                 $tag = strtolower( $tag );
4243                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4244                 $this->mTagHooks[$tag] = $callback;
4245                 $this->mStripList[] = $tag;
4246
4247                 return $oldVal;
4248         }
4249
4250         function setTransparentTagHook( $tag, $callback ) {
4251                 $tag = strtolower( $tag );
4252                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4253                 $this->mTransparentTagHooks[$tag] = $callback;
4254
4255                 return $oldVal;
4256         }
4257
4258         /**
4259          * Create a function, e.g. {{sum:1|2|3}}
4260          * The callback function should have the form:
4261          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4262          *
4263          * The callback may either return the text result of the function, or an array with the text
4264          * in element 0, and a number of flags in the other elements. The names of the flags are
4265          * specified in the keys. Valid flags are:
4266          *   found                     The text returned is valid, stop processing the template. This
4267          *                             is on by default.
4268          *   nowiki                    Wiki markup in the return value should be escaped
4269          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4270          *   noargs                    Don't replace triple-brace arguments in the return value
4271          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4272          *
4273          * @public
4274          *
4275          * @param string $id The magic word ID
4276          * @param mixed $callback The callback function (and object) to use
4277          * @param integer $flags a combination of the following flags:
4278          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4279          *
4280          * @return The old callback function for this name, if any
4281          */
4282         function setFunctionHook( $id, $callback, $flags = 0 ) {
4283                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4284                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4285
4286                 # Add to function cache
4287                 $mw = MagicWord::get( $id );
4288                 if( !$mw )
4289                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4290
4291                 $synonyms = $mw->getSynonyms();
4292                 $sensitive = intval( $mw->isCaseSensitive() );
4293
4294                 foreach ( $synonyms as $syn ) {
4295                         # Case
4296                         if ( !$sensitive ) {
4297                                 $syn = strtolower( $syn );
4298                         }
4299                         # Add leading hash
4300                         if ( !( $flags & SFH_NO_HASH ) ) {
4301                                 $syn = '#' . $syn;
4302                         }
4303                         # Remove trailing colon
4304                         if ( substr( $syn, -1, 1 ) == ':' ) {
4305                                 $syn = substr( $syn, 0, -1 );
4306                         }
4307                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4308                 }
4309                 return $oldVal;
4310         }
4311
4312         /**
4313          * Get all registered function hook identifiers
4314          *
4315          * @return array
4316          */
4317         function getFunctionHooks() {
4318                 return array_keys( $this->mFunctionHooks );
4319         }
4320
4321         /**
4322          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4323          * Placeholders created in Skin::makeLinkObj()
4324          * Returns an array of link CSS classes, indexed by PDBK.
4325          * $options is a bit field, RLH_FOR_UPDATE to select for update
4326          */
4327         function replaceLinkHolders( &$text, $options = 0 ) {
4328                 global $wgUser;
4329                 global $wgContLang;
4330
4331                 $fname = 'Parser::replaceLinkHolders';
4332                 wfProfileIn( $fname );
4333
4334                 $pdbks = array();
4335                 $colours = array();
4336                 $linkcolour_ids = array();
4337                 $sk = $this->mOptions->getSkin();
4338                 $linkCache =& LinkCache::singleton();
4339
4340                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4341                         wfProfileIn( $fname.'-check' );
4342                         $dbr = wfGetDB( DB_SLAVE );
4343                         $page = $dbr->tableName( 'page' );
4344                         $threshold = $wgUser->getOption('stubthreshold');
4345
4346                         # Sort by namespace
4347                         asort( $this->mLinkHolders['namespaces'] );
4348
4349                         # Generate query
4350                         $query = false;
4351                         $current = null;
4352                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4353                                 # Make title object
4354                                 $title = $this->mLinkHolders['titles'][$key];
4355
4356                                 # Skip invalid entries.
4357                                 # Result will be ugly, but prevents crash.
4358                                 if ( is_null( $title ) ) {
4359                                         continue;
4360                                 }
4361                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4362
4363                                 # Check if it's a static known link, e.g. interwiki
4364                                 if ( $title->isAlwaysKnown() ) {
4365                                         $colours[$pdbk] = '';
4366                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4367                                         $colours[$pdbk] = '';
4368                                         $this->mOutput->addLink( $title, $id );
4369                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4370                                         $colours[$pdbk] = 'new';
4371                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4372                                         $colours[$pdbk] = 'new';
4373                                 } else {
4374                                         # Not in the link cache, add it to the query
4375                                         if ( !isset( $current ) ) {
4376                                                 $current = $ns;
4377                                                 $query =  "SELECT page_id, page_namespace, page_title";
4378                                                 if ( $threshold > 0 ) {
4379                                                         $query .= ', page_len, page_is_redirect';
4380                                                 }
4381                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4382                                         } elseif ( $current != $ns ) {
4383                                                 $current = $ns;
4384                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4385                                         } else {
4386                                                 $query .= ', ';
4387                                         }
4388
4389                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4390                                 }
4391                         }
4392                         if ( $query ) {
4393                                 $query .= '))';
4394                                 if ( $options & RLH_FOR_UPDATE ) {
4395                                         $query .= ' FOR UPDATE';
4396                                 }
4397
4398                                 $res = $dbr->query( $query, $fname );
4399
4400                                 # Fetch data and form into an associative array
4401                                 # non-existent = broken
4402                                 while ( $s = $dbr->fetchObject($res) ) {
4403                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4404                                         $pdbk = $title->getPrefixedDBkey();
4405                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4406                                         $this->mOutput->addLink( $title, $s->page_id );
4407                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4408                                         //add id to the extension todolist
4409                                         $linkcolour_ids[$s->page_id] = $pdbk;
4410                                 }
4411                                 //pass an array of page_ids to an extension
4412                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4413                         }
4414                         wfProfileOut( $fname.'-check' );
4415
4416                         # Do a second query for different language variants of links and categories
4417                         if($wgContLang->hasVariants()){
4418                                 $linkBatch = new LinkBatch();
4419                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4420                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4421                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4422
4423                                 $categories = $this->mOutput->getCategoryLinks();
4424
4425                                 // Add variants of links to link batch
4426                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4427                                         $title = $this->mLinkHolders['titles'][$key];
4428                                         if ( is_null( $title ) )
4429                                                 continue;
4430
4431                                         $pdbk = $title->getPrefixedDBkey();
4432                                         $titleText = $title->getText();
4433
4434                                         // generate all variants of the link title text
4435                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4436
4437                                         // if link was not found (in first query), add all variants to query
4438                                         if ( !isset($colours[$pdbk]) ){
4439                                                 foreach($allTextVariants as $textVariant){
4440                                                         if($textVariant != $titleText){
4441                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4442                                                                 if(is_null($variantTitle)) continue;
4443                                                                 $linkBatch->addObj( $variantTitle );
4444                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4445                                                         }
4446                                                 }
4447                                         }
4448                                 }
4449
4450                                 // process categories, check if a category exists in some variant
4451                                 foreach( $categories as $category ){
4452                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4453                                         foreach($variants as $variant){
4454                                                 if($variant != $category){
4455                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4456                                                         if(is_null($variantTitle)) continue;
4457                                                         $linkBatch->addObj( $variantTitle );
4458                                                         $categoryMap[$variant] = $category;
4459                                                 }
4460                                         }
4461                                 }
4462
4463
4464                                 if(!$linkBatch->isEmpty()){
4465                                         // construct query
4466                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4467
4468                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4469                                         if ( $threshold > 0 ) {
4470                                                 $variantQuery .= ', page_len, page_is_redirect';
4471                                         }
4472
4473                                         $variantQuery .= " FROM $page WHERE $titleClause";
4474                                         if ( $options & RLH_FOR_UPDATE ) {
4475                                                 $variantQuery .= ' FOR UPDATE';
4476                                         }
4477
4478                                         $varRes = $dbr->query( $variantQuery, $fname );
4479
4480                                         // for each found variants, figure out link holders and replace
4481                                         while ( $s = $dbr->fetchObject($varRes) ) {
4482
4483                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4484                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4485                                                 $vardbk = $variantTitle->getDBkey();
4486
4487                                                 $holderKeys = array();
4488                                                 if(isset($variantMap[$varPdbk])){
4489                                                         $holderKeys = $variantMap[$varPdbk];
4490                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4491                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4492                                                 }
4493
4494                                                 // loop over link holders
4495                                                 foreach($holderKeys as $key){
4496                                                         $title = $this->mLinkHolders['titles'][$key];
4497                                                         if ( is_null( $title ) ) continue;
4498
4499                                                         $pdbk = $title->getPrefixedDBkey();
4500
4501                                                         if(!isset($colours[$pdbk])){
4502                                                                 // found link in some of the variants, replace the link holder data
4503                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4504                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4505
4506                                                                 // set pdbk and colour
4507                                                                 $pdbks[$key] = $varPdbk;
4508                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4509                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4510                                                         }
4511                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4512                                                 }
4513
4514                                                 // check if the object is a variant of a category
4515                                                 if(isset($categoryMap[$vardbk])){
4516                                                         $oldkey = $categoryMap[$vardbk];
4517                                                         if($oldkey != $vardbk)
4518                                                                 $varCategories[$oldkey]=$vardbk;
4519                                                 }
4520                                         }
4521
4522                                         // rebuild the categories in original order (if there are replacements)
4523                                         if(count($varCategories)>0){
4524                                                 $newCats = array();
4525                                                 $originalCats = $this->mOutput->getCategories();
4526                                                 foreach($originalCats as $cat => $sortkey){
4527                                                         // make the replacement
4528                                                         if( array_key_exists($cat,$varCategories) )
4529                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4530                                                         else $newCats[$cat] = $sortkey;
4531                                                 }
4532                                                 $this->mOutput->setCategoryLinks($newCats);
4533                                         }
4534                                 }
4535                         }
4536
4537                         # Construct search and replace arrays
4538                         wfProfileIn( $fname.'-construct' );
4539                         $replacePairs = array();
4540                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4541                                 $pdbk = $pdbks[$key];
4542                                 $searchkey = "<!--LINK $key-->";
4543                                 $title = $this->mLinkHolders['titles'][$key];
4544                                 if ( !isset( $colours[$pdbk] ) ) {
4545                                         $linkCache->addBadLinkObj( $title );
4546                                         $colours[$pdbk] = 'new';
4547                                         $this->mOutput->addLink( $title, 0 );
4548                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4549                                                                         $this->mLinkHolders['texts'][$key],
4550                                                                         $this->mLinkHolders['queries'][$key] );
4551                                 } else {
4552                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4553                                                                         $this->mLinkHolders['texts'][$key],
4554                                                                         $this->mLinkHolders['queries'][$key] );
4555                                 }
4556                         }
4557                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4558                         wfProfileOut( $fname.'-construct' );
4559
4560                         # Do the thing
4561                         wfProfileIn( $fname.'-replace' );
4562                         $text = preg_replace_callback(
4563                                 '/(<!--LINK .*?-->)/',
4564                                 $replacer->cb(),
4565                                 $text);
4566
4567                         wfProfileOut( $fname.'-replace' );
4568                 }
4569
4570                 # Now process interwiki link holders
4571                 # This is quite a bit simpler than internal links
4572                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4573                         wfProfileIn( $fname.'-interwiki' );
4574                         # Make interwiki link HTML
4575                         $replacePairs = array();
4576                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4577                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4578                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4579                         }
4580                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4581
4582                         $text = preg_replace_callback(
4583                                 '/<!--IWLINK (.*?)-->/',
4584                                 $replacer->cb(),
4585                                 $text );
4586                         wfProfileOut( $fname.'-interwiki' );
4587                 }
4588
4589                 wfProfileOut( $fname );
4590                 return $colours;
4591         }
4592
4593         /**
4594          * Replace <!--LINK--> link placeholders with plain text of links
4595          * (not HTML-formatted).
4596          * @param string $text
4597          * @return string
4598          */
4599         function replaceLinkHoldersText( $text ) {
4600                 $fname = 'Parser::replaceLinkHoldersText';
4601                 wfProfileIn( $fname );
4602
4603                 $text = preg_replace_callback(
4604                         '/<!--(LINK|IWLINK) (.*?)-->/',
4605                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4606                         $text );
4607
4608                 wfProfileOut( $fname );
4609                 return $text;
4610         }
4611
4612         /**
4613          * @param array $matches
4614          * @return string
4615          * @private
4616          */
4617         function replaceLinkHoldersTextCallback( $matches ) {
4618                 $type = $matches[1];
4619                 $key  = $matches[2];
4620                 if( $type == 'LINK' ) {
4621                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4622                                 return $this->mLinkHolders['texts'][$key];
4623                         }
4624                 } elseif( $type == 'IWLINK' ) {
4625                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4626                                 return $this->mInterwikiLinkHolders['texts'][$key];
4627                         }
4628                 }
4629                 return $matches[0];
4630         }
4631
4632         /**
4633          * Tag hook handler for 'pre'.
4634          */
4635         function renderPreTag( $text, $attribs ) {
4636                 // Backwards-compatibility hack
4637                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4638
4639                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4640                 return wfOpenElement( 'pre', $attribs ) .
4641                         Xml::escapeTagsOnly( $content ) .
4642                         '</pre>';
4643         }
4644
4645         /**
4646          * Renders an image gallery from a text with one line per image.
4647          * text labels may be given by using |-style alternative text. E.g.
4648          *   Image:one.jpg|The number "1"
4649          *   Image:tree.jpg|A tree
4650          * given as text will return the HTML of a gallery with two images,
4651          * labeled 'The number "1"' and
4652          * 'A tree'.
4653          */
4654         function renderImageGallery( $text, $params ) {
4655                 $ig = new ImageGallery();
4656                 $ig->setContextTitle( $this->mTitle );
4657                 $ig->setShowBytes( false );
4658                 $ig->setShowFilename( false );
4659                 $ig->setParser( $this );
4660                 $ig->setHideBadImages();
4661                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4662                 $ig->useSkin( $this->mOptions->getSkin() );
4663                 $ig->mRevisionId = $this->mRevisionId;
4664
4665                 if( isset( $params['caption'] ) ) {
4666                         $caption = $params['caption'];
4667                         $caption = htmlspecialchars( $caption );
4668                         $caption = $this->replaceInternalLinks( $caption );
4669                         $ig->setCaptionHtml( $caption );
4670                 }
4671                 if( isset( $params['perrow'] ) ) {
4672                         $ig->setPerRow( $params['perrow'] );
4673                 }
4674                 if( isset( $params['widths'] ) ) {
4675                         $ig->setWidths( $params['widths'] );
4676                 }
4677                 if( isset( $params['heights'] ) ) {
4678                         $ig->setHeights( $params['heights'] );
4679                 }
4680
4681                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4682
4683                 $lines = explode( "\n", $text );
4684                 foreach ( $lines as $line ) {
4685                         # match lines like these:
4686                         # Image:someimage.jpg|This is some image
4687                         $matches = array();
4688                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4689                         # Skip empty lines
4690                         if ( count( $matches ) == 0 ) {
4691                                 continue;
4692                         }
4693                         $tp = Title::newFromText( $matches[1] );
4694                         $nt =& $tp;
4695                         if( is_null( $nt ) ) {
4696                                 # Bogus title. Ignore these so we don't bomb out later.
4697                                 continue;
4698                         }
4699                         if ( isset( $matches[3] ) ) {
4700                                 $label = $matches[3];
4701                         } else {
4702                                 $label = '';
4703                         }
4704
4705                         $pout = $this->parse( $label,
4706                                 $this->mTitle,
4707                                 $this->mOptions,
4708                                 false, // Strip whitespace...?
4709                                 false  // Don't clear state!
4710                         );
4711                         $html = $pout->getText();
4712
4713                         $ig->add( $nt, $html );
4714
4715                         # Only add real images (bug #5586)
4716                         if ( $nt->getNamespace() == NS_IMAGE ) {
4717                                 $this->mOutput->addImage( $nt->getDBkey() );
4718                         }
4719                 }
4720                 return $ig->toHTML();
4721         }
4722
4723         function getImageParams( $handler ) {
4724                 if ( $handler ) {
4725                         $handlerClass = get_class( $handler );
4726                 } else {
4727                         $handlerClass = '';
4728                 }
4729                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4730                         // Initialise static lists
4731                         static $internalParamNames = array(
4732                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4733                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4734                                         'bottom', 'text-bottom' ),
4735                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4736                                         'upright', 'border' ),
4737                         );
4738                         static $internalParamMap;
4739                         if ( !$internalParamMap ) {
4740                                 $internalParamMap = array();
4741                                 foreach ( $internalParamNames as $type => $names ) {
4742                                         foreach ( $names as $name ) {
4743                                                 $magicName = str_replace( '-', '_', "img_$name" );
4744                                                 $internalParamMap[$magicName] = array( $type, $name );
4745                                         }
4746                                 }
4747                         }
4748
4749                         // Add handler params
4750                         $paramMap = $internalParamMap;
4751                         if ( $handler ) {
4752                                 $handlerParamMap = $handler->getParamMap();
4753                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4754                                         $paramMap[$magic] = array( 'handler', $paramName );
4755                                 }
4756                         }
4757                         $this->mImageParams[$handlerClass] = $paramMap;
4758                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4759                 }
4760                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4761         }
4762
4763         /**
4764          * Parse image options text and use it to make an image
4765          */
4766         function makeImage( $title, $options ) {
4767                 # @TODO: let the MediaHandler specify its transform parameters
4768                 #
4769                 # Check if the options text is of the form "options|alt text"
4770                 # Options are:
4771                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4772                 #  * left               no resizing, just left align. label is used for alt= only
4773                 #  * right              same, but right aligned
4774                 #  * none               same, but not aligned
4775                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4776                 #  * center             center the image
4777                 #  * framed             Keep original image size, no magnify-button.
4778                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4779                 #  * upright            reduce width for upright images, rounded to full __0 px
4780                 #  * border             draw a 1px border around the image
4781                 # vertical-align values (no % or length right now):
4782                 #  * baseline
4783                 #  * sub
4784                 #  * super
4785                 #  * top
4786                 #  * text-top
4787                 #  * middle
4788                 #  * bottom
4789                 #  * text-bottom
4790
4791                 $parts = array_map( 'trim', explode( '|', $options) );
4792                 $sk = $this->mOptions->getSkin();
4793
4794                 # Give extensions a chance to select the file revision for us
4795                 $skip = $time = false;
4796                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4797
4798                 if ( $skip ) {
4799                         return $sk->makeLinkObj( $title );
4800                 }
4801
4802                 # Get parameter map
4803                 $file = wfFindFile( $title, $time );
4804                 $handler = $file ? $file->getHandler() : false;
4805
4806                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4807
4808                 # Process the input parameters
4809                 $caption = '';
4810                 $params = array( 'frame' => array(), 'handler' => array(),
4811                         'horizAlign' => array(), 'vertAlign' => array() );
4812                 foreach( $parts as $part ) {
4813                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4814                         if ( isset( $paramMap[$magicName] ) ) {
4815                                 list( $type, $paramName ) = $paramMap[$magicName];
4816                                 $params[$type][$paramName] = $value;
4817
4818                                 // Special case; width and height come in one variable together
4819                                 if( $type == 'handler' && $paramName == 'width' ) {
4820                                         $m = array();
4821                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4822                                                 $params[$type]['width'] = intval( $m[1] );
4823                                                 $params[$type]['height'] = intval( $m[2] );
4824                                         } else {
4825                                                 $params[$type]['width'] = intval( $value );
4826                                         }
4827                                 }
4828                         } else {
4829                                 $caption = $part;
4830                         }
4831                 }
4832
4833                 # Process alignment parameters
4834                 if ( $params['horizAlign'] ) {
4835                         $params['frame']['align'] = key( $params['horizAlign'] );
4836                 }
4837                 if ( $params['vertAlign'] ) {
4838                         $params['frame']['valign'] = key( $params['vertAlign'] );
4839                 }
4840
4841                 # Validate the handler parameters
4842                 if ( $handler ) {
4843                         foreach ( $params['handler'] as $name => $value ) {
4844                                 if ( !$handler->validateParam( $name, $value ) ) {
4845                                         unset( $params['handler'][$name] );
4846                                 }
4847                         }
4848                 }
4849
4850                 # Strip bad stuff out of the alt text
4851                 $alt = $this->replaceLinkHoldersText( $caption );
4852
4853                 # make sure there are no placeholders in thumbnail attributes
4854                 # that are later expanded to html- so expand them now and
4855                 # remove the tags
4856                 $alt = $this->mStripState->unstripBoth( $alt );
4857                 $alt = Sanitizer::stripAllTags( $alt );
4858
4859                 $params['frame']['alt'] = $alt;
4860                 $params['frame']['caption'] = $caption;
4861
4862                 # Linker does the rest
4863                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4864
4865                 # Give the handler a chance to modify the parser object
4866                 if ( $handler ) {
4867                         $handler->parserTransformHook( $this, $file );
4868                 }
4869
4870                 return $ret;
4871         }
4872
4873         /**
4874          * Set a flag in the output object indicating that the content is dynamic and
4875          * shouldn't be cached.
4876          */
4877         function disableCache() {
4878                 wfDebug( "Parser output marked as uncacheable.\n" );
4879                 $this->mOutput->mCacheTime = -1;
4880         }
4881
4882         /**#@+
4883          * Callback from the Sanitizer for expanding items found in HTML attribute
4884          * values, so they can be safely tested and escaped.
4885          * @param string $text
4886          * @param PPFrame $frame
4887          * @return string
4888          * @private
4889          */
4890         function attributeStripCallback( &$text, $frame = false ) {
4891                 $text = $this->replaceVariables( $text, $frame );
4892                 $text = $this->mStripState->unstripBoth( $text );
4893                 return $text;
4894         }
4895
4896         /**#@-*/
4897
4898         /**#@+
4899          * Accessor/mutator
4900          */
4901         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4902         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4903         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4904         /**#@-*/
4905
4906         /**#@+
4907          * Accessor
4908          */
4909         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4910         /**#@-*/
4911
4912
4913         /**
4914          * Break wikitext input into sections, and either pull or replace
4915          * some particular section's text.
4916          *
4917          * External callers should use the getSection and replaceSection methods.
4918          *
4919          * @param $text Page wikitext
4920          * @param $section Numbered section. 0 pulls the text before the first
4921          *                 heading; other numbers will pull the given section
4922          *                 along with its lower-level subsections. If the section is
4923          *                 not found, $mode=get will return $newtext, and
4924          *                 $mode=replace will return $text.
4925          * @param $mode One of "get" or "replace"
4926          * @param $newText Replacement text for section data.
4927          * @return string for "get", the extracted section text.
4928          *                for "replace", the whole page with the section replaced.
4929          */
4930         private function extractSections( $text, $section, $mode, $newText='' ) {
4931                 $this->clearState();
4932                 $this->mOptions = new ParserOptions;
4933                 $this->setOutputType( OT_WIKI );
4934                 $curIndex = 0;
4935                 $outText = '';
4936                 $frame = new PPFrame( $this );
4937
4938                 // Preprocess the text
4939                 $dom = $this->preprocessToDom( $text );
4940                 $root = $dom->documentElement;
4941
4942                 // <h> nodes indicate section breaks
4943                 // They can only occur at the top level, so we can find them by iterating the root's children
4944                 $node = $root->firstChild;
4945
4946                 // Find the target section
4947                 if ( $section == 0 ) {
4948                         // Section zero doesn't nest, level=big
4949                         $targetLevel = 1000;
4950                 } else {
4951                         while ( $node ) {
4952                                 if ( $node->nodeName == 'h' ) {
4953                                         if ( $curIndex + 1 == $section ) {
4954                                                 break;
4955                                         }
4956                                         $curIndex++;
4957                                 }
4958                                 if ( $mode == 'replace' ) {
4959                                         $outText .= $frame->expand( $node );
4960                                 }
4961                                 $node = $node->nextSibling;
4962                         }
4963                         if ( $node ) {
4964                                 $targetLevel = $node->getAttribute( 'level' );
4965                         }
4966                 }
4967
4968                 if ( !$node ) {
4969                         // Not found
4970                         if ( $mode == 'get' ) {
4971                                 return $newText;
4972                         } else {
4973                                 return $text;
4974                         }
4975                 }
4976
4977                 // Find the end of the section, including nested sections
4978                 do {
4979                         if ( $node->nodeName == 'h' ) {
4980                                 $curIndex++;
4981                                 $curLevel = $node->getAttribute( 'level' );
4982                                 if ( $curIndex != $section && $curLevel <= $targetLevel ) {
4983                                         break;
4984                                 }
4985                         }
4986                         if ( $mode == 'get' ) {
4987                                 $outText .= $frame->expand( $node );
4988                         }
4989                         $node = $node->nextSibling;
4990                 } while ( $node );
4991
4992                 // Write out the remainder (in replace mode only)
4993                 if ( $mode == 'replace' ) {
4994                         // Output the replacement text
4995                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4996                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4997                         $outText .= $newText . "\n\n";
4998                         while ( $node ) {
4999                                 $outText .= $frame->expand( $node );
5000                                 $node = $node->nextSibling;
5001                         }
5002                 }
5003
5004                 if ( is_string( $outText ) ) {
5005                         // Re-insert stripped tags
5006                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5007                 }
5008
5009                 return $outText;
5010         }
5011
5012         /**
5013          * This function returns the text of a section, specified by a number ($section).
5014          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5015          * the first section before any such heading (section 0).
5016          *
5017          * If a section contains subsections, these are also returned.
5018          *
5019          * @param $text String: text to look in
5020          * @param $section Integer: section number
5021          * @param $deftext: default to return if section is not found
5022          * @return string text of the requested section
5023          */
5024         public function getSection( $text, $section, $deftext='' ) {
5025                 return $this->extractSections( $text, $section, "get", $deftext );
5026         }
5027
5028         public function replaceSection( $oldtext, $section, $text ) {
5029                 return $this->extractSections( $oldtext, $section, "replace", $text );
5030         }
5031
5032         /**
5033          * Get the timestamp associated with the current revision, adjusted for
5034          * the default server-local timestamp
5035          */
5036         function getRevisionTimestamp() {
5037                 if ( is_null( $this->mRevisionTimestamp ) ) {
5038                         wfProfileIn( __METHOD__ );
5039                         global $wgContLang;
5040                         $dbr = wfGetDB( DB_SLAVE );
5041                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5042                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5043
5044                         // Normalize timestamp to internal MW format for timezone processing.
5045                         // This has the added side-effect of replacing a null value with
5046                         // the current time, which gives us more sensible behavior for
5047                         // previews.
5048                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5049
5050                         // The cryptic '' timezone parameter tells to use the site-default
5051                         // timezone offset instead of the user settings.
5052                         //
5053                         // Since this value will be saved into the parser cache, served
5054                         // to other users, and potentially even used inside links and such,
5055                         // it needs to be consistent for all visitors.
5056                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5057
5058                         wfProfileOut( __METHOD__ );
5059                 }
5060                 return $this->mRevisionTimestamp;
5061         }
5062
5063         /**
5064          * Mutator for $mDefaultSort
5065          *
5066          * @param $sort New value
5067          */
5068         public function setDefaultSort( $sort ) {
5069                 $this->mDefaultSort = $sort;
5070         }
5071
5072         /**
5073          * Accessor for $mDefaultSort
5074          * Will use the title/prefixed title if none is set
5075          *
5076          * @return string
5077          */
5078         public function getDefaultSort() {
5079                 if( $this->mDefaultSort !== false ) {
5080                         return $this->mDefaultSort;
5081                 } else {
5082                         return $this->mTitle->getNamespace() == NS_CATEGORY
5083                                         ? $this->mTitle->getText()
5084                                         : $this->mTitle->getPrefixedText();
5085                 }
5086         }
5087
5088         /**
5089          * Try to guess the section anchor name based on a wikitext fragment
5090          * presumably extracted from a heading, for example "Header" from
5091          * "== Header ==".
5092          */
5093         public function guessSectionNameFromWikiText( $text ) {
5094                 # Strip out wikitext links(they break the anchor)
5095                 $text = $this->stripSectionName( $text );
5096                 $headline = Sanitizer::decodeCharReferences( $text );
5097                 # strip out HTML
5098                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5099                 $headline = trim( $headline );
5100                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5101                 $replacearray = array(
5102                         '%3A' => ':',
5103                         '%' => '.'
5104                 );
5105                 return str_replace(
5106                         array_keys( $replacearray ),
5107                         array_values( $replacearray ),
5108                         $sectionanchor );
5109         }
5110
5111         /**
5112          * Strips a text string of wikitext for use in a section anchor
5113          *
5114          * Accepts a text string and then removes all wikitext from the
5115          * string and leaves only the resultant text (i.e. the result of
5116          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5117          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5118          * to create valid section anchors by mimicing the output of the
5119          * parser when headings are parsed.
5120          *
5121          * @param $text string Text string to be stripped of wikitext
5122          * for use in a Section anchor
5123          * @return Filtered text string
5124          */
5125         public function stripSectionName( $text ) {
5126                 # Strip internal link markup
5127                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5128                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5129
5130                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5131                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5132                 # on how many empty links there are on the page - need to figure that out.
5133                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5134
5135                 # Parse wikitext quotes (italics & bold)
5136                 $text = $this->doQuotes($text);
5137
5138                 # Strip HTML tags
5139                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5140                 return $text;
5141         }
5142
5143         /**
5144          * strip/replaceVariables/unstrip for preprocessor regression testing
5145          */
5146         function srvus( $text ) {
5147                 $text = $this->replaceVariables( $text );
5148                 $text = $this->mStripState->unstripBoth( $text );
5149                 return $text;
5150         }
5151 }
5152
5153 /**
5154  * @todo document, briefly.
5155  * @addtogroup Parser
5156  */
5157 class OnlyIncludeReplacer {
5158         var $output = '';
5159
5160         function replace( $matches ) {
5161                 if ( substr( $matches[1], -1 ) == "\n" ) {
5162                         $this->output .= substr( $matches[1], 0, -1 );
5163                 } else {
5164                         $this->output .= $matches[1];
5165                 }
5166         }
5167 }
5168
5169 /**
5170  * @todo document, briefly.
5171  * @addtogroup Parser
5172  */
5173 class StripState {
5174         var $general, $nowiki;
5175
5176         function __construct() {
5177                 $this->general = new ReplacementArray;
5178                 $this->nowiki = new ReplacementArray;
5179         }
5180
5181         function unstripGeneral( $text ) {
5182                 wfProfileIn( __METHOD__ );
5183                 do {
5184                         $oldText = $text;
5185                         $text = $this->general->replace( $text );
5186                 } while ( $text != $oldText );
5187                 wfProfileOut( __METHOD__ );
5188                 return $text;
5189         }
5190
5191         function unstripNoWiki( $text ) {
5192                 wfProfileIn( __METHOD__ );
5193                 do {
5194                         $oldText = $text;
5195                         $text = $this->nowiki->replace( $text );
5196                 } while ( $text != $oldText );
5197                 wfProfileOut( __METHOD__ );
5198                 return $text;
5199         }
5200
5201         function unstripBoth( $text ) {
5202                 wfProfileIn( __METHOD__ );
5203                 do {
5204                         $oldText = $text;
5205                         $text = $this->general->replace( $text );
5206                         $text = $this->nowiki->replace( $text );
5207                 } while ( $text != $oldText );
5208                 wfProfileOut( __METHOD__ );
5209                 return $text;
5210         }
5211 }
5212
5213 /**
5214  * An expansion frame, used as a context to expand the result of preprocessToDom()
5215  */
5216 class PPFrame {
5217         var $parser, $title;
5218
5219         const NO_ARGS = 1;
5220         const NO_TEMPLATES = 2;
5221
5222         /**
5223          * Construct a new preprocessor frame.
5224          * @param Parser $parser The parent parser
5225          * @param Title $title The context title, or false if there isn't one
5226          */
5227         function __construct( $parser ) {
5228                 $this->parser = $parser;
5229                 $this->title = $parser->mTitle;
5230         }
5231
5232         /**
5233          * Create a new child frame
5234          * $args is optionally a DOMNodeList containing the template arguments
5235          */
5236         function newChild( $args = false, $title = false ) {
5237                 $assocArgs = array();
5238                 if ( $title === false ) {
5239                         $title = $this->title;
5240                 }
5241                 if ( $args !== false ) {
5242                         $xpath = false;
5243                         foreach ( $args as $arg ) {
5244                                 if ( !$xpath ) {
5245                                         $xpath = new DOMXPath( $arg->ownerDocument );
5246                                 }
5247
5248                                 $nameNodes = $xpath->query( 'name', $arg );
5249                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5250                                         // Numbered parameter
5251                                         $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5252                                 } else {
5253                                         // Named parameter
5254                                         $name = $this->expand( $nameNodes->item( 0 ) );
5255                                 }
5256
5257                                 $value = $xpath->query( 'value', $arg );
5258                                 $assocArgs[$name] = $value->item( 0 );
5259                         }
5260                 }
5261                 return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
5262         }
5263
5264         /**
5265          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5266          * using the current context
5267          * @param $root the node
5268          */
5269         function expand( $root, $shallowFlags = 0, $deepFlags = 0 ) {
5270                 if ( is_string( $root ) ) {
5271                         return $root;
5272                 }
5273
5274                 if ( $this->parser->ot['html']
5275                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5276                 {
5277                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5278                 }
5279                 $flags = $shallowFlags | $deepFlags;
5280
5281                 if ( is_array( $root ) ) {
5282                         $s = '';
5283                         foreach ( $root as $node ) {
5284                                 $s .= $this->expand( $node, 0, $deepFlags );
5285                         }
5286                 } elseif ( $root instanceof DOMNodeList ) {
5287                         $s = '';
5288                         foreach ( $root as $node ) {
5289                                 $s .= $this->expand( $node, 0, $deepFlags );
5290                         }
5291                 } elseif ( $root instanceof DOMNode ) {
5292                         if ( $root->nodeType == XML_TEXT_NODE ) {
5293                                 $s = $root->nodeValue;
5294                         } elseif ( $root->nodeName == 'template' ) {
5295                                 # Double-brace expansion
5296                                 $xpath = new DOMXPath( $root->ownerDocument );
5297                                 $titles = $xpath->query( 'title', $root );
5298                                 $title = $titles->item( 0 );
5299                                 $parts = $xpath->query( 'part', $root );
5300                                 if ( $flags & self::NO_TEMPLATES ) {
5301                                         $s = '{{' . $this->implodeWithFlags( '|', 0, $deepFlags, $title, $parts ) . '}}';
5302                                 } else {
5303                                         $lineStart = $root->getAttribute( 'lineStart' );
5304                                         $params = array(
5305                                                 'title' => $title,
5306                                                 'parts' => $parts,
5307                                                 'lineStart' => $lineStart,
5308                                                 'text' => 'FIXME' );
5309                                         $s = $this->parser->braceSubstitution( $params, $this );
5310                                 }
5311                         } elseif ( $root->nodeName == 'tplarg' ) {
5312                                 # Triple-brace expansion
5313                                 $xpath = new DOMXPath( $root->ownerDocument );
5314                                 $titles = $xpath->query( 'title', $root );
5315                                 $title = $titles->item( 0 );
5316                                 $parts = $xpath->query( 'part', $root );
5317                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5318                                         $s = '{{{' . $this->implode( '|', 0, $deepFlags, $title, $parts ) . '}}}';
5319                                 } else {
5320                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5321                                         $s = $this->parser->argSubstitution( $params, $this );
5322                                 }
5323                         } elseif ( $root->nodeName == 'ext' ) {
5324                                 # Extension tag
5325                                 $xpath = new DOMXPath( $root->ownerDocument );
5326                                 $names = $xpath->query( 'name', $root );
5327                                 $attrs = $xpath->query( 'attr', $root );
5328                                 $inners = $xpath->query( 'inner', $root );
5329                                 $closes = $xpath->query( 'close', $root );
5330                                 $params = array(
5331                                         'name' => $names->item( 0 ),
5332                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5333                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5334                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5335                                 );
5336                                 $s = $this->parser->extensionSubstitution( $params, $this );
5337                         } elseif ( $root->nodeName == 'h' ) {
5338                                 # Heading
5339                                 $s = $this->expand( $root->childNodes, 0, $deepFlags );
5340
5341                                 if ( $this->parser->ot['html'] ) {
5342                                         # Insert heading index marker
5343                                         $headingIndex = $root->getAttribute( 'i' );
5344                                         $titleText = $this->title->getPrefixedDBkey();
5345                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5346                                         $serial = count( $this->parser->mHeadings ) - 1;
5347                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5348                                         $count = $root->getAttribute( 'level' );
5349
5350                                         // FIXME: bug-for-bug with old parser
5351                                         // Lose whitespace for no apparent reason
5352                                         // Remove this after differential testing is done
5353                                         if ( true ) {
5354                                                 // Good version
5355                                                 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5356                                         } else {
5357                                                 // Bad version
5358                                                 if ( preg_match( '/^(={1,6})(.*?)(={1,6})\s*?$/', $s, $m ) ) {
5359                                                         if ( $m[2] != '' ) {
5360                                                                 $s = $m[1] . $marker . $m[2] . $m[3];
5361                                                         }
5362                                                 }
5363                                         }
5364                                         $this->parser->mStripState->general->setPair( $marker, '' );
5365                                 }
5366                         } else {
5367                                 # Generic recursive expansion
5368                                 $s = '';
5369                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5370                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5371                                                 $s .= $node->nodeValue;
5372                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5373                                                 $s .= $this->expand( $node, 0, $deepFlags );
5374                                         }
5375                                 }
5376                         }
5377                 } else {
5378                         throw new MWException( __METHOD__.': Invalid parameter type' );
5379                 }
5380                 return $s;
5381         }
5382
5383         function implodeWithFlags( $sep, $shallowFlags, $deepFlags /*, ... */ ) {
5384                 $args = array_slice( func_get_args(), 3 );
5385
5386                 $first = true;
5387                 $s = '';
5388                 foreach ( $args as $root ) {
5389                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5390                                 $root = array( $root );
5391                         }
5392                         foreach ( $root as $node ) {
5393                                 if ( $first ) {
5394                                         $first = false;
5395                                 } else {
5396                                         $s .= $sep;
5397                                 }
5398                                 $s .= $this->expand( $node, $shallowFlags, $deepFlags );
5399                         }
5400                 }
5401                 return $s;
5402         }
5403
5404         function implode( $sep /*, ... */ ) {
5405                 $args = func_get_args();
5406                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0, 0 ), array_slice( $args, 1 ) );
5407                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5408         }
5409
5410         function __toString() {
5411                 return 'frame{}';
5412         }
5413 }
5414
5415 /**
5416  * Expansion frame with template arguments
5417  */
5418 class PPTemplateFrame extends PPFrame {
5419         public $parser, $args, $parent, $serial;
5420
5421         function __construct( $parser, $parent = false, $args = array(), $title = false ) {
5422                 $this->parser = $parser;
5423                 $this->parent = $parent;
5424                 $this->args = $args;
5425                 $this->title = $title;
5426         }
5427
5428         function __toString() {
5429                 $s = 'tplframe{';
5430                 $first = true;
5431                 foreach ( $this->args as $name => $value ) {
5432                         if ( $first ) {
5433                                 $first = false;
5434                         } else {
5435                                 $s .= ', ';
5436                         }
5437                         $s .= "\"$name\":\"" .
5438                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5439                 }
5440                 $s .= '}';
5441                 return $s;
5442         }
5443 }
5444