includes/parser/Parser_OldPP.php

   1 <?php
   2 /**
   3  * Parser with old preprocessor
   4  * @ingroup Parser
   5  */
   6 class Parser_OldPP
   7 {
   8         /**
   9          * Update this version number when the ParserOutput format
  10          * changes in an incompatible way, so the parser cache
  11          * can automatically discard old data.
  12          */
  13         const VERSION = '1.6.4';
  14
  15         # Flags for Parser::setFunctionHook
  16         # Also available as global constants from Defines.php
  17         const SFH_NO_HASH = 1;
  18
  19         # Constants needed for external link processing
  20         # Everything except bracket, space, or control characters
  21         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  22         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)\\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/S';
  23
  24         // State constants for the definition list colon extraction
  25         const COLON_STATE_TEXT = 0;
  26         const COLON_STATE_TAG = 1;
  27         const COLON_STATE_TAGSTART = 2;
  28         const COLON_STATE_CLOSETAG = 3;
  29         const COLON_STATE_TAGSLASH = 4;
  30         const COLON_STATE_COMMENT = 5;
  31         const COLON_STATE_COMMENTDASH = 6;
  32         const COLON_STATE_COMMENTDASHDASH = 7;
  33
  34         // Allowed values for $this->mOutputType
  35         // Parameter to startExternalParse().
  36         const OT_HTML = 1;
  37         const OT_WIKI = 2;
  38         const OT_PREPROCESS = 3;
  39         const OT_MSG = 4;
  40
  41         /**#@+
  42          * @private
  43          */
  44         # Persistent:
  45         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  46                 $mImageParams, $mImageParamsMagicArray, $mExtLinkBracketedRegex;
  47
  48         # Cleared with clearState():
  49         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  50         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  51         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
  52         var $mIncludeSizes, $mDefaultSort;
  53         var $mTemplates,        // cache of already loaded templates, avoids
  54                                 // multiple SQL queries for the same string
  55             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  56                                 // in this path. Used for loop detection.
  57
  58         # Temporary
  59         # These are variables reset at least once per parse regardless of $clearState
  60         var $mOptions,      // ParserOptions object
  61                 $mTitle,        // Title context, used for self-link rendering and similar things
  62                 $mOutputType,   // Output type, one of the OT_xxx constants
  63                 $ot,            // Shortcut alias, see setOutputType()
  64                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  65                 $mRevisionTimestamp, // The timestamp of the specified revision ID
  66                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
  67
  68         /**#@-*/
  69
  70         /**
  71          * Constructor
  72          *
  73          * @public
  74          */
  75         function __construct( $conf = array() ) {
  76                 $this->mTagHooks = array();
  77                 $this->mTransparentTagHooks = array();
  78                 $this->mFunctionHooks = array();
  79                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
  80                 $this->mFirstCall = true;
  81                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
  82                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
  83         }
  84
  85         /**
  86          * Do various kinds of initialisation on the first call of the parser
  87          */
  88         function firstCallInit() {
  89                 if ( !$this->mFirstCall ) {
  90                         return;
  91                 }
  92                 $this->mFirstCall = false;
  93
  94                 wfProfileIn( __METHOD__ );
  95                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
  96
  97                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
  98
  99                 # Syntax for arguments (see self::setFunctionHook):
 100                 #  "name for lookup in localized magic words array",
 101                 #  function callback,
 102                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 103                 #    instead of {{#int:...}})
 104                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 105                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 106                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 107                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 108                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 109                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 110                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 111                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 112                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 113                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 114                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 115                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 116                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 117                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 118                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 119                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 120                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 121                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 122                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 123                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 124                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 125                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 126                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 127                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 128                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 129                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 130                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 131
 132                 if ( $wgAllowDisplayTitle ) {
 133                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 134                 }
 135                 if ( $wgAllowSlowParserFunctions ) {
 136                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 137                 }
 138
 139                 $this->initialiseVariables();
 140
 141                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 142                 wfProfileOut( __METHOD__ );
 143         }
 144
 145         /**
 146          * Clear Parser state
 147          *
 148          * @private
 149          */
 150         function clearState() {
 151                 wfProfileIn( __METHOD__ );
 152                 if ( $this->mFirstCall ) {
 153                         $this->firstCallInit();
 154                 }
 155                 $this->mOutput = new ParserOutput;
 156                 $this->mAutonumber = 0;
 157                 $this->mLastSection = '';
 158                 $this->mDTopen = false;
 159                 $this->mIncludeCount = array();
 160                 $this->mStripState = new StripState;
 161                 $this->mArgStack = array();
 162                 $this->mInPre = false;
 163                 $this->mInterwikiLinkHolders = array(
 164                         'texts' => array(),
 165                         'titles' => array()
 166                 );
 167                 $this->mLinkHolders = array(
 168                         'namespaces' => array(),
 169                         'dbkeys' => array(),
 170                         'queries' => array(),
 171                         'texts' => array(),
 172                         'titles' => array()
 173                 );
 174                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 175
 176                 /**
 177                  * Prefix for temporary replacement strings for the multipass parser.
 178                  * \x07 should never appear in input as it's disallowed in XML.
 179                  * Using it at the front also gives us a little extra robustness
 180                  * since it shouldn't match when butted up against identifier-like
 181                  * string constructs.
 182                  */
 183                 $this->mUniqPrefix = "\x07UNIQ" . self::getRandomString();
 184
 185                 # Clear these on every parse, bug 4549
 186                 $this->mTemplates = array();
 187                 $this->mTemplatePath = array();
 188
 189                 $this->mShowToc = true;
 190                 $this->mForceTocPosition = false;
 191                 $this->mIncludeSizes = array(
 192                         'pre-expand' => 0,
 193                         'post-expand' => 0,
 194                         'arg' => 0
 195                 );
 196                 $this->mDefaultSort = false;
 197
 198                 wfRunHooks( 'ParserClearState', array( &$this ) );
 199                 wfProfileOut( __METHOD__ );
 200         }
 201
 202         function setOutputType( $ot ) {
 203                 $this->mOutputType = $ot;
 204                 // Shortcut alias
 205                 $this->ot = array(
 206                         'html' => $ot == self::OT_HTML,
 207                         'wiki' => $ot == self::OT_WIKI,
 208                         'msg' => $ot == self::OT_MSG,
 209                         'pre' => $ot == self::OT_PREPROCESS,
 210                 );
 211         }
 212
 213         /**
 214          * Accessor for mUniqPrefix.
 215          *
 216          * @public
 217          */
 218         function uniqPrefix() {
 219                 return $this->mUniqPrefix;
 220         }
 221
 222         /**
 223          * Convert wikitext to HTML
 224          * Do not call this function recursively.
 225          *
 226          * @param string $text Text we want to parse
 227          * @param Title &$title A title object
 228          * @param array $options
 229          * @param boolean $linestart
 230          * @param boolean $clearState
 231          * @param int $revid number to pass in {{REVISIONID}}
 232          * @return ParserOutput a ParserOutput
 233          */
 234         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 235                 /**
 236                  * First pass--just handle <nowiki> sections, pass the rest off
 237                  * to internalParse() which does all the real work.
 238                  */
 239
 240                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 241                 $fname = 'Parser::parse-' . wfGetCaller();
 242                 wfProfileIn( __METHOD__ );
 243                 wfProfileIn( $fname );
 244
 245                 if ( $clearState ) {
 246                         $this->clearState();
 247                 }
 248
 249                 $this->mOptions = $options;
 250                 $this->mTitle =& $title;
 251                 $oldRevisionId = $this->mRevisionId;
 252                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 253                 if( $revid !== null ) {
 254                         $this->mRevisionId = $revid;
 255                         $this->mRevisionTimestamp = null;
 256                 }
 257                 $this->setOutputType( self::OT_HTML );
 258                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 259                 $text = $this->strip( $text, $this->mStripState );
 260                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 261                 $text = $this->internalParse( $text );
 262                 $text = $this->mStripState->unstripGeneral( $text );
 263
 264                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 265                 $fixtags = array(
 266                         # french spaces, last one Guillemet-left
 267                         # only if there is something before the space
 268                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 269                         # french spaces, Guillemet-right
 270                         '/(\\302\\253) /' => '\\1&nbsp;',
 271                 );
 272                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 273
 274                 # only once and last
 275                 $text = $this->doBlockLevels( $text, $linestart );
 276
 277                 $this->replaceLinkHolders( $text );
 278
 279                 # the position of the parserConvert() call should not be changed. it
 280                 # assumes that the links are all replaced and the only thing left
 281                 # is the <nowiki> mark.
 282                 # Side-effects: this calls $this->mOutput->setTitleText()
 283                 $text = $wgContLang->parserConvert( $text, $this );
 284
 285                 $text = $this->mStripState->unstripNoWiki( $text );
 286
 287                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 288
 289 //!JF Move to its own function
 290
 291                 $uniq_prefix = $this->mUniqPrefix;
 292                 $matches = array();
 293                 $elements = array_keys( $this->mTransparentTagHooks );
 294                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 295
 296                 foreach( $matches as $marker => $data ) {
 297                         list( $element, $content, $params, $tag ) = $data;
 298                         $tagName = strtolower( $element );
 299                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 300                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 301                                         array( $content, $params, $this ) );
 302                         } else {
 303                                 $output = $tag;
 304                         }
 305                         $this->mStripState->general->setPair( $marker, $output );
 306                 }
 307                 $text = $this->mStripState->unstripGeneral( $text );
 308
 309                 $text = Sanitizer::normalizeCharReferences( $text );
 310
 311                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 312                         $text = self::tidy($text);
 313                 } else {
 314                         # attempt to sanitize at least some nesting problems
 315                         # (bug #2702 and quite a few others)
 316                         $tidyregs = array(
 317                                 # ''Something [http://www.cool.com cool''] -->
 318                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 319                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 320                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 321                                 # fix up an anchor inside another anchor, only
 322                                 # at least for a single single nested link (bug 3695)
 323                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 324                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 325                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 326                                 # contains a div, so fix it up here; replace
 327                                 # div with escaped text
 328                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 329                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 330                                 # remove empty italic or bold tag pairs, some
 331                                 # introduced by rules above
 332                                 '/<([bi])><\/\\1>/' => '',
 333                         );
 334
 335                         $text = preg_replace(
 336                                 array_keys( $tidyregs ),
 337                                 array_values( $tidyregs ),
 338                                 $text );
 339                 }
 340
 341                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 342
 343                 # Information on include size limits, for the benefit of users who try to skirt them
 344                 if ( $this->mOptions->getEnableLimitReport() ) {
 345                         $max = $this->mOptions->getMaxIncludeSize();
 346                         $limitReport =
 347                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']}/$max bytes\n" .
 348                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 349                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 350                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 351                         $text .= "<!-- \n$limitReport-->\n";
 352                 }
 353                 $this->mOutput->setText( $text );
 354                 $this->mRevisionId = $oldRevisionId;
 355                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 356                 wfProfileOut( $fname );
 357                 wfProfileOut( __METHOD__ );
 358
 359                 return $this->mOutput;
 360         }
 361
 362         /**
 363          * Recursive parser entry point that can be called from an extension tag
 364          * hook.
 365          */
 366         function recursiveTagParse( $text ) {
 367                 wfProfileIn( __METHOD__ );
 368                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 369                 $text = $this->strip( $text, $this->mStripState );
 370                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 371                 $text = $this->internalParse( $text );
 372                 wfProfileOut( __METHOD__ );
 373                 return $text;
 374         }
 375
 376         /**
 377          * Expand templates and variables in the text, producing valid, static wikitext.
 378          * Also removes comments.
 379          */
 380         function preprocess( $text, $title, $options, $revid = null ) {
 381                 wfProfileIn( __METHOD__ );
 382                 $this->clearState();
 383                 $this->setOutputType( self::OT_PREPROCESS );
 384                 $this->mOptions = $options;
 385                 $this->mTitle = $title;
 386                 if( $revid !== null ) {
 387                         $this->mRevisionId = $revid;
 388                 }
 389                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 390                 $text = $this->strip( $text, $this->mStripState );
 391                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 392                 if ( $this->mOptions->getRemoveComments() ) {
 393                         $text = Sanitizer::removeHTMLcomments( $text );
 394                 }
 395                 $text = $this->replaceVariables( $text );
 396                 $text = $this->mStripState->unstripBoth( $text );
 397                 wfProfileOut( __METHOD__ );
 398                 return $text;
 399         }
 400
 401         /**
 402          * Get a random string
 403          *
 404          * @private
 405          * @static
 406          */
 407         function getRandomString() {
 408                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 409         }
 410
 411         function &getTitle() { return $this->mTitle; }
 412         function getOptions() { return $this->mOptions; }
 413         function getRevisionId() { return $this->mRevisionId; }
 414
 415         function getFunctionLang() {
 416                 global $wgLang, $wgContLang;
 417                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 418         }
 419
 420         /**
 421          * Replaces all occurrences of HTML-style comments and the given tags
 422          * in the text with a random marker and returns teh next text. The output
 423          * parameter $matches will be an associative array filled with data in
 424          * the form:
 425          *   'UNIQ-xxxxx' => array(
 426          *     'element',
 427          *     'tag content',
 428          *     array( 'param' => 'x' ),
 429          *     '<element param="x">tag content</element>' ) )
 430          *
 431          * @param $elements list of element names. Comments are always extracted.
 432          * @param $text Source text string.
 433          * @param $uniq_prefix
 434          *
 435          * @public
 436          * @static
 437          */
 438         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 439                 static $n = 1;
 440                 $stripped = '';
 441                 $matches = array();
 442
 443                 $taglist = implode( '|', $elements );
 444                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 445
 446                 while ( '' != $text ) {
 447                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 448                         $stripped .= $p[0];
 449                         if( count( $p ) < 5 ) {
 450                                 break;
 451                         }
 452                         if( count( $p ) > 5 ) {
 453                                 // comment
 454                                 $element    = $p[4];
 455                                 $attributes = '';
 456                                 $close      = '';
 457                                 $inside     = $p[5];
 458                         } else {
 459                                 // tag
 460                                 $element    = $p[1];
 461                                 $attributes = $p[2];
 462                                 $close      = $p[3];
 463                                 $inside     = $p[4];
 464                         }
 465
 466                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 467                         $stripped .= $marker;
 468
 469                         if ( $close === '/>' ) {
 470                                 // Empty element tag, <tag />
 471                                 $content = null;
 472                                 $text = $inside;
 473                                 $tail = null;
 474                         } else {
 475                                 if( $element == '!--' ) {
 476                                         $end = '/(-->)/';
 477                                 } else {
 478                                         $end = "/(<\\/$element\\s*>)/i";
 479                                 }
 480                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 481                                 $content = $q[0];
 482                                 if( count( $q ) < 3 ) {
 483                                         # No end tag -- let it run out to the end of the text.
 484                                         $tail = '';
 485                                         $text = '';
 486                                 } else {
 487                                         $tail = $q[1];
 488                                         $text = $q[2];
 489                                 }
 490                         }
 491
 492                         $matches[$marker] = array( $element,
 493                                 $content,
 494                                 Sanitizer::decodeTagAttributes( $attributes ),
 495                                 "<$element$attributes$close$content$tail" );
 496                 }
 497                 return $stripped;
 498         }
 499
 500         /**
 501          * Strips and renders nowiki, pre, math, hiero
 502          * If $render is set, performs necessary rendering operations on plugins
 503          * Returns the text, and fills an array with data needed in unstrip()
 504          *
 505          * @param StripState $state
 506          *
 507          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 508          *  will be stripped in addition to other tags. This is important
 509          *  for section editing, where these comments cause confusion when
 510          *  counting the sections in the wikisource
 511          *
 512          * @param array dontstrip contains tags which should not be stripped;
 513          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 514          *
 515          * @private
 516          */
 517         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 518                 global $wgContLang;
 519                 wfProfileIn( __METHOD__ );
 520                 $render = ($this->mOutputType == self::OT_HTML);
 521
 522                 $uniq_prefix = $this->mUniqPrefix;
 523                 $commentState = new ReplacementArray;
 524                 $nowikiItems = array();
 525                 $generalItems = array();
 526
 527                 $elements = array_merge(
 528                         array( 'nowiki', 'gallery' ),
 529                         array_keys( $this->mTagHooks ) );
 530                 global $wgRawHtml;
 531                 if( $wgRawHtml ) {
 532                         $elements[] = 'html';
 533                 }
 534                 if( $this->mOptions->getUseTeX() ) {
 535                         $elements[] = 'math';
 536                 }
 537
 538                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 539                 foreach ( $elements AS $k => $v ) {
 540                         if ( !in_array ( $v , $dontstrip ) ) continue;
 541                         unset ( $elements[$k] );
 542                 }
 543
 544                 $matches = array();
 545                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 546
 547                 foreach( $matches as $marker => $data ) {
 548                         list( $element, $content, $params, $tag ) = $data;
 549                         if( $render ) {
 550                                 $tagName = strtolower( $element );
 551                                 wfProfileIn( __METHOD__."-render-$tagName" );
 552                                 switch( $tagName ) {
 553                                 case '!--':
 554                                         // Comment
 555                                         if( substr( $tag, -3 ) == '-->' ) {
 556                                                 $output = $tag;
 557                                         } else {
 558                                                 // Unclosed comment in input.
 559                                                 // Close it so later stripping can remove it
 560                                                 $output = "$tag-->";
 561                                         }
 562                                         break;
 563                                 case 'html':
 564                                         if( $wgRawHtml ) {
 565                                                 $output = $content;
 566                                                 break;
 567                                         }
 568                                         // Shouldn't happen otherwise. :)
 569                                 case 'nowiki':
 570                                         $output = Xml::escapeTagsOnly( $content );
 571                                         break;
 572                                 case 'math':
 573                                         $output = $wgContLang->armourMath(
 574                                                 MathRenderer::renderMath( $content, $params ) );
 575                                         break;
 576                                 case 'gallery':
 577                                         $output = $this->renderImageGallery( $content, $params );
 578                                         break;
 579                                 default:
 580                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 581                                                 # Workaround for PHP bug 35229 and similar
 582                                                 if ( !is_callable( $this->mTagHooks[$tagName] ) ) {
 583                                                         throw new MWException( "Tag hook for $tagName is not callable\n" );
 584                                                 }
 585                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 586                                                         array( $content, $params, $this ) );
 587                                         } else {
 588                                                 throw new MWException( "Invalid call hook $element" );
 589                                         }
 590                                 }
 591                                 wfProfileOut( __METHOD__."-render-$tagName" );
 592                         } else {
 593                                 // Just stripping tags; keep the source
 594                                 $output = $tag;
 595                         }
 596
 597                         // Unstrip the output, to support recursive strip() calls
 598                         $output = $state->unstripBoth( $output );
 599
 600                         if( !$stripcomments && $element == '!--' ) {
 601                                 $commentState->setPair( $marker, $output );
 602                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 603                                 $nowikiItems[$marker] = $output;
 604                         } else {
 605                                 $generalItems[$marker] = $output;
 606                         }
 607                 }
 608                 # Add the new items to the state
 609                 # We do this after the loop instead of during it to avoid slowing
 610                 # down the recursive unstrip
 611                 $state->nowiki->mergeArray( $nowikiItems );
 612                 $state->general->mergeArray( $generalItems );
 613
 614                 # Unstrip comments unless explicitly told otherwise.
 615                 # (The comments are always stripped prior to this point, so as to
 616                 # not invoke any extension tags / parser hooks contained within
 617                 # a comment.)
 618                 if ( !$stripcomments ) {
 619                         // Put them all back and forget them
 620                         $text = $commentState->replace( $text );
 621                 }
 622
 623                 wfProfileOut( __METHOD__ );
 624                 return $text;
 625         }
 626
 627         /**
 628          * Restores pre, math, and other extensions removed by strip()
 629          *
 630          * always call unstripNoWiki() after this one
 631          * @private
 632          * @deprecated use $this->mStripState->unstrip()
 633          */
 634         function unstrip( $text, $state ) {
 635                 return $state->unstripGeneral( $text );
 636         }
 637
 638         /**
 639          * Always call this after unstrip() to preserve the order
 640          *
 641          * @private
 642          * @deprecated use $this->mStripState->unstrip()
 643          */
 644         function unstripNoWiki( $text, $state ) {
 645                 return $state->unstripNoWiki( $text );
 646         }
 647
 648         /**
 649          * @deprecated use $this->mStripState->unstripBoth()
 650          */
 651         function unstripForHTML( $text ) {
 652                 return $this->mStripState->unstripBoth( $text );
 653         }
 654
 655         /**
 656          * Add an item to the strip state
 657          * Returns the unique tag which must be inserted into the stripped text
 658          * The tag will be replaced with the original text in unstrip()
 659          *
 660          * @private
 661          */
 662         function insertStripItem( $text, &$state ) {
 663                 $rnd = $this->mUniqPrefix . '-item' . self::getRandomString();
 664                 $state->general->setPair( $rnd, $text );
 665                 return $rnd;
 666         }
 667
 668         /**
 669          * Interface with html tidy, used if $wgUseTidy = true.
 670          * If tidy isn't able to correct the markup, the original will be
 671          * returned in all its glory with a warning comment appended.
 672          *
 673          * Either the external tidy program or the in-process tidy extension
 674          * will be used depending on availability. Override the default
 675          * $wgTidyInternal setting to disable the internal if it's not working.
 676          *
 677          * @param string $text Hideous HTML input
 678          * @return string Corrected HTML output
 679          * @public
 680          * @static
 681          */
 682         function tidy( $text ) {
 683                 global $wgTidyInternal;
 684                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 685 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 686 '<head><title>test</title></head><body>'.$text.'</body></html>';
 687                 if( $wgTidyInternal ) {
 688                         $correctedtext = self::internalTidy( $wrappedtext );
 689                 } else {
 690                         $correctedtext = self::externalTidy( $wrappedtext );
 691                 }
 692                 if( is_null( $correctedtext ) ) {
 693                         wfDebug( "Tidy error detected!\n" );
 694                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 695                 }
 696                 return $correctedtext;
 697         }
 698
 699         /**
 700          * Spawn an external HTML tidy process and get corrected markup back from it.
 701          *
 702          * @private
 703          * @static
 704          */
 705         function externalTidy( $text ) {
 706                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 707                 $fname = 'Parser::externalTidy';
 708                 wfProfileIn( $fname );
 709
 710                 $cleansource = '';
 711                 $opts = ' -utf8';
 712
 713                 $descriptorspec = array(
 714                         0 => array('pipe', 'r'),
 715                         1 => array('pipe', 'w'),
 716                         2 => array('file', wfGetNull(), 'a')
 717                 );
 718                 $pipes = array();
 719                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 720                 if (is_resource($process)) {
 721                         // Theoretically, this style of communication could cause a deadlock
 722                         // here. If the stdout buffer fills up, then writes to stdin could
 723                         // block. This doesn't appear to happen with tidy, because tidy only
 724                         // writes to stdout after it's finished reading from stdin. Search
 725                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 726                         fwrite($pipes[0], $text);
 727                         fclose($pipes[0]);
 728                         while (!feof($pipes[1])) {
 729                                 $cleansource .= fgets($pipes[1], 1024);
 730                         }
 731                         fclose($pipes[1]);
 732                         proc_close($process);
 733                 }
 734
 735                 wfProfileOut( $fname );
 736
 737                 if( $cleansource == '' && $text != '') {
 738                         // Some kind of error happened, so we couldn't get the corrected text.
 739                         // Just give up; we'll use the source text and append a warning.
 740                         return null;
 741                 } else {
 742                         return $cleansource;
 743                 }
 744         }
 745
 746         /**
 747          * Use the HTML tidy PECL extension to use the tidy library in-process,
 748          * saving the overhead of spawning a new process.
 749          *
 750          * 'pear install tidy' should be able to compile the extension module.
 751          *
 752          * @private
 753          * @static
 754          */
 755         function internalTidy( $text ) {
 756                 global $wgTidyConf, $IP;
 757                 $fname = 'Parser::internalTidy';
 758                 wfProfileIn( $fname );
 759
 760                 $tidy = new tidy;
 761                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 762                 $tidy->cleanRepair();
 763                 if( $tidy->getStatus() == 2 ) {
 764                         // 2 is magic number for fatal error
 765                         // http://www.php.net/manual/en/function.tidy-get-status.php
 766                         $cleansource = null;
 767                 } else {
 768                         $cleansource = tidy_get_output( $tidy );
 769                 }
 770                 wfProfileOut( $fname );
 771                 return $cleansource;
 772         }
 773
 774         /**
 775          * parse the wiki syntax used to render tables
 776          *
 777          * @private
 778          */
 779         function doTableStuff ( $text ) {
 780                 $fname = 'Parser::doTableStuff';
 781                 wfProfileIn( $fname );
 782
 783                 $lines = explode ( "\n" , $text );
 784                 $td_history = array (); // Is currently a td tag open?
 785                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 786                 $tr_history = array (); // Is currently a tr tag open?
 787                 $tr_attributes = array (); // history of tr attributes
 788                 $has_opened_tr = array(); // Did this table open a <tr> element?
 789                 $indent_level = 0; // indent level of the table
 790                 foreach ( $lines as $key => $line )
 791                 {
 792                         $line = trim ( $line );
 793
 794                         if( $line == '' ) { // empty line, go to next line
 795                                 continue;
 796                         }
 797                         $first_character = $line{0};
 798                         $matches = array();
 799
 800                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 801                                 // First check if we are starting a new table
 802                                 $indent_level = strlen( $matches[1] );
 803
 804                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 805                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 806
 807                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 808                                 array_push ( $td_history , false );
 809                                 array_push ( $last_tag_history , '' );
 810                                 array_push ( $tr_history , false );
 811                                 array_push ( $tr_attributes , '' );
 812                                 array_push ( $has_opened_tr , false );
 813                         } else if ( count ( $td_history ) == 0 ) {
 814                                 // Don't do any of the following
 815                                 continue;
 816                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 817                                 // We are ending a table
 818                                 $line = '</table>' . substr ( $line , 2 );
 819                                 $last_tag = array_pop ( $last_tag_history );
 820
 821                                 if ( !array_pop ( $has_opened_tr ) ) {
 822                                         $line = "<tr><td></td></tr>{$line}";
 823                                 }
 824
 825                                 if ( array_pop ( $tr_history ) ) {
 826                                         $line = "</tr>{$line}";
 827                                 }
 828
 829                                 if ( array_pop ( $td_history ) ) {
 830                                         $line = "</{$last_tag}>{$line}";
 831                                 }
 832                                 array_pop ( $tr_attributes );
 833                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 834                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 835                                 // Now we have a table row
 836                                 $line = preg_replace( '#^\|-+#', '', $line );
 837
 838                                 // Whats after the tag is now only attributes
 839                                 $attributes = $this->mStripState->unstripBoth( $line );
 840                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 841                                 array_pop ( $tr_attributes );
 842                                 array_push ( $tr_attributes , $attributes );
 843
 844                                 $line = '';
 845                                 $last_tag = array_pop ( $last_tag_history );
 846                                 array_pop ( $has_opened_tr );
 847                                 array_push ( $has_opened_tr , true );
 848
 849                                 if ( array_pop ( $tr_history ) ) {
 850                                         $line = '</tr>';
 851                                 }
 852
 853                                 if ( array_pop ( $td_history ) ) {
 854                                         $line = "</{$last_tag}>{$line}";
 855                                 }
 856
 857                                 $lines[$key] = $line;
 858                                 array_push ( $tr_history , false );
 859                                 array_push ( $td_history , false );
 860                                 array_push ( $last_tag_history , '' );
 861                         }
 862                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 863                                 // This might be cell elements, td, th or captions
 864                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 865                                         $first_character = '+';
 866                                         $line = substr ( $line , 1 );
 867                                 }
 868
 869                                 $line = substr ( $line , 1 );
 870
 871                                 if ( $first_character == '!' ) {
 872                                         $line = str_replace ( '!!' , '||' , $line );
 873                                 }
 874
 875                                 // Split up multiple cells on the same line.
 876                                 // FIXME : This can result in improper nesting of tags processed
 877                                 // by earlier parser steps, but should avoid splitting up eg
 878                                 // attribute values containing literal "||".
 879                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 880
 881                                 $lines[$key] = '';
 882
 883                                 // Loop through each table cell
 884                                 foreach ( $cells as $cell )
 885                                 {
 886                                         $previous = '';
 887                                         if ( $first_character != '+' )
 888                                         {
 889                                                 $tr_after = array_pop ( $tr_attributes );
 890                                                 if ( !array_pop ( $tr_history ) ) {
 891                                                         $previous = "<tr{$tr_after}>\n";
 892                                                 }
 893                                                 array_push ( $tr_history , true );
 894                                                 array_push ( $tr_attributes , '' );
 895                                                 array_pop ( $has_opened_tr );
 896                                                 array_push ( $has_opened_tr , true );
 897                                         }
 898
 899                                         $last_tag = array_pop ( $last_tag_history );
 900
 901                                         if ( array_pop ( $td_history ) ) {
 902                                                 $previous = "</{$last_tag}>{$previous}";
 903                                         }
 904
 905                                         if ( $first_character == '|' ) {
 906                                                 $last_tag = 'td';
 907                                         } else if ( $first_character == '!' ) {
 908                                                 $last_tag = 'th';
 909                                         } else if ( $first_character == '+' ) {
 910                                                 $last_tag = 'caption';
 911                                         } else {
 912                                                 $last_tag = '';
 913                                         }
 914
 915                                         array_push ( $last_tag_history , $last_tag );
 916
 917                                         // A cell could contain both parameters and data
 918                                         $cell_data = explode ( '|' , $cell , 2 );
 919
 920                                         // Bug 553: Note that a '|' inside an invalid link should not
 921                                         // be mistaken as delimiting cell parameters
 922                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 923                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 924                                         } else if ( count ( $cell_data ) == 1 )
 925                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 926                                         else {
 927                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 928                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 929                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 930                                         }
 931
 932                                         $lines[$key] .= $cell;
 933                                         array_push ( $td_history , true );
 934                                 }
 935                         }
 936                 }
 937
 938                 // Closing open td, tr && table
 939                 while ( count ( $td_history ) > 0 )
 940                 {
 941                         if ( array_pop ( $td_history ) ) {
 942                                 $lines[] = '</td>' ;
 943                         }
 944                         if ( array_pop ( $tr_history ) ) {
 945                                 $lines[] = '</tr>' ;
 946                         }
 947                         if ( !array_pop ( $has_opened_tr ) ) {
 948                                 $lines[] = "<tr><td></td></tr>" ;
 949                         }
 950
 951                         $lines[] = '</table>' ;
 952                 }
 953
 954                 $output = implode ( "\n" , $lines ) ;
 955
 956                 // special case: don't return empty table
 957                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 958                         $output = '';
 959                 }
 960
 961                 wfProfileOut( $fname );
 962
 963                 return $output;
 964         }
 965
 966         /**
 967          * Helper function for parse() that transforms wiki markup into
 968          * HTML. Only called for $mOutputType == OT_HTML.
 969          *
 970          * @private
 971          */
 972         function internalParse( $text ) {
 973                 $args = array();
 974                 $isMain = true;
 975                 $fname = 'Parser::internalParse';
 976                 wfProfileIn( $fname );
 977
 978                 # Hook to suspend the parser in this state
 979                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 980                         wfProfileOut( $fname );
 981                         return $text ;
 982                 }
 983
 984                 # Remove <noinclude> tags and <includeonly> sections
 985                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 986                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 987                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 988
 989                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
 990
 991                 $text = $this->replaceVariables( $text, $args );
 992                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 993
 994                 // Tables need to come after variable replacement for things to work
 995                 // properly; putting them before other transformations should keep
 996                 // exciting things like link expansions from showing up in surprising
 997                 // places.
 998                 $text = $this->doTableStuff( $text );
 999
1000                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1001
1002                 $text = $this->stripToc( $text );
1003                 $this->stripNoGallery( $text );
1004                 $text = $this->doHeadings( $text );
1005                 if($this->mOptions->getUseDynamicDates()) {
1006                         $df =& DateFormatter::getInstance();
1007                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1008                 }
1009                 $text = $this->doAllQuotes( $text );
1010                 $text = $this->replaceInternalLinks( $text );
1011                 $text = $this->replaceExternalLinks( $text );
1012
1013                 # replaceInternalLinks may sometimes leave behind
1014                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1015                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1016
1017                 $text = $this->doMagicLinks( $text );
1018                 $text = $this->formatHeadings( $text, $isMain );
1019
1020                 wfProfileOut( $fname );
1021                 return $text;
1022         }
1023
1024         /**
1025          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1026          * magic external links.
1027          *
1028          * @private
1029          */
1030         function &doMagicLinks( &$text ) {
1031                 wfProfileIn( __METHOD__ );
1032                 $text = preg_replace_callback(
1033                         '!(?:                           # Start cases
1034                             <a.*?</a> |                 # Skip link text
1035                             <.*?> |                     # Skip stuff inside HTML elements
1036                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1037                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1038                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1039                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1040                                      [0-9Xx]                 # check digit
1041                                    \b)
1042                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1043                 wfProfileOut( __METHOD__ );
1044                 return $text;
1045         }
1046
1047         function magicLinkCallback( $m ) {
1048                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1049                         # Skip HTML element
1050                         return $m[0];
1051                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1052                         $isbn = $m[2];
1053                         $num = strtr( $isbn, array(
1054                                 '-' => '',
1055                                 ' ' => '',
1056                                 'x' => 'X',
1057                         ));
1058                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1059                         $text = '<a href="' .
1060                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1061                                 "\" class=\"internal\">ISBN $isbn</a>";
1062                 } else {
1063                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1064                                 $keyword = 'RFC';
1065                                 $urlmsg = 'rfcurl';
1066                                 $id = $m[1];
1067                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1068                                 $keyword = 'PMID';
1069                                 $urlmsg = 'pubmedurl';
1070                                 $id = $m[1];
1071                         } else {
1072                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1073                                         substr($m[0], 0, 20 ) . '"' );
1074                         }
1075
1076                         $url = wfMsg( $urlmsg, $id);
1077                         $sk = $this->mOptions->getSkin();
1078                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1079                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1080                 }
1081                 return $text;
1082         }
1083
1084         /**
1085          * Parse headers and return html
1086          *
1087          * @private
1088          */
1089         function doHeadings( $text ) {
1090                 $fname = 'Parser::doHeadings';
1091                 wfProfileIn( $fname );
1092                 for ( $i = 6; $i >= 1; --$i ) {
1093                         $h = str_repeat( '=', $i );
1094                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1095                           "<h{$i}>\\1</h{$i}>\\2", $text );
1096                 }
1097                 wfProfileOut( $fname );
1098                 return $text;
1099         }
1100
1101         /**
1102          * Replace single quotes with HTML markup
1103          * @private
1104          * @return string the altered text
1105          */
1106         function doAllQuotes( $text ) {
1107                 $fname = 'Parser::doAllQuotes';
1108                 wfProfileIn( $fname );
1109                 $outtext = '';
1110                 $lines = explode( "\n", $text );
1111                 foreach ( $lines as $line ) {
1112                         $outtext .= $this->doQuotes ( $line ) . "\n";
1113                 }
1114                 $outtext = substr($outtext, 0,-1);
1115                 wfProfileOut( $fname );
1116                 return $outtext;
1117         }
1118
1119         /**
1120          * Helper function for doAllQuotes()
1121          */
1122         public function doQuotes( $text ) {
1123                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1124                 if ( count( $arr ) == 1 )
1125                         return $text;
1126                 else
1127                 {
1128                         # First, do some preliminary work. This may shift some apostrophes from
1129                         # being mark-up to being text. It also counts the number of occurrences
1130                         # of bold and italics mark-ups.
1131                         $i = 0;
1132                         $numbold = 0;
1133                         $numitalics = 0;
1134                         foreach ( $arr as $r )
1135                         {
1136                                 if ( ( $i % 2 ) == 1 )
1137                                 {
1138                                         # If there are ever four apostrophes, assume the first is supposed to
1139                                         # be text, and the remaining three constitute mark-up for bold text.
1140                                         if ( strlen( $arr[$i] ) == 4 )
1141                                         {
1142                                                 $arr[$i-1] .= "'";
1143                                                 $arr[$i] = "'''";
1144                                         }
1145                                         # If there are more than 5 apostrophes in a row, assume they're all
1146                                         # text except for the last 5.
1147                                         else if ( strlen( $arr[$i] ) > 5 )
1148                                         {
1149                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1150                                                 $arr[$i] = "'''''";
1151                                         }
1152                                         # Count the number of occurrences of bold and italics mark-ups.
1153                                         # We are not counting sequences of five apostrophes.
1154                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1155                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1156                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1157                                 }
1158                                 $i++;
1159                         }
1160
1161                         # If there is an odd number of both bold and italics, it is likely
1162                         # that one of the bold ones was meant to be an apostrophe followed
1163                         # by italics. Which one we cannot know for certain, but it is more
1164                         # likely to be one that has a single-letter word before it.
1165                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1166                         {
1167                                 $i = 0;
1168                                 $firstsingleletterword = -1;
1169                                 $firstmultiletterword = -1;
1170                                 $firstspace = -1;
1171                                 foreach ( $arr as $r )
1172                                 {
1173                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1174                                         {
1175                                                 $x1 = substr ($arr[$i-1], -1);
1176                                                 $x2 = substr ($arr[$i-1], -2, 1);
1177                                                 if ($x1 == ' ') {
1178                                                         if ($firstspace == -1) $firstspace = $i;
1179                                                 } else if ($x2 == ' ') {
1180                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1181                                                 } else {
1182                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1183                                                 }
1184                                         }
1185                                         $i++;
1186                                 }
1187
1188                                 # If there is a single-letter word, use it!
1189                                 if ($firstsingleletterword > -1)
1190                                 {
1191                                         $arr [ $firstsingleletterword ] = "''";
1192                                         $arr [ $firstsingleletterword-1 ] .= "'";
1193                                 }
1194                                 # If not, but there's a multi-letter word, use that one.
1195                                 else if ($firstmultiletterword > -1)
1196                                 {
1197                                         $arr [ $firstmultiletterword ] = "''";
1198                                         $arr [ $firstmultiletterword-1 ] .= "'";
1199                                 }
1200                                 # ... otherwise use the first one that has neither.
1201                                 # (notice that it is possible for all three to be -1 if, for example,
1202                                 # there is only one pentuple-apostrophe in the line)
1203                                 else if ($firstspace > -1)
1204                                 {
1205                                         $arr [ $firstspace ] = "''";
1206                                         $arr [ $firstspace-1 ] .= "'";
1207                                 }
1208                         }
1209
1210                         # Now let's actually convert our apostrophic mush to HTML!
1211                         $output = '';
1212                         $buffer = '';
1213                         $state = '';
1214                         $i = 0;
1215                         foreach ($arr as $r)
1216                         {
1217                                 if (($i % 2) == 0)
1218                                 {
1219                                         if ($state == 'both')
1220                                                 $buffer .= $r;
1221                                         else
1222                                                 $output .= $r;
1223                                 }
1224                                 else
1225                                 {
1226                                         if (strlen ($r) == 2)
1227                                         {
1228                                                 if ($state == 'i')
1229                                                 { $output .= '</i>'; $state = ''; }
1230                                                 else if ($state == 'bi')
1231                                                 { $output .= '</i>'; $state = 'b'; }
1232                                                 else if ($state == 'ib')
1233                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1234                                                 else if ($state == 'both')
1235                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1236                                                 else # $state can be 'b' or ''
1237                                                 { $output .= '<i>'; $state .= 'i'; }
1238                                         }
1239                                         else if (strlen ($r) == 3)
1240                                         {
1241                                                 if ($state == 'b')
1242                                                 { $output .= '</b>'; $state = ''; }
1243                                                 else if ($state == 'bi')
1244                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1245                                                 else if ($state == 'ib')
1246                                                 { $output .= '</b>'; $state = 'i'; }
1247                                                 else if ($state == 'both')
1248                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1249                                                 else # $state can be 'i' or ''
1250                                                 { $output .= '<b>'; $state .= 'b'; }
1251                                         }
1252                                         else if (strlen ($r) == 5)
1253                                         {
1254                                                 if ($state == 'b')
1255                                                 { $output .= '</b><i>'; $state = 'i'; }
1256                                                 else if ($state == 'i')
1257                                                 { $output .= '</i><b>'; $state = 'b'; }
1258                                                 else if ($state == 'bi')
1259                                                 { $output .= '</i></b>'; $state = ''; }
1260                                                 else if ($state == 'ib')
1261                                                 { $output .= '</b></i>'; $state = ''; }
1262                                                 else if ($state == 'both')
1263                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1264                                                 else # ($state == '')
1265                                                 { $buffer = ''; $state = 'both'; }
1266                                         }
1267                                 }
1268                                 $i++;
1269                         }
1270                         # Now close all remaining tags.  Notice that the order is important.
1271                         if ($state == 'b' || $state == 'ib')
1272                                 $output .= '</b>';
1273                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1274                                 $output .= '</i>';
1275                         if ($state == 'bi')
1276                                 $output .= '</b>';
1277                         # There might be lonely ''''', so make sure we have a buffer
1278                         if ($state == 'both' && $buffer)
1279                                 $output .= '<b><i>'.$buffer.'</i></b>';
1280                         return $output;
1281                 }
1282         }
1283
1284         /**
1285          * Replace external links
1286          *
1287          * Note: this is all very hackish and the order of execution matters a lot.
1288          * Make sure to run maintenance/parserTests.php if you change this code.
1289          *
1290          * @private
1291          */
1292         function replaceExternalLinks( $text ) {
1293                 global $wgContLang;
1294                 $fname = 'Parser::replaceExternalLinks';
1295                 wfProfileIn( $fname );
1296
1297                 $sk = $this->mOptions->getSkin();
1298
1299                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1300
1301                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1302
1303                 $i = 0;
1304                 while ( $i<count( $bits ) ) {
1305                         $url = $bits[$i++];
1306                         $protocol = $bits[$i++];
1307                         $text = $bits[$i++];
1308                         $trail = $bits[$i++];
1309
1310                         # The characters '<' and '>' (which were escaped by
1311                         # removeHTMLtags()) should not be included in
1312                         # URLs, per RFC 2396.
1313                         $m2 = array();
1314                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1315                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1316                                 $url = substr($url, 0, $m2[0][1]);
1317                         }
1318
1319                         # If the link text is an image URL, replace it with an <img> tag
1320                         # This happened by accident in the original parser, but some people used it extensively
1321                         $img = $this->maybeMakeExternalImage( $text );
1322                         if ( $img !== false ) {
1323                                 $text = $img;
1324                         }
1325
1326                         $dtrail = '';
1327
1328                         # Set linktype for CSS - if URL==text, link is essentially free
1329                         $linktype = ($text == $url) ? 'free' : 'text';
1330
1331                         # No link text, e.g. [http://domain.tld/some.link]
1332                         if ( $text == '' ) {
1333                                 # Autonumber if allowed. See bug #5918
1334                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1335                                         $text = '[' . ++$this->mAutonumber . ']';
1336                                         $linktype = 'autonumber';
1337                                 } else {
1338                                         # Otherwise just use the URL
1339                                         $text = htmlspecialchars( $url );
1340                                         $linktype = 'free';
1341                                 }
1342                         } else {
1343                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1344                                 # Check for trail
1345                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1346                         }
1347
1348                         $text = $wgContLang->markNoConversion($text);
1349
1350                         $url = Sanitizer::cleanUrl( $url );
1351
1352                         # Process the trail (i.e. everything after this link up until start of the next link),
1353                         # replacing any non-bracketed links
1354                         $trail = $this->replaceFreeExternalLinks( $trail );
1355
1356                         # Use the encoded URL
1357                         # This means that users can paste URLs directly into the text
1358                         # Funny characters like &ouml; aren't valid in URLs anyway
1359                         # This was changed in August 2004
1360                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->getExternalLinkAttribs() ) . $dtrail . $trail;
1361
1362                         # Register link in the output object.
1363                         # Replace unnecessary URL escape codes with the referenced character
1364                         # This prevents spammers from hiding links from the filters
1365                         $pasteurized = self::replaceUnusualEscapes( $url );
1366                         $this->mOutput->addExternalLink( $pasteurized );
1367                 }
1368
1369                 wfProfileOut( $fname );
1370                 return $s;
1371         }
1372
1373         function getExternalLinkAttribs() {
1374                 $attribs = array();
1375                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1376                 $ns = $this->mTitle->getNamespace();
1377                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1378                         $attribs['rel'] = 'nofollow';
1379                 }
1380                 if ( $this->mOptions->getExternalLinkTarget() ) {
1381                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1382                 }
1383                 return $attribs;
1384         }
1385
1386         /**
1387          * Replace anything that looks like a URL with a link
1388          * @private
1389          */
1390         function replaceFreeExternalLinks( $text ) {
1391                 global $wgContLang;
1392                 $fname = 'Parser::replaceFreeExternalLinks';
1393                 wfProfileIn( $fname );
1394
1395                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1396                 $s = array_shift( $bits );
1397                 $i = 0;
1398
1399                 $sk = $this->mOptions->getSkin();
1400
1401                 while ( $i < count( $bits ) ){
1402                         $protocol = $bits[$i++];
1403                         $remainder = $bits[$i++];
1404
1405                         $m = array();
1406                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1407                                 # Found some characters after the protocol that look promising
1408                                 $url = $protocol . $m[1];
1409                                 $trail = $m[2];
1410
1411                                 # special case: handle urls as url args:
1412                                 # http://www.example.com/foo?=http://www.example.com/bar
1413                                 if(strlen($trail) == 0 &&
1414                                         isset($bits[$i]) &&
1415                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1416                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1417                                 {
1418                                         # add protocol, arg
1419                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1420                                         $i += 2;
1421                                         $trail = $m[2];
1422                                 }
1423
1424                                 # The characters '<' and '>' (which were escaped by
1425                                 # removeHTMLtags()) should not be included in
1426                                 # URLs, per RFC 2396.
1427                                 $m2 = array();
1428                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1429                                         $trail = substr($url, $m2[0][1]) . $trail;
1430                                         $url = substr($url, 0, $m2[0][1]);
1431                                 }
1432
1433                                 # Move trailing punctuation to $trail
1434                                 $sep = ',;\.:!?';
1435                                 # If there is no left bracket, then consider right brackets fair game too
1436                                 if ( strpos( $url, '(' ) === false ) {
1437                                         $sep .= ')';
1438                                 }
1439
1440                                 $numSepChars = strspn( strrev( $url ), $sep );
1441                                 if ( $numSepChars ) {
1442                                         $trail = substr( $url, -$numSepChars ) . $trail;
1443                                         $url = substr( $url, 0, -$numSepChars );
1444                                 }
1445
1446                                 $url = Sanitizer::cleanUrl( $url );
1447
1448                                 # Is this an external image?
1449                                 $text = $this->maybeMakeExternalImage( $url );
1450                                 if ( $text === false ) {
1451                                         # Not an image, make a link
1452                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1453                                                 $this->getExternalLinkAttribs() );
1454                                         # Register it in the output object...
1455                                         # Replace unnecessary URL escape codes with their equivalent characters
1456                                         $pasteurized = self::replaceUnusualEscapes( $url );
1457                                         $this->mOutput->addExternalLink( $pasteurized );
1458                                 }
1459                                 $s .= $text . $trail;
1460                         } else {
1461                                 $s .= $protocol . $remainder;
1462                         }
1463                 }
1464                 wfProfileOut( $fname );
1465                 return $s;
1466         }
1467
1468         /**
1469          * Replace unusual URL escape codes with their equivalent characters
1470          * @param string
1471          * @return string
1472          * @static
1473          * @todo  This can merge genuinely required bits in the path or query string,
1474          *        breaking legit URLs. A proper fix would treat the various parts of
1475          *        the URL differently; as a workaround, just use the output for
1476          *        statistical records, not for actual linking/output.
1477          */
1478         static function replaceUnusualEscapes( $url ) {
1479                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1480                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1481         }
1482
1483         /**
1484          * Callback function used in replaceUnusualEscapes().
1485          * Replaces unusual URL escape codes with their equivalent character
1486          * @static
1487          * @private
1488          */
1489         private static function replaceUnusualEscapesCallback( $matches ) {
1490                 $char = urldecode( $matches[0] );
1491                 $ord = ord( $char );
1492                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1493                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1494                         // No, shouldn't be escaped
1495                         return $char;
1496                 } else {
1497                         // Yes, leave it escaped
1498                         return $matches[0];
1499                 }
1500         }
1501
1502         /**
1503          * make an image if it's allowed, either through the global
1504          * option or through the exception
1505          * @private
1506          */
1507         function maybeMakeExternalImage( $url ) {
1508                 $sk = $this->mOptions->getSkin();
1509                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1510                 $imagesexception = !empty($imagesfrom);
1511                 $text = false;
1512                 if ( $this->mOptions->getAllowExternalImages()
1513                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1514                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1515                                 # Image found
1516                                 $text = $sk->makeExternalImage( $url );
1517                         }
1518                 }
1519                 return $text;
1520         }
1521
1522         /**
1523          * Process [[ ]] wikilinks
1524          *
1525          * @private
1526          */
1527         function replaceInternalLinks( $s ) {
1528                 global $wgContLang;
1529                 static $fname = 'Parser::replaceInternalLinks' ;
1530
1531                 wfProfileIn( $fname );
1532
1533                 wfProfileIn( $fname.'-setup' );
1534                 static $tc = FALSE;
1535                 # the % is needed to support urlencoded titles as well
1536                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1537
1538                 $sk = $this->mOptions->getSkin();
1539
1540                 #split the entire text string on occurences of [[
1541                 $a = explode( '[[', ' ' . $s );
1542                 #get the first element (all text up to first [[), and remove the space we added
1543                 $s = array_shift( $a );
1544                 $s = substr( $s, 1 );
1545
1546                 # Match a link having the form [[namespace:link|alternate]]trail
1547                 static $e1 = FALSE;
1548                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1549                 # Match cases where there is no "]]", which might still be images
1550                 static $e1_img = FALSE;
1551                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1552                 # Match the end of a line for a word that's not followed by whitespace,
1553                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1554                 $e2 = wfMsgForContent( 'linkprefix' );
1555
1556                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1557                 if( is_null( $this->mTitle ) ) {
1558                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1559                 }
1560                 $nottalk = !$this->mTitle->isTalkPage();
1561
1562                 if ( $useLinkPrefixExtension ) {
1563                         $m = array();
1564                         if ( preg_match( $e2, $s, $m ) ) {
1565                                 $first_prefix = $m[2];
1566                         } else {
1567                                 $first_prefix = false;
1568                         }
1569                 } else {
1570                         $prefix = '';
1571                 }
1572
1573                 if($wgContLang->hasVariants()) {
1574                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1575                 } else {
1576                         $selflink = array($this->mTitle->getPrefixedText());
1577                 }
1578                 $useSubpages = $this->areSubpagesAllowed();
1579                 wfProfileOut( $fname.'-setup' );
1580
1581                 # Loop for each link
1582                 for ($k = 0; isset( $a[$k] ); $k++) {
1583                         $line = $a[$k];
1584                         if ( $useLinkPrefixExtension ) {
1585                                 wfProfileIn( $fname.'-prefixhandling' );
1586                                 if ( preg_match( $e2, $s, $m ) ) {
1587                                         $prefix = $m[2];
1588                                         $s = $m[1];
1589                                 } else {
1590                                         $prefix='';
1591                                 }
1592                                 # first link
1593                                 if($first_prefix) {
1594                                         $prefix = $first_prefix;
1595                                         $first_prefix = false;
1596                                 }
1597                                 wfProfileOut( $fname.'-prefixhandling' );
1598                         }
1599
1600                         $might_be_img = false;
1601
1602                         wfProfileIn( "$fname-e1" );
1603                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1604                                 $text = $m[2];
1605                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1606                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1607                                 # the real problem is with the $e1 regex
1608                                 # See bug 1300.
1609                                 #
1610                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1611                                 # and no image is in sight. See bug 2095.
1612                                 #
1613                                 if( $text !== '' &&
1614                                         substr( $m[3], 0, 1 ) === ']' &&
1615                                         strpos($text, '[') !== false
1616                                 )
1617                                 {
1618                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1619                                         $m[3] = substr( $m[3], 1 );
1620                                 }
1621                                 # fix up urlencoded title texts
1622                                 if( strpos( $m[1], '%' ) !== false ) {
1623                                         # Should anchors '#' also be rejected?
1624                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1625                                 }
1626                                 $trail = $m[3];
1627                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1628                                 $might_be_img = true;
1629                                 $text = $m[2];
1630                                 if ( strpos( $m[1], '%' ) !== false ) {
1631                                         $m[1] = urldecode($m[1]);
1632                                 }
1633                                 $trail = "";
1634                         } else { # Invalid form; output directly
1635                                 $s .= $prefix . '[[' . $line ;
1636                                 wfProfileOut( "$fname-e1" );
1637                                 continue;
1638                         }
1639                         wfProfileOut( "$fname-e1" );
1640                         wfProfileIn( "$fname-misc" );
1641
1642                         # Don't allow internal links to pages containing
1643                         # PROTO: where PROTO is a valid URL protocol; these
1644                         # should be external links.
1645                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1646                                 $s .= $prefix . '[[' . $line ;
1647                                 continue;
1648                         }
1649
1650                         # Make subpage if necessary
1651                         if( $useSubpages ) {
1652                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1653                         } else {
1654                                 $link = $m[1];
1655                         }
1656
1657                         $noforce = (substr($m[1], 0, 1) != ':');
1658                         if (!$noforce) {
1659                                 # Strip off leading ':'
1660                                 $link = substr($link, 1);
1661                         }
1662
1663                         wfProfileOut( "$fname-misc" );
1664                         wfProfileIn( "$fname-title" );
1665                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1666                         if( !$nt ) {
1667                                 $s .= $prefix . '[[' . $line;
1668                                 wfProfileOut( "$fname-title" );
1669                                 continue;
1670                         }
1671
1672                         $ns = $nt->getNamespace();
1673                         $iw = $nt->getInterWiki();
1674                         wfProfileOut( "$fname-title" );
1675
1676                         if ($might_be_img) { # if this is actually an invalid link
1677                                 wfProfileIn( "$fname-might_be_img" );
1678                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1679                                         $found = false;
1680                                         while (isset ($a[$k+1]) ) {
1681                                                 #look at the next 'line' to see if we can close it there
1682                                                 $spliced = array_splice( $a, $k + 1, 1 );
1683                                                 $next_line = array_shift( $spliced );
1684                                                 $m = explode( ']]', $next_line, 3 );
1685                                                 if ( count( $m ) == 3 ) {
1686                                                         # the first ]] closes the inner link, the second the image
1687                                                         $found = true;
1688                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1689                                                         $trail = $m[2];
1690                                                         break;
1691                                                 } elseif ( count( $m ) == 2 ) {
1692                                                         #if there's exactly one ]] that's fine, we'll keep looking
1693                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1694                                                 } else {
1695                                                         #if $next_line is invalid too, we need look no further
1696                                                         $text .= '[[' . $next_line;
1697                                                         break;
1698                                                 }
1699                                         }
1700                                         if ( !$found ) {
1701                                                 # we couldn't find the end of this imageLink, so output it raw
1702                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1703                                                 $text = $this->replaceInternalLinks($text);
1704                                                 $s .= "{$prefix}[[$link|$text";
1705                                                 # note: no $trail, because without an end, there *is* no trail
1706                                                 wfProfileOut( "$fname-might_be_img" );
1707                                                 continue;
1708                                         }
1709                                 } else { #it's not an image, so output it raw
1710                                         $s .= "{$prefix}[[$link|$text";
1711                                         # note: no $trail, because without an end, there *is* no trail
1712                                         wfProfileOut( "$fname-might_be_img" );
1713                                         continue;
1714                                 }
1715                                 wfProfileOut( "$fname-might_be_img" );
1716                         }
1717
1718                         $wasblank = ( '' == $text );
1719                         if( $wasblank ) $text = $link;
1720
1721                         # Link not escaped by : , create the various objects
1722                         if( $noforce ) {
1723
1724                                 # Interwikis
1725                                 wfProfileIn( "$fname-interwiki" );
1726                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1727                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1728                                         $s = rtrim($s . $prefix);
1729                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1730                                         wfProfileOut( "$fname-interwiki" );
1731                                         continue;
1732                                 }
1733                                 wfProfileOut( "$fname-interwiki" );
1734
1735                                 if ( $ns == NS_IMAGE ) {
1736                                         wfProfileIn( "$fname-image" );
1737                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1738                                                 # recursively parse links inside the image caption
1739                                                 # actually, this will parse them in any other parameters, too,
1740                                                 # but it might be hard to fix that, and it doesn't matter ATM
1741                                                 $text = $this->replaceExternalLinks($text);
1742                                                 $text = $this->replaceInternalLinks($text);
1743
1744                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1745                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1746                                                 $this->mOutput->addImage( $nt->getDBkey() );
1747
1748                                                 wfProfileOut( "$fname-image" );
1749                                                 continue;
1750                                         } else {
1751                                                 # We still need to record the image's presence on the page
1752                                                 $this->mOutput->addImage( $nt->getDBkey() );
1753                                         }
1754                                         wfProfileOut( "$fname-image" );
1755
1756                                 }
1757
1758                                 if ( $ns == NS_CATEGORY ) {
1759                                         wfProfileIn( "$fname-category" );
1760                                         $s = rtrim($s . "\n"); # bug 87
1761
1762                                         if ( $wasblank ) {
1763                                                 $sortkey = $this->getDefaultSort();
1764                                         } else {
1765                                                 $sortkey = $text;
1766                                         }
1767                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1768                                         $sortkey = str_replace( "\n", '', $sortkey );
1769                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1770                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1771
1772                                         /**
1773                                          * Strip the whitespace Category links produce, see bug 87
1774                                          * @todo We might want to use trim($tmp, "\n") here.
1775                                          */
1776                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1777
1778                                         wfProfileOut( "$fname-category" );
1779                                         continue;
1780                                 }
1781                         }
1782
1783                         # Self-link checking
1784                         if( $nt->getFragment() === '' ) {
1785                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1786                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1787                                         continue;
1788                                 }
1789                         }
1790
1791                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1792                         if( $ns == NS_MEDIA ) {
1793                                 # Give extensions a chance to select the file revision for us
1794                                 $skip = $time = false;
1795                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1796                                 if ( $skip ) {
1797                                         $link = $sk->makeLinkObj( $nt );
1798                                 } else {
1799                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1800                                 }
1801                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1802                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1803                                 $this->mOutput->addImage( $nt->getDBkey() );
1804                                 continue;
1805                         } elseif( $ns == NS_SPECIAL ) {
1806                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1807                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1808                                 } else {
1809                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1810                                 }
1811                                 continue;
1812                         } elseif( $ns == NS_IMAGE ) {
1813                                 $img = wfFindFile( $nt );
1814                                 if( $img ) {
1815                                         // Force a blue link if the file exists; may be a remote
1816                                         // upload on the shared repository, and we want to see its
1817                                         // auto-generated page.
1818                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1819                                         $this->mOutput->addLink( $nt );
1820                                         continue;
1821                                 }
1822                         }
1823                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1824                 }
1825                 wfProfileOut( $fname );
1826                 return $s;
1827         }
1828
1829         /**
1830          * Make a link placeholder. The text returned can be later resolved to a real link with
1831          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1832          * parsing of interwiki links, and secondly to allow all existence checks and
1833          * article length checks (for stub links) to be bundled into a single query.
1834          *
1835          */
1836         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1837                 wfProfileIn( __METHOD__ );
1838                 if ( ! is_object($nt) ) {
1839                         # Fail gracefully
1840                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1841                 } else {
1842                         # Separate the link trail from the rest of the link
1843                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1844
1845                         if ( $nt->isExternal() ) {
1846                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1847                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1848                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1849                         } else {
1850                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1851                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1852                                 $this->mLinkHolders['queries'][] = $query;
1853                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1854                                 $this->mLinkHolders['titles'][] = $nt;
1855
1856                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1857                         }
1858                 }
1859                 wfProfileOut( __METHOD__ );
1860                 return $retVal;
1861         }
1862
1863         /**
1864          * Render a forced-blue link inline; protect against double expansion of
1865          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1866          * Since this little disaster has to split off the trail text to avoid
1867          * breaking URLs in the following text without breaking trails on the
1868          * wiki links, it's been made into a horrible function.
1869          *
1870          * @param Title $nt
1871          * @param string $text
1872          * @param string $query
1873          * @param string $trail
1874          * @param string $prefix
1875          * @return string HTML-wikitext mix oh yuck
1876          */
1877         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1878                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1879                 $sk = $this->mOptions->getSkin();
1880                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1881                 return $this->armorLinks( $link ) . $trail;
1882         }
1883
1884         /**
1885          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1886          * going to go through further parsing steps before inline URL expansion.
1887          *
1888          * In particular this is important when using action=render, which causes
1889          * full URLs to be included.
1890          *
1891          * Oh man I hate our multi-layer parser!
1892          *
1893          * @param string more-or-less HTML
1894          * @return string less-or-more HTML with NOPARSE bits
1895          */
1896         function armorLinks( $text ) {
1897                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1898                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1899         }
1900
1901         /**
1902          * Return true if subpage links should be expanded on this page.
1903          * @return bool
1904          */
1905         function areSubpagesAllowed() {
1906                 # Some namespaces don't allow subpages
1907                 global $wgNamespacesWithSubpages;
1908                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1909         }
1910
1911         /**
1912          * Handle link to subpage if necessary
1913          * @param string $target the source of the link
1914          * @param string &$text the link text, modified as necessary
1915          * @return string the full name of the link
1916          * @private
1917          */
1918         function maybeDoSubpageLink($target, &$text) {
1919                 # Valid link forms:
1920                 # Foobar -- normal
1921                 # :Foobar -- override special treatment of prefix (images, language links)
1922                 # /Foobar -- convert to CurrentPage/Foobar
1923                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1924                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1925                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1926
1927                 $fname = 'Parser::maybeDoSubpageLink';
1928                 wfProfileIn( $fname );
1929                 $ret = $target; # default return value is no change
1930
1931                 # Some namespaces don't allow subpages,
1932                 # so only perform processing if subpages are allowed
1933                 if( $this->areSubpagesAllowed() ) {
1934                         $hash = strpos( $target, '#' );
1935                         if( $hash !== false ) {
1936                                 $suffix = substr( $target, $hash );
1937                                 $target = substr( $target, 0, $hash );
1938                         } else {
1939                                 $suffix = '';
1940                         }
1941                         # bug 7425
1942                         $target = trim( $target );
1943                         # Look at the first character
1944                         if( $target != '' && $target{0} == '/' ) {
1945                                 # / at end means we don't want the slash to be shown
1946                                 $m = array();
1947                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1948                                 if( $trailingSlashes ) {
1949                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1950                                 } else {
1951                                         $noslash = substr( $target, 1 );
1952                                 }
1953
1954                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1955                                 if( '' === $text ) {
1956                                         $text = $target . $suffix;
1957                                 } # this might be changed for ugliness reasons
1958                         } else {
1959                                 # check for .. subpage backlinks
1960                                 $dotdotcount = 0;
1961                                 $nodotdot = $target;
1962                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1963                                         ++$dotdotcount;
1964                                         $nodotdot = substr( $nodotdot, 3 );
1965                                 }
1966                                 if($dotdotcount > 0) {
1967                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1968                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1969                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1970                                                 # / at the end means don't show full path
1971                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1972                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1973                                                         if( '' === $text ) {
1974                                                                 $text = $nodotdot . $suffix;
1975                                                         }
1976                                                 }
1977                                                 $nodotdot = trim( $nodotdot );
1978                                                 if( $nodotdot != '' ) {
1979                                                         $ret .= '/' . $nodotdot;
1980                                                 }
1981                                                 $ret .= $suffix;
1982                                         }
1983                                 }
1984                         }
1985                 }
1986
1987                 wfProfileOut( $fname );
1988                 return $ret;
1989         }
1990
1991         /**#@+
1992          * Used by doBlockLevels()
1993          * @private
1994          */
1995         /* private */ function closeParagraph() {
1996                 $result = '';
1997                 if ( '' != $this->mLastSection ) {
1998                         $result = '</' . $this->mLastSection  . ">\n";
1999                 }
2000                 $this->mInPre = false;
2001                 $this->mLastSection = '';
2002                 return $result;
2003         }
2004         # getCommon() returns the length of the longest common substring
2005         # of both arguments, starting at the beginning of both.
2006         #
2007         /* private */ function getCommon( $st1, $st2 ) {
2008                 $fl = strlen( $st1 );
2009                 $shorter = strlen( $st2 );
2010                 if ( $fl < $shorter ) { $shorter = $fl; }
2011
2012                 for ( $i = 0; $i < $shorter; ++$i ) {
2013                         if ( $st1{$i} != $st2{$i} ) { break; }
2014                 }
2015                 return $i;
2016         }
2017         # These next three functions open, continue, and close the list
2018         # element appropriate to the prefix character passed into them.
2019         #
2020         /* private */ function openList( $char ) {
2021                 $result = $this->closeParagraph();
2022
2023                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2024                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2025                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2026                 else if ( ';' == $char ) {
2027                         $result .= '<dl><dt>';
2028                         $this->mDTopen = true;
2029                 }
2030                 else { $result = '<!-- ERR 1 -->'; }
2031
2032                 return $result;
2033         }
2034
2035         /* private */ function nextItem( $char ) {
2036                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2037                 else if ( ':' == $char || ';' == $char ) {
2038                         $close = '</dd>';
2039                         if ( $this->mDTopen ) { $close = '</dt>'; }
2040                         if ( ';' == $char ) {
2041                                 $this->mDTopen = true;
2042                                 return $close . '<dt>';
2043                         } else {
2044                                 $this->mDTopen = false;
2045                                 return $close . '<dd>';
2046                         }
2047                 }
2048                 return '<!-- ERR 2 -->';
2049         }
2050
2051         /* private */ function closeList( $char ) {
2052                 if ( '*' == $char ) { $text = '</li></ul>'; }
2053                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2054                 else if ( ':' == $char ) {
2055                         if ( $this->mDTopen ) {
2056                                 $this->mDTopen = false;
2057                                 $text = '</dt></dl>';
2058                         } else {
2059                                 $text = '</dd></dl>';
2060                         }
2061                 }
2062                 else {  return '<!-- ERR 3 -->'; }
2063                 return $text."\n";
2064         }
2065         /**#@-*/
2066
2067         /**
2068          * Make lists from lines starting with ':', '*', '#', etc.
2069          *
2070          * @private
2071          * @return string the lists rendered as HTML
2072          */
2073         function doBlockLevels( $text, $linestart ) {
2074                 $fname = 'Parser::doBlockLevels';
2075                 wfProfileIn( $fname );
2076
2077                 # Parsing through the text line by line.  The main thing
2078                 # happening here is handling of block-level elements p, pre,
2079                 # and making lists from lines starting with * # : etc.
2080                 #
2081                 $textLines = explode( "\n", $text );
2082
2083                 $lastPrefix = $output = '';
2084                 $this->mDTopen = $inBlockElem = false;
2085                 $prefixLength = 0;
2086                 $paragraphStack = false;
2087
2088                 if ( !$linestart ) {
2089                         $output .= array_shift( $textLines );
2090                 }
2091                 foreach ( $textLines as $oLine ) {
2092                         $lastPrefixLength = strlen( $lastPrefix );
2093                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2094                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2095                         if ( !$this->mInPre ) {
2096                                 # Multiple prefixes may abut each other for nested lists.
2097                                 $prefixLength = strspn( $oLine, '*#:;' );
2098                                 $pref = substr( $oLine, 0, $prefixLength );
2099
2100                                 # eh?
2101                                 $pref2 = str_replace( ';', ':', $pref );
2102                                 $t = substr( $oLine, $prefixLength );
2103                                 $this->mInPre = !empty($preOpenMatch);
2104                         } else {
2105                                 # Don't interpret any other prefixes in preformatted text
2106                                 $prefixLength = 0;
2107                                 $pref = $pref2 = '';
2108                                 $t = $oLine;
2109                         }
2110
2111                         # List generation
2112                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2113                                 # Same as the last item, so no need to deal with nesting or opening stuff
2114                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2115                                 $paragraphStack = false;
2116
2117                                 if ( substr( $pref, -1 ) == ';') {
2118                                         # The one nasty exception: definition lists work like this:
2119                                         # ; title : definition text
2120                                         # So we check for : in the remainder text to split up the
2121                                         # title and definition, without b0rking links.
2122                                         $term = $t2 = '';
2123                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2124                                                 $t = $t2;
2125                                                 $output .= $term . $this->nextItem( ':' );
2126                                         }
2127                                 }
2128                         } elseif( $prefixLength || $lastPrefixLength ) {
2129                                 # Either open or close a level...
2130                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2131                                 $paragraphStack = false;
2132
2133                                 while( $commonPrefixLength < $lastPrefixLength ) {
2134                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2135                                         --$lastPrefixLength;
2136                                 }
2137                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2138                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2139                                 }
2140                                 while ( $prefixLength > $commonPrefixLength ) {
2141                                         $char = substr( $pref, $commonPrefixLength, 1 );
2142                                         $output .= $this->openList( $char );
2143
2144                                         if ( ';' == $char ) {
2145                                                 # FIXME: This is dupe of code above
2146                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2147                                                         $t = $t2;
2148                                                         $output .= $term . $this->nextItem( ':' );
2149                                                 }
2150                                         }
2151                                         ++$commonPrefixLength;
2152                                 }
2153                                 $lastPrefix = $pref2;
2154                         }
2155                         if( 0 == $prefixLength ) {
2156                                 wfProfileIn( "$fname-paragraph" );
2157                                 # No prefix (not in list)--go to paragraph mode
2158                                 // XXX: use a stack for nestable elements like span, table and div
2159                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2160                                 $closematch = preg_match(
2161                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2162                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2163                                 if ( $openmatch or $closematch ) {
2164                                         $paragraphStack = false;
2165                                         # TODO bug 5718: paragraph closed
2166                                         $output .= $this->closeParagraph();
2167                                         if ( $preOpenMatch and !$preCloseMatch ) {
2168                                                 $this->mInPre = true;
2169                                         }
2170                                         if ( $closematch ) {
2171                                                 $inBlockElem = false;
2172                                         } else {
2173                                                 $inBlockElem = true;
2174                                         }
2175                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2176                                         if ( '' != $t and ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2177                                                 // pre
2178                                                 if ($this->mLastSection != 'pre') {
2179                                                         $paragraphStack = false;
2180                                                         $output .= $this->closeParagraph().'<pre>';
2181                                                         $this->mLastSection = 'pre';
2182                                                 }
2183                                                 $t = substr( $t, 1 );
2184                                         } else {
2185                                                 // paragraph
2186                                                 if ( '' == trim($t) ) {
2187                                                         if ( $paragraphStack ) {
2188                                                                 $output .= $paragraphStack.'<br />';
2189                                                                 $paragraphStack = false;
2190                                                                 $this->mLastSection = 'p';
2191                                                         } else {
2192                                                                 if ($this->mLastSection != 'p' ) {
2193                                                                         $output .= $this->closeParagraph();
2194                                                                         $this->mLastSection = '';
2195                                                                         $paragraphStack = '<p>';
2196                                                                 } else {
2197                                                                         $paragraphStack = '</p><p>';
2198                                                                 }
2199                                                         }
2200                                                 } else {
2201                                                         if ( $paragraphStack ) {
2202                                                                 $output .= $paragraphStack;
2203                                                                 $paragraphStack = false;
2204                                                                 $this->mLastSection = 'p';
2205                                                         } else if ($this->mLastSection != 'p') {
2206                                                                 $output .= $this->closeParagraph().'<p>';
2207                                                                 $this->mLastSection = 'p';
2208                                                         }
2209                                                 }
2210                                         }
2211                                 }
2212                                 wfProfileOut( "$fname-paragraph" );
2213                         }
2214                         // somewhere above we forget to get out of pre block (bug 785)
2215                         if($preCloseMatch && $this->mInPre) {
2216                                 $this->mInPre = false;
2217                         }
2218                         if ($paragraphStack === false) {
2219                                 $output .= $t."\n";
2220                         }
2221                 }
2222                 while ( $prefixLength ) {
2223                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2224                         --$prefixLength;
2225                 }
2226                 if ( '' != $this->mLastSection ) {
2227                         $output .= '</' . $this->mLastSection . '>';
2228                         $this->mLastSection = '';
2229                 }
2230
2231                 wfProfileOut( $fname );
2232                 return $output;
2233         }
2234
2235         /**
2236          * Split up a string on ':', ignoring any occurences inside tags
2237          * to prevent illegal overlapping.
2238          * @param string $str the string to split
2239          * @param string &$before set to everything before the ':'
2240          * @param string &$after set to everything after the ':'
2241          * return string the position of the ':', or false if none found
2242          */
2243         function findColonNoLinks($str, &$before, &$after) {
2244                 $fname = 'Parser::findColonNoLinks';
2245                 wfProfileIn( $fname );
2246
2247                 $pos = strpos( $str, ':' );
2248                 if( $pos === false ) {
2249                         // Nothing to find!
2250                         wfProfileOut( $fname );
2251                         return false;
2252                 }
2253
2254                 $lt = strpos( $str, '<' );
2255                 if( $lt === false || $lt > $pos ) {
2256                         // Easy; no tag nesting to worry about
2257                         $before = substr( $str, 0, $pos );
2258                         $after = substr( $str, $pos+1 );
2259                         wfProfileOut( $fname );
2260                         return $pos;
2261                 }
2262
2263                 // Ugly state machine to walk through avoiding tags.
2264                 $state = self::COLON_STATE_TEXT;
2265                 $stack = 0;
2266                 $len = strlen( $str );
2267                 for( $i = 0; $i < $len; $i++ ) {
2268                         $c = $str{$i};
2269
2270                         switch( $state ) {
2271                         // (Using the number is a performance hack for common cases)
2272                         case 0: // self::COLON_STATE_TEXT:
2273                                 switch( $c ) {
2274                                 case "<":
2275                                         // Could be either a <start> tag or an </end> tag
2276                                         $state = self::COLON_STATE_TAGSTART;
2277                                         break;
2278                                 case ":":
2279                                         if( $stack == 0 ) {
2280                                                 // We found it!
2281                                                 $before = substr( $str, 0, $i );
2282                                                 $after = substr( $str, $i + 1 );
2283                                                 wfProfileOut( $fname );
2284                                                 return $i;
2285                                         }
2286                                         // Embedded in a tag; don't break it.
2287                                         break;
2288                                 default:
2289                                         // Skip ahead looking for something interesting
2290                                         $colon = strpos( $str, ':', $i );
2291                                         if( $colon === false ) {
2292                                                 // Nothing else interesting
2293                                                 wfProfileOut( $fname );
2294                                                 return false;
2295                                         }
2296                                         $lt = strpos( $str, '<', $i );
2297                                         if( $stack === 0 ) {
2298                                                 if( $lt === false || $colon < $lt ) {
2299                                                         // We found it!
2300                                                         $before = substr( $str, 0, $colon );
2301                                                         $after = substr( $str, $colon + 1 );
2302                                                         wfProfileOut( $fname );
2303                                                         return $i;
2304                                                 }
2305                                         }
2306                                         if( $lt === false ) {
2307                                                 // Nothing else interesting to find; abort!
2308                                                 // We're nested, but there's no close tags left. Abort!
2309                                                 break 2;
2310                                         }
2311                                         // Skip ahead to next tag start
2312                                         $i = $lt;
2313                                         $state = self::COLON_STATE_TAGSTART;
2314                                 }
2315                                 break;
2316                         case 1: // self::COLON_STATE_TAG:
2317                                 // In a <tag>
2318                                 switch( $c ) {
2319                                 case ">":
2320                                         $stack++;
2321                                         $state = self::COLON_STATE_TEXT;
2322                                         break;
2323                                 case "/":
2324                                         // Slash may be followed by >?
2325                                         $state = self::COLON_STATE_TAGSLASH;
2326                                         break;
2327                                 default:
2328                                         // ignore
2329                                 }
2330                                 break;
2331                         case 2: // self::COLON_STATE_TAGSTART:
2332                                 switch( $c ) {
2333                                 case "/":
2334                                         $state = self::COLON_STATE_CLOSETAG;
2335                                         break;
2336                                 case "!":
2337                                         $state = self::COLON_STATE_COMMENT;
2338                                         break;
2339                                 case ">":
2340                                         // Illegal early close? This shouldn't happen D:
2341                                         $state = self::COLON_STATE_TEXT;
2342                                         break;
2343                                 default:
2344                                         $state = self::COLON_STATE_TAG;
2345                                 }
2346                                 break;
2347                         case 3: // self::COLON_STATE_CLOSETAG:
2348                                 // In a </tag>
2349                                 if( $c == ">" ) {
2350                                         $stack--;
2351                                         if( $stack < 0 ) {
2352                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2353                                                 wfProfileOut( $fname );
2354                                                 return false;
2355                                         }
2356                                         $state = self::COLON_STATE_TEXT;
2357                                 }
2358                                 break;
2359                         case self::COLON_STATE_TAGSLASH:
2360                                 if( $c == ">" ) {
2361                                         // Yes, a self-closed tag <blah/>
2362                                         $state = self::COLON_STATE_TEXT;
2363                                 } else {
2364                                         // Probably we're jumping the gun, and this is an attribute
2365                                         $state = self::COLON_STATE_TAG;
2366                                 }
2367                                 break;
2368                         case 5: // self::COLON_STATE_COMMENT:
2369                                 if( $c == "-" ) {
2370                                         $state = self::COLON_STATE_COMMENTDASH;
2371                                 }
2372                                 break;
2373                         case self::COLON_STATE_COMMENTDASH:
2374                                 if( $c == "-" ) {
2375                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2376                                 } else {
2377                                         $state = self::COLON_STATE_COMMENT;
2378                                 }
2379                                 break;
2380                         case self::COLON_STATE_COMMENTDASHDASH:
2381                                 if( $c == ">" ) {
2382                                         $state = self::COLON_STATE_TEXT;
2383                                 } else {
2384                                         $state = self::COLON_STATE_COMMENT;
2385                                 }
2386                                 break;
2387                         default:
2388                                 throw new MWException( "State machine error in $fname" );
2389                         }
2390                 }
2391                 if( $stack > 0 ) {
2392                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2393                         return false;
2394                 }
2395                 wfProfileOut( $fname );
2396                 return false;
2397         }
2398
2399         /**
2400          * Return value of a magic variable (like PAGENAME)
2401          *
2402          * @private
2403          */
2404         function getVariableValue( $index ) {
2405                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2406
2407                 /**
2408                  * Some of these require message or data lookups and can be
2409                  * expensive to check many times.
2410                  */
2411                 static $varCache = array();
2412                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2413                         if ( isset( $varCache[$index] ) ) {
2414                                 return $varCache[$index];
2415                         }
2416                 }
2417
2418                 $ts = time();
2419                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2420
2421                 # Use the time zone
2422                 global $wgLocaltimezone;
2423                 if ( isset( $wgLocaltimezone ) ) {
2424                         $oldtz = getenv( 'TZ' );
2425                         putenv( 'TZ='.$wgLocaltimezone );
2426                 }
2427
2428                 wfSuppressWarnings(); // E_STRICT system time bitching
2429                 $localTimestamp = date( 'YmdHis', $ts );
2430                 $localMonth = date( 'm', $ts );
2431                 $localMonthName = date( 'n', $ts );
2432                 $localDay = date( 'j', $ts );
2433                 $localDay2 = date( 'd', $ts );
2434                 $localDayOfWeek = date( 'w', $ts );
2435                 $localWeek = date( 'W', $ts );
2436                 $localYear = date( 'Y', $ts );
2437                 $localHour = date( 'H', $ts );
2438                 if ( isset( $wgLocaltimezone ) ) {
2439                         putenv( 'TZ='.$oldtz );
2440                 }
2441                 wfRestoreWarnings();
2442
2443                 switch ( $index ) {
2444                         case 'currentmonth':
2445                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2446                         case 'currentmonthname':
2447                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2448                         case 'currentmonthnamegen':
2449                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2450                         case 'currentmonthabbrev':
2451                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2452                         case 'currentday':
2453                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2454                         case 'currentday2':
2455                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2456                         case 'localmonth':
2457                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2458                         case 'localmonthname':
2459                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2460                         case 'localmonthnamegen':
2461                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2462                         case 'localmonthabbrev':
2463                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2464                         case 'localday':
2465                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2466                         case 'localday2':
2467                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2468                         case 'pagename':
2469                                 return wfEscapeWikiText( $this->mTitle->getText() );
2470                         case 'pagenamee':
2471                                 return $this->mTitle->getPartialURL();
2472                         case 'fullpagename':
2473                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2474                         case 'fullpagenamee':
2475                                 return $this->mTitle->getPrefixedURL();
2476                         case 'subpagename':
2477                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2478                         case 'subpagenamee':
2479                                 return $this->mTitle->getSubpageUrlForm();
2480                         case 'basepagename':
2481                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2482                         case 'basepagenamee':
2483                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2484                         case 'talkpagename':
2485                                 if( $this->mTitle->canTalk() ) {
2486                                         $talkPage = $this->mTitle->getTalkPage();
2487                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2488                                 } else {
2489                                         return '';
2490                                 }
2491                         case 'talkpagenamee':
2492                                 if( $this->mTitle->canTalk() ) {
2493                                         $talkPage = $this->mTitle->getTalkPage();
2494                                         return $talkPage->getPrefixedUrl();
2495                                 } else {
2496                                         return '';
2497                                 }
2498                         case 'subjectpagename':
2499                                 $subjPage = $this->mTitle->getSubjectPage();
2500                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2501                         case 'subjectpagenamee':
2502                                 $subjPage = $this->mTitle->getSubjectPage();
2503                                 return $subjPage->getPrefixedUrl();
2504                         case 'revisionid':
2505                                 return $this->mRevisionId;
2506                         case 'revisionday':
2507                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2508                         case 'revisionday2':
2509                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2510                         case 'revisionmonth':
2511                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2512                         case 'revisionyear':
2513                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2514                         case 'revisiontimestamp':
2515                                 return $this->getRevisionTimestamp();
2516                         case 'namespace':
2517                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2518                         case 'namespacee':
2519                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2520                         case 'talkspace':
2521                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2522                         case 'talkspacee':
2523                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2524                         case 'subjectspace':
2525                                 return $this->mTitle->getSubjectNsText();
2526                         case 'subjectspacee':
2527                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2528                         case 'currentdayname':
2529                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2530                         case 'currentyear':
2531                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2532                         case 'currenttime':
2533                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2534                         case 'currenthour':
2535                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2536                         case 'currentweek':
2537                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2538                                 // int to remove the padding
2539                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2540                         case 'currentdow':
2541                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2542                         case 'localdayname':
2543                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2544                         case 'localyear':
2545                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2546                         case 'localtime':
2547                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2548                         case 'localhour':
2549                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2550                         case 'localweek':
2551                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2552                                 // int to remove the padding
2553                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2554                         case 'localdow':
2555                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2556                         case 'numberofarticles':
2557                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2558                         case 'numberoffiles':
2559                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2560                         case 'numberofusers':
2561                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2562                         case 'numberofpages':
2563                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2564                         case 'numberofadmins':
2565                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2566                         case 'numberofedits':
2567                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2568                         case 'currenttimestamp':
2569                                 return $varCache[$index] = wfTimestampNow();
2570                         case 'localtimestamp':
2571                                 return $varCache[$index] = $localTimestamp;
2572                         case 'currentversion':
2573                                 return $varCache[$index] = SpecialVersion::getVersion();
2574                         case 'sitename':
2575                                 return $wgSitename;
2576                         case 'server':
2577                                 return $wgServer;
2578                         case 'servername':
2579                                 return $wgServerName;
2580                         case 'scriptpath':
2581                                 return $wgScriptPath;
2582                         case 'directionmark':
2583                                 return $wgContLang->getDirMark();
2584                         case 'contentlanguage':
2585                                 global $wgContLanguageCode;
2586                                 return $wgContLanguageCode;
2587                         default:
2588                                 $ret = null;
2589                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2590                                         return $ret;
2591                                 else
2592                                         return null;
2593                 }
2594         }
2595
2596         /**
2597          * initialise the magic variables (like CURRENTMONTHNAME)
2598          *
2599          * @private
2600          */
2601         function initialiseVariables() {
2602                 $fname = 'Parser::initialiseVariables';
2603                 wfProfileIn( $fname );
2604                 $variableIDs = MagicWord::getVariableIDs();
2605
2606                 $this->mVariables = array();
2607                 foreach ( $variableIDs as $id ) {
2608                         $mw =& MagicWord::get( $id );
2609                         $mw->addToArray( $this->mVariables, $id );
2610                 }
2611                 wfProfileOut( $fname );
2612         }
2613
2614         /**
2615          * parse any parentheses in format ((title|part|part))
2616          * and call callbacks to get a replacement text for any found piece
2617          *
2618          * @param string $text The text to parse
2619          * @param array $callbacks rules in form:
2620          *     '{' => array(                            # opening parentheses
2621          *                                      'end' => '}',   # closing parentheses
2622          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2623          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2624          *                                                                )
2625          *                                      )
2626          *                                      'min' => 2,     # Minimum parenthesis count in cb
2627          *                                      'max' => 3,     # Maximum parenthesis count in cb
2628          * @private
2629          */
2630         function replace_callback ($text, $callbacks) {
2631                 wfProfileIn( __METHOD__ );
2632                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2633                 $lastOpeningBrace = -1;                 # last not closed parentheses
2634
2635                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2636
2637                 $i = 0;
2638                 while ( $i < strlen( $text ) ) {
2639                         # Find next opening brace, closing brace or pipe
2640                         if ( $lastOpeningBrace == -1 ) {
2641                                 $currentClosing = '';
2642                                 $search = $validOpeningBraces;
2643                         } else {
2644                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2645                                 $search = $validOpeningBraces . '|' . $currentClosing;
2646                         }
2647                         $rule = null;
2648                         $i += strcspn( $text, $search, $i );
2649                         if ( $i < strlen( $text ) ) {
2650                                 if ( $text[$i] == '|' ) {
2651                                         $found = 'pipe';
2652                                 } elseif ( $text[$i] == $currentClosing ) {
2653                                         $found = 'close';
2654                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2655                                         $found = 'open';
2656                                         $rule = $callbacks[$text[$i]];
2657                                 } else {
2658                                         # Some versions of PHP have a strcspn which stops on null characters
2659                                         # Ignore and continue
2660                                         ++$i;
2661                                         continue;
2662                                 }
2663                         } else {
2664                                 # All done
2665                                 break;
2666                         }
2667
2668                         if ( $found == 'open' ) {
2669                                 # found opening brace, let's add it to parentheses stack
2670                                 $piece = array('brace' => $text[$i],
2671                                                            'braceEnd' => $rule['end'],
2672                                                            'title' => '',
2673                                                            'parts' => null);
2674
2675                                 # count opening brace characters
2676                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2677                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2678                                 $i += $piece['count'];
2679
2680                                 # we need to add to stack only if opening brace count is enough for one of the rules
2681                                 if ( $piece['count'] >= $rule['min'] ) {
2682                                         $lastOpeningBrace ++;
2683                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2684                                 }
2685                         } elseif ( $found == 'close' ) {
2686                                 # lets check if it is enough characters for closing brace
2687                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2688                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2689
2690                                 # check for maximum matching characters (if there are 5 closing
2691                                 # characters, we will probably need only 3 - depending on the rules)
2692                                 $matchingCount = 0;
2693                                 $matchingCallback = null;
2694                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2695                                 if ( $count > $cbType['max'] ) {
2696                                         # The specified maximum exists in the callback array, unless the caller
2697                                         # has made an error
2698                                         $matchingCount = $cbType['max'];
2699                                 } else {
2700                                         # Count is less than the maximum
2701                                         # Skip any gaps in the callback array to find the true largest match
2702                                         # Need to use array_key_exists not isset because the callback can be null
2703                                         $matchingCount = $count;
2704                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2705                                                 --$matchingCount;
2706                                         }
2707                                 }
2708
2709                                 if ($matchingCount <= 0) {
2710                                         $i += $count;
2711                                         continue;
2712                                 }
2713                                 $matchingCallback = $cbType['cb'][$matchingCount];
2714
2715                                 # let's set a title or last part (if '|' was found)
2716                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2717                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2718                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2719                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2720                                 } else {
2721                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2722                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2723                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2724                                 }
2725
2726                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2727                                 $pieceEnd = $i + $matchingCount;
2728
2729                                 if( is_callable( $matchingCallback ) ) {
2730                                         $cbArgs = array (
2731                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2732                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2733                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2734                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2735                                                                          );
2736                                         # finally we can call a user callback and replace piece of text
2737                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2738                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2739                                         $i = $pieceStart + strlen($replaceWith);
2740                                 } else {
2741                                         # null value for callback means that parentheses should be parsed, but not replaced
2742                                         $i += $matchingCount;
2743                                 }
2744
2745                                 # reset last opening parentheses, but keep it in case there are unused characters
2746                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2747                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2748                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2749                                                            'title' => '',
2750                                                            'parts' => null,
2751                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2752                                 $openingBraceStack[$lastOpeningBrace--] = null;
2753
2754                                 if ($matchingCount < $piece['count']) {
2755                                         $piece['count'] -= $matchingCount;
2756                                         $piece['startAt'] -= $matchingCount;
2757                                         $piece['partStart'] = $piece['startAt'];
2758                                         # do we still qualify for any callback with remaining count?
2759                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2760                                         while ( $piece['count'] ) {
2761                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2762                                                         $lastOpeningBrace++;
2763                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2764                                                         break;
2765                                                 }
2766                                                 --$piece['count'];
2767                                         }
2768                                 }
2769                         } elseif ( $found == 'pipe' ) {
2770                                 # lets set a title if it is a first separator, or next part otherwise
2771                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2772                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2773                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2774                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2775                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2776                                 } else {
2777                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2778                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2779                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2780                                 }
2781                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2782                         }
2783                 }
2784
2785                 wfProfileOut( __METHOD__ );
2786                 return $text;
2787         }
2788
2789         /**
2790          * Replace magic variables, templates, and template arguments
2791          * with the appropriate text. Templates are substituted recursively,
2792          * taking care to avoid infinite loops.
2793          *
2794          * Note that the substitution depends on value of $mOutputType:
2795          *  self::OT_WIKI: only {{subst:}} templates
2796          *  self::OT_MSG: only magic variables
2797          *  self::OT_HTML: all templates and magic variables
2798          *
2799          * @param string $tex The text to transform
2800          * @param array $args Key-value pairs representing template parameters to substitute
2801          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2802          * @private
2803          */
2804         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2805                 # Prevent too big inclusions
2806                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2807                         return $text;
2808                 }
2809
2810                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2811                 wfProfileIn( $fname );
2812
2813                 # This function is called recursively. To keep track of arguments we need a stack:
2814                 array_push( $this->mArgStack, $args );
2815
2816                 $braceCallbacks = array();
2817                 if ( !$argsOnly ) {
2818                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2819                 }
2820                 if ( $this->mOutputType != self::OT_MSG ) {
2821                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2822                 }
2823                 if ( $braceCallbacks ) {
2824                         $callbacks = array(
2825                                 '{' => array(
2826                                         'end' => '}',
2827                                         'cb' => $braceCallbacks,
2828                                         'min' => $argsOnly ? 3 : 2,
2829                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2830                                 ),
2831                                 '[' => array(
2832                                         'end' => ']',
2833                                         'cb' => array(2=>null),
2834                                         'min' => 2,
2835                                         'max' => 2,
2836                                 )
2837                         );
2838                         $text = $this->replace_callback ($text, $callbacks);
2839
2840                         array_pop( $this->mArgStack );
2841                 }
2842                 wfProfileOut( $fname );
2843                 return $text;
2844         }
2845
2846         /**
2847          * Replace magic variables
2848          * @private
2849          */
2850         function variableSubstitution( $matches ) {
2851                 global $wgContLang;
2852                 $fname = 'Parser::variableSubstitution';
2853                 $varname = $wgContLang->lc($matches[1]);
2854                 wfProfileIn( $fname );
2855                 $skip = false;
2856                 if ( $this->mOutputType == self::OT_WIKI ) {
2857                         # Do only magic variables prefixed by SUBST
2858                         $mwSubst =& MagicWord::get( 'subst' );
2859                         if (!$mwSubst->matchStartAndRemove( $varname ))
2860                                 $skip = true;
2861                         # Note that if we don't substitute the variable below,
2862                         # we don't remove the {{subst:}} magic word, in case
2863                         # it is a template rather than a magic variable.
2864                 }
2865                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2866                         $id = $this->mVariables[$varname];
2867                         # Now check if we did really match, case sensitive or not
2868                         $mw =& MagicWord::get( $id );
2869                         if ($mw->match($matches[1])) {
2870                                 $text = $this->getVariableValue( $id );
2871                                 if (MagicWord::getCacheTTL($id)>-1)
2872                                         $this->mOutput->mContainsOldMagic = true;
2873                         } else {
2874                                 $text = $matches[0];
2875                         }
2876                 } else {
2877                         $text = $matches[0];
2878                 }
2879                 wfProfileOut( $fname );
2880                 return $text;
2881         }
2882
2883
2884         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2885         static function createAssocArgs( $args ) {
2886                 $assocArgs = array();
2887                 $index = 1;
2888                 foreach( $args as $arg ) {
2889                         $eqpos = strpos( $arg, '=' );
2890                         if ( $eqpos === false ) {
2891                                 $assocArgs[$index++] = $arg;
2892                         } else {
2893                                 $name = trim( substr( $arg, 0, $eqpos ) );
2894                                 $value = trim( substr( $arg, $eqpos+1 ) );
2895                                 if ( $value === false ) {
2896                                         $value = '';
2897                                 }
2898                                 if ( $name !== false ) {
2899                                         $assocArgs[$name] = $value;
2900                                 }
2901                         }
2902                 }
2903
2904                 return $assocArgs;
2905         }
2906
2907         /**
2908          * Return the text of a template, after recursively
2909          * replacing any variables or templates within the template.
2910          *
2911          * @param array $piece The parts of the template
2912          *  $piece['text']: matched text
2913          *  $piece['title']: the title, i.e. the part before the |
2914          *  $piece['parts']: the parameter array
2915          * @return string the text of the template
2916          * @private
2917          */
2918         function braceSubstitution( $piece ) {
2919                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2920                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2921                 wfProfileIn( $fname );
2922                 wfProfileIn( __METHOD__.'-setup' );
2923
2924                 # Flags
2925                 $found = false;             # $text has been filled
2926                 $nowiki = false;            # wiki markup in $text should be escaped
2927                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2928                 $noargs = false;            # Don't replace triple-brace arguments in $text
2929                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2930                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2931                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2932                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2933
2934                 # Title object, where $text came from
2935                 $title = NULL;
2936
2937                 $linestart = '';
2938
2939
2940                 # $part1 is the bit before the first |, and must contain only title characters
2941                 # $args is a list of arguments, starting from index 0, not including $part1
2942
2943                 $titleText = $part1 = $piece['title'];
2944                 # If the third subpattern matched anything, it will start with |
2945
2946                 if (null == $piece['parts']) {
2947                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2948                         if ($replaceWith != $piece['text']) {
2949                                 $text = $replaceWith;
2950                                 $found = true;
2951                                 $noparse = true;
2952                                 $noargs = true;
2953                         }
2954                 }
2955
2956                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2957                 wfProfileOut( __METHOD__.'-setup' );
2958
2959                 # SUBST
2960                 wfProfileIn( __METHOD__.'-modifiers' );
2961                 if ( !$found ) {
2962                         $mwSubst =& MagicWord::get( 'subst' );
2963                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2964                                 # One of two possibilities is true:
2965                                 # 1) Found SUBST but not in the PST phase
2966                                 # 2) Didn't find SUBST and in the PST phase
2967                                 # In either case, return without further processing
2968                                 $text = $piece['text'];
2969                                 $found = true;
2970                                 $noparse = true;
2971                                 $noargs = true;
2972                         }
2973                 }
2974
2975                 # MSG, MSGNW and RAW
2976                 if ( !$found ) {
2977                         # Check for MSGNW:
2978                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2979                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2980                                 $nowiki = true;
2981                         } else {
2982                                 # Remove obsolete MSG:
2983                                 $mwMsg =& MagicWord::get( 'msg' );
2984                                 $mwMsg->matchStartAndRemove( $part1 );
2985                         }
2986
2987                         # Check for RAW:
2988                         $mwRaw =& MagicWord::get( 'raw' );
2989                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2990                                 $forceRawInterwiki = true;
2991                         }
2992                 }
2993                 wfProfileOut( __METHOD__.'-modifiers' );
2994
2995                 //save path level before recursing into functions & templates.
2996                 $lastPathLevel = $this->mTemplatePath;
2997
2998                 # Parser functions
2999                 if ( !$found ) {
3000                         wfProfileIn( __METHOD__ . '-pfunc' );
3001
3002                         $colonPos = strpos( $part1, ':' );
3003                         if ( $colonPos !== false ) {
3004                                 # Case sensitive functions
3005                                 $function = substr( $part1, 0, $colonPos );
3006                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3007                                         $function = $this->mFunctionSynonyms[1][$function];
3008                                 } else {
3009                                         # Case insensitive functions
3010                                         $function = strtolower( $function );
3011                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3012                                                 $function = $this->mFunctionSynonyms[0][$function];
3013                                         } else {
3014                                                 $function = false;
3015                                         }
3016                                 }
3017                                 if ( $function ) {
3018                                         $funcArgs = array_map( 'trim', $args );
3019                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
3020
3021                                         # Workaround for PHP bug 35229 and similar
3022                                         if ( !is_callable( $this->mFunctionHooks[$function] ) ) {
3023                                                 throw new MWException( "Function hook for $function is not callable\n" );
3024                                         }
3025                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
3026                                         $found = true;
3027
3028                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3029                                         //$noargs = true;
3030                                         //$noparse = true;
3031
3032                                         if ( is_array( $result ) ) {
3033                                                 if ( isset( $result[0] ) ) {
3034                                                         $text = $linestart . $result[0];
3035                                                         unset( $result[0] );
3036                                                 }
3037
3038                                                 // Extract flags into the local scope
3039                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3040                                                 extract( $result );
3041                                         } else {
3042                                                 $text = $linestart . $result;
3043                                         }
3044                                 }
3045                         }
3046                         wfProfileOut( __METHOD__ . '-pfunc' );
3047                 }
3048
3049                 # Template table test
3050
3051                 # Did we encounter this template already? If yes, it is in the cache
3052                 # and we need to check for loops.
3053                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3054                         $found = true;
3055
3056                         # Infinite loop test
3057                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3058                                 $noparse = true;
3059                                 $noargs = true;
3060                                 $found = true;
3061                                 $text = $linestart .
3062                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3063                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3064                         } else {
3065                                 # set $text to cached message.
3066                                 $text = $linestart . $this->mTemplates[$piece['title']];
3067                                 #treat title for cached page the same as others
3068                                 $ns = NS_TEMPLATE;
3069                                 $subpage = '';
3070                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3071                                 if ($subpage !== '') {
3072                                   $ns = $this->mTitle->getNamespace();
3073                                 }
3074                                 $title = Title::newFromText( $part1, $ns );
3075                                 //used by include size checking
3076                                 $titleText = $title->getPrefixedText();
3077                                 //used by edit section links
3078                                 $replaceHeadings = true;
3079
3080                         }
3081                 }
3082
3083                 # Load from database
3084                 if ( !$found ) {
3085                         wfProfileIn( __METHOD__ . '-loadtpl' );
3086                         $ns = NS_TEMPLATE;
3087                         # declaring $subpage directly in the function call
3088                         # does not work correctly with references and breaks
3089                         # {{/subpage}}-style inclusions
3090                         $subpage = '';
3091                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3092                         if ($subpage !== '') {
3093                                 $ns = $this->mTitle->getNamespace();
3094                         }
3095                         $title = Title::newFromText( $part1, $ns );
3096
3097
3098                         if ( !is_null( $title ) ) {
3099                                 $titleText = $title->getPrefixedText();
3100                                 # Check for language variants if the template is not found
3101                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3102                                         $wgContLang->findVariantLink( $part1, $title, true );
3103                                 }
3104
3105                                 if ( !$title->isExternal() ) {
3106                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3107                                                 $text = SpecialPage::capturePath( $title );
3108                                                 if ( is_string( $text ) ) {
3109                                                         $found = true;
3110                                                         $noparse = true;
3111                                                         $noargs = true;
3112                                                         $isHTML = true;
3113                                                         $this->disableCache();
3114                                                 }
3115                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3116                                                 $found = false; //access denied
3117                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3118                                         } else {
3119                                                 list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3120                                                 if ( $articleContent !== false ) {
3121                                                         $found = true;
3122                                                         $text = $articleContent;
3123                                                         $replaceHeadings = true;
3124                                                 }
3125                                         }
3126
3127                                         # If the title is valid but undisplayable, make a link to it
3128                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3129                                                 $text = "[[:$titleText]]";
3130                                                 $found = true;
3131                                         }
3132                                 } elseif ( $title->isTrans() ) {
3133                                         // Interwiki transclusion
3134                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3135                                                 $text = $this->interwikiTransclude( $title, 'render' );
3136                                                 $isHTML = true;
3137                                                 $noparse = true;
3138                                         } else {
3139                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3140                                                 $replaceHeadings = true;
3141                                         }
3142                                         $found = true;
3143                                 }
3144
3145                                 # Template cache array insertion
3146                                 # Use the original $piece['title'] not the mangled $part1, so that
3147                                 # modifiers such as RAW: produce separate cache entries
3148                                 if( $found ) {
3149                                         if( $isHTML ) {
3150                                                 // A special page; don't store it in the template cache.
3151                                         } else {
3152                                                 $this->mTemplates[$piece['title']] = $text;
3153                                         }
3154                                         $text = $linestart . $text;
3155                                 }
3156                         }
3157                         wfProfileOut( __METHOD__ . '-loadtpl' );
3158                 }
3159
3160                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3161                         # Error, oversize inclusion
3162                         $text = $linestart .
3163                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3164                         $noparse = true;
3165                         $noargs = true;
3166                 }
3167
3168                 # Recursive parsing, escaping and link table handling
3169                 # Only for HTML output
3170                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3171                         $text = wfEscapeWikiText( $text );
3172                 } elseif ( !$this->ot['msg'] && $found ) {
3173                         if ( $noargs ) {
3174                                 $assocArgs = array();
3175                         } else {
3176                                 # Clean up argument array
3177                                 $assocArgs = self::createAssocArgs($args);
3178                                 # Add a new element to the templace recursion path
3179                                 $this->mTemplatePath[$part1] = 1;
3180                         }
3181
3182                         if ( !$noparse ) {
3183                                 # If there are any <onlyinclude> tags, only include them
3184                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3185                                         $replacer = new OnlyIncludeReplacer;
3186                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3187                                                 array( &$replacer, 'replace' ), $text );
3188                                         $text = $replacer->output;
3189                                 }
3190                                 # Remove <noinclude> sections and <includeonly> tags
3191                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3192                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3193
3194                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3195                                         # Strip <nowiki>, <pre>, etc.
3196                                         $text = $this->strip( $text, $this->mStripState );
3197                                         if ( $this->ot['html'] ) {
3198                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3199                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3200                                                 $text = Sanitizer::removeHTMLcomments( $text );
3201                                         }
3202                                 }
3203                                 $text = $this->replaceVariables( $text, $assocArgs );
3204
3205                                 # If the template begins with a table or block-level
3206                                 # element, it should be treated as beginning a new line.
3207                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3208                                         $text = "\n" . $text;
3209                                 }
3210                         } elseif ( !$noargs ) {
3211                                 # $noparse and !$noargs
3212                                 # Just replace the arguments, not any double-brace items
3213                                 # This is used for rendered interwiki transclusion
3214                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3215                         }
3216                 }
3217                 # Prune lower levels off the recursion check path
3218                 $this->mTemplatePath = $lastPathLevel;
3219
3220                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3221                         # Error, oversize inclusion
3222                         $text = $linestart .
3223                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3224                         $noparse = true;
3225                         $noargs = true;
3226                 }
3227
3228                 if ( !$found ) {
3229                         wfProfileOut( $fname );
3230                         return $piece['text'];
3231                 } else {
3232                         wfProfileIn( __METHOD__ . '-placeholders' );
3233                         if ( $isHTML ) {
3234                                 # Replace raw HTML by a placeholder
3235                                 # Add a blank line preceding, to prevent it from mucking up
3236                                 # immediately preceding headings
3237                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3238                         } else {
3239                                 # replace ==section headers==
3240                                 # XXX this needs to go away once we have a better parser.
3241                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3242                                         if( !is_null( $title ) )
3243                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3244                                         else
3245                                                 $encodedname = base64_encode("");
3246                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3247                                                 PREG_SPLIT_DELIM_CAPTURE);
3248                                         $text = '';
3249                                         $nsec = $headingOffset;
3250
3251                                         for( $i = 0; $i < count($m); $i += 2 ) {
3252                                                 $text .= $m[$i];
3253                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3254                                                 $hl = $m[$i + 1];
3255                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3256                                                         $text .= $hl;
3257                                                         continue;
3258                                                 }
3259                                                 $m2 = array();
3260                                                 preg_match('/^(={1,6})(.*?)(={1,6}\s*?)$/m', $hl, $m2);
3261                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3262                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3263
3264                                                 $nsec++;
3265                                         }
3266                                 }
3267                         }
3268                         wfProfileOut( __METHOD__ . '-placeholders' );
3269                 }
3270
3271                 # Prune lower levels off the recursion check path
3272                 $this->mTemplatePath = $lastPathLevel;
3273
3274                 if ( !$found ) {
3275                         wfProfileOut( $fname );
3276                         return $piece['text'];
3277                 } else {
3278                         wfProfileOut( $fname );
3279                         return $text;
3280                 }
3281         }
3282
3283         /**
3284          * Fetch the unparsed text of a template and register a reference to it.
3285          */
3286         function fetchTemplateAndTitle( $title ) {
3287                 $templateCb = $this->mOptions->getTemplateCallback();
3288                 $stuff = call_user_func( $templateCb, $title, $this );
3289                 $text = $stuff['text'];
3290                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3291                 if ( isset( $stuff['deps'] ) ) {
3292                         foreach ( $stuff['deps'] as $dep ) {
3293                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3294                         }
3295                 }
3296                 return array($text,$finalTitle);
3297         }
3298
3299         function fetchTemplate( $title ) {
3300                 $rv = $this->fetchTemplateAndtitle($title);
3301                 return $rv[0];
3302         }
3303
3304         /**
3305          * Static function to get a template
3306          * Can be overridden via ParserOptions::setTemplateCallback().
3307          *
3308          * Returns an associative array:
3309          *    text          The unparsed template text
3310          *    finalTitle    (Optional) The title after following redirects
3311          *    deps          (Optional) An array of associative array dependencies:
3312          *                       title:    The dependency title, to be registered in templatelinks
3313          *                       page_id:  The page_id of the title
3314          *                       rev_id:   The revision ID loaded
3315          */
3316         static function statelessFetchTemplate( $title, $parser=false ) {
3317                 $text = $skip = false;
3318                 $finalTitle = $title;
3319                 $deps = array();
3320
3321                 // Loop to fetch the article, with up to 1 redirect
3322                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3323                         # Give extensions a chance to select the revision instead
3324                         $id = false; // Assume current
3325                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3326
3327                         if( $skip ) {
3328                                 $text = false;
3329                                 $deps[] = array(
3330                                         'title' => $title,
3331                                         'page_id' => $title->getArticleID(),
3332                                         'rev_id' => null );
3333                                 break;
3334                         }
3335                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3336                         $rev_id = $rev ? $rev->getId() : 0;
3337
3338                         $deps[] = array(
3339                                 'title' => $title,
3340                                 'page_id' => $title->getArticleID(),
3341                                 'rev_id' => $rev_id );
3342
3343                         if( $rev ) {
3344                                 $text = $rev->getText();
3345                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3346                                 global $wgLang;
3347                                 $message = $wgLang->lcfirst( $title->getText() );
3348                                 $text = wfMsgForContentNoTrans( $message );
3349                                 if( wfEmptyMsg( $message, $text ) ) {
3350                                         $text = false;
3351                                         break;
3352                                 }
3353                         } else {
3354                                 break;
3355                         }
3356                         if ( $text === false ) {
3357                                 break;
3358                         }
3359                         // Redirect?
3360                         $finalTitle = $title;
3361                         $title = Title::newFromRedirect( $text );
3362                 }
3363                 return array(
3364                         'text' => $text,
3365                         'finalTitle' => $finalTitle,
3366                         'deps' => $deps );
3367         }
3368
3369         /**
3370          * Transclude an interwiki link.
3371          */
3372         function interwikiTransclude( $title, $action ) {
3373                 global $wgEnableScaryTranscluding;
3374
3375                 if (!$wgEnableScaryTranscluding)
3376                         return wfMsg('scarytranscludedisabled');
3377
3378                 $url = $title->getFullUrl( "action=$action" );
3379
3380                 if (strlen($url) > 255)
3381                         return wfMsg('scarytranscludetoolong');
3382                 return $this->fetchScaryTemplateMaybeFromCache($url);
3383         }
3384
3385         function fetchScaryTemplateMaybeFromCache($url) {
3386                 global $wgTranscludeCacheExpiry;
3387                 $dbr = wfGetDB(DB_SLAVE);
3388                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3389                                 array('tc_url' => $url));
3390                 if ($obj) {
3391                         $time = $obj->tc_time;
3392                         $text = $obj->tc_contents;
3393                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3394                                 return $text;
3395                         }
3396                 }
3397
3398                 $text = Http::get($url);
3399                 if (!$text)
3400                         return wfMsg('scarytranscludefailed', $url);
3401
3402                 $dbw = wfGetDB(DB_MASTER);
3403                 $dbw->replace('transcache', array('tc_url'), array(
3404                         'tc_url' => $url,
3405                         'tc_time' => time(),
3406                         'tc_contents' => $text));
3407                 return $text;
3408         }
3409
3410
3411         /**
3412          * Triple brace replacement -- used for template arguments
3413          * @private
3414          */
3415         function argSubstitution( $matches ) {
3416                 $arg = trim( $matches['title'] );
3417                 $text = $matches['text'];
3418                 $inputArgs = end( $this->mArgStack );
3419
3420                 if ( array_key_exists( $arg, $inputArgs ) ) {
3421                         $text = $inputArgs[$arg];
3422                 } else if (($this->mOutputType == self::OT_HTML || $this->mOutputType == self::OT_PREPROCESS ) &&
3423                 null != $matches['parts'] && count($matches['parts']) > 0) {
3424                         $text = $matches['parts'][0];
3425                 }
3426                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3427                         $text = $matches['text'] .
3428                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3429                 }
3430
3431                 return $text;
3432         }
3433
3434         /**
3435          * Increment an include size counter
3436          *
3437          * @param string $type The type of expansion
3438          * @param integer $size The size of the text
3439          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3440          */
3441         function incrementIncludeSize( $type, $size ) {
3442                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3443                         return false;
3444                 } else {
3445                         $this->mIncludeSizes[$type] += $size;
3446                         return true;
3447                 }
3448         }
3449
3450         /**
3451          * Detect __NOGALLERY__ magic word and set a placeholder
3452          */
3453         function stripNoGallery( &$text ) {
3454                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3455                 # do not add TOC
3456                 $mw = MagicWord::get( 'nogallery' );
3457                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3458         }
3459
3460         /**
3461          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3462          * placeholder that will then be replaced by the real TOC in
3463          * ->formatHeadings, this works because at this points real
3464          * comments will have already been discarded by the sanitizer.
3465          *
3466          * Any additional __TOC__ magic words left over will be discarded
3467          * as there can only be one TOC on the page.
3468          */
3469         function stripToc( $text ) {
3470                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3471                 # do not add TOC
3472                 $mw = MagicWord::get( 'notoc' );
3473                 if( $mw->matchAndRemove( $text ) ) {
3474                         $this->mShowToc = false;
3475                 }
3476
3477                 $mw = MagicWord::get( 'toc' );
3478                 if( $mw->match( $text ) ) {
3479                         $this->mShowToc = true;
3480                         $this->mForceTocPosition = true;
3481
3482                         // Set a placeholder. At the end we'll fill it in with the TOC.
3483                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3484
3485                         // Only keep the first one.
3486                         $text = $mw->replace( '', $text );
3487                 }
3488                 return $text;
3489         }
3490
3491         /**
3492          * This function accomplishes several tasks:
3493          * 1) Auto-number headings if that option is enabled
3494          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3495          * 3) Add a Table of contents on the top for users who have enabled the option
3496          * 4) Auto-anchor headings
3497          *
3498          * It loops through all headlines, collects the necessary data, then splits up the
3499          * string and re-inserts the newly formatted headlines.
3500          *
3501          * @param string $text
3502          * @param boolean $isMain
3503          * @private
3504          */
3505         function formatHeadings( $text, $isMain=true ) {
3506                 global $wgMaxTocLevel, $wgContLang;
3507
3508                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3509                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3510                         $showEditLink = 0;
3511                 } else {
3512                         $showEditLink = $this->mOptions->getEditSection();
3513                 }
3514
3515                 # Inhibit editsection links if requested in the page
3516                 $esw =& MagicWord::get( 'noeditsection' );
3517                 if( $esw->matchAndRemove( $text ) ) {
3518                         $showEditLink = 0;
3519                 }
3520
3521                 # Get all headlines for numbering them and adding funky stuff like [edit]
3522                 # links - this is for later, but we need the number of headlines right now
3523                 $matches = array();
3524                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3525
3526                 # if there are fewer than 4 headlines in the article, do not show TOC
3527                 # unless it's been explicitly enabled.
3528                 $enoughToc = $this->mShowToc &&
3529                         (($numMatches >= 4) || $this->mForceTocPosition);
3530
3531                 # Allow user to stipulate that a page should have a "new section"
3532                 # link added via __NEWSECTIONLINK__
3533                 $mw =& MagicWord::get( 'newsectionlink' );
3534                 if( $mw->matchAndRemove( $text ) )
3535                         $this->mOutput->setNewSection( true );
3536
3537                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3538                 # override above conditions and always show TOC above first header
3539                 $mw =& MagicWord::get( 'forcetoc' );
3540                 if ($mw->matchAndRemove( $text ) ) {
3541                         $this->mShowToc = true;
3542                         $enoughToc = true;
3543                 }
3544
3545                 # We need this to perform operations on the HTML
3546                 $sk = $this->mOptions->getSkin();
3547
3548                 # headline counter
3549                 $headlineCount = 0;
3550                 $sectionCount = 0; # headlineCount excluding template sections
3551                 $numVisible = 0;
3552
3553                 # Ugh .. the TOC should have neat indentation levels which can be
3554                 # passed to the skin functions. These are determined here
3555                 $toc = '';
3556                 $full = '';
3557                 $head = array();
3558                 $sublevelCount = array();
3559                 $levelCount = array();
3560                 $toclevel = 0;
3561                 $level = 0;
3562                 $prevlevel = 0;
3563                 $toclevel = 0;
3564                 $prevtoclevel = 0;
3565                 $tocraw = array();
3566
3567                 foreach( $matches[3] as $headline ) {
3568                         $istemplate = 0;
3569                         $templatetitle = '';
3570                         $templatesection = 0;
3571                         $numbering = '';
3572                         $mat = array();
3573                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3574                                 $istemplate = 1;
3575                                 $templatetitle = base64_decode($mat[1]);
3576                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3577                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3578                         }
3579
3580                         if( $toclevel ) {
3581                                 $prevlevel = $level;
3582                                 $prevtoclevel = $toclevel;
3583                         }
3584                         $level = $matches[1][$headlineCount];
3585
3586                         if( $doNumberHeadings || $enoughToc ) {
3587
3588                                 if ( $level > $prevlevel ) {
3589                                         # Increase TOC level
3590                                         $toclevel++;
3591                                         $sublevelCount[$toclevel] = 0;
3592                                         if( $toclevel<$wgMaxTocLevel ) {
3593                                                 $prevtoclevel = $toclevel;
3594                                                 $toc .= $sk->tocIndent();
3595                                                 $numVisible++;
3596                                         }
3597                                 }
3598                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3599                                         # Decrease TOC level, find level to jump to
3600
3601                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3602                                                 # Can only go down to level 1
3603                                                 $toclevel = 1;
3604                                         } else {
3605                                                 for ($i = $toclevel; $i > 0; $i--) {
3606                                                         if ( $levelCount[$i] == $level ) {
3607                                                                 # Found last matching level
3608                                                                 $toclevel = $i;
3609                                                                 break;
3610                                                         }
3611                                                         elseif ( $levelCount[$i] < $level ) {
3612                                                                 # Found first matching level below current level
3613                                                                 $toclevel = $i + 1;
3614                                                                 break;
3615                                                         }
3616                                                 }
3617                                         }
3618                                         if( $toclevel<$wgMaxTocLevel ) {
3619                                                 if($prevtoclevel < $wgMaxTocLevel) {
3620                                                         # Unindent only if the previous toc level was shown :p
3621                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3622                                                 } else {
3623                                                         $toc .= $sk->tocLineEnd();
3624                                                 }
3625                                         }
3626                                 }
3627                                 else {
3628                                         # No change in level, end TOC line
3629                                         if( $toclevel<$wgMaxTocLevel ) {
3630                                                 $toc .= $sk->tocLineEnd();
3631                                         }
3632                                 }
3633
3634                                 $levelCount[$toclevel] = $level;
3635
3636                                 # count number of headlines for each level
3637                                 @$sublevelCount[$toclevel]++;
3638                                 $dot = 0;
3639                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3640                                         if( !empty( $sublevelCount[$i] ) ) {
3641                                                 if( $dot ) {
3642                                                         $numbering .= '.';
3643                                                 }
3644                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3645                                                 $dot = 1;
3646                                         }
3647                                 }
3648                         }
3649
3650                         # The canonized header is a version of the header text safe to use for links
3651                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3652                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3653
3654                         # Remove link placeholders by the link text.
3655                         #     <!--LINK number-->
3656                         # turns into
3657                         #     link text with suffix
3658                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3659                                                             "\$this->mLinkHolders['texts'][\$1]",
3660                                                             $canonized_headline );
3661                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3662                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3663                                                             $canonized_headline );
3664
3665                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3666                         $tocline = preg_replace(
3667                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3668                                 array( '',                          '<$1>'),
3669                                 $canonized_headline
3670                         );
3671                         $tocline = trim( $tocline );
3672
3673                         # For the anchor, strip out HTML-y stuff period
3674                         $canonized_headline = preg_replace( '/<.*?'.'>/', '', $canonized_headline );
3675                         $canonized_headline = trim( $canonized_headline );
3676
3677                         # Save headline for section edit hint before it's escaped
3678                         $headline_hint = $canonized_headline;
3679                         $canonized_headline = Sanitizer::escapeId( $canonized_headline );
3680                         $refers[$headlineCount] = $canonized_headline;
3681
3682                         # count how many in assoc. array so we can track dupes in anchors
3683                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3684                         $refcount[$headlineCount]=$refers[$canonized_headline];
3685
3686                         # Don't number the heading if it is the only one (looks silly)
3687                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3688                                 # the two are different if the line contains a link
3689                                 $headline=$numbering . ' ' . $headline;
3690                         }
3691
3692                         # Create the anchor for linking from the TOC to the section
3693                         $anchor = $canonized_headline;
3694                         if($refcount[$headlineCount] > 1 ) {
3695                                 $anchor .= '_' . $refcount[$headlineCount];
3696                         }
3697                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3698                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3699                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3700                         }
3701                         # give headline the correct <h#> tag
3702                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3703                                 if( $istemplate )
3704                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3705                                 else
3706                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3707                         } else {
3708                                 $editlink = '';
3709                         }
3710                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3711
3712                         $headlineCount++;
3713                         if( !$istemplate )
3714                                 $sectionCount++;
3715                 }
3716
3717                 $this->mOutput->setSections( $tocraw );
3718
3719                 # Never ever show TOC if no headers
3720                 if( $numVisible < 1 ) {
3721                         $enoughToc = false;
3722                 }
3723
3724                 if( $enoughToc ) {
3725                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3726                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3727                         }
3728                         $toc = $sk->tocList( $toc );
3729                 }
3730
3731                 # split up and insert constructed headlines
3732
3733                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3734                 $i = 0;
3735
3736                 foreach( $blocks as $block ) {
3737                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3738                                 # This is the [edit] link that appears for the top block of text when
3739                                 # section editing is enabled
3740
3741                                 # Disabled because it broke block formatting
3742                                 # For example, a bullet point in the top line
3743                                 # $full .= $sk->editSectionLink(0);
3744                         }
3745                         $full .= $block;
3746                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3747                                 # Top anchor now in skin
3748                                 $full = $full.$toc;
3749                         }
3750
3751                         if( !empty( $head[$i] ) ) {
3752                                 $full .= $head[$i];
3753                         }
3754                         $i++;
3755                 }
3756                 if( $this->mForceTocPosition ) {
3757                         return str_replace( '<!--MWTOC-->', $toc, $full );
3758                 } else {
3759                         return $full;
3760                 }
3761         }
3762
3763         /**
3764          * Transform wiki markup when saving a page by doing \r\n -> \n
3765          * conversion, substitting signatures, {{subst:}} templates, etc.
3766          *
3767          * @param string $text the text to transform
3768          * @param Title &$title the Title object for the current article
3769          * @param User &$user the User object describing the current user
3770          * @param ParserOptions $options parsing options
3771          * @param bool $clearState whether to clear the parser state first
3772          * @return string the altered wiki markup
3773          * @public
3774          */
3775         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3776                 $this->mOptions = $options;
3777                 $this->mTitle =& $title;
3778                 $this->setOutputType( self::OT_WIKI );
3779
3780                 if ( $clearState ) {
3781                         $this->clearState();
3782                 }
3783
3784                 $stripState = new StripState;
3785                 $pairs = array(
3786                         "\r\n" => "\n",
3787                 );
3788                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3789                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3790                 $text = $this->pstPass2( $text, $stripState, $user );
3791                 $text = $stripState->unstripBoth( $text );
3792                 return $text;
3793         }
3794
3795         /**
3796          * Pre-save transform helper function
3797          * @private
3798          */
3799         function pstPass2( $text, &$stripState, $user ) {
3800                 global $wgContLang, $wgLocaltimezone;
3801
3802                 /* Note: This is the timestamp saved as hardcoded wikitext to
3803                  * the database, we use $wgContLang here in order to give
3804                  * everyone the same signature and use the default one rather
3805                  * than the one selected in each user's preferences.
3806                  */
3807                 if ( isset( $wgLocaltimezone ) ) {
3808                         $oldtz = getenv( 'TZ' );
3809                         putenv( 'TZ='.$wgLocaltimezone );
3810                 }
3811                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3812                   ' (' . date( 'T' ) . ')';
3813                 if ( isset( $wgLocaltimezone ) ) {
3814                         putenv( 'TZ='.$oldtz );
3815                 }
3816
3817                 # Variable replacement
3818                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3819                 $text = $this->replaceVariables( $text );
3820
3821                 # Strip out <nowiki> etc. added via replaceVariables
3822                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3823
3824                 # Signatures
3825                 $sigText = $this->getUserSig( $user );
3826                 $text = strtr( $text, array(
3827                         '~~~~~' => $d,
3828                         '~~~~' => "$sigText $d",
3829                         '~~~' => $sigText
3830                 ) );
3831
3832                 # Context links: [[|name]] and [[name (context)|]]
3833                 #
3834                 global $wgLegalTitleChars;
3835                 $tc = "[$wgLegalTitleChars]";
3836                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3837
3838                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3839                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3840                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3841
3842                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3843                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3844                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3845
3846                 $t = $this->mTitle->getText();
3847                 $m = array();
3848                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3849                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3850                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3851                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3852                 } else {
3853                         # if there's no context, don't bother duplicating the title
3854                         $text = preg_replace( $p2, '[[\\1]]', $text );
3855                 }
3856
3857                 # Trim trailing whitespace
3858                 $text = rtrim( $text );
3859
3860                 return $text;
3861         }
3862
3863         /**
3864          * Fetch the user's signature text, if any, and normalize to
3865          * validated, ready-to-insert wikitext.
3866          *
3867          * @param User $user
3868          * @return string
3869          * @private
3870          */
3871         function getUserSig( &$user ) {
3872                 global $wgMaxSigChars;
3873
3874                 $username = $user->getName();
3875                 $nickname = $user->getOption( 'nickname' );
3876                 $nickname = $nickname === '' ? $username : $nickname;
3877
3878                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3879                         $nickname = $username;
3880                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3881                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3882                         # Sig. might contain markup; validate this
3883                         if( $this->validateSig( $nickname ) !== false ) {
3884                                 # Validated; clean up (if needed) and return it
3885                                 return $this->cleanSig( $nickname, true );
3886                         } else {
3887                                 # Failed to validate; fall back to the default
3888                                 $nickname = $username;
3889                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3890                         }
3891                 }
3892
3893                 // Make sure nickname doesnt get a sig in a sig
3894                 $nickname = $this->cleanSigInSig( $nickname );
3895
3896                 # If we're still here, make it a link to the user page
3897                 $userText = wfEscapeWikiText( $username );
3898                 $nickText = wfEscapeWikiText( $nickname );
3899                 if ( $user->isAnon() )  {
3900                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3901                 } else {
3902                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3903                 }
3904         }
3905
3906         /**
3907          * Check that the user's signature contains no bad XML
3908          *
3909          * @param string $text
3910          * @return mixed An expanded string, or false if invalid.
3911          */
3912         function validateSig( $text ) {
3913                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3914         }
3915
3916         /**
3917          * Clean up signature text
3918          *
3919          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3920          * 2) Substitute all transclusions
3921          *
3922          * @param string $text
3923          * @param $parsing Whether we're cleaning (preferences save) or parsing
3924          * @return string Signature text
3925          */
3926         function cleanSig( $text, $parsing = false ) {
3927                 global $wgTitle;
3928                 $this->startExternalParse( $this->mTitle, new ParserOptions(), $parsing ? self::OT_WIKI : self::OT_MSG );
3929
3930                 $substWord = MagicWord::get( 'subst' );
3931                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3932                 $substText = '{{' . $substWord->getSynonym( 0 );
3933
3934                 $text = preg_replace( $substRegex, $substText, $text );
3935                 $text = $this->cleanSigInSig( $text );
3936                 $text = $this->replaceVariables( $text );
3937
3938                 $this->clearState();
3939                 return $text;
3940         }
3941
3942         /**
3943          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3944          * @param string $text
3945          * @return string Signature text with /~{3,5}/ removed
3946          */
3947         function cleanSigInSig( $text ) {
3948                 $text = preg_replace( '/~{3,5}/', '', $text );
3949                 return $text;
3950         }
3951
3952         /**
3953          * Set up some variables which are usually set up in parse()
3954          * so that an external function can call some class members with confidence
3955          * @public
3956          */
3957         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3958                 $this->mTitle =& $title;
3959                 $this->mOptions = $options;
3960                 $this->setOutputType( $outputType );
3961                 if ( $clearState ) {
3962                         $this->clearState();
3963                 }
3964         }
3965
3966         /**
3967          * Transform a MediaWiki message by replacing magic variables.
3968          *
3969          * @param string $text the text to transform
3970          * @param ParserOptions $options  options
3971          * @return string the text with variables substituted
3972          * @public
3973          */
3974         function transformMsg( $text, $options ) {
3975                 global $wgTitle;
3976                 static $executing = false;
3977
3978                 $fname = "Parser::transformMsg";
3979
3980                 # Guard against infinite recursion
3981                 if ( $executing ) {
3982                         return $text;
3983                 }
3984                 $executing = true;
3985
3986                 wfProfileIn($fname);
3987
3988                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3989                         $this->mTitle = $wgTitle;
3990                 } else {
3991                         $this->mTitle = Title::newFromText('msg');
3992                 }
3993                 $this->mOptions = $options;
3994                 $this->setOutputType( self::OT_MSG );
3995                 $this->clearState();
3996                 $text = $this->replaceVariables( $text );
3997
3998                 $executing = false;
3999                 wfProfileOut($fname);
4000                 return $text;
4001         }
4002
4003         /**
4004          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4005          * The callback should have the following form:
4006          *    function myParserHook( $text, $params, &$parser ) { ... }
4007          *
4008          * Transform and return $text. Use $parser for any required context, e.g. use
4009          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4010          *
4011          * @public
4012          *
4013          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4014          * @param mixed $callback The callback function (and object) to use for the tag
4015          *
4016          * @return The old value of the mTagHooks array associated with the hook
4017          */
4018         function setHook( $tag, $callback ) {
4019                 $tag = strtolower( $tag );
4020                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4021                 $this->mTagHooks[$tag] = $callback;
4022
4023                 return $oldVal;
4024         }
4025
4026         function setTransparentTagHook( $tag, $callback ) {
4027                 $tag = strtolower( $tag );
4028                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4029                 $this->mTransparentTagHooks[$tag] = $callback;
4030
4031                 return $oldVal;
4032         }
4033
4034         /**
4035          * Create a function, e.g. {{sum:1|2|3}}
4036          * The callback function should have the form:
4037          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4038          *
4039          * The callback may either return the text result of the function, or an array with the text
4040          * in element 0, and a number of flags in the other elements. The names of the flags are
4041          * specified in the keys. Valid flags are:
4042          *   found                     The text returned is valid, stop processing the template. This
4043          *                             is on by default.
4044          *   nowiki                    Wiki markup in the return value should be escaped
4045          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4046          *   noargs                    Don't replace triple-brace arguments in the return value
4047          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4048          *
4049          * @public
4050          *
4051          * @param string $id The magic word ID
4052          * @param mixed $callback The callback function (and object) to use
4053          * @param integer $flags a combination of the following flags:
4054          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4055          *
4056          * @return The old callback function for this name, if any
4057          */
4058         function setFunctionHook( $id, $callback, $flags = 0 ) {
4059                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
4060                 $this->mFunctionHooks[$id] = $callback;
4061
4062                 # Add to function cache
4063                 $mw = MagicWord::get( $id );
4064                 if( !$mw )
4065                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4066
4067                 $synonyms = $mw->getSynonyms();
4068                 $sensitive = intval( $mw->isCaseSensitive() );
4069
4070                 foreach ( $synonyms as $syn ) {
4071                         # Case
4072                         if ( !$sensitive ) {
4073                                 $syn = strtolower( $syn );
4074                         }
4075                         # Add leading hash
4076                         if ( !( $flags & SFH_NO_HASH ) ) {
4077                                 $syn = '#' . $syn;
4078                         }
4079                         # Remove trailing colon
4080                         if ( substr( $syn, -1, 1 ) == ':' ) {
4081                                 $syn = substr( $syn, 0, -1 );
4082                         }
4083                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4084                 }
4085                 return $oldVal;
4086         }
4087
4088         /**
4089          * Get all registered function hook identifiers
4090          *
4091          * @return array
4092          */
4093         function getFunctionHooks() {
4094                 return array_keys( $this->mFunctionHooks );
4095         }
4096
4097         /**
4098          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4099          * Placeholders created in Skin::makeLinkObj()
4100          * Returns an array of links found, indexed by PDBK:
4101          *  0 - broken
4102          *  1 - normal link
4103          *  2 - stub
4104          * $options is a bit field, RLH_FOR_UPDATE to select for update
4105          */
4106         function replaceLinkHolders( &$text, $options = 0 ) {
4107                 global $wgUser;
4108                 global $wgContLang;
4109
4110                 $fname = 'Parser::replaceLinkHolders';
4111                 wfProfileIn( $fname );
4112
4113                 $pdbks = array();
4114                 $colours = array();
4115                 $sk = $this->mOptions->getSkin();
4116                 $linkCache = LinkCache::singleton();
4117
4118                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4119                         wfProfileIn( $fname.'-check' );
4120                         $dbr = wfGetDB( DB_SLAVE );
4121                         $page = $dbr->tableName( 'page' );
4122                         $threshold = $wgUser->getOption('stubthreshold');
4123
4124                         # Sort by namespace
4125                         asort( $this->mLinkHolders['namespaces'] );
4126
4127                         # Generate query
4128                         $query = false;
4129                         $current = null;
4130                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4131                                 # Make title object
4132                                 $title = $this->mLinkHolders['titles'][$key];
4133
4134                                 # Skip invalid entries.
4135                                 # Result will be ugly, but prevents crash.
4136                                 if ( is_null( $title ) ) {
4137                                         continue;
4138                                 }
4139                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4140
4141                                 # Check if it's a static known link, e.g. interwiki
4142                                 if ( $title->isAlwaysKnown() ) {
4143                                         $colours[$pdbk] = 1;
4144                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4145                                         $colours[$pdbk] = 1;
4146                                         $this->mOutput->addLink( $title, $id );
4147                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4148                                         $colours[$pdbk] = 0;
4149                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4150                                         $colours[$pdbk] = 0;
4151                                 } else {
4152                                         # Not in the link cache, add it to the query
4153                                         if ( !isset( $current ) ) {
4154                                                 $current = $ns;
4155                                                 $query =  "SELECT page_id, page_namespace, page_title, page_len, page_is_redirect";
4156                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4157                                         } elseif ( $current != $ns ) {
4158                                                 $current = $ns;
4159                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4160                                         } else {
4161                                                 $query .= ', ';
4162                                         }
4163
4164                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4165                                 }
4166                         }
4167                         if ( $query ) {
4168                                 $query .= '))';
4169                                 if ( $options & RLH_FOR_UPDATE ) {
4170                                         $query .= ' FOR UPDATE';
4171                                 }
4172
4173                                 $res = $dbr->query( $query, $fname );
4174
4175                                 # Fetch data and form into an associative array
4176                                 # non-existent = broken
4177                                 # 1 = known
4178                                 # 2 = stub
4179                                 while ( $s = $dbr->fetchObject($res) ) {
4180                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4181                                         $pdbk = $title->getPrefixedDBkey();
4182                                         $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
4183                                         $this->mOutput->addLink( $title, $s->page_id );
4184
4185                                         $colours[$pdbk] = ( $threshold == 0 || (
4186                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4187                                                                 $s->page_is_redirect ||
4188                                                                 !MWNamespace::isContent( $s->page_namespace ) )
4189                                                             ? 1 : 2 );
4190                                 }
4191                         }
4192                         wfProfileOut( $fname.'-check' );
4193
4194                         # Do a second query for different language variants of links and categories
4195                         if( $wgContLang->hasVariants() ) {
4196                                 $linkBatch = new LinkBatch();
4197                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4198                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4199                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4200
4201                                 $categories = $this->mOutput->getCategoryLinks();
4202
4203                                 // Add variants of links to link batch
4204                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4205                                         $title = $this->mLinkHolders['titles'][$key];
4206                                         if ( is_null( $title ) )
4207                                                 continue;
4208
4209                                         $pdbk = $title->getPrefixedDBkey();
4210                                         $titleText = $title->getText();
4211
4212                                         // generate all variants of the link title text
4213                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4214
4215                                         // if link was not found (in first query), add all variants to query
4216                                         if ( !isset($colours[$pdbk]) ){
4217                                                 foreach($allTextVariants as $textVariant){
4218                                                         if($textVariant != $titleText){
4219                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4220                                                                 if(is_null($variantTitle)) continue;
4221                                                                 $linkBatch->addObj( $variantTitle );
4222                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4223                                                         }
4224                                                 }
4225                                         }
4226                                 }
4227
4228                                 // process categories, check if a category exists in some variant
4229                                 foreach( $categories as $category ){
4230                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4231                                         foreach($variants as $variant){
4232                                                 if($variant != $category){
4233                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4234                                                         if(is_null($variantTitle)) continue;
4235                                                         $linkBatch->addObj( $variantTitle );
4236                                                         $categoryMap[$variant] = $category;
4237                                                 }
4238                                         }
4239                                 }
4240
4241
4242                                 if ( !$linkBatch->isEmpty() ){
4243                                         // construct query
4244                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4245
4246                                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_len, page_is_redirect";
4247
4248                                         $variantQuery .= " FROM $page WHERE $titleClause";
4249                                         if ( $options & RLH_FOR_UPDATE ) {
4250                                                 $variantQuery .= ' FOR UPDATE';
4251                                         }
4252
4253                                         $varRes = $dbr->query( $variantQuery, $fname );
4254
4255                                         // for each found variants, figure out link holders and replace
4256                                         while ( $s = $dbr->fetchObject($varRes) ) {
4257
4258                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4259                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4260                                                 $vardbk = $variantTitle->getDBkey();
4261
4262                                                 $holderKeys = array();
4263                                                 if(isset($variantMap[$varPdbk])){
4264                                                         $holderKeys = $variantMap[$varPdbk];
4265                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
4266                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4267                                                 }
4268
4269                                                 // loop over link holders
4270                                                 foreach($holderKeys as $key){
4271                                                         $title = $this->mLinkHolders['titles'][$key];
4272                                                         if ( is_null( $title ) ) continue;
4273
4274                                                         $pdbk = $title->getPrefixedDBkey();
4275
4276                                                         if(!isset($colours[$pdbk])){
4277                                                                 // found link in some of the variants, replace the link holder data
4278                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4279                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4280
4281                                                                 // set pdbk and colour
4282                                                                 $pdbks[$key] = $varPdbk;
4283                                                                 if ( $threshold >  0 ) {
4284                                                                         $size = $s->page_len;
4285                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4286                                                                                 $colours[$varPdbk] = 1;
4287                                                                         } else {
4288                                                                                 $colours[$varPdbk] = 2;
4289                                                                         }
4290                                                                 }
4291                                                                 else {
4292                                                                         $colours[$varPdbk] = 1;
4293                                                                 }
4294                                                         }
4295                                                 }
4296
4297                                                 // check if the object is a variant of a category
4298                                                 if(isset($categoryMap[$vardbk])){
4299                                                         $oldkey = $categoryMap[$vardbk];
4300                                                         if($oldkey != $vardbk)
4301                                                                 $varCategories[$oldkey]=$vardbk;
4302                                                 }
4303                                         }
4304
4305                                         // rebuild the categories in original order (if there are replacements)
4306                                         if(count($varCategories)>0){
4307                                                 $newCats = array();
4308                                                 $originalCats = $this->mOutput->getCategories();
4309                                                 foreach($originalCats as $cat => $sortkey){
4310                                                         // make the replacement
4311                                                         if( array_key_exists($cat,$varCategories) )
4312                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4313                                                         else $newCats[$cat] = $sortkey;
4314                                                 }
4315                                                 $this->mOutput->setCategoryLinks($newCats);
4316                                         }
4317                                 }
4318                         }
4319
4320                         # Construct search and replace arrays
4321                         wfProfileIn( $fname.'-construct' );
4322                         $replacePairs = array();
4323                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4324                                 $pdbk = $pdbks[$key];
4325                                 $searchkey = "<!--LINK $key-->";
4326                                 $title = $this->mLinkHolders['titles'][$key];
4327                                 if ( empty( $colours[$pdbk] ) ) {
4328                                         $linkCache->addBadLinkObj( $title );
4329                                         $colours[$pdbk] = 0;
4330                                         $this->mOutput->addLink( $title, 0 );
4331                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4332                                                                         $this->mLinkHolders['texts'][$key],
4333                                                                         $this->mLinkHolders['queries'][$key] );
4334                                 } elseif ( $colours[$pdbk] == 1 ) {
4335                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4336                                                                         $this->mLinkHolders['texts'][$key],
4337                                                                         $this->mLinkHolders['queries'][$key] );
4338                                 } elseif ( $colours[$pdbk] == 2 ) {
4339                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4340                                                                         $this->mLinkHolders['texts'][$key],
4341                                                                         $this->mLinkHolders['queries'][$key] );
4342                                 }
4343                         }
4344                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4345                         wfProfileOut( $fname.'-construct' );
4346
4347                         # Do the thing
4348                         wfProfileIn( $fname.'-replace' );
4349                         $text = preg_replace_callback(
4350                                 '/(<!--LINK .*?-->)/',
4351                                 $replacer->cb(),
4352                                 $text);
4353
4354                         wfProfileOut( $fname.'-replace' );
4355                 }
4356
4357                 # Now process interwiki link holders
4358                 # This is quite a bit simpler than internal links
4359                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4360                         wfProfileIn( $fname.'-interwiki' );
4361                         # Make interwiki link HTML
4362                         $replacePairs = array();
4363                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4364                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4365                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4366                         }
4367                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4368
4369                         $text = preg_replace_callback(
4370                                 '/<!--IWLINK (.*?)-->/',
4371                                 $replacer->cb(),
4372                                 $text );
4373                         wfProfileOut( $fname.'-interwiki' );
4374                 }
4375
4376                 wfProfileOut( $fname );
4377                 return $colours;
4378         }
4379
4380         /**
4381          * Replace <!--LINK--> link placeholders with plain text of links
4382          * (not HTML-formatted).
4383          * @param string $text
4384          * @return string
4385          */
4386         function replaceLinkHoldersText( $text ) {
4387                 $fname = 'Parser::replaceLinkHoldersText';
4388                 wfProfileIn( $fname );
4389
4390                 $text = preg_replace_callback(
4391                         '/<!--(LINK|IWLINK) (.*?)-->/',
4392                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4393                         $text );
4394
4395                 wfProfileOut( $fname );
4396                 return $text;
4397         }
4398
4399         /**
4400          * @param array $matches
4401          * @return string
4402          * @private
4403          */
4404         function replaceLinkHoldersTextCallback( $matches ) {
4405                 $type = $matches[1];
4406                 $key  = $matches[2];
4407                 if( $type == 'LINK' ) {
4408                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4409                                 return $this->mLinkHolders['texts'][$key];
4410                         }
4411                 } elseif( $type == 'IWLINK' ) {
4412                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4413                                 return $this->mInterwikiLinkHolders['texts'][$key];
4414                         }
4415                 }
4416                 return $matches[0];
4417         }
4418
4419         /**
4420          * Tag hook handler for 'pre'.
4421          */
4422         function renderPreTag( $text, $attribs ) {
4423                 // Backwards-compatibility hack
4424                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4425
4426                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4427                 return wfOpenElement( 'pre', $attribs ) .
4428                         Xml::escapeTagsOnly( $content ) .
4429                         '</pre>';
4430         }
4431
4432         /**
4433          * Renders an image gallery from a text with one line per image.
4434          * text labels may be given by using |-style alternative text. E.g.
4435          *   Image:one.jpg|The number "1"
4436          *   Image:tree.jpg|A tree
4437          * given as text will return the HTML of a gallery with two images,
4438          * labeled 'The number "1"' and
4439          * 'A tree'.
4440          */
4441         function renderImageGallery( $text, $params ) {
4442                 $ig = new ImageGallery();
4443                 $ig->setContextTitle( $this->mTitle );
4444                 $ig->setShowBytes( false );
4445                 $ig->setShowFilename( false );
4446                 $ig->setParser( $this );
4447                 $ig->setHideBadImages();
4448                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4449                 $ig->useSkin( $this->mOptions->getSkin() );
4450                 $ig->mRevisionId = $this->mRevisionId;
4451
4452                 if( isset( $params['caption'] ) ) {
4453                         $caption = $params['caption'];
4454                         $caption = htmlspecialchars( $caption );
4455                         $caption = $this->replaceInternalLinks( $caption );
4456                         $ig->setCaptionHtml( $caption );
4457                 }
4458                 if( isset( $params['perrow'] ) ) {
4459                         $ig->setPerRow( $params['perrow'] );
4460                 }
4461                 if( isset( $params['widths'] ) ) {
4462                         $ig->setWidths( $params['widths'] );
4463                 }
4464                 if( isset( $params['heights'] ) ) {
4465                         $ig->setHeights( $params['heights'] );
4466                 }
4467
4468                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4469
4470                 $lines = explode( "\n", $text );
4471                 foreach ( $lines as $line ) {
4472                         # match lines like these:
4473                         # Image:someimage.jpg|This is some image
4474                         $matches = array();
4475                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4476                         # Skip empty lines
4477                         if ( count( $matches ) == 0 ) {
4478                                 continue;
4479                         }
4480                         $tp = Title::newFromText( $matches[1] );
4481                         $nt =& $tp;
4482                         if( is_null( $nt ) ) {
4483                                 # Bogus title. Ignore these so we don't bomb out later.
4484                                 continue;
4485                         }
4486                         if ( isset( $matches[3] ) ) {
4487                                 $label = $matches[3];
4488                         } else {
4489                                 $label = '';
4490                         }
4491
4492                         $pout = $this->parse( $label,
4493                                 $this->mTitle,
4494                                 $this->mOptions,
4495                                 false, // Strip whitespace...?
4496                                 false  // Don't clear state!
4497                         );
4498                         $html = $pout->getText();
4499
4500                         $ig->add( $nt, $html );
4501
4502                         # Only add real images (bug #5586)
4503                         if ( $nt->getNamespace() == NS_IMAGE ) {
4504                                 $this->mOutput->addImage( $nt->getDBkey() );
4505                         }
4506                 }
4507                 return $ig->toHTML();
4508         }
4509
4510         function getImageParams( $handler ) {
4511                 if ( $handler ) {
4512                         $handlerClass = get_class( $handler );
4513                 } else {
4514                         $handlerClass = '';
4515                 }
4516                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4517                         // Initialise static lists
4518                         static $internalParamNames = array(
4519                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4520                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4521                                         'bottom', 'text-bottom' ),
4522                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4523                                         'upright', 'border' ),
4524                         );
4525                         static $internalParamMap;
4526                         if ( !$internalParamMap ) {
4527                                 $internalParamMap = array();
4528                                 foreach ( $internalParamNames as $type => $names ) {
4529                                         foreach ( $names as $name ) {
4530                                                 $magicName = str_replace( '-', '_', "img_$name" );
4531                                                 $internalParamMap[$magicName] = array( $type, $name );
4532                                         }
4533                                 }
4534                         }
4535
4536                         // Add handler params
4537                         $paramMap = $internalParamMap;
4538                         if ( $handler ) {
4539                                 $handlerParamMap = $handler->getParamMap();
4540                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4541                                         $paramMap[$magic] = array( 'handler', $paramName );
4542                                 }
4543                         }
4544                         $this->mImageParams[$handlerClass] = $paramMap;
4545                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4546                 }
4547                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4548         }
4549
4550         /**
4551          * Parse image options text and use it to make an image
4552          */
4553         function makeImage( $title, $options ) {
4554                 # @TODO: let the MediaHandler specify its transform parameters
4555                 #
4556                 # Check if the options text is of the form "options|alt text"
4557                 # Options are:
4558                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4559                 #  * left               no resizing, just left align. label is used for alt= only
4560                 #  * right              same, but right aligned
4561                 #  * none               same, but not aligned
4562                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4563                 #  * center             center the image
4564                 #  * framed             Keep original image size, no magnify-button.
4565                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4566                 #  * upright            reduce width for upright images, rounded to full __0 px
4567                 #  * border             draw a 1px border around the image
4568                 # vertical-align values (no % or length right now):
4569                 #  * baseline
4570                 #  * sub
4571                 #  * super
4572                 #  * top
4573                 #  * text-top
4574                 #  * middle
4575                 #  * bottom
4576                 #  * text-bottom
4577
4578                 $parts = array_map( 'trim', explode( '|', $options) );
4579                 $sk = $this->mOptions->getSkin();
4580
4581                 # Give extensions a chance to select the file revision for us
4582                 $skip = $time = false;
4583                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4584
4585                 if ( $skip ) {
4586                         return $sk->makeLinkObj( $title );
4587                 }
4588
4589                 # Get parameter map
4590                 $file = wfFindFile( $title, $time );
4591                 $handler = $file ? $file->getHandler() : false;
4592
4593                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4594
4595                 # Process the input parameters
4596                 $caption = '';
4597                 $params = array( 'frame' => array(), 'handler' => array(),
4598                         'horizAlign' => array(), 'vertAlign' => array() );
4599                 foreach( $parts as $part ) {
4600                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4601                         if ( isset( $paramMap[$magicName] ) ) {
4602                                 list( $type, $paramName ) = $paramMap[$magicName];
4603                                 $params[$type][$paramName] = $value;
4604
4605                                 // Special case; width and height come in one variable together
4606                                 if( $type == 'handler' && $paramName == 'width' ) {
4607                                         $m = array();
4608                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4609                                                 $params[$type]['width'] = intval( $m[1] );
4610                                                 $params[$type]['height'] = intval( $m[2] );
4611                                         } else {
4612                                                 $params[$type]['width'] = intval( $value );
4613                                         }
4614                                 }
4615                         } else {
4616                                 $caption = $part;
4617                         }
4618                 }
4619
4620                 # Process alignment parameters
4621                 if ( $params['horizAlign'] ) {
4622                         $params['frame']['align'] = key( $params['horizAlign'] );
4623                 }
4624                 if ( $params['vertAlign'] ) {
4625                         $params['frame']['valign'] = key( $params['vertAlign'] );
4626                 }
4627
4628                 # Validate the handler parameters
4629                 if ( $handler ) {
4630                         foreach ( $params['handler'] as $name => $value ) {
4631                                 if ( !$handler->validateParam( $name, $value ) ) {
4632                                         unset( $params['handler'][$name] );
4633                                 }
4634                         }
4635                 }
4636
4637                 # Strip bad stuff out of the alt text
4638                 $alt = $this->replaceLinkHoldersText( $caption );
4639
4640                 # make sure there are no placeholders in thumbnail attributes
4641                 # that are later expanded to html- so expand them now and
4642                 # remove the tags
4643                 $alt = $this->mStripState->unstripBoth( $alt );
4644                 $alt = Sanitizer::stripAllTags( $alt );
4645
4646                 $params['frame']['alt'] = $alt;
4647                 $params['frame']['caption'] = $caption;
4648
4649                 # Linker does the rest
4650                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4651
4652                 # Give the handler a chance to modify the parser object
4653                 if ( $handler ) {
4654                         $handler->parserTransformHook( $this, $file );
4655                 }
4656
4657                 return $ret;
4658         }
4659
4660         /**
4661          * Set a flag in the output object indicating that the content is dynamic and
4662          * shouldn't be cached.
4663          */
4664         function disableCache() {
4665                 wfDebug( "Parser output marked as uncacheable.\n" );
4666                 $this->mOutput->mCacheTime = -1;
4667         }
4668
4669         /**#@+
4670          * Callback from the Sanitizer for expanding items found in HTML attribute
4671          * values, so they can be safely tested and escaped.
4672          * @param string $text
4673          * @param array $args
4674          * @return string
4675          * @private
4676          */
4677         function attributeStripCallback( &$text, $args ) {
4678                 $text = $this->replaceVariables( $text, $args );
4679                 $text = $this->mStripState->unstripBoth( $text );
4680                 return $text;
4681         }
4682
4683         /**#@-*/
4684
4685         /**#@+
4686          * Accessor/mutator
4687          */
4688         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4689         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4690         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4691         /**#@-*/
4692
4693         /**#@+
4694          * Accessor
4695          */
4696         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4697         /**#@-*/
4698
4699
4700         /**
4701          * Break wikitext input into sections, and either pull or replace
4702          * some particular section's text.
4703          *
4704          * External callers should use the getSection and replaceSection methods.
4705          *
4706          * @param $text Page wikitext
4707          * @param $section Numbered section. 0 pulls the text before the first
4708          *                 heading; other numbers will pull the given section
4709          *                 along with its lower-level subsections.
4710          * @param $mode One of "get" or "replace"
4711          * @param $newtext Replacement text for section data.
4712          * @return string for "get", the extracted section text.
4713          *                for "replace", the whole page with the section replaced.
4714          */
4715         private function extractSections( $text, $section, $mode, $newtext='' ) {
4716                 # I.... _hope_ this is right.
4717                 # Otherwise, sometimes we don't have things initialized properly.
4718                 $this->clearState();
4719
4720                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4721                 # comments to be stripped as well)
4722                 $stripState = new StripState;
4723
4724                 $oldOutputType = $this->mOutputType;
4725                 $oldOptions = $this->mOptions;
4726                 $this->mOptions = new ParserOptions();
4727                 $this->setOutputType( self::OT_WIKI );
4728
4729                 $striptext = $this->strip( $text, $stripState, true );
4730
4731                 $this->setOutputType( $oldOutputType );
4732                 $this->mOptions = $oldOptions;
4733
4734                 # now that we can be sure that no pseudo-sections are in the source,
4735                 # split it up by section
4736                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4737                 $comment = "(?:$uniq-!--.*?QINU\x07)";
4738                 $secs = preg_split(
4739                         "/
4740                         (
4741                                 ^
4742                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4743                                 (=+) # Should this be limited to 6?
4744                                 .+?  # Section title...
4745                                 \\2  # Ending = count must match start
4746                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4747                                 $
4748                         |
4749                                 <h([1-6])\b.*?>
4750                                 .*?
4751                                 <\/h\\3\s*>
4752                         )
4753                         /mix",
4754                         $striptext, -1,
4755                         PREG_SPLIT_DELIM_CAPTURE);
4756
4757                 if( $mode == "get" ) {
4758                         if( $section == 0 ) {
4759                                 // "Section 0" returns the content before any other section.
4760                                 $rv = $secs[0];
4761                         } else {
4762                                 //track missing section, will replace if found.
4763                                 $rv = $newtext;
4764                         }
4765                 } elseif( $mode == "replace" ) {
4766                         if( $section == 0 ) {
4767                                 $rv = $newtext . "\n\n";
4768                                 $remainder = true;
4769                         } else {
4770                                 $rv = $secs[0];
4771                                 $remainder = false;
4772                         }
4773                 }
4774                 $count = 0;
4775                 $sectionLevel = 0;
4776                 for( $index = 1; $index < count( $secs ); ) {
4777                         $headerLine = $secs[$index++];
4778                         if( $secs[$index] ) {
4779                                 // A wiki header
4780                                 $headerLevel = strlen( $secs[$index++] );
4781                         } else {
4782                                 // An HTML header
4783                                 $index++;
4784                                 $headerLevel = intval( $secs[$index++] );
4785                         }
4786                         $content = $secs[$index++];
4787
4788                         $count++;
4789                         if( $mode == "get" ) {
4790                                 if( $count == $section ) {
4791                                         $rv = $headerLine . $content;
4792                                         $sectionLevel = $headerLevel;
4793                                 } elseif( $count > $section ) {
4794                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4795                                                 $rv .= $headerLine . $content;
4796                                         } else {
4797                                                 // Broke out to a higher-level section
4798                                                 break;
4799                                         }
4800                                 }
4801                         } elseif( $mode == "replace" ) {
4802                                 if( $count < $section ) {
4803                                         $rv .= $headerLine . $content;
4804                                 } elseif( $count == $section ) {
4805                                         $rv .= $newtext . "\n\n";
4806                                         $sectionLevel = $headerLevel;
4807                                 } elseif( $count > $section ) {
4808                                         if( $headerLevel <= $sectionLevel ) {
4809                                                 // Passed the section's sub-parts.
4810                                                 $remainder = true;
4811                                         }
4812                                         if( $remainder ) {
4813                                                 $rv .= $headerLine . $content;
4814                                         }
4815                                 }
4816                         }
4817                 }
4818                 if (is_string($rv))
4819                         # reinsert stripped tags
4820                         $rv = trim( $stripState->unstripBoth( $rv ) );
4821
4822                 return $rv;
4823         }
4824
4825         /**
4826          * This function returns the text of a section, specified by a number ($section).
4827          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4828          * the first section before any such heading (section 0).
4829          *
4830          * If a section contains subsections, these are also returned.
4831          *
4832          * @param $text String: text to look in
4833          * @param $section Integer: section number
4834          * @param $deftext: default to return if section is not found
4835          * @return string text of the requested section
4836          */
4837         public function getSection( $text, $section, $deftext='' ) {
4838                 return $this->extractSections( $text, $section, "get", $deftext );
4839         }
4840
4841         public function replaceSection( $oldtext, $section, $text ) {
4842                 return $this->extractSections( $oldtext, $section, "replace", $text );
4843         }
4844
4845         /**
4846          * Get the timestamp associated with the current revision, adjusted for
4847          * the default server-local timestamp
4848          */
4849         function getRevisionTimestamp() {
4850                 if ( is_null( $this->mRevisionTimestamp ) ) {
4851                         wfProfileIn( __METHOD__ );
4852                         global $wgContLang;
4853                         $dbr = wfGetDB( DB_SLAVE );
4854                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4855                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4856
4857                         // Normalize timestamp to internal MW format for timezone processing.
4858                         // This has the added side-effect of replacing a null value with
4859                         // the current time, which gives us more sensible behavior for
4860                         // previews.
4861                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4862
4863                         // The cryptic '' timezone parameter tells to use the site-default
4864                         // timezone offset instead of the user settings.
4865                         //
4866                         // Since this value will be saved into the parser cache, served
4867                         // to other users, and potentially even used inside links and such,
4868                         // it needs to be consistent for all visitors.
4869                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4870
4871                         wfProfileOut( __METHOD__ );
4872                 }
4873                 return $this->mRevisionTimestamp;
4874         }
4875
4876         /**
4877          * Mutator for $mDefaultSort
4878          *
4879          * @param $sort New value
4880          */
4881         public function setDefaultSort( $sort ) {
4882                 $this->mDefaultSort = $sort;
4883         }
4884
4885         /**
4886          * Accessor for $mDefaultSort
4887          * Will use the title/prefixed title if none is set
4888          *
4889          * @return string
4890          */
4891         public function getDefaultSort() {
4892                 if( $this->mDefaultSort !== false ) {
4893                         return $this->mDefaultSort;
4894                 } else {
4895                         return $this->mTitle->getNamespace() == NS_CATEGORY
4896                                         ? $this->mTitle->getText()
4897                                         : $this->mTitle->getPrefixedText();
4898                 }
4899         }
4900
4901         /**
4902          * Try to guess the section anchor name based on a wikitext fragment
4903          * presumably extracted from a heading, for example "Header" from
4904          * "== Header ==".
4905          */
4906         public function guessSectionNameFromWikiText( $text ) {
4907                 # Strip out wikitext links(they break the anchor)
4908                 $text = $this->stripSectionName( $text );
4909                 $headline = Sanitizer::decodeCharReferences( $text );
4910                 # strip out HTML
4911                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4912                 $headline = trim( $headline );
4913                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4914                 $replacearray = array(
4915                         '%3A' => ':',
4916                         '%' => '.'
4917                 );
4918                 return str_replace(
4919                         array_keys( $replacearray ),
4920                         array_values( $replacearray ),
4921                         $sectionanchor );
4922         }
4923
4924         /**
4925          * Strips a text string of wikitext for use in a section anchor
4926          *
4927          * Accepts a text string and then removes all wikitext from the
4928          * string and leaves only the resultant text (i.e. the result of
4929          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4930          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4931          * to create valid section anchors by mimicing the output of the
4932          * parser when headings are parsed.
4933          *
4934          * @param $text string Text string to be stripped of wikitext
4935          * for use in a Section anchor
4936          * @return Filtered text string
4937          */
4938         public function stripSectionName( $text ) {
4939                 # Strip internal link markup
4940                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4941                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4942
4943                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4944                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4945                 # on how many empty links there are on the page - need to figure that out.
4946                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4947
4948                 # Parse wikitext quotes (italics & bold)
4949                 $text = $this->doQuotes($text);
4950
4951                 # Strip HTML tags
4952                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4953                 return $text;
4954         }
4955
4956         /**
4957          * strip/replaceVariables/unstrip for preprocessor regression testing
4958          */
4959         function srvus( $text ) {
4960                 $text = $this->strip( $text, $this->mStripState );
4961                 $text = Sanitizer::removeHTMLtags( $text );
4962                 $text = $this->replaceVariables( $text );
4963                 $text = preg_replace( '/<!--MWTEMPLATESECTION.*?-->/', '', $text );
4964                 $text = $this->mStripState->unstripBoth( $text );
4965                 return $text;
4966         }
4967 }