includes/parser/Parser.php

   1 <?php
   2 /**
   3  * PHP parser that converts wiki markup to HTML.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Parser
  22  */
  23
  24 /**
  25  * @defgroup Parser Parser
  26  */
  27
  28 /**
  29  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  30  * syntax, such as "[[link]]" for making links), and provides a one-way
  31  * transformation of that wiki markup it into XHTML output / markup
  32  * (which in turn the browser understands, and can display).
  33  *
  34  * <pre>
  35  * There are five main entry points into the Parser class:
  36  * parse()
  37  *     produces HTML output
  38  * preSaveTransform().
  39  *     produces altered wiki markup.
  40  * preprocess()
  41  *     removes HTML comments and expands templates
  42  * cleanSig() / cleanSigInSig()
  43  *     Cleans a signature before saving it to preferences
  44  * getSection()
  45  *     Return the content of a section from an article for section editing
  46  * replaceSection()
  47  *     Replaces a section by number inside an article
  48  * getPreloadText()
  49  *     Removes <noinclude> sections, and <includeonly> tags.
  50  *
  51  * Globals used:
  52  *    object: $wgContLang
  53  *
  54  * NOT $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
  55  *
  56  * settings:
  57  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  58  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  59  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  60  *  $wgMaxArticleSize*
  61  *
  62  *  * only within ParserOptions
  63  * </pre>
  64  *
  65  * @ingroup Parser
  66  */
  67 class Parser {
  68         /**
  69          * Update this version number when the ParserOutput format
  70          * changes in an incompatible way, so the parser cache
  71          * can automatically discard old data.
  72          */
  73         const VERSION = '1.6.4';
  74
  75         /**
  76          * Update this version number when the output of serialiseHalfParsedText()
  77          * changes in an incompatible way
  78          */
  79         const HALF_PARSED_VERSION = 2;
  80
  81         # Flags for Parser::setFunctionHook
  82         # Also available as global constants from Defines.php
  83         const SFH_NO_HASH = 1;
  84         const SFH_OBJECT_ARGS = 2;
  85
  86         # Constants needed for external link processing
  87         # Everything except bracket, space, or control characters
  88         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  89         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  90         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  91         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  92                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
  93
  94         # State constants for the definition list colon extraction
  95         const COLON_STATE_TEXT = 0;
  96         const COLON_STATE_TAG = 1;
  97         const COLON_STATE_TAGSTART = 2;
  98         const COLON_STATE_CLOSETAG = 3;
  99         const COLON_STATE_TAGSLASH = 4;
 100         const COLON_STATE_COMMENT = 5;
 101         const COLON_STATE_COMMENTDASH = 6;
 102         const COLON_STATE_COMMENTDASHDASH = 7;
 103
 104         # Flags for preprocessToDom
 105         const PTD_FOR_INCLUSION = 1;
 106
 107         # Allowed values for $this->mOutputType
 108         # Parameter to startExternalParse().
 109         const OT_HTML = 1; # like parse()
 110         const OT_WIKI = 2; # like preSaveTransform()
 111         const OT_PREPROCESS = 3; # like preprocess()
 112         const OT_MSG = 3;
 113         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
 114
 115         # Marker Suffix needs to be accessible staticly.
 116         const MARKER_SUFFIX = "-QINU\x7f";
 117
 118         # Persistent:
 119         var $mTagHooks = array();
 120         var $mTransparentTagHooks = array();
 121         var $mFunctionHooks = array();
 122         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 123         var $mFunctionTagHooks = array();
 124         var $mStripList  = array();
 125         var $mDefaultStripList  = array();
 126         var $mVarCache = array();
 127         var $mImageParams = array();
 128         var $mImageParamsMagicArray = array();
 129         var $mMarkerIndex = 0;
 130         var $mFirstCall = true;
 131
 132         # Initialised by initialiseVariables()
 133
 134         /**
 135          * @var MagicWordArray
 136          */
 137         var $mVariables;
 138
 139         /**
 140          * @var MagicWordArray
 141          */
 142         var $mSubstWords;
 143         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 144
 145         # Cleared with clearState():
 146         /**
 147          * @var ParserOutput
 148          */
 149         var $mOutput;
 150         var $mAutonumber, $mDTopen;
 151
 152         /**
 153          * @var StripState
 154          */
 155         var $mStripState;
 156
 157         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 158         /**
 159          * @var LinkHolderArray
 160          */
 161         var $mLinkHolders;
 162
 163         var $mLinkID;
 164         var $mIncludeSizes, $mPPNodeCount, $mHighestExpansionDepth, $mDefaultSort;
 165         var $mTplExpandCache; # empty-frame expansion cache
 166         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 167         var $mExpensiveFunctionCount; # number of expensive parser function calls
 168         var $mShowToc, $mForceTocPosition;
 169
 170         /**
 171          * @var User
 172          */
 173         var $mUser; # User object; only used when doing pre-save transform
 174
 175         # Temporary
 176         # These are variables reset at least once per parse regardless of $clearState
 177
 178         /**
 179          * @var ParserOptions
 180          */
 181         var $mOptions;
 182
 183         /**
 184          * @var Title
 185          */
 186         var $mTitle;        # Title context, used for self-link rendering and similar things
 187         var $mOutputType;   # Output type, one of the OT_xxx constants
 188         var $ot;            # Shortcut alias, see setOutputType()
 189         var $mRevisionObject; # The revision object of the specified revision ID
 190         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 191         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 192         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 193         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 194
 195         /**
 196          * @var string
 197          */
 198         var $mUniqPrefix;
 199
 200         /**
 201          * Constructor
 202          *
 203          * @param $conf array
 204          */
 205         public function __construct( $conf = array() ) {
 206                 $this->mConf = $conf;
 207                 $this->mUrlProtocols = wfUrlProtocols();
 208                 $this->mExtLinkBracketedRegex = '/\[((' . $this->mUrlProtocols . ')'.
 209                         self::EXT_LINK_URL_CLASS.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 210                 if ( isset( $conf['preprocessorClass'] ) ) {
 211                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 212                 } elseif ( defined( 'MW_COMPILED' ) ) {
 213                         # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
 214                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 215                 } elseif ( extension_loaded( 'domxml' ) ) {
 216                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 217                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 218                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 219                 } elseif ( extension_loaded( 'dom' ) ) {
 220                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 221                 } else {
 222                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 223                 }
 224                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 225         }
 226
 227         /**
 228          * Reduce memory usage to reduce the impact of circular references
 229          */
 230         function __destruct() {
 231                 if ( isset( $this->mLinkHolders ) ) {
 232                         unset( $this->mLinkHolders );
 233                 }
 234                 foreach ( $this as $name => $value ) {
 235                         unset( $this->$name );
 236                 }
 237         }
 238
 239         /**
 240          * Do various kinds of initialisation on the first call of the parser
 241          */
 242         function firstCallInit() {
 243                 if ( !$this->mFirstCall ) {
 244                         return;
 245                 }
 246                 $this->mFirstCall = false;
 247
 248                 wfProfileIn( __METHOD__ );
 249
 250                 CoreParserFunctions::register( $this );
 251                 CoreTagHooks::register( $this );
 252                 $this->initialiseVariables();
 253
 254                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 255                 wfProfileOut( __METHOD__ );
 256         }
 257
 258         /**
 259          * Clear Parser state
 260          *
 261          * @private
 262          */
 263         function clearState() {
 264                 wfProfileIn( __METHOD__ );
 265                 if ( $this->mFirstCall ) {
 266                         $this->firstCallInit();
 267                 }
 268                 $this->mOutput = new ParserOutput;
 269                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 270                 $this->mAutonumber = 0;
 271                 $this->mLastSection = '';
 272                 $this->mDTopen = false;
 273                 $this->mIncludeCount = array();
 274                 $this->mArgStack = false;
 275                 $this->mInPre = false;
 276                 $this->mLinkHolders = new LinkHolderArray( $this );
 277                 $this->mLinkID = 0;
 278                 $this->mRevisionObject = $this->mRevisionTimestamp =
 279                         $this->mRevisionId = $this->mRevisionUser = null;
 280                 $this->mVarCache = array();
 281                 $this->mUser = null;
 282
 283                 /**
 284                  * Prefix for temporary replacement strings for the multipass parser.
 285                  * \x07 should never appear in input as it's disallowed in XML.
 286                  * Using it at the front also gives us a little extra robustness
 287                  * since it shouldn't match when butted up against identifier-like
 288                  * string constructs.
 289                  *
 290                  * Must not consist of all title characters, or else it will change
 291                  * the behaviour of <nowiki> in a link.
 292                  */
 293                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 294                 $this->mStripState = new StripState( $this->mUniqPrefix );
 295
 296
 297                 # Clear these on every parse, bug 4549
 298                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 299
 300                 $this->mShowToc = true;
 301                 $this->mForceTocPosition = false;
 302                 $this->mIncludeSizes = array(
 303                         'post-expand' => 0,
 304                         'arg' => 0,
 305                 );
 306                 $this->mPPNodeCount = 0;
 307                 $this->mHighestExpansionDepth = 0;
 308                 $this->mDefaultSort = false;
 309                 $this->mHeadings = array();
 310                 $this->mDoubleUnderscores = array();
 311                 $this->mExpensiveFunctionCount = 0;
 312
 313                 # Fix cloning
 314                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 315                         $this->mPreprocessor = null;
 316                 }
 317
 318                 wfRunHooks( 'ParserClearState', array( &$this ) );
 319                 wfProfileOut( __METHOD__ );
 320         }
 321
 322         /**
 323          * Convert wikitext to HTML
 324          * Do not call this function recursively.
 325          *
 326          * @param $text String: text we want to parse
 327          * @param $title Title object
 328          * @param $options ParserOptions
 329          * @param $linestart boolean
 330          * @param $clearState boolean
 331          * @param $revid Int: number to pass in {{REVISIONID}}
 332          * @return ParserOutput a ParserOutput
 333          */
 334         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 335                 /**
 336                  * First pass--just handle <nowiki> sections, pass the rest off
 337                  * to internalParse() which does all the real work.
 338                  */
 339
 340                 global $wgUseTidy, $wgAlwaysUseTidy, $wgDisableLangConversion, $wgDisableTitleConversion;
 341                 $fname = __METHOD__.'-' . wfGetCaller();
 342                 wfProfileIn( __METHOD__ );
 343                 wfProfileIn( $fname );
 344
 345                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 346
 347                 # Remove the strip marker tag prefix from the input, if present.
 348                 if ( $clearState ) {
 349                         $text = str_replace( $this->mUniqPrefix, '', $text );
 350                 }
 351
 352                 $oldRevisionId = $this->mRevisionId;
 353                 $oldRevisionObject = $this->mRevisionObject;
 354                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 355                 $oldRevisionUser = $this->mRevisionUser;
 356                 if ( $revid !== null ) {
 357                         $this->mRevisionId = $revid;
 358                         $this->mRevisionObject = null;
 359                         $this->mRevisionTimestamp = null;
 360                         $this->mRevisionUser = null;
 361                 }
 362
 363                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 364                 # No more strip!
 365                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 366                 $text = $this->internalParse( $text );
 367                 wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
 368
 369                 $text = $this->mStripState->unstripGeneral( $text );
 370
 371                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 372                 $fixtags = array(
 373                         # french spaces, last one Guillemet-left
 374                         # only if there is something before the space
 375                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
 376                         # french spaces, Guillemet-right
 377                         '/(\\302\\253) /' => '\\1&#160;',
 378                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 379                 );
 380                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 381
 382                 $text = $this->doBlockLevels( $text, $linestart );
 383
 384                 $this->replaceLinkHolders( $text );
 385
 386                 /**
 387                  * The input doesn't get language converted if
 388                  * a) It's disabled
 389                  * b) Content isn't converted
 390                  * c) It's a conversion table
 391                  * d) it is an interface message (which is in the user language)
 392                  */
 393                 if ( !( $wgDisableLangConversion
 394                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 395                                 || $this->mTitle->isConversionTable() ) )
 396                 {
 397                         # Run convert unconditionally in 1.18-compatible mode
 398                         global $wgBug34832TransitionalRollback;
 399                         if ( $wgBug34832TransitionalRollback || !$this->mOptions->getInterfaceMessage() ) {
 400                                 # The position of the convert() call should not be changed. it
 401                                 # assumes that the links are all replaced and the only thing left
 402                                 # is the <nowiki> mark.
 403                                 $text = $this->getConverterLanguage()->convert( $text );
 404                         }
 405                 }
 406
 407                 /**
 408                  * A converted title will be provided in the output object if title and
 409                  * content conversion are enabled, the article text does not contain
 410                  * a conversion-suppressing double-underscore tag, and no
 411                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 412                  * automatic link conversion.
 413                  */
 414                 if ( !( $wgDisableLangConversion
 415                                 || $wgDisableTitleConversion
 416                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 417                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 418                                 || $this->mOutput->getDisplayTitle() !== false ) )
 419                 {
 420                         $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
 421                         if ( $convruletitle ) {
 422                                 $this->mOutput->setTitleText( $convruletitle );
 423                         } else {
 424                                 $titleText = $this->getConverterLanguage()->convertTitle( $title );
 425                                 $this->mOutput->setTitleText( $titleText );
 426                         }
 427                 }
 428
 429                 $text = $this->mStripState->unstripNoWiki( $text );
 430
 431                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 432
 433                 $text = $this->replaceTransparentTags( $text );
 434                 $text = $this->mStripState->unstripGeneral( $text );
 435
 436                 $text = Sanitizer::normalizeCharReferences( $text );
 437
 438                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 439                         $text = MWTidy::tidy( $text );
 440                 } else {
 441                         # attempt to sanitize at least some nesting problems
 442                         # (bug #2702 and quite a few others)
 443                         $tidyregs = array(
 444                                 # ''Something [http://www.cool.com cool''] -->
 445                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 446                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 447                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 448                                 # fix up an anchor inside another anchor, only
 449                                 # at least for a single single nested link (bug 3695)
 450                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 451                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 452                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 453                                 # contains a div, so fix it up here; replace
 454                                 # div with escaped text
 455                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 456                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 457                                 # remove empty italic or bold tag pairs, some
 458                                 # introduced by rules above
 459                                 '/<([bi])><\/\\1>/' => '',
 460                         );
 461
 462                         $text = preg_replace(
 463                                 array_keys( $tidyregs ),
 464                                 array_values( $tidyregs ),
 465                                 $text );
 466                 }
 467
 468                 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
 469                         $this->limitationWarn( 'expensive-parserfunction',
 470                                 $this->mExpensiveFunctionCount,
 471                                 $this->mOptions->getExpensiveParserFunctionLimit()
 472                         );
 473                 }
 474
 475                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 476
 477                 # Information on include size limits, for the benefit of users who try to skirt them
 478                 if ( $this->mOptions->getEnableLimitReport() ) {
 479                         $max = $this->mOptions->getMaxIncludeSize();
 480                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/{$this->mOptions->getExpensiveParserFunctionLimit()}\n";
 481                         $limitReport =
 482                                 "NewPP limit report\n" .
 483                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 484                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 485                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 486                                 "Highest expansion depth: {$this->mHighestExpansionDepth}/{$this->mOptions->getMaxPPExpandDepth()}\n".
 487                                 $PFreport;
 488                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 489                         $text .= "\n<!-- \n$limitReport-->\n";
 490                 }
 491                 $this->mOutput->setText( $text );
 492
 493                 $this->mRevisionId = $oldRevisionId;
 494                 $this->mRevisionObject = $oldRevisionObject;
 495                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 496                 $this->mRevisionUser = $oldRevisionUser;
 497                 wfProfileOut( $fname );
 498                 wfProfileOut( __METHOD__ );
 499
 500                 return $this->mOutput;
 501         }
 502
 503         /**
 504          * Recursive parser entry point that can be called from an extension tag
 505          * hook.
 506          *
 507          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 508          *
 509          * @param $text String: text extension wants to have parsed
 510          * @param $frame PPFrame: The frame to use for expanding any template variables
 511          *
 512          * @return string
 513          */
 514         function recursiveTagParse( $text, $frame=false ) {
 515                 wfProfileIn( __METHOD__ );
 516                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 517                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 518                 $text = $this->internalParse( $text, false, $frame );
 519                 wfProfileOut( __METHOD__ );
 520                 return $text;
 521         }
 522
 523         /**
 524          * Expand templates and variables in the text, producing valid, static wikitext.
 525          * Also removes comments.
 526          * @return mixed|string
 527          */
 528         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 529                 wfProfileIn( __METHOD__ );
 530                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 531                 if ( $revid !== null ) {
 532                         $this->mRevisionId = $revid;
 533                 }
 534                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 535                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 536                 $text = $this->replaceVariables( $text );
 537                 $text = $this->mStripState->unstripBoth( $text );
 538                 wfProfileOut( __METHOD__ );
 539                 return $text;
 540         }
 541
 542         /**
 543          * Recursive parser entry point that can be called from an extension tag
 544          * hook.
 545          *
 546          * @param $text String: text to be expanded
 547          * @param $frame PPFrame: The frame to use for expanding any template variables
 548          * @return String
 549          * @since 1.19
 550          */
 551         public function recursivePreprocess( $text, $frame = false ) {
 552                 wfProfileIn( __METHOD__ );
 553                 $text = $this->replaceVariables( $text, $frame );
 554                 $text = $this->mStripState->unstripBoth( $text );
 555                 wfProfileOut( __METHOD__ );
 556                 return $text;
 557         }
 558
 559         /**
 560          * Process the wikitext for the "?preload=" feature. (bug 5210)
 561          *
 562          * "<noinclude>", "<includeonly>" etc. are parsed as for template
 563          * transclusion, comments, templates, arguments, tags hooks and parser
 564          * functions are untouched.
 565          *
 566          * @param $text String
 567          * @param $title Title
 568          * @param $options ParserOptions
 569          * @return String
 570          */
 571         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 572                 # Parser (re)initialisation
 573                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 574
 575                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 576                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 577                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 578                 $text = $this->mStripState->unstripBoth( $text );
 579                 return $text;
 580         }
 581
 582         /**
 583          * Get a random string
 584          *
 585          * @return string
 586          */
 587         static public function getRandomString() {
 588                 return wfRandomString( 16 );
 589         }
 590
 591         /**
 592          * Set the current user.
 593          * Should only be used when doing pre-save transform.
 594          *
 595          * @param $user Mixed: User object or null (to reset)
 596          */
 597         function setUser( $user ) {
 598                 $this->mUser = $user;
 599         }
 600
 601         /**
 602          * Accessor for mUniqPrefix.
 603          *
 604          * @return String
 605          */
 606         public function uniqPrefix() {
 607                 if ( !isset( $this->mUniqPrefix ) ) {
 608                         # @todo FIXME: This is probably *horribly wrong*
 609                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 610                         # if this is called for a parser cache hit, the parser may not
 611                         # have ever been initialized in the first place.
 612                         # Not really sure what the heck is supposed to be going on here.
 613                         return '';
 614                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 615                 }
 616                 return $this->mUniqPrefix;
 617         }
 618
 619         /**
 620          * Set the context title
 621          *
 622          * @param $t Title
 623          */
 624         function setTitle( $t ) {
 625                 if ( !$t || $t instanceof FakeTitle ) {
 626                         $t = Title::newFromText( 'NO TITLE' );
 627                 }
 628
 629                 if ( strval( $t->getFragment() ) !== '' ) {
 630                         # Strip the fragment to avoid various odd effects
 631                         $this->mTitle = clone $t;
 632                         $this->mTitle->setFragment( '' );
 633                 } else {
 634                         $this->mTitle = $t;
 635                 }
 636         }
 637
 638         /**
 639          * Accessor for the Title object
 640          *
 641          * @return Title object
 642          */
 643         function getTitle() {
 644                 return $this->mTitle;
 645         }
 646
 647         /**
 648          * Accessor/mutator for the Title object
 649          *
 650          * @param $x Title object or null to just get the current one
 651          * @return Title object
 652          */
 653         function Title( $x = null ) {
 654                 return wfSetVar( $this->mTitle, $x );
 655         }
 656
 657         /**
 658          * Set the output type
 659          *
 660          * @param $ot Integer: new value
 661          */
 662         function setOutputType( $ot ) {
 663                 $this->mOutputType = $ot;
 664                 # Shortcut alias
 665                 $this->ot = array(
 666                         'html' => $ot == self::OT_HTML,
 667                         'wiki' => $ot == self::OT_WIKI,
 668                         'pre' => $ot == self::OT_PREPROCESS,
 669                         'plain' => $ot == self::OT_PLAIN,
 670                 );
 671         }
 672
 673         /**
 674          * Accessor/mutator for the output type
 675          *
 676          * @param $x int|null New value or null to just get the current one
 677          * @return Integer
 678          */
 679         function OutputType( $x = null ) {
 680                 return wfSetVar( $this->mOutputType, $x );
 681         }
 682
 683         /**
 684          * Get the ParserOutput object
 685          *
 686          * @return ParserOutput object
 687          */
 688         function getOutput() {
 689                 return $this->mOutput;
 690         }
 691
 692         /**
 693          * Get the ParserOptions object
 694          *
 695          * @return ParserOptions object
 696          */
 697         function getOptions() {
 698                 return $this->mOptions;
 699         }
 700
 701         /**
 702          * Accessor/mutator for the ParserOptions object
 703          *
 704          * @param $x ParserOptions New value or null to just get the current one
 705          * @return ParserOptions Current ParserOptions object
 706          */
 707         function Options( $x = null ) {
 708                 return wfSetVar( $this->mOptions, $x );
 709         }
 710
 711         /**
 712          * @return int
 713          */
 714         function nextLinkID() {
 715                 return $this->mLinkID++;
 716         }
 717
 718         /**
 719          * @param $id int
 720          */
 721         function setLinkID( $id ) {
 722                 $this->mLinkID = $id;
 723         }
 724
 725         /**
 726          * Get a language object for use in parser functions such as {{FORMATNUM:}}
 727          * @return Language
 728          */
 729         function getFunctionLang() {
 730                 return $this->getTargetLanguage();
 731         }
 732
 733         /**
 734          * Get the target language for the content being parsed. This is usually the
 735          * language that the content is in.
 736          */
 737         function getTargetLanguage() {
 738                 $target = $this->mOptions->getTargetLanguage();
 739                 if ( $target !== null ) {
 740                         return $target;
 741                 } elseif( $this->mOptions->getInterfaceMessage() ) {
 742                         return $this->mOptions->getUserLangObj();
 743                 } elseif( is_null( $this->mTitle ) ) {
 744                         throw new MWException( __METHOD__.': $this->mTitle is null' );
 745                 }
 746                 return $this->mTitle->getPageLanguage();
 747         }
 748
 749         /**
 750          * Get the language object for language conversion
 751          */
 752         function getConverterLanguage() {
 753                 global $wgBug34832TransitionalRollback, $wgContLang;
 754                 if ( $wgBug34832TransitionalRollback ) {
 755                         return $wgContLang;
 756                 } else {
 757                         return $this->getTargetLanguage();
 758                 }
 759         }
 760
 761         /**
 762          * Get a User object either from $this->mUser, if set, or from the
 763          * ParserOptions object otherwise
 764          *
 765          * @return User object
 766          */
 767         function getUser() {
 768                 if ( !is_null( $this->mUser ) ) {
 769                         return $this->mUser;
 770                 }
 771                 return $this->mOptions->getUser();
 772         }
 773
 774         /**
 775          * Get a preprocessor object
 776          *
 777          * @return Preprocessor instance
 778          */
 779         function getPreprocessor() {
 780                 if ( !isset( $this->mPreprocessor ) ) {
 781                         $class = $this->mPreprocessorClass;
 782                         $this->mPreprocessor = new $class( $this );
 783                 }
 784                 return $this->mPreprocessor;
 785         }
 786
 787         /**
 788          * Replaces all occurrences of HTML-style comments and the given tags
 789          * in the text with a random marker and returns the next text. The output
 790          * parameter $matches will be an associative array filled with data in
 791          * the form:
 792          *
 793          * @code
 794          *   'UNIQ-xxxxx' => array(
 795          *     'element',
 796          *     'tag content',
 797          *     array( 'param' => 'x' ),
 798          *     '<element param="x">tag content</element>' ) )
 799          * @endcode
 800          *
 801          * @param $elements array list of element names. Comments are always extracted.
 802          * @param $text string Source text string.
 803          * @param $matches array Out parameter, Array: extracted tags
 804          * @param $uniq_prefix string
 805          * @return String: stripped text
 806          */
 807         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 808                 static $n = 1;
 809                 $stripped = '';
 810                 $matches = array();
 811
 812                 $taglist = implode( '|', $elements );
 813                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 814
 815                 while ( $text != '' ) {
 816                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 817                         $stripped .= $p[0];
 818                         if ( count( $p ) < 5 ) {
 819                                 break;
 820                         }
 821                         if ( count( $p ) > 5 ) {
 822                                 # comment
 823                                 $element    = $p[4];
 824                                 $attributes = '';
 825                                 $close      = '';
 826                                 $inside     = $p[5];
 827                         } else {
 828                                 # tag
 829                                 $element    = $p[1];
 830                                 $attributes = $p[2];
 831                                 $close      = $p[3];
 832                                 $inside     = $p[4];
 833                         }
 834
 835                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 836                         $stripped .= $marker;
 837
 838                         if ( $close === '/>' ) {
 839                                 # Empty element tag, <tag />
 840                                 $content = null;
 841                                 $text = $inside;
 842                                 $tail = null;
 843                         } else {
 844                                 if ( $element === '!--' ) {
 845                                         $end = '/(-->)/';
 846                                 } else {
 847                                         $end = "/(<\\/$element\\s*>)/i";
 848                                 }
 849                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 850                                 $content = $q[0];
 851                                 if ( count( $q ) < 3 ) {
 852                                         # No end tag -- let it run out to the end of the text.
 853                                         $tail = '';
 854                                         $text = '';
 855                                 } else {
 856                                         $tail = $q[1];
 857                                         $text = $q[2];
 858                                 }
 859                         }
 860
 861                         $matches[$marker] = array( $element,
 862                                 $content,
 863                                 Sanitizer::decodeTagAttributes( $attributes ),
 864                                 "<$element$attributes$close$content$tail" );
 865                 }
 866                 return $stripped;
 867         }
 868
 869         /**
 870          * Get a list of strippable XML-like elements
 871          *
 872          * @return array
 873          */
 874         function getStripList() {
 875                 return $this->mStripList;
 876         }
 877
 878         /**
 879          * Add an item to the strip state
 880          * Returns the unique tag which must be inserted into the stripped text
 881          * The tag will be replaced with the original text in unstrip()
 882          *
 883          * @param $text string
 884          *
 885          * @return string
 886          */
 887         function insertStripItem( $text ) {
 888                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 889                 $this->mMarkerIndex++;
 890                 $this->mStripState->addGeneral( $rnd, $text );
 891                 return $rnd;
 892         }
 893
 894         /**
 895          * parse the wiki syntax used to render tables
 896          *
 897          * @private
 898          * @return string
 899          */
 900         function doTableStuff( $text ) {
 901                 wfProfileIn( __METHOD__ );
 902
 903                 $lines = StringUtils::explode( "\n", $text );
 904                 $out = '';
 905                 $td_history = array(); # Is currently a td tag open?
 906                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 907                 $tr_history = array(); # Is currently a tr tag open?
 908                 $tr_attributes = array(); # history of tr attributes
 909                 $has_opened_tr = array(); # Did this table open a <tr> element?
 910                 $indent_level = 0; # indent level of the table
 911
 912                 foreach ( $lines as $outLine ) {
 913                         $line = trim( $outLine );
 914
 915                         if ( $line === '' ) { # empty line, go to next line
 916                                 $out .= $outLine."\n";
 917                                 continue;
 918                         }
 919
 920                         $first_character = $line[0];
 921                         $matches = array();
 922
 923                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 924                                 # First check if we are starting a new table
 925                                 $indent_level = strlen( $matches[1] );
 926
 927                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 928                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 929
 930                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 931                                 array_push( $td_history , false );
 932                                 array_push( $last_tag_history , '' );
 933                                 array_push( $tr_history , false );
 934                                 array_push( $tr_attributes , '' );
 935                                 array_push( $has_opened_tr , false );
 936                         } elseif ( count( $td_history ) == 0 ) {
 937                                 # Don't do any of the following
 938                                 $out .= $outLine."\n";
 939                                 continue;
 940                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 941                                 # We are ending a table
 942                                 $line = '</table>' . substr( $line , 2 );
 943                                 $last_tag = array_pop( $last_tag_history );
 944
 945                                 if ( !array_pop( $has_opened_tr ) ) {
 946                                         $line = "<tr><td></td></tr>{$line}";
 947                                 }
 948
 949                                 if ( array_pop( $tr_history ) ) {
 950                                         $line = "</tr>{$line}";
 951                                 }
 952
 953                                 if ( array_pop( $td_history ) ) {
 954                                         $line = "</{$last_tag}>{$line}";
 955                                 }
 956                                 array_pop( $tr_attributes );
 957                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 958                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 959                                 # Now we have a table row
 960                                 $line = preg_replace( '#^\|-+#', '', $line );
 961
 962                                 # Whats after the tag is now only attributes
 963                                 $attributes = $this->mStripState->unstripBoth( $line );
 964                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 965                                 array_pop( $tr_attributes );
 966                                 array_push( $tr_attributes, $attributes );
 967
 968                                 $line = '';
 969                                 $last_tag = array_pop( $last_tag_history );
 970                                 array_pop( $has_opened_tr );
 971                                 array_push( $has_opened_tr , true );
 972
 973                                 if ( array_pop( $tr_history ) ) {
 974                                         $line = '</tr>';
 975                                 }
 976
 977                                 if ( array_pop( $td_history ) ) {
 978                                         $line = "</{$last_tag}>{$line}";
 979                                 }
 980
 981                                 $outLine = $line;
 982                                 array_push( $tr_history , false );
 983                                 array_push( $td_history , false );
 984                                 array_push( $last_tag_history , '' );
 985                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 986                                 # This might be cell elements, td, th or captions
 987                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 988                                         $first_character = '+';
 989                                         $line = substr( $line , 1 );
 990                                 }
 991
 992                                 $line = substr( $line , 1 );
 993
 994                                 if ( $first_character === '!' ) {
 995                                         $line = str_replace( '!!' , '||' , $line );
 996                                 }
 997
 998                                 # Split up multiple cells on the same line.
 999                                 # FIXME : This can result in improper nesting of tags processed
1000                                 # by earlier parser steps, but should avoid splitting up eg
1001                                 # attribute values containing literal "||".
1002                                 $cells = StringUtils::explodeMarkup( '||' , $line );
1003
1004                                 $outLine = '';
1005
1006                                 # Loop through each table cell
1007                                 foreach ( $cells as $cell ) {
1008                                         $previous = '';
1009                                         if ( $first_character !== '+' ) {
1010                                                 $tr_after = array_pop( $tr_attributes );
1011                                                 if ( !array_pop( $tr_history ) ) {
1012                                                         $previous = "<tr{$tr_after}>\n";
1013                                                 }
1014                                                 array_push( $tr_history , true );
1015                                                 array_push( $tr_attributes , '' );
1016                                                 array_pop( $has_opened_tr );
1017                                                 array_push( $has_opened_tr , true );
1018                                         }
1019
1020                                         $last_tag = array_pop( $last_tag_history );
1021
1022                                         if ( array_pop( $td_history ) ) {
1023                                                 $previous = "</{$last_tag}>\n{$previous}";
1024                                         }
1025
1026                                         if ( $first_character === '|' ) {
1027                                                 $last_tag = 'td';
1028                                         } elseif ( $first_character === '!' ) {
1029                                                 $last_tag = 'th';
1030                                         } elseif ( $first_character === '+' ) {
1031                                                 $last_tag = 'caption';
1032                                         } else {
1033                                                 $last_tag = '';
1034                                         }
1035
1036                                         array_push( $last_tag_history , $last_tag );
1037
1038                                         # A cell could contain both parameters and data
1039                                         $cell_data = explode( '|' , $cell , 2 );
1040
1041                                         # Bug 553: Note that a '|' inside an invalid link should not
1042                                         # be mistaken as delimiting cell parameters
1043                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
1044                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
1045                                         } elseif ( count( $cell_data ) == 1 ) {
1046                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1047                                         } else {
1048                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1049                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
1050                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1051                                         }
1052
1053                                         $outLine .= $cell;
1054                                         array_push( $td_history , true );
1055                                 }
1056                         }
1057                         $out .= $outLine . "\n";
1058                 }
1059
1060                 # Closing open td, tr && table
1061                 while ( count( $td_history ) > 0 ) {
1062                         if ( array_pop( $td_history ) ) {
1063                                 $out .= "</td>\n";
1064                         }
1065                         if ( array_pop( $tr_history ) ) {
1066                                 $out .= "</tr>\n";
1067                         }
1068                         if ( !array_pop( $has_opened_tr ) ) {
1069                                 $out .= "<tr><td></td></tr>\n" ;
1070                         }
1071
1072                         $out .= "</table>\n";
1073                 }
1074
1075                 # Remove trailing line-ending (b/c)
1076                 if ( substr( $out, -1 ) === "\n" ) {
1077                         $out = substr( $out, 0, -1 );
1078                 }
1079
1080                 # special case: don't return empty table
1081                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1082                         $out = '';
1083                 }
1084
1085                 wfProfileOut( __METHOD__ );
1086
1087                 return $out;
1088         }
1089
1090         /**
1091          * Helper function for parse() that transforms wiki markup into
1092          * HTML. Only called for $mOutputType == self::OT_HTML.
1093          *
1094          * @private
1095          *
1096          * @param $text string
1097          * @param $isMain bool
1098          * @param $frame bool
1099          *
1100          * @return string
1101          */
1102         function internalParse( $text, $isMain = true, $frame = false ) {
1103                 wfProfileIn( __METHOD__ );
1104
1105                 $origText = $text;
1106
1107                 # Hook to suspend the parser in this state
1108                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1109                         wfProfileOut( __METHOD__ );
1110                         return $text ;
1111                 }
1112
1113                 # if $frame is provided, then use $frame for replacing any variables
1114                 if ( $frame ) {
1115                         # use frame depth to infer how include/noinclude tags should be handled
1116                         # depth=0 means this is the top-level document; otherwise it's an included document
1117                         if ( !$frame->depth ) {
1118                                 $flag = 0;
1119                         } else {
1120                                 $flag = Parser::PTD_FOR_INCLUSION;
1121                         }
1122                         $dom = $this->preprocessToDom( $text, $flag );
1123                         $text = $frame->expand( $dom );
1124                 } else {
1125                         # if $frame is not provided, then use old-style replaceVariables
1126                         $text = $this->replaceVariables( $text );
1127                 }
1128
1129                 wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1130                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1131                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1132
1133                 # Tables need to come after variable replacement for things to work
1134                 # properly; putting them before other transformations should keep
1135                 # exciting things like link expansions from showing up in surprising
1136                 # places.
1137                 $text = $this->doTableStuff( $text );
1138
1139                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1140
1141                 $text = $this->doDoubleUnderscore( $text );
1142
1143                 $text = $this->doHeadings( $text );
1144                 if ( $this->mOptions->getUseDynamicDates() ) {
1145                         $df = DateFormatter::getInstance();
1146                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1147                 }
1148                 $text = $this->replaceInternalLinks( $text );
1149                 $text = $this->doAllQuotes( $text );
1150                 $text = $this->replaceExternalLinks( $text );
1151
1152                 # replaceInternalLinks may sometimes leave behind
1153                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1154                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1155
1156                 $text = $this->doMagicLinks( $text );
1157                 $text = $this->formatHeadings( $text, $origText, $isMain );
1158
1159                 wfProfileOut( __METHOD__ );
1160                 return $text;
1161         }
1162
1163         /**
1164          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1165          * magic external links.
1166          *
1167          * DML
1168          * @private
1169          *
1170          * @param $text string
1171          *
1172          * @return string
1173          */
1174         function doMagicLinks( $text ) {
1175                 wfProfileIn( __METHOD__ );
1176                 $prots = wfUrlProtocolsWithoutProtRel();
1177                 $urlChar = self::EXT_LINK_URL_CLASS;
1178                 $text = preg_replace_callback(
1179                         '!(?:                           # Start cases
1180                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1181                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1182                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1183                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1184                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1185                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1186                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1187                                         [0-9Xx]                 # check digit
1188                                         \b)
1189                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1190                 wfProfileOut( __METHOD__ );
1191                 return $text;
1192         }
1193
1194         /**
1195          * @throws MWException
1196          * @param $m array
1197          * @return HTML|string
1198          */
1199         function magicLinkCallback( $m ) {
1200                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1201                         # Skip anchor
1202                         return $m[0];
1203                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1204                         # Skip HTML element
1205                         return $m[0];
1206                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1207                         # Free external link
1208                         return $this->makeFreeExternalLink( $m[0] );
1209                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1210                         # RFC or PMID
1211                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1212                                 $keyword = 'RFC';
1213                                 $urlmsg = 'rfcurl';
1214                                 $CssClass = 'mw-magiclink-rfc';
1215                                 $id = $m[4];
1216                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1217                                 $keyword = 'PMID';
1218                                 $urlmsg = 'pubmedurl';
1219                                 $CssClass = 'mw-magiclink-pmid';
1220                                 $id = $m[4];
1221                         } else {
1222                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1223                                         substr( $m[0], 0, 20 ) . '"' );
1224                         }
1225                         $url = wfMsgForContent( $urlmsg, $id );
1226                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1227                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1228                         # ISBN
1229                         $isbn = $m[5];
1230                         $num = strtr( $isbn, array(
1231                                 '-' => '',
1232                                 ' ' => '',
1233                                 'x' => 'X',
1234                         ));
1235                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1236                         return'<a href="' .
1237                                 htmlspecialchars( $titleObj->getLocalUrl() ) .
1238                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1239                 } else {
1240                         return $m[0];
1241                 }
1242         }
1243
1244         /**
1245          * Make a free external link, given a user-supplied URL
1246          *
1247          * @param $url string
1248          *
1249          * @return string HTML
1250          * @private
1251          */
1252         function makeFreeExternalLink( $url ) {
1253                 wfProfileIn( __METHOD__ );
1254
1255                 $trail = '';
1256
1257                 # The characters '<' and '>' (which were escaped by
1258                 # removeHTMLtags()) should not be included in
1259                 # URLs, per RFC 2396.
1260                 $m2 = array();
1261                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1262                         $trail = substr( $url, $m2[0][1] ) . $trail;
1263                         $url = substr( $url, 0, $m2[0][1] );
1264                 }
1265
1266                 # Move trailing punctuation to $trail
1267                 $sep = ',;\.:!?';
1268                 # If there is no left bracket, then consider right brackets fair game too
1269                 if ( strpos( $url, '(' ) === false ) {
1270                         $sep .= ')';
1271                 }
1272
1273                 $numSepChars = strspn( strrev( $url ), $sep );
1274                 if ( $numSepChars ) {
1275                         $trail = substr( $url, -$numSepChars ) . $trail;
1276                         $url = substr( $url, 0, -$numSepChars );
1277                 }
1278
1279                 $url = Sanitizer::cleanUrl( $url );
1280
1281                 # Is this an external image?
1282                 $text = $this->maybeMakeExternalImage( $url );
1283                 if ( $text === false ) {
1284                         # Not an image, make a link
1285                         $text = Linker::makeExternalLink( $url,
1286                                 $this->getConverterLanguage()->markNoConversion($url), true, 'free',
1287                                 $this->getExternalLinkAttribs( $url ) );
1288                         # Register it in the output object...
1289                         # Replace unnecessary URL escape codes with their equivalent characters
1290                         $pasteurized = self::replaceUnusualEscapes( $url );
1291                         $this->mOutput->addExternalLink( $pasteurized );
1292                 }
1293                 wfProfileOut( __METHOD__ );
1294                 return $text . $trail;
1295         }
1296
1297
1298         /**
1299          * Parse headers and return html
1300          *
1301          * @private
1302          *
1303          * @param $text string
1304          *
1305          * @return string
1306          */
1307         function doHeadings( $text ) {
1308                 wfProfileIn( __METHOD__ );
1309                 for ( $i = 6; $i >= 1; --$i ) {
1310                         $h = str_repeat( '=', $i );
1311                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1312                           "<h$i>\\1</h$i>", $text );
1313                 }
1314                 wfProfileOut( __METHOD__ );
1315                 return $text;
1316         }
1317
1318         /**
1319          * Replace single quotes with HTML markup
1320          * @private
1321          *
1322          * @param $text string
1323          *
1324          * @return string the altered text
1325          */
1326         function doAllQuotes( $text ) {
1327                 wfProfileIn( __METHOD__ );
1328                 $outtext = '';
1329                 $lines = StringUtils::explode( "\n", $text );
1330                 foreach ( $lines as $line ) {
1331                         $outtext .= $this->doQuotes( $line ) . "\n";
1332                 }
1333                 $outtext = substr( $outtext, 0,-1 );
1334                 wfProfileOut( __METHOD__ );
1335                 return $outtext;
1336         }
1337
1338         /**
1339          * Helper function for doAllQuotes()
1340          *
1341          * @param $text string
1342          *
1343          * @return string
1344          */
1345         public function doQuotes( $text ) {
1346                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1347                 if ( count( $arr ) == 1 ) {
1348                         return $text;
1349                 } else {
1350                         # First, do some preliminary work. This may shift some apostrophes from
1351                         # being mark-up to being text. It also counts the number of occurrences
1352                         # of bold and italics mark-ups.
1353                         $numbold = 0;
1354                         $numitalics = 0;
1355                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1356                                 if ( ( $i % 2 ) == 1 ) {
1357                                         # If there are ever four apostrophes, assume the first is supposed to
1358                                         # be text, and the remaining three constitute mark-up for bold text.
1359                                         if ( strlen( $arr[$i] ) == 4 ) {
1360                                                 $arr[$i-1] .= "'";
1361                                                 $arr[$i] = "'''";
1362                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1363                                                 # If there are more than 5 apostrophes in a row, assume they're all
1364                                                 # text except for the last 5.
1365                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1366                                                 $arr[$i] = "'''''";
1367                                         }
1368                                         # Count the number of occurrences of bold and italics mark-ups.
1369                                         # We are not counting sequences of five apostrophes.
1370                                         if ( strlen( $arr[$i] ) == 2 ) {
1371                                                 $numitalics++;
1372                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1373                                                 $numbold++;
1374                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1375                                                 $numitalics++;
1376                                                 $numbold++;
1377                                         }
1378                                 }
1379                         }
1380
1381                         # If there is an odd number of both bold and italics, it is likely
1382                         # that one of the bold ones was meant to be an apostrophe followed
1383                         # by italics. Which one we cannot know for certain, but it is more
1384                         # likely to be one that has a single-letter word before it.
1385                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1386                                 $i = 0;
1387                                 $firstsingleletterword = -1;
1388                                 $firstmultiletterword = -1;
1389                                 $firstspace = -1;
1390                                 foreach ( $arr as $r ) {
1391                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1392                                                 $x1 = substr( $arr[$i-1], -1 );
1393                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1394                                                 if ( $x1 === ' ' ) {
1395                                                         if ( $firstspace == -1 ) {
1396                                                                 $firstspace = $i;
1397                                                         }
1398                                                 } elseif ( $x2 === ' ') {
1399                                                         if ( $firstsingleletterword == -1 ) {
1400                                                                 $firstsingleletterword = $i;
1401                                                         }
1402                                                 } else {
1403                                                         if ( $firstmultiletterword == -1 ) {
1404                                                                 $firstmultiletterword = $i;
1405                                                         }
1406                                                 }
1407                                         }
1408                                         $i++;
1409                                 }
1410
1411                                 # If there is a single-letter word, use it!
1412                                 if ( $firstsingleletterword > -1 ) {
1413                                         $arr[$firstsingleletterword] = "''";
1414                                         $arr[$firstsingleletterword-1] .= "'";
1415                                 } elseif ( $firstmultiletterword > -1 ) {
1416                                         # If not, but there's a multi-letter word, use that one.
1417                                         $arr[$firstmultiletterword] = "''";
1418                                         $arr[$firstmultiletterword-1] .= "'";
1419                                 } elseif ( $firstspace > -1 ) {
1420                                         # ... otherwise use the first one that has neither.
1421                                         # (notice that it is possible for all three to be -1 if, for example,
1422                                         # there is only one pentuple-apostrophe in the line)
1423                                         $arr[$firstspace] = "''";
1424                                         $arr[$firstspace-1] .= "'";
1425                                 }
1426                         }
1427
1428                         # Now let's actually convert our apostrophic mush to HTML!
1429                         $output = '';
1430                         $buffer = '';
1431                         $state = '';
1432                         $i = 0;
1433                         foreach ( $arr as $r ) {
1434                                 if ( ( $i % 2 ) == 0 ) {
1435                                         if ( $state === 'both' ) {
1436                                                 $buffer .= $r;
1437                                         } else {
1438                                                 $output .= $r;
1439                                         }
1440                                 } else {
1441                                         if ( strlen( $r ) == 2 ) {
1442                                                 if ( $state === 'i' ) {
1443                                                         $output .= '</i>'; $state = '';
1444                                                 } elseif ( $state === 'bi' ) {
1445                                                         $output .= '</i>'; $state = 'b';
1446                                                 } elseif ( $state === 'ib' ) {
1447                                                         $output .= '</b></i><b>'; $state = 'b';
1448                                                 } elseif ( $state === 'both' ) {
1449                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1450                                                 } else { # $state can be 'b' or ''
1451                                                         $output .= '<i>'; $state .= 'i';
1452                                                 }
1453                                         } elseif ( strlen( $r ) == 3 ) {
1454                                                 if ( $state === 'b' ) {
1455                                                         $output .= '</b>'; $state = '';
1456                                                 } elseif ( $state === 'bi' ) {
1457                                                         $output .= '</i></b><i>'; $state = 'i';
1458                                                 } elseif ( $state === 'ib' ) {
1459                                                         $output .= '</b>'; $state = 'i';
1460                                                 } elseif ( $state === 'both' ) {
1461                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1462                                                 } else { # $state can be 'i' or ''
1463                                                         $output .= '<b>'; $state .= 'b';
1464                                                 }
1465                                         } elseif ( strlen( $r ) == 5 ) {
1466                                                 if ( $state === 'b' ) {
1467                                                         $output .= '</b><i>'; $state = 'i';
1468                                                 } elseif ( $state === 'i' ) {
1469                                                         $output .= '</i><b>'; $state = 'b';
1470                                                 } elseif ( $state === 'bi' ) {
1471                                                         $output .= '</i></b>'; $state = '';
1472                                                 } elseif ( $state === 'ib' ) {
1473                                                         $output .= '</b></i>'; $state = '';
1474                                                 } elseif ( $state === 'both' ) {
1475                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1476                                                 } else { # ($state == '')
1477                                                         $buffer = ''; $state = 'both';
1478                                                 }
1479                                         }
1480                                 }
1481                                 $i++;
1482                         }
1483                         # Now close all remaining tags.  Notice that the order is important.
1484                         if ( $state === 'b' || $state === 'ib' ) {
1485                                 $output .= '</b>';
1486                         }
1487                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1488                                 $output .= '</i>';
1489                         }
1490                         if ( $state === 'bi' ) {
1491                                 $output .= '</b>';
1492                         }
1493                         # There might be lonely ''''', so make sure we have a buffer
1494                         if ( $state === 'both' && $buffer ) {
1495                                 $output .= '<b><i>'.$buffer.'</i></b>';
1496                         }
1497                         return $output;
1498                 }
1499         }
1500
1501         /**
1502          * Replace external links (REL)
1503          *
1504          * Note: this is all very hackish and the order of execution matters a lot.
1505          * Make sure to run maintenance/parserTests.php if you change this code.
1506          *
1507          * @private
1508          *
1509          * @param $text string
1510          *
1511          * @return string
1512          */
1513         function replaceExternalLinks( $text ) {
1514                 wfProfileIn( __METHOD__ );
1515
1516                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1517                 $s = array_shift( $bits );
1518
1519                 $i = 0;
1520                 while ( $i<count( $bits ) ) {
1521                         $url = $bits[$i++];
1522                         $protocol = $bits[$i++];
1523                         $text = $bits[$i++];
1524                         $trail = $bits[$i++];
1525
1526                         # The characters '<' and '>' (which were escaped by
1527                         # removeHTMLtags()) should not be included in
1528                         # URLs, per RFC 2396.
1529                         $m2 = array();
1530                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1531                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1532                                 $url = substr( $url, 0, $m2[0][1] );
1533                         }
1534
1535                         # If the link text is an image URL, replace it with an <img> tag
1536                         # This happened by accident in the original parser, but some people used it extensively
1537                         $img = $this->maybeMakeExternalImage( $text );
1538                         if ( $img !== false ) {
1539                                 $text = $img;
1540                         }
1541
1542                         $dtrail = '';
1543
1544                         # Set linktype for CSS - if URL==text, link is essentially free
1545                         $linktype = ( $text === $url ) ? 'free' : 'text';
1546
1547                         # No link text, e.g. [http://domain.tld/some.link]
1548                         if ( $text == '' ) {
1549                                 # Autonumber
1550                                 $langObj = $this->getTargetLanguage();
1551                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1552                                 $linktype = 'autonumber';
1553                         } else {
1554                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1555                                 # Check for trail
1556                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1557                         }
1558
1559                         $text = $this->getConverterLanguage()->markNoConversion( $text );
1560
1561                         $url = Sanitizer::cleanUrl( $url );
1562
1563                         # Use the encoded URL
1564                         # This means that users can paste URLs directly into the text
1565                         # Funny characters like ö aren't valid in URLs anyway
1566                         # This was changed in August 2004
1567                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1568                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1569
1570                         # Register link in the output object.
1571                         # Replace unnecessary URL escape codes with the referenced character
1572                         # This prevents spammers from hiding links from the filters
1573                         $pasteurized = self::replaceUnusualEscapes( $url );
1574                         $this->mOutput->addExternalLink( $pasteurized );
1575                 }
1576
1577                 wfProfileOut( __METHOD__ );
1578                 return $s;
1579         }
1580
1581         /**
1582          * Get an associative array of additional HTML attributes appropriate for a
1583          * particular external link.  This currently may include rel => nofollow
1584          * (depending on configuration, namespace, and the URL's domain) and/or a
1585          * target attribute (depending on configuration).
1586          *
1587          * @param $url String|bool optional URL, to extract the domain from for rel =>
1588          *   nofollow if appropriate
1589          * @return Array associative array of HTML attributes
1590          */
1591         function getExternalLinkAttribs( $url = false ) {
1592                 $attribs = array();
1593                 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1594                 $ns = $this->mTitle->getNamespace();
1595                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1596                                 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1597                 {
1598                         $attribs['rel'] = 'nofollow';
1599                 }
1600                 if ( $this->mOptions->getExternalLinkTarget() ) {
1601                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1602                 }
1603                 return $attribs;
1604         }
1605
1606         /**
1607          * Replace unusual URL escape codes with their equivalent characters
1608          *
1609          * @param $url String
1610          * @return String
1611          *
1612          * @todo  This can merge genuinely required bits in the path or query string,
1613          *        breaking legit URLs. A proper fix would treat the various parts of
1614          *        the URL differently; as a workaround, just use the output for
1615          *        statistical records, not for actual linking/output.
1616          */
1617         static function replaceUnusualEscapes( $url ) {
1618                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1619                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1620         }
1621
1622         /**
1623          * Callback function used in replaceUnusualEscapes().
1624          * Replaces unusual URL escape codes with their equivalent character
1625          *
1626          * @param $matches array
1627          *
1628          * @return string
1629          */
1630         private static function replaceUnusualEscapesCallback( $matches ) {
1631                 $char = urldecode( $matches[0] );
1632                 $ord = ord( $char );
1633                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1634                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1635                         # No, shouldn't be escaped
1636                         return $char;
1637                 } else {
1638                         # Yes, leave it escaped
1639                         return $matches[0];
1640                 }
1641         }
1642
1643         /**
1644          * make an image if it's allowed, either through the global
1645          * option, through the exception, or through the on-wiki whitelist
1646          * @private
1647          *
1648          * $param $url string
1649          *
1650          * @return string
1651          */
1652         function maybeMakeExternalImage( $url ) {
1653                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1654                 $imagesexception = !empty( $imagesfrom );
1655                 $text = false;
1656                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1657                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1658                         $imagematch = false;
1659                         foreach ( $imagesfrom as $match ) {
1660                                 if ( strpos( $url, $match ) === 0 ) {
1661                                         $imagematch = true;
1662                                         break;
1663                                 }
1664                         }
1665                 } elseif ( $imagesexception ) {
1666                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1667                 } else {
1668                         $imagematch = false;
1669                 }
1670                 if ( $this->mOptions->getAllowExternalImages()
1671                          || ( $imagesexception && $imagematch ) ) {
1672                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1673                                 # Image found
1674                                 $text = Linker::makeExternalImage( $url );
1675                         }
1676                 }
1677                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1678                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1679                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1680                         foreach ( $whitelist as $entry ) {
1681                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1682                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1683                                         continue;
1684                                 }
1685                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1686                                         # Image matches a whitelist entry
1687                                         $text = Linker::makeExternalImage( $url );
1688                                         break;
1689                                 }
1690                         }
1691                 }
1692                 return $text;
1693         }
1694
1695         /**
1696          * Process [[ ]] wikilinks
1697          *
1698          * @param $s string
1699          *
1700          * @return String: processed text
1701          *
1702          * @private
1703          */
1704         function replaceInternalLinks( $s ) {
1705                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1706                 return $s;
1707         }
1708
1709         /**
1710          * Process [[ ]] wikilinks (RIL)
1711          * @return LinkHolderArray
1712          *
1713          * @private
1714          */
1715         function replaceInternalLinks2( &$s ) {
1716                 wfProfileIn( __METHOD__ );
1717
1718                 wfProfileIn( __METHOD__.'-setup' );
1719                 static $tc = FALSE, $e1, $e1_img;
1720                 # the % is needed to support urlencoded titles as well
1721                 if ( !$tc ) {
1722                         $tc = Title::legalChars() . '#%';
1723                         # Match a link having the form [[namespace:link|alternate]]trail
1724                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1725                         # Match cases where there is no "]]", which might still be images
1726                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1727                 }
1728
1729                 $holders = new LinkHolderArray( $this );
1730
1731                 # split the entire text string on occurrences of [[
1732                 $a = StringUtils::explode( '[[', ' ' . $s );
1733                 # get the first element (all text up to first [[), and remove the space we added
1734                 $s = $a->current();
1735                 $a->next();
1736                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1737                 $s = substr( $s, 1 );
1738
1739                 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1740                 $e2 = null;
1741                 if ( $useLinkPrefixExtension ) {
1742                         # Match the end of a line for a word that's not followed by whitespace,
1743                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1744                         $e2 = wfMsgForContent( 'linkprefix' );
1745                 }
1746
1747                 if ( is_null( $this->mTitle ) ) {
1748                         wfProfileOut( __METHOD__.'-setup' );
1749                         wfProfileOut( __METHOD__ );
1750                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1751                 }
1752                 $nottalk = !$this->mTitle->isTalkPage();
1753
1754                 if ( $useLinkPrefixExtension ) {
1755                         $m = array();
1756                         if ( preg_match( $e2, $s, $m ) ) {
1757                                 $first_prefix = $m[2];
1758                         } else {
1759                                 $first_prefix = false;
1760                         }
1761                 } else {
1762                         $prefix = '';
1763                 }
1764
1765                 if ( $this->getConverterLanguage()->hasVariants() ) {
1766                         $selflink = $this->getConverterLanguage()->autoConvertToAllVariants(
1767                                 $this->mTitle->getPrefixedText() );
1768                 } else {
1769                         $selflink = array( $this->mTitle->getPrefixedText() );
1770                 }
1771                 $useSubpages = $this->areSubpagesAllowed();
1772                 wfProfileOut( __METHOD__.'-setup' );
1773
1774                 # Loop for each link
1775                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1776                         # Check for excessive memory usage
1777                         if ( $holders->isBig() ) {
1778                                 # Too big
1779                                 # Do the existence check, replace the link holders and clear the array
1780                                 $holders->replace( $s );
1781                                 $holders->clear();
1782                         }
1783
1784                         if ( $useLinkPrefixExtension ) {
1785                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1786                                 if ( preg_match( $e2, $s, $m ) ) {
1787                                         $prefix = $m[2];
1788                                         $s = $m[1];
1789                                 } else {
1790                                         $prefix='';
1791                                 }
1792                                 # first link
1793                                 if ( $first_prefix ) {
1794                                         $prefix = $first_prefix;
1795                                         $first_prefix = false;
1796                                 }
1797                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1798                         }
1799
1800                         $might_be_img = false;
1801
1802                         wfProfileIn( __METHOD__."-e1" );
1803                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1804                                 $text = $m[2];
1805                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1806                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1807                                 # the real problem is with the $e1 regex
1808                                 # See bug 1300.
1809                                 #
1810                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1811                                 # and no image is in sight. See bug 2095.
1812                                 #
1813                                 if ( $text !== '' &&
1814                                         substr( $m[3], 0, 1 ) === ']' &&
1815                                         strpos( $text, '[' ) !== false
1816                                 )
1817                                 {
1818                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1819                                         $m[3] = substr( $m[3], 1 );
1820                                 }
1821                                 # fix up urlencoded title texts
1822                                 if ( strpos( $m[1], '%' ) !== false ) {
1823                                         # Should anchors '#' also be rejected?
1824                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1825                                 }
1826                                 $trail = $m[3];
1827                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1828                                 $might_be_img = true;
1829                                 $text = $m[2];
1830                                 if ( strpos( $m[1], '%' ) !== false ) {
1831                                         $m[1] = rawurldecode( $m[1] );
1832                                 }
1833                                 $trail = "";
1834                         } else { # Invalid form; output directly
1835                                 $s .= $prefix . '[[' . $line ;
1836                                 wfProfileOut( __METHOD__."-e1" );
1837                                 continue;
1838                         }
1839                         wfProfileOut( __METHOD__."-e1" );
1840                         wfProfileIn( __METHOD__."-misc" );
1841
1842                         # Don't allow internal links to pages containing
1843                         # PROTO: where PROTO is a valid URL protocol; these
1844                         # should be external links.
1845                         if ( preg_match( '/^(?:' . $this->mUrlProtocols . ')/', $m[1] ) ) {
1846                                 $s .= $prefix . '[[' . $line ;
1847                                 wfProfileOut( __METHOD__."-misc" );
1848                                 continue;
1849                         }
1850
1851                         # Make subpage if necessary
1852                         if ( $useSubpages ) {
1853                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1854                         } else {
1855                                 $link = $m[1];
1856                         }
1857
1858                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1859                         if ( !$noforce ) {
1860                                 # Strip off leading ':'
1861                                 $link = substr( $link, 1 );
1862                         }
1863
1864                         wfProfileOut( __METHOD__."-misc" );
1865                         wfProfileIn( __METHOD__."-title" );
1866                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1867                         if ( $nt === null ) {
1868                                 $s .= $prefix . '[[' . $line;
1869                                 wfProfileOut( __METHOD__."-title" );
1870                                 continue;
1871                         }
1872
1873                         $ns = $nt->getNamespace();
1874                         $iw = $nt->getInterWiki();
1875                         wfProfileOut( __METHOD__."-title" );
1876
1877                         if ( $might_be_img ) { # if this is actually an invalid link
1878                                 wfProfileIn( __METHOD__."-might_be_img" );
1879                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1880                                         $found = false;
1881                                         while ( true ) {
1882                                                 # look at the next 'line' to see if we can close it there
1883                                                 $a->next();
1884                                                 $next_line = $a->current();
1885                                                 if ( $next_line === false || $next_line === null ) {
1886                                                         break;
1887                                                 }
1888                                                 $m = explode( ']]', $next_line, 3 );
1889                                                 if ( count( $m ) == 3 ) {
1890                                                         # the first ]] closes the inner link, the second the image
1891                                                         $found = true;
1892                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1893                                                         $trail = $m[2];
1894                                                         break;
1895                                                 } elseif ( count( $m ) == 2 ) {
1896                                                         # if there's exactly one ]] that's fine, we'll keep looking
1897                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1898                                                 } else {
1899                                                         # if $next_line is invalid too, we need look no further
1900                                                         $text .= '[[' . $next_line;
1901                                                         break;
1902                                                 }
1903                                         }
1904                                         if ( !$found ) {
1905                                                 # we couldn't find the end of this imageLink, so output it raw
1906                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1907                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1908                                                 $s .= "{$prefix}[[$link|$text";
1909                                                 # note: no $trail, because without an end, there *is* no trail
1910                                                 wfProfileOut( __METHOD__."-might_be_img" );
1911                                                 continue;
1912                                         }
1913                                 } else { # it's not an image, so output it raw
1914                                         $s .= "{$prefix}[[$link|$text";
1915                                         # note: no $trail, because without an end, there *is* no trail
1916                                         wfProfileOut( __METHOD__."-might_be_img" );
1917                                         continue;
1918                                 }
1919                                 wfProfileOut( __METHOD__."-might_be_img" );
1920                         }
1921
1922                         $wasblank = ( $text  == '' );
1923                         if ( $wasblank ) {
1924                                 $text = $link;
1925                         } else {
1926                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1927                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1928                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1929                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1930                                 $text = $this->doQuotes( $text );
1931                         }
1932
1933                         # Link not escaped by : , create the various objects
1934                         if ( $noforce ) {
1935                                 # Interwikis
1936                                 wfProfileIn( __METHOD__."-interwiki" );
1937                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && Language::fetchLanguageName( $iw, null, 'mw' ) ) {
1938                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1939                                         $s = rtrim( $s . $prefix );
1940                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1941                                         wfProfileOut( __METHOD__."-interwiki" );
1942                                         continue;
1943                                 }
1944                                 wfProfileOut( __METHOD__."-interwiki" );
1945
1946                                 if ( $ns == NS_FILE ) {
1947                                         wfProfileIn( __METHOD__."-image" );
1948                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1949                                                 if ( $wasblank ) {
1950                                                         # if no parameters were passed, $text
1951                                                         # becomes something like "File:Foo.png",
1952                                                         # which we don't want to pass on to the
1953                                                         # image generator
1954                                                         $text = '';
1955                                                 } else {
1956                                                         # recursively parse links inside the image caption
1957                                                         # actually, this will parse them in any other parameters, too,
1958                                                         # but it might be hard to fix that, and it doesn't matter ATM
1959                                                         $text = $this->replaceExternalLinks( $text );
1960                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1961                                                 }
1962                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1963                                                 $s .= $prefix . $this->armorLinks(
1964                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1965                                         } else {
1966                                                 $s .= $prefix . $trail;
1967                                         }
1968                                         wfProfileOut( __METHOD__."-image" );
1969                                         continue;
1970                                 }
1971
1972                                 if ( $ns == NS_CATEGORY ) {
1973                                         wfProfileIn( __METHOD__."-category" );
1974                                         $s = rtrim( $s . "\n" ); # bug 87
1975
1976                                         if ( $wasblank ) {
1977                                                 $sortkey = $this->getDefaultSort();
1978                                         } else {
1979                                                 $sortkey = $text;
1980                                         }
1981                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1982                                         $sortkey = str_replace( "\n", '', $sortkey );
1983                                         $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
1984                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1985
1986                                         /**
1987                                          * Strip the whitespace Category links produce, see bug 87
1988                                          * @todo We might want to use trim($tmp, "\n") here.
1989                                          */
1990                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1991
1992                                         wfProfileOut( __METHOD__."-category" );
1993                                         continue;
1994                                 }
1995                         }
1996
1997                         # Self-link checking
1998                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1999                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2000                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2001                                         continue;
2002                                 }
2003                         }
2004
2005                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2006                         # @todo FIXME: Should do batch file existence checks, see comment below
2007                         if ( $ns == NS_MEDIA ) {
2008                                 wfProfileIn( __METHOD__."-media" );
2009                                 # Give extensions a chance to select the file revision for us
2010                                 $options = array();
2011                                 $descQuery = false;
2012                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2013                                         array( $this, $nt, &$options, &$descQuery ) );
2014                                 # Fetch and register the file (file title may be different via hooks)
2015                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2016                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2017                                 $s .= $prefix . $this->armorLinks(
2018                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2019                                 wfProfileOut( __METHOD__."-media" );
2020                                 continue;
2021                         }
2022
2023                         wfProfileIn( __METHOD__."-always_known" );
2024                         # Some titles, such as valid special pages or files in foreign repos, should
2025                         # be shown as bluelinks even though they're not included in the page table
2026                         #
2027                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2028                         # batch file existence checks for NS_FILE and NS_MEDIA
2029                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2030                                 $this->mOutput->addLink( $nt );
2031                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2032                         } else {
2033                                 # Links will be added to the output link list after checking
2034                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2035                         }
2036                         wfProfileOut( __METHOD__."-always_known" );
2037                 }
2038                 wfProfileOut( __METHOD__ );
2039                 return $holders;
2040         }
2041
2042         /**
2043          * Render a forced-blue link inline; protect against double expansion of
2044          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2045          * Since this little disaster has to split off the trail text to avoid
2046          * breaking URLs in the following text without breaking trails on the
2047          * wiki links, it's been made into a horrible function.
2048          *
2049          * @param $nt Title
2050          * @param $text String
2051          * @param $query Array or String
2052          * @param $trail String
2053          * @param $prefix String
2054          * @return String: HTML-wikitext mix oh yuck
2055          */
2056         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2057                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2058
2059                 if ( is_string( $query ) ) {
2060                         $query = wfCgiToArray( $query );
2061                 }
2062                 if ( $text == '' ) {
2063                         $text = htmlspecialchars( $nt->getPrefixedText() );
2064                 }
2065
2066                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2067
2068                 return $this->armorLinks( $link ) . $trail;
2069         }
2070
2071         /**
2072          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2073          * going to go through further parsing steps before inline URL expansion.
2074          *
2075          * Not needed quite as much as it used to be since free links are a bit
2076          * more sensible these days. But bracketed links are still an issue.
2077          *
2078          * @param $text String: more-or-less HTML
2079          * @return String: less-or-more HTML with NOPARSE bits
2080          */
2081         function armorLinks( $text ) {
2082                 return preg_replace( '/\b(' . $this->mUrlProtocols . ')/',
2083                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2084         }
2085
2086         /**
2087          * Return true if subpage links should be expanded on this page.
2088          * @return Boolean
2089          */
2090         function areSubpagesAllowed() {
2091                 # Some namespaces don't allow subpages
2092                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2093         }
2094
2095         /**
2096          * Handle link to subpage if necessary
2097          *
2098          * @param $target String: the source of the link
2099          * @param &$text String: the link text, modified as necessary
2100          * @return string the full name of the link
2101          * @private
2102          */
2103         function maybeDoSubpageLink( $target, &$text ) {
2104                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2105         }
2106
2107         /**#@+
2108          * Used by doBlockLevels()
2109          * @private
2110          *
2111          * @return string
2112          */
2113         function closeParagraph() {
2114                 $result = '';
2115                 if ( $this->mLastSection != '' ) {
2116                         $result = '</' . $this->mLastSection  . ">\n";
2117                 }
2118                 $this->mInPre = false;
2119                 $this->mLastSection = '';
2120                 return $result;
2121         }
2122
2123         /**
2124          * getCommon() returns the length of the longest common substring
2125          * of both arguments, starting at the beginning of both.
2126          * @private
2127          *
2128          * @param $st1 string
2129          * @param $st2 string
2130          *
2131          * @return int
2132          */
2133         function getCommon( $st1, $st2 ) {
2134                 $fl = strlen( $st1 );
2135                 $shorter = strlen( $st2 );
2136                 if ( $fl < $shorter ) {
2137                         $shorter = $fl;
2138                 }
2139
2140                 for ( $i = 0; $i < $shorter; ++$i ) {
2141                         if ( $st1[$i] != $st2[$i] ) {
2142                                 break;
2143                         }
2144                 }
2145                 return $i;
2146         }
2147
2148         /**
2149          * These next three functions open, continue, and close the list
2150          * element appropriate to the prefix character passed into them.
2151          * @private
2152          *
2153          * @param $char string
2154          *
2155          * @return string
2156          */
2157         function openList( $char ) {
2158                 $result = $this->closeParagraph();
2159
2160                 if ( '*' === $char ) {
2161                         $result .= '<ul><li>';
2162                 } elseif ( '#' === $char ) {
2163                         $result .= '<ol><li>';
2164                 } elseif ( ':' === $char ) {
2165                         $result .= '<dl><dd>';
2166                 } elseif ( ';' === $char ) {
2167                         $result .= '<dl><dt>';
2168                         $this->mDTopen = true;
2169                 } else {
2170                         $result = '<!-- ERR 1 -->';
2171                 }
2172
2173                 return $result;
2174         }
2175
2176         /**
2177          * TODO: document
2178          * @param $char String
2179          * @private
2180          *
2181          * @return string
2182          */
2183         function nextItem( $char ) {
2184                 if ( '*' === $char || '#' === $char ) {
2185                         return '</li><li>';
2186                 } elseif ( ':' === $char || ';' === $char ) {
2187                         $close = '</dd>';
2188                         if ( $this->mDTopen ) {
2189                                 $close = '</dt>';
2190                         }
2191                         if ( ';' === $char ) {
2192                                 $this->mDTopen = true;
2193                                 return $close . '<dt>';
2194                         } else {
2195                                 $this->mDTopen = false;
2196                                 return $close . '<dd>';
2197                         }
2198                 }
2199                 return '<!-- ERR 2 -->';
2200         }
2201
2202         /**
2203          * TODO: document
2204          * @param $char String
2205          * @private
2206          *
2207          * @return string
2208          */
2209         function closeList( $char ) {
2210                 if ( '*' === $char ) {
2211                         $text = '</li></ul>';
2212                 } elseif ( '#' === $char ) {
2213                         $text = '</li></ol>';
2214                 } elseif ( ':' === $char ) {
2215                         if ( $this->mDTopen ) {
2216                                 $this->mDTopen = false;
2217                                 $text = '</dt></dl>';
2218                         } else {
2219                                 $text = '</dd></dl>';
2220                         }
2221                 } else {
2222                         return '<!-- ERR 3 -->';
2223                 }
2224                 return $text."\n";
2225         }
2226         /**#@-*/
2227
2228         /**
2229          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2230          *
2231          * @param $text String
2232          * @param $linestart Boolean: whether or not this is at the start of a line.
2233          * @private
2234          * @return string the lists rendered as HTML
2235          */
2236         function doBlockLevels( $text, $linestart ) {
2237                 wfProfileIn( __METHOD__ );
2238
2239                 # Parsing through the text line by line.  The main thing
2240                 # happening here is handling of block-level elements p, pre,
2241                 # and making lists from lines starting with * # : etc.
2242                 #
2243                 $textLines = StringUtils::explode( "\n", $text );
2244
2245                 $lastPrefix = $output = '';
2246                 $this->mDTopen = $inBlockElem = false;
2247                 $prefixLength = 0;
2248                 $paragraphStack = false;
2249
2250                 foreach ( $textLines as $oLine ) {
2251                         # Fix up $linestart
2252                         if ( !$linestart ) {
2253                                 $output .= $oLine;
2254                                 $linestart = true;
2255                                 continue;
2256                         }
2257                         # * = ul
2258                         # # = ol
2259                         # ; = dt
2260                         # : = dd
2261
2262                         $lastPrefixLength = strlen( $lastPrefix );
2263                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2264                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2265                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2266                         if ( !$this->mInPre ) {
2267                                 # Multiple prefixes may abut each other for nested lists.
2268                                 $prefixLength = strspn( $oLine, '*#:;' );
2269                                 $prefix = substr( $oLine, 0, $prefixLength );
2270
2271                                 # eh?
2272                                 # ; and : are both from definition-lists, so they're equivalent
2273                                 #  for the purposes of determining whether or not we need to open/close
2274                                 #  elements.
2275                                 $prefix2 = str_replace( ';', ':', $prefix );
2276                                 $t = substr( $oLine, $prefixLength );
2277                                 $this->mInPre = (bool)$preOpenMatch;
2278                         } else {
2279                                 # Don't interpret any other prefixes in preformatted text
2280                                 $prefixLength = 0;
2281                                 $prefix = $prefix2 = '';
2282                                 $t = $oLine;
2283                         }
2284
2285                         # List generation
2286                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2287                                 # Same as the last item, so no need to deal with nesting or opening stuff
2288                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2289                                 $paragraphStack = false;
2290
2291                                 if ( substr( $prefix, -1 ) === ';') {
2292                                         # The one nasty exception: definition lists work like this:
2293                                         # ; title : definition text
2294                                         # So we check for : in the remainder text to split up the
2295                                         # title and definition, without b0rking links.
2296                                         $term = $t2 = '';
2297                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2298                                                 $t = $t2;
2299                                                 $output .= $term . $this->nextItem( ':' );
2300                                         }
2301                                 }
2302                         } elseif ( $prefixLength || $lastPrefixLength ) {
2303                                 # We need to open or close prefixes, or both.
2304
2305                                 # Either open or close a level...
2306                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2307                                 $paragraphStack = false;
2308
2309                                 # Close all the prefixes which aren't shared.
2310                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2311                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2312                                         --$lastPrefixLength;
2313                                 }
2314
2315                                 # Continue the current prefix if appropriate.
2316                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2317                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2318                                 }
2319
2320                                 # Open prefixes where appropriate.
2321                                 while ( $prefixLength > $commonPrefixLength ) {
2322                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2323                                         $output .= $this->openList( $char );
2324
2325                                         if ( ';' === $char ) {
2326                                                 # @todo FIXME: This is dupe of code above
2327                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2328                                                         $t = $t2;
2329                                                         $output .= $term . $this->nextItem( ':' );
2330                                                 }
2331                                         }
2332                                         ++$commonPrefixLength;
2333                                 }
2334                                 $lastPrefix = $prefix2;
2335                         }
2336
2337                         # If we have no prefixes, go to paragraph mode.
2338                         if ( 0 == $prefixLength ) {
2339                                 wfProfileIn( __METHOD__."-paragraph" );
2340                                 # No prefix (not in list)--go to paragraph mode
2341                                 # XXX: use a stack for nestable elements like span, table and div
2342                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2343                                 $closematch = preg_match(
2344                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2345                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2346                                 if ( $openmatch or $closematch ) {
2347                                         $paragraphStack = false;
2348                                         # TODO bug 5718: paragraph closed
2349                                         $output .= $this->closeParagraph();
2350                                         if ( $preOpenMatch and !$preCloseMatch ) {
2351                                                 $this->mInPre = true;
2352                                         }
2353                                         $inBlockElem = !$closematch;
2354                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2355                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2356                                                 # pre
2357                                                 if ( $this->mLastSection !== 'pre' ) {
2358                                                         $paragraphStack = false;
2359                                                         $output .= $this->closeParagraph().'<pre>';
2360                                                         $this->mLastSection = 'pre';
2361                                                 }
2362                                                 $t = substr( $t, 1 );
2363                                         } else {
2364                                                 # paragraph
2365                                                 if ( trim( $t ) === '' ) {
2366                                                         if ( $paragraphStack ) {
2367                                                                 $output .= $paragraphStack.'<br />';
2368                                                                 $paragraphStack = false;
2369                                                                 $this->mLastSection = 'p';
2370                                                         } else {
2371                                                                 if ( $this->mLastSection !== 'p' ) {
2372                                                                         $output .= $this->closeParagraph();
2373                                                                         $this->mLastSection = '';
2374                                                                         $paragraphStack = '<p>';
2375                                                                 } else {
2376                                                                         $paragraphStack = '</p><p>';
2377                                                                 }
2378                                                         }
2379                                                 } else {
2380                                                         if ( $paragraphStack ) {
2381                                                                 $output .= $paragraphStack;
2382                                                                 $paragraphStack = false;
2383                                                                 $this->mLastSection = 'p';
2384                                                         } elseif ( $this->mLastSection !== 'p' ) {
2385                                                                 $output .= $this->closeParagraph().'<p>';
2386                                                                 $this->mLastSection = 'p';
2387                                                         }
2388                                                 }
2389                                         }
2390                                 }
2391                                 wfProfileOut( __METHOD__."-paragraph" );
2392                         }
2393                         # somewhere above we forget to get out of pre block (bug 785)
2394                         if ( $preCloseMatch && $this->mInPre ) {
2395                                 $this->mInPre = false;
2396                         }
2397                         if ( $paragraphStack === false ) {
2398                                 $output .= $t."\n";
2399                         }
2400                 }
2401                 while ( $prefixLength ) {
2402                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2403                         --$prefixLength;
2404                 }
2405                 if ( $this->mLastSection != '' ) {
2406                         $output .= '</' . $this->mLastSection . '>';
2407                         $this->mLastSection = '';
2408                 }
2409
2410                 wfProfileOut( __METHOD__ );
2411                 return $output;
2412         }
2413
2414         /**
2415          * Split up a string on ':', ignoring any occurrences inside tags
2416          * to prevent illegal overlapping.
2417          *
2418          * @param $str String the string to split
2419          * @param &$before String set to everything before the ':'
2420          * @param &$after String set to everything after the ':'
2421          * @return String the position of the ':', or false if none found
2422          */
2423         function findColonNoLinks( $str, &$before, &$after ) {
2424                 wfProfileIn( __METHOD__ );
2425
2426                 $pos = strpos( $str, ':' );
2427                 if ( $pos === false ) {
2428                         # Nothing to find!
2429                         wfProfileOut( __METHOD__ );
2430                         return false;
2431                 }
2432
2433                 $lt = strpos( $str, '<' );
2434                 if ( $lt === false || $lt > $pos ) {
2435                         # Easy; no tag nesting to worry about
2436                         $before = substr( $str, 0, $pos );
2437                         $after = substr( $str, $pos+1 );
2438                         wfProfileOut( __METHOD__ );
2439                         return $pos;
2440                 }
2441
2442                 # Ugly state machine to walk through avoiding tags.
2443                 $state = self::COLON_STATE_TEXT;
2444                 $stack = 0;
2445                 $len = strlen( $str );
2446                 for( $i = 0; $i < $len; $i++ ) {
2447                         $c = $str[$i];
2448
2449                         switch( $state ) {
2450                         # (Using the number is a performance hack for common cases)
2451                         case 0: # self::COLON_STATE_TEXT:
2452                                 switch( $c ) {
2453                                 case "<":
2454                                         # Could be either a <start> tag or an </end> tag
2455                                         $state = self::COLON_STATE_TAGSTART;
2456                                         break;
2457                                 case ":":
2458                                         if ( $stack == 0 ) {
2459                                                 # We found it!
2460                                                 $before = substr( $str, 0, $i );
2461                                                 $after = substr( $str, $i + 1 );
2462                                                 wfProfileOut( __METHOD__ );
2463                                                 return $i;
2464                                         }
2465                                         # Embedded in a tag; don't break it.
2466                                         break;
2467                                 default:
2468                                         # Skip ahead looking for something interesting
2469                                         $colon = strpos( $str, ':', $i );
2470                                         if ( $colon === false ) {
2471                                                 # Nothing else interesting
2472                                                 wfProfileOut( __METHOD__ );
2473                                                 return false;
2474                                         }
2475                                         $lt = strpos( $str, '<', $i );
2476                                         if ( $stack === 0 ) {
2477                                                 if ( $lt === false || $colon < $lt ) {
2478                                                         # We found it!
2479                                                         $before = substr( $str, 0, $colon );
2480                                                         $after = substr( $str, $colon + 1 );
2481                                                         wfProfileOut( __METHOD__ );
2482                                                         return $i;
2483                                                 }
2484                                         }
2485                                         if ( $lt === false ) {
2486                                                 # Nothing else interesting to find; abort!
2487                                                 # We're nested, but there's no close tags left. Abort!
2488                                                 break 2;
2489                                         }
2490                                         # Skip ahead to next tag start
2491                                         $i = $lt;
2492                                         $state = self::COLON_STATE_TAGSTART;
2493                                 }
2494                                 break;
2495                         case 1: # self::COLON_STATE_TAG:
2496                                 # In a <tag>
2497                                 switch( $c ) {
2498                                 case ">":
2499                                         $stack++;
2500                                         $state = self::COLON_STATE_TEXT;
2501                                         break;
2502                                 case "/":
2503                                         # Slash may be followed by >?
2504                                         $state = self::COLON_STATE_TAGSLASH;
2505                                         break;
2506                                 default:
2507                                         # ignore
2508                                 }
2509                                 break;
2510                         case 2: # self::COLON_STATE_TAGSTART:
2511                                 switch( $c ) {
2512                                 case "/":
2513                                         $state = self::COLON_STATE_CLOSETAG;
2514                                         break;
2515                                 case "!":
2516                                         $state = self::COLON_STATE_COMMENT;
2517                                         break;
2518                                 case ">":
2519                                         # Illegal early close? This shouldn't happen D:
2520                                         $state = self::COLON_STATE_TEXT;
2521                                         break;
2522                                 default:
2523                                         $state = self::COLON_STATE_TAG;
2524                                 }
2525                                 break;
2526                         case 3: # self::COLON_STATE_CLOSETAG:
2527                                 # In a </tag>
2528                                 if ( $c === ">" ) {
2529                                         $stack--;
2530                                         if ( $stack < 0 ) {
2531                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2532                                                 wfProfileOut( __METHOD__ );
2533                                                 return false;
2534                                         }
2535                                         $state = self::COLON_STATE_TEXT;
2536                                 }
2537                                 break;
2538                         case self::COLON_STATE_TAGSLASH:
2539                                 if ( $c === ">" ) {
2540                                         # Yes, a self-closed tag <blah/>
2541                                         $state = self::COLON_STATE_TEXT;
2542                                 } else {
2543                                         # Probably we're jumping the gun, and this is an attribute
2544                                         $state = self::COLON_STATE_TAG;
2545                                 }
2546                                 break;
2547                         case 5: # self::COLON_STATE_COMMENT:
2548                                 if ( $c === "-" ) {
2549                                         $state = self::COLON_STATE_COMMENTDASH;
2550                                 }
2551                                 break;
2552                         case self::COLON_STATE_COMMENTDASH:
2553                                 if ( $c === "-" ) {
2554                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2555                                 } else {
2556                                         $state = self::COLON_STATE_COMMENT;
2557                                 }
2558                                 break;
2559                         case self::COLON_STATE_COMMENTDASHDASH:
2560                                 if ( $c === ">" ) {
2561                                         $state = self::COLON_STATE_TEXT;
2562                                 } else {
2563                                         $state = self::COLON_STATE_COMMENT;
2564                                 }
2565                                 break;
2566                         default:
2567                                 throw new MWException( "State machine error in " . __METHOD__ );
2568                         }
2569                 }
2570                 if ( $stack > 0 ) {
2571                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2572                         wfProfileOut( __METHOD__ );
2573                         return false;
2574                 }
2575                 wfProfileOut( __METHOD__ );
2576                 return false;
2577         }
2578
2579         /**
2580          * Return value of a magic variable (like PAGENAME)
2581          *
2582          * @private
2583          *
2584          * @param $index integer
2585          * @param $frame PPFrame
2586          *
2587          * @return string
2588          */
2589         function getVariableValue( $index, $frame = false ) {
2590                 global $wgContLang, $wgSitename, $wgServer;
2591                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2592
2593                 if ( is_null( $this->mTitle ) ) {
2594                         // If no title set, bad things are going to happen
2595                         // later. Title should always be set since this
2596                         // should only be called in the middle of a parse
2597                         // operation (but the unit-tests do funky stuff)
2598                         throw new MWException( __METHOD__ . ' Should only be '
2599                                 . ' called while parsing (no title set)' );
2600                 }
2601
2602                 /**
2603                  * Some of these require message or data lookups and can be
2604                  * expensive to check many times.
2605                  */
2606                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2607                         if ( isset( $this->mVarCache[$index] ) ) {
2608                                 return $this->mVarCache[$index];
2609                         }
2610                 }
2611
2612                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2613                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2614
2615                 # Use the time zone
2616                 global $wgLocaltimezone;
2617                 if ( isset( $wgLocaltimezone ) ) {
2618                         $oldtz = date_default_timezone_get();
2619                         date_default_timezone_set( $wgLocaltimezone );
2620                 }
2621
2622                 $localTimestamp = date( 'YmdHis', $ts );
2623                 $localMonth = date( 'm', $ts );
2624                 $localMonth1 = date( 'n', $ts );
2625                 $localMonthName = date( 'n', $ts );
2626                 $localDay = date( 'j', $ts );
2627                 $localDay2 = date( 'd', $ts );
2628                 $localDayOfWeek = date( 'w', $ts );
2629                 $localWeek = date( 'W', $ts );
2630                 $localYear = date( 'Y', $ts );
2631                 $localHour = date( 'H', $ts );
2632                 if ( isset( $wgLocaltimezone ) ) {
2633                         date_default_timezone_set( $oldtz );
2634                 }
2635
2636                 $pageLang = $this->getFunctionLang();
2637
2638                 switch ( $index ) {
2639                         case 'currentmonth':
2640                                 $value = $pageLang->formatNum( gmdate( 'm', $ts ) );
2641                                 break;
2642                         case 'currentmonth1':
2643                                 $value = $pageLang->formatNum( gmdate( 'n', $ts ) );
2644                                 break;
2645                         case 'currentmonthname':
2646                                 $value = $pageLang->getMonthName( gmdate( 'n', $ts ) );
2647                                 break;
2648                         case 'currentmonthnamegen':
2649                                 $value = $pageLang->getMonthNameGen( gmdate( 'n', $ts ) );
2650                                 break;
2651                         case 'currentmonthabbrev':
2652                                 $value = $pageLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2653                                 break;
2654                         case 'currentday':
2655                                 $value = $pageLang->formatNum( gmdate( 'j', $ts ) );
2656                                 break;
2657                         case 'currentday2':
2658                                 $value = $pageLang->formatNum( gmdate( 'd', $ts ) );
2659                                 break;
2660                         case 'localmonth':
2661                                 $value = $pageLang->formatNum( $localMonth );
2662                                 break;
2663                         case 'localmonth1':
2664                                 $value = $pageLang->formatNum( $localMonth1 );
2665                                 break;
2666                         case 'localmonthname':
2667                                 $value = $pageLang->getMonthName( $localMonthName );
2668                                 break;
2669                         case 'localmonthnamegen':
2670                                 $value = $pageLang->getMonthNameGen( $localMonthName );
2671                                 break;
2672                         case 'localmonthabbrev':
2673                                 $value = $pageLang->getMonthAbbreviation( $localMonthName );
2674                                 break;
2675                         case 'localday':
2676                                 $value = $pageLang->formatNum( $localDay );
2677                                 break;
2678                         case 'localday2':
2679                                 $value = $pageLang->formatNum( $localDay2 );
2680                                 break;
2681                         case 'pagename':
2682                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2683                                 break;
2684                         case 'pagenamee':
2685                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2686                                 break;
2687                         case 'fullpagename':
2688                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2689                                 break;
2690                         case 'fullpagenamee':
2691                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2692                                 break;
2693                         case 'subpagename':
2694                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2695                                 break;
2696                         case 'subpagenamee':
2697                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2698                                 break;
2699                         case 'basepagename':
2700                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2701                                 break;
2702                         case 'basepagenamee':
2703                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2704                                 break;
2705                         case 'talkpagename':
2706                                 if ( $this->mTitle->canTalk() ) {
2707                                         $talkPage = $this->mTitle->getTalkPage();
2708                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2709                                 } else {
2710                                         $value = '';
2711                                 }
2712                                 break;
2713                         case 'talkpagenamee':
2714                                 if ( $this->mTitle->canTalk() ) {
2715                                         $talkPage = $this->mTitle->getTalkPage();
2716                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2717                                 } else {
2718                                         $value = '';
2719                                 }
2720                                 break;
2721                         case 'subjectpagename':
2722                                 $subjPage = $this->mTitle->getSubjectPage();
2723                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2724                                 break;
2725                         case 'subjectpagenamee':
2726                                 $subjPage = $this->mTitle->getSubjectPage();
2727                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2728                                 break;
2729                         case 'pageid': // requested in bug 23427
2730                                 $pageid = $this->getTitle()->getArticleId();
2731                                 if( $pageid == 0 ) {
2732                                         # 0 means the page doesn't exist in the database,
2733                                         # which means the user is previewing a new page.
2734                                         # The vary-revision flag must be set, because the magic word
2735                                         # will have a different value once the page is saved.
2736                                         $this->mOutput->setFlag( 'vary-revision' );
2737                                         wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2738                                 }
2739                                 $value = $pageid ? $pageid : null;
2740                                 break;
2741                         case 'revisionid':
2742                                 # Let the edit saving system know we should parse the page
2743                                 # *after* a revision ID has been assigned.
2744                                 $this->mOutput->setFlag( 'vary-revision' );
2745                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2746                                 $value = $this->mRevisionId;
2747                                 break;
2748                         case 'revisionday':
2749                                 # Let the edit saving system know we should parse the page
2750                                 # *after* a revision ID has been assigned. This is for null edits.
2751                                 $this->mOutput->setFlag( 'vary-revision' );
2752                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2753                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2754                                 break;
2755                         case 'revisionday2':
2756                                 # Let the edit saving system know we should parse the page
2757                                 # *after* a revision ID has been assigned. This is for null edits.
2758                                 $this->mOutput->setFlag( 'vary-revision' );
2759                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2760                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2761                                 break;
2762                         case 'revisionmonth':
2763                                 # Let the edit saving system know we should parse the page
2764                                 # *after* a revision ID has been assigned. This is for null edits.
2765                                 $this->mOutput->setFlag( 'vary-revision' );
2766                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2767                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2768                                 break;
2769                         case 'revisionmonth1':
2770                                 # Let the edit saving system know we should parse the page
2771                                 # *after* a revision ID has been assigned. This is for null edits.
2772                                 $this->mOutput->setFlag( 'vary-revision' );
2773                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2774                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2775                                 break;
2776                         case 'revisionyear':
2777                                 # Let the edit saving system know we should parse the page
2778                                 # *after* a revision ID has been assigned. This is for null edits.
2779                                 $this->mOutput->setFlag( 'vary-revision' );
2780                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2781                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2782                                 break;
2783                         case 'revisiontimestamp':
2784                                 # Let the edit saving system know we should parse the page
2785                                 # *after* a revision ID has been assigned. This is for null edits.
2786                                 $this->mOutput->setFlag( 'vary-revision' );
2787                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2788                                 $value = $this->getRevisionTimestamp();
2789                                 break;
2790                         case 'revisionuser':
2791                                 # Let the edit saving system know we should parse the page
2792                                 # *after* a revision ID has been assigned. This is for null edits.
2793                                 $this->mOutput->setFlag( 'vary-revision' );
2794                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2795                                 $value = $this->getRevisionUser();
2796                                 break;
2797                         case 'namespace':
2798                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2799                                 break;
2800                         case 'namespacee':
2801                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2802                                 break;
2803                         case 'namespacenumber':
2804                                 $value = $this->mTitle->getNamespace();
2805                                 break;
2806                         case 'talkspace':
2807                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2808                                 break;
2809                         case 'talkspacee':
2810                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2811                                 break;
2812                         case 'subjectspace':
2813                                 $value = $this->mTitle->getSubjectNsText();
2814                                 break;
2815                         case 'subjectspacee':
2816                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2817                                 break;
2818                         case 'currentdayname':
2819                                 $value = $pageLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2820                                 break;
2821                         case 'currentyear':
2822                                 $value = $pageLang->formatNum( gmdate( 'Y', $ts ), true );
2823                                 break;
2824                         case 'currenttime':
2825                                 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2826                                 break;
2827                         case 'currenthour':
2828                                 $value = $pageLang->formatNum( gmdate( 'H', $ts ), true );
2829                                 break;
2830                         case 'currentweek':
2831                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2832                                 # int to remove the padding
2833                                 $value = $pageLang->formatNum( (int)gmdate( 'W', $ts ) );
2834                                 break;
2835                         case 'currentdow':
2836                                 $value = $pageLang->formatNum( gmdate( 'w', $ts ) );
2837                                 break;
2838                         case 'localdayname':
2839                                 $value = $pageLang->getWeekdayName( $localDayOfWeek + 1 );
2840                                 break;
2841                         case 'localyear':
2842                                 $value = $pageLang->formatNum( $localYear, true );
2843                                 break;
2844                         case 'localtime':
2845                                 $value = $pageLang->time( $localTimestamp, false, false );
2846                                 break;
2847                         case 'localhour':
2848                                 $value = $pageLang->formatNum( $localHour, true );
2849                                 break;
2850                         case 'localweek':
2851                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2852                                 # int to remove the padding
2853                                 $value = $pageLang->formatNum( (int)$localWeek );
2854                                 break;
2855                         case 'localdow':
2856                                 $value = $pageLang->formatNum( $localDayOfWeek );
2857                                 break;
2858                         case 'numberofarticles':
2859                                 $value = $pageLang->formatNum( SiteStats::articles() );
2860                                 break;
2861                         case 'numberoffiles':
2862                                 $value = $pageLang->formatNum( SiteStats::images() );
2863                                 break;
2864                         case 'numberofusers':
2865                                 $value = $pageLang->formatNum( SiteStats::users() );
2866                                 break;
2867                         case 'numberofactiveusers':
2868                                 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2869                                 break;
2870                         case 'numberofpages':
2871                                 $value = $pageLang->formatNum( SiteStats::pages() );
2872                                 break;
2873                         case 'numberofadmins':
2874                                 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2875                                 break;
2876                         case 'numberofedits':
2877                                 $value = $pageLang->formatNum( SiteStats::edits() );
2878                                 break;
2879                         case 'numberofviews':
2880                                 global $wgDisableCounters;
2881                                 $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
2882                                 break;
2883                         case 'currenttimestamp':
2884                                 $value = wfTimestamp( TS_MW, $ts );
2885                                 break;
2886                         case 'localtimestamp':
2887                                 $value = $localTimestamp;
2888                                 break;
2889                         case 'currentversion':
2890                                 $value = SpecialVersion::getVersion();
2891                                 break;
2892                         case 'articlepath':
2893                                 return $wgArticlePath;
2894                         case 'sitename':
2895                                 return $wgSitename;
2896                         case 'server':
2897                                 return $wgServer;
2898                         case 'servername':
2899                                 $serverParts = wfParseUrl( $wgServer );
2900                                 return $serverParts && isset( $serverParts['host'] ) ? $serverParts['host'] : $wgServer;
2901                         case 'scriptpath':
2902                                 return $wgScriptPath;
2903                         case 'stylepath':
2904                                 return $wgStylePath;
2905                         case 'directionmark':
2906                                 return $pageLang->getDirMark();
2907                         case 'contentlanguage':
2908                                 global $wgLanguageCode;
2909                                 return $wgLanguageCode;
2910                         default:
2911                                 $ret = null;
2912                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2913                                         return $ret;
2914                                 } else {
2915                                         return null;
2916                                 }
2917                 }
2918
2919                 if ( $index ) {
2920                         $this->mVarCache[$index] = $value;
2921                 }
2922
2923                 return $value;
2924         }
2925
2926         /**
2927          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2928          *
2929          * @private
2930          */
2931         function initialiseVariables() {
2932                 wfProfileIn( __METHOD__ );
2933                 $variableIDs = MagicWord::getVariableIDs();
2934                 $substIDs = MagicWord::getSubstIDs();
2935
2936                 $this->mVariables = new MagicWordArray( $variableIDs );
2937                 $this->mSubstWords = new MagicWordArray( $substIDs );
2938                 wfProfileOut( __METHOD__ );
2939         }
2940
2941         /**
2942          * Preprocess some wikitext and return the document tree.
2943          * This is the ghost of replace_variables().
2944          *
2945          * @param $text String: The text to parse
2946          * @param $flags Integer: bitwise combination of:
2947          *          self::PTD_FOR_INCLUSION    Handle "<noinclude>" and "<includeonly>" as if the text is being
2948          *                                     included. Default is to assume a direct page view.
2949          *
2950          * The generated DOM tree must depend only on the input text and the flags.
2951          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2952          *
2953          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2954          * change in the DOM tree for a given text, must be passed through the section identifier
2955          * in the section edit link and thus back to extractSections().
2956          *
2957          * The output of this function is currently only cached in process memory, but a persistent
2958          * cache may be implemented at a later date which takes further advantage of these strict
2959          * dependency requirements.
2960          *
2961          * @private
2962          *
2963          * @return PPNode
2964          */
2965         function preprocessToDom( $text, $flags = 0 ) {
2966                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2967                 return $dom;
2968         }
2969
2970         /**
2971          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2972          *
2973          * @param $s string
2974          *
2975          * @return array
2976          */
2977         public static function splitWhitespace( $s ) {
2978                 $ltrimmed = ltrim( $s );
2979                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2980                 $trimmed = rtrim( $ltrimmed );
2981                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2982                 if ( $diff > 0 ) {
2983                         $w2 = substr( $ltrimmed, -$diff );
2984                 } else {
2985                         $w2 = '';
2986                 }
2987                 return array( $w1, $trimmed, $w2 );
2988         }
2989
2990         /**
2991          * Replace magic variables, templates, and template arguments
2992          * with the appropriate text. Templates are substituted recursively,
2993          * taking care to avoid infinite loops.
2994          *
2995          * Note that the substitution depends on value of $mOutputType:
2996          *  self::OT_WIKI: only {{subst:}} templates
2997          *  self::OT_PREPROCESS: templates but not extension tags
2998          *  self::OT_HTML: all templates and extension tags
2999          *
3000          * @param $text String the text to transform
3001          * @param $frame PPFrame Object describing the arguments passed to the template.
3002          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
3003          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
3004          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
3005          * @private
3006          *
3007          * @return string
3008          */
3009         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3010                 # Is there any text? Also, Prevent too big inclusions!
3011                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3012                         return $text;
3013                 }
3014                 wfProfileIn( __METHOD__ );
3015
3016                 if ( $frame === false ) {
3017                         $frame = $this->getPreprocessor()->newFrame();
3018                 } elseif ( !( $frame instanceof PPFrame ) ) {
3019                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
3020                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3021                 }
3022
3023                 $dom = $this->preprocessToDom( $text );
3024                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3025                 $text = $frame->expand( $dom, $flags );
3026
3027                 wfProfileOut( __METHOD__ );
3028                 return $text;
3029         }
3030
3031         /**
3032          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3033          *
3034          * @param $args array
3035          *
3036          * @return array
3037          */
3038         static function createAssocArgs( $args ) {
3039                 $assocArgs = array();
3040                 $index = 1;
3041                 foreach ( $args as $arg ) {
3042                         $eqpos = strpos( $arg, '=' );
3043                         if ( $eqpos === false ) {
3044                                 $assocArgs[$index++] = $arg;
3045                         } else {
3046                                 $name = trim( substr( $arg, 0, $eqpos ) );
3047                                 $value = trim( substr( $arg, $eqpos+1 ) );
3048                                 if ( $value === false ) {
3049                                         $value = '';
3050                                 }
3051                                 if ( $name !== false ) {
3052                                         $assocArgs[$name] = $value;
3053                                 }
3054                         }
3055                 }
3056
3057                 return $assocArgs;
3058         }
3059
3060         /**
3061          * Warn the user when a parser limitation is reached
3062          * Will warn at most once the user per limitation type
3063          *
3064          * @param $limitationType String: should be one of:
3065          *   'expensive-parserfunction' (corresponding messages:
3066          *       'expensive-parserfunction-warning',
3067          *       'expensive-parserfunction-category')
3068          *   'post-expand-template-argument' (corresponding messages:
3069          *       'post-expand-template-argument-warning',
3070          *       'post-expand-template-argument-category')
3071          *   'post-expand-template-inclusion' (corresponding messages:
3072          *       'post-expand-template-inclusion-warning',
3073          *       'post-expand-template-inclusion-category')
3074          * @param $current int|null Current value
3075          * @param $max int|null Maximum allowed, when an explicit limit has been
3076          *       exceeded, provide the values (optional)
3077          */
3078         function limitationWarn( $limitationType, $current = null, $max = null) {
3079                 # does no harm if $current and $max are present but are unnecessary for the message
3080                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3081                 $this->mOutput->addWarning( $warning );
3082                 $this->addTrackingCategory( "$limitationType-category" );
3083         }
3084
3085         /**
3086          * Return the text of a template, after recursively
3087          * replacing any variables or templates within the template.
3088          *
3089          * @param $piece Array: the parts of the template
3090          *  $piece['title']: the title, i.e. the part before the |
3091          *  $piece['parts']: the parameter array
3092          *  $piece['lineStart']: whether the brace was at the start of a line
3093          * @param $frame PPFrame The current frame, contains template arguments
3094          * @return String: the text of the template
3095          * @private
3096          */
3097         function braceSubstitution( $piece, $frame ) {
3098                 global $wgContLang;
3099                 wfProfileIn( __METHOD__ );
3100                 wfProfileIn( __METHOD__.'-setup' );
3101
3102                 # Flags
3103                 $found = false;             # $text has been filled
3104                 $nowiki = false;            # wiki markup in $text should be escaped
3105                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3106                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3107                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3108                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3109
3110                 # Title object, where $text came from
3111                 $title = false;
3112
3113                 # $part1 is the bit before the first |, and must contain only title characters.
3114                 # Various prefixes will be stripped from it later.
3115                 $titleWithSpaces = $frame->expand( $piece['title'] );
3116                 $part1 = trim( $titleWithSpaces );
3117                 $titleText = false;
3118
3119                 # Original title text preserved for various purposes
3120                 $originalTitle = $part1;
3121
3122                 # $args is a list of argument nodes, starting from index 0, not including $part1
3123                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3124                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3125                 wfProfileOut( __METHOD__.'-setup' );
3126
3127                 $titleProfileIn = null; // profile templates
3128
3129                 # SUBST
3130                 wfProfileIn( __METHOD__.'-modifiers' );
3131                 if ( !$found ) {
3132
3133                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3134
3135                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3136                         # Decide whether to expand template or keep wikitext as-is.
3137                         if ( $this->ot['wiki'] ) {
3138                                 if ( $substMatch === false ) {
3139                                         $literal = true;  # literal when in PST with no prefix
3140                                 } else {
3141                                         $literal = false; # expand when in PST with subst: or safesubst:
3142                                 }
3143                         } else {
3144                                 if ( $substMatch == 'subst' ) {
3145                                         $literal = true;  # literal when not in PST with plain subst:
3146                                 } else {
3147                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3148                                 }
3149                         }
3150                         if ( $literal ) {
3151                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3152                                 $isLocalObj = true;
3153                                 $found = true;
3154                         }
3155                 }
3156
3157                 # Variables
3158                 if ( !$found && $args->getLength() == 0 ) {
3159                         $id = $this->mVariables->matchStartToEnd( $part1 );
3160                         if ( $id !== false ) {
3161                                 $text = $this->getVariableValue( $id, $frame );
3162                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3163                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3164                                 }
3165                                 $found = true;
3166                         }
3167                 }
3168
3169                 # MSG, MSGNW and RAW
3170                 if ( !$found ) {
3171                         # Check for MSGNW:
3172                         $mwMsgnw = MagicWord::get( 'msgnw' );
3173                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3174                                 $nowiki = true;
3175                         } else {
3176                                 # Remove obsolete MSG:
3177                                 $mwMsg = MagicWord::get( 'msg' );
3178                                 $mwMsg->matchStartAndRemove( $part1 );
3179                         }
3180
3181                         # Check for RAW:
3182                         $mwRaw = MagicWord::get( 'raw' );
3183                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3184                                 $forceRawInterwiki = true;
3185                         }
3186                 }
3187                 wfProfileOut( __METHOD__.'-modifiers' );
3188
3189                 # Parser functions
3190                 if ( !$found ) {
3191                         wfProfileIn( __METHOD__ . '-pfunc' );
3192
3193                         $colonPos = strpos( $part1, ':' );
3194                         if ( $colonPos !== false ) {
3195                                 # Case sensitive functions
3196                                 $function = substr( $part1, 0, $colonPos );
3197                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3198                                         $function = $this->mFunctionSynonyms[1][$function];
3199                                 } else {
3200                                         # Case insensitive functions
3201                                         $function = $wgContLang->lc( $function );
3202                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3203                                                 $function = $this->mFunctionSynonyms[0][$function];
3204                                         } else {
3205                                                 $function = false;
3206                                         }
3207                                 }
3208                                 if ( $function ) {
3209                                         wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3210                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3211                                         $initialArgs = array( &$this );
3212                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3213                                         if ( $flags & SFH_OBJECT_ARGS ) {
3214                                                 # Add a frame parameter, and pass the arguments as an array
3215                                                 $allArgs = $initialArgs;
3216                                                 $allArgs[] = $frame;
3217                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3218                                                         $funcArgs[] = $args->item( $i );
3219                                                 }
3220                                                 $allArgs[] = $funcArgs;
3221                                         } else {
3222                                                 # Convert arguments to plain text
3223                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3224                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3225                                                 }
3226                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3227                                         }
3228
3229                                         # Workaround for PHP bug 35229 and similar
3230                                         if ( !is_callable( $callback ) ) {
3231                                                 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3232                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3233                                                 wfProfileOut( __METHOD__ );
3234                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3235                                         }
3236                                         $result = call_user_func_array( $callback, $allArgs );
3237                                         $found = true;
3238                                         $noparse = true;
3239                                         $preprocessFlags = 0;
3240
3241                                         if ( is_array( $result ) ) {
3242                                                 if ( isset( $result[0] ) ) {
3243                                                         $text = $result[0];
3244                                                         unset( $result[0] );
3245                                                 }
3246
3247                                                 # Extract flags into the local scope
3248                                                 # This allows callers to set flags such as nowiki, found, etc.
3249                                                 extract( $result );
3250                                         } else {
3251                                                 $text = $result;
3252                                         }
3253                                         if ( !$noparse ) {
3254                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3255                                                 $isChildObj = true;
3256                                         }
3257                                         wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3258                                 }
3259                         }
3260                         wfProfileOut( __METHOD__ . '-pfunc' );
3261                 }
3262
3263                 # Finish mangling title and then check for loops.
3264                 # Set $title to a Title object and $titleText to the PDBK
3265                 if ( !$found ) {
3266                         $ns = NS_TEMPLATE;
3267                         # Split the title into page and subpage
3268                         $subpage = '';
3269                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3270                         if ( $subpage !== '' ) {
3271                                 $ns = $this->mTitle->getNamespace();
3272                         }
3273                         $title = Title::newFromText( $part1, $ns );
3274                         if ( $title ) {
3275                                 $titleText = $title->getPrefixedText();
3276                                 # Check for language variants if the template is not found
3277                                 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3278                                         $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3279                                 }
3280                                 # Do recursion depth check
3281                                 $limit = $this->mOptions->getMaxTemplateDepth();
3282                                 if ( $frame->depth >= $limit ) {
3283                                         $found = true;
3284                                         $text = '<span class="error">'
3285                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3286                                                 . '</span>';
3287                                 }
3288                         }
3289                 }
3290
3291                 # Load from database
3292                 if ( !$found && $title ) {
3293                         if ( !Profiler::instance()->isPersistent() ) {
3294                                 # Too many unique items can kill profiling DBs/collectors
3295                                 $titleProfileIn = __METHOD__ . "-title-" . $title->getDBKey();
3296                                 wfProfileIn( $titleProfileIn ); // template in
3297                         }
3298                         wfProfileIn( __METHOD__ . '-loadtpl' );
3299                         if ( !$title->isExternal() ) {
3300                                 if ( $title->isSpecialPage()
3301                                         && $this->mOptions->getAllowSpecialInclusion()
3302                                         && $this->ot['html'] )
3303                                 {
3304                                         // Pass the template arguments as URL parameters.
3305                                         // "uselang" will have no effect since the Language object
3306                                         // is forced to the one defined in ParserOptions.
3307                                         $pageArgs = array();
3308                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3309                                                 $bits = $args->item( $i )->splitArg();
3310                                                 if ( strval( $bits['index'] ) === '' ) {
3311                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3312                                                         $value = trim( $frame->expand( $bits['value'] ) );
3313                                                         $pageArgs[$name] = $value;
3314                                                 }
3315                                         }
3316
3317                                         // Create a new context to execute the special page
3318                                         $context = new RequestContext;
3319                                         $context->setTitle( $title );
3320                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3321                                         $context->setUser( $this->getUser() );
3322                                         $context->setLanguage( $this->mOptions->getUserLangObj() );
3323                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3324                                         if ( $ret ) {
3325                                                 $text = $context->getOutput()->getHTML();
3326                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3327                                                 $found = true;
3328                                                 $isHTML = true;
3329                                                 $this->disableCache();
3330                                         }
3331                                 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3332                                         $found = false; # access denied
3333                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3334                                 } else {
3335                                         list( $text, $title ) = $this->getTemplateDom( $title );
3336                                         if ( $text !== false ) {
3337                                                 $found = true;
3338                                                 $isChildObj = true;
3339                                         }
3340                                 }
3341
3342                                 # If the title is valid but undisplayable, make a link to it
3343                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3344                                         $text = "[[:$titleText]]";
3345                                         $found = true;
3346                                 }
3347                         } elseif ( $title->isTrans() ) {
3348                                 # Interwiki transclusion
3349                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3350                                         $text = $this->interwikiTransclude( $title, 'render' );
3351                                         $isHTML = true;
3352                                 } else {
3353                                         $text = $this->interwikiTransclude( $title, 'raw' );
3354                                         # Preprocess it like a template
3355                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3356                                         $isChildObj = true;
3357                                 }
3358                                 $found = true;
3359                         }
3360
3361                         # Do infinite loop check
3362                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3363                         if ( !$frame->loopCheck( $title ) ) {
3364                                 $found = true;
3365                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3366                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3367                         }
3368                         wfProfileOut( __METHOD__ . '-loadtpl' );
3369                 }
3370
3371                 # If we haven't found text to substitute by now, we're done
3372                 # Recover the source wikitext and return it
3373                 if ( !$found ) {
3374                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3375                         if ( $titleProfileIn ) {
3376                                 wfProfileOut( $titleProfileIn ); // template out
3377                         }
3378                         wfProfileOut( __METHOD__ );
3379                         return array( 'object' => $text );
3380                 }
3381
3382                 # Expand DOM-style return values in a child frame
3383                 if ( $isChildObj ) {
3384                         # Clean up argument array
3385                         $newFrame = $frame->newChild( $args, $title );
3386
3387                         if ( $nowiki ) {
3388                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3389                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3390                                 # Expansion is eligible for the empty-frame cache
3391                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3392                                         $text = $this->mTplExpandCache[$titleText];
3393                                 } else {
3394                                         $text = $newFrame->expand( $text );
3395                                         $this->mTplExpandCache[$titleText] = $text;
3396                                 }
3397                         } else {
3398                                 # Uncached expansion
3399                                 $text = $newFrame->expand( $text );
3400                         }
3401                 }
3402                 if ( $isLocalObj && $nowiki ) {
3403                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3404                         $isLocalObj = false;
3405                 }
3406
3407                 if ( $titleProfileIn ) {
3408                         wfProfileOut( $titleProfileIn ); // template out
3409                 }
3410
3411                 # Replace raw HTML by a placeholder
3412                 if ( $isHTML ) {
3413                         $text = $this->insertStripItem( $text );
3414                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3415                         # Escape nowiki-style return values
3416                         $text = wfEscapeWikiText( $text );
3417                 } elseif ( is_string( $text )
3418                         && !$piece['lineStart']
3419                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3420                 {
3421                         # Bug 529: if the template begins with a table or block-level
3422                         # element, it should be treated as beginning a new line.
3423                         # This behaviour is somewhat controversial.
3424                         $text = "\n" . $text;
3425                 }
3426
3427                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3428                         # Error, oversize inclusion
3429                         if ( $titleText !== false ) {
3430                                 # Make a working, properly escaped link if possible (bug 23588)
3431                                 $text = "[[:$titleText]]";
3432                         } else {
3433                                 # This will probably not be a working link, but at least it may
3434                                 # provide some hint of where the problem is
3435                                 preg_replace( '/^:/', '', $originalTitle );
3436                                 $text = "[[:$originalTitle]]";
3437                         }
3438                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3439                         $this->limitationWarn( 'post-expand-template-inclusion' );
3440                 }
3441
3442                 if ( $isLocalObj ) {
3443                         $ret = array( 'object' => $text );
3444                 } else {
3445                         $ret = array( 'text' => $text );
3446                 }
3447
3448                 wfProfileOut( __METHOD__ );
3449                 return $ret;
3450         }
3451
3452         /**
3453          * Get the semi-parsed DOM representation of a template with a given title,
3454          * and its redirect destination title. Cached.
3455          *
3456          * @param $title Title
3457          *
3458          * @return array
3459          */
3460         function getTemplateDom( $title ) {
3461                 $cacheTitle = $title;
3462                 $titleText = $title->getPrefixedDBkey();
3463
3464                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3465                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3466                         $title = Title::makeTitle( $ns, $dbk );
3467                         $titleText = $title->getPrefixedDBkey();
3468                 }
3469                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3470                         return array( $this->mTplDomCache[$titleText], $title );
3471                 }
3472
3473                 # Cache miss, go to the database
3474                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3475
3476                 if ( $text === false ) {
3477                         $this->mTplDomCache[$titleText] = false;
3478                         return array( false, $title );
3479                 }
3480
3481                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3482                 $this->mTplDomCache[ $titleText ] = $dom;
3483
3484                 if ( !$title->equals( $cacheTitle ) ) {
3485                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3486                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3487                 }
3488
3489                 return array( $dom, $title );
3490         }
3491
3492         /**
3493          * Fetch the unparsed text of a template and register a reference to it.
3494          * @param Title $title
3495          * @return Array ( string or false, Title )
3496          */
3497         function fetchTemplateAndTitle( $title ) {
3498                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3499                 $stuff = call_user_func( $templateCb, $title, $this );
3500                 $text = $stuff['text'];
3501                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3502                 if ( isset( $stuff['deps'] ) ) {
3503                         foreach ( $stuff['deps'] as $dep ) {
3504                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3505                         }
3506                 }
3507                 return array( $text, $finalTitle );
3508         }
3509
3510         /**
3511          * Fetch the unparsed text of a template and register a reference to it.
3512          * @param Title $title
3513          * @return mixed string or false
3514          */
3515         function fetchTemplate( $title ) {
3516                 $rv = $this->fetchTemplateAndTitle( $title );
3517                 return $rv[0];
3518         }
3519
3520         /**
3521          * Static function to get a template
3522          * Can be overridden via ParserOptions::setTemplateCallback().
3523          *
3524          * @parma $title Title
3525          * @param $parser Parser
3526          *
3527          * @return array
3528          */
3529         static function statelessFetchTemplate( $title, $parser = false ) {
3530                 $text = $skip = false;
3531                 $finalTitle = $title;
3532                 $deps = array();
3533
3534                 # Loop to fetch the article, with up to 1 redirect
3535                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3536                         # Give extensions a chance to select the revision instead
3537                         $id = false; # Assume current
3538                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3539                                 array( $parser, $title, &$skip, &$id ) );
3540
3541                         if ( $skip ) {
3542                                 $text = false;
3543                                 $deps[] = array(
3544                                         'title'         => $title,
3545                                         'page_id'       => $title->getArticleID(),
3546                                         'rev_id'        => null
3547                                 );
3548                                 break;
3549                         }
3550                         # Get the revision
3551                         $rev = $id
3552                                 ? Revision::newFromId( $id )
3553                                 : Revision::newFromTitle( $title );
3554                         $rev_id = $rev ? $rev->getId() : 0;
3555                         # If there is no current revision, there is no page
3556                         if ( $id === false && !$rev ) {
3557                                 $linkCache = LinkCache::singleton();
3558                                 $linkCache->addBadLinkObj( $title );
3559                         }
3560
3561                         $deps[] = array(
3562                                 'title'         => $title,
3563                                 'page_id'       => $title->getArticleID(),
3564                                 'rev_id'        => $rev_id );
3565                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3566                                 # We fetched a rev from a different title; register it too...
3567                                 $deps[] = array(
3568                                         'title'         => $rev->getTitle(),
3569                                         'page_id'       => $rev->getPage(),
3570                                         'rev_id'        => $rev_id );
3571                         }
3572
3573                         if ( $rev ) {
3574                                 $text = $rev->getText();
3575                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3576                                 global $wgContLang;
3577                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3578                                 if ( !$message->exists() ) {
3579                                         $text = false;
3580                                         break;
3581                                 }
3582                                 $text = $message->plain();
3583                         } else {
3584                                 break;
3585                         }
3586                         if ( $text === false ) {
3587                                 break;
3588                         }
3589                         # Redirect?
3590                         $finalTitle = $title;
3591                         $title = Title::newFromRedirect( $text );
3592                 }
3593                 return array(
3594                         'text' => $text,
3595                         'finalTitle' => $finalTitle,
3596                         'deps' => $deps );
3597         }
3598
3599         /**
3600          * Fetch a file and its title and register a reference to it.
3601          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3602          * @param Title $title
3603          * @param Array $options Array of options to RepoGroup::findFile
3604          * @return File|bool
3605          */
3606         function fetchFile( $title, $options = array() ) {
3607                 $res = $this->fetchFileAndTitle( $title, $options );
3608                 return $res[0];
3609         }
3610
3611         /**
3612          * Fetch a file and its title and register a reference to it.
3613          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3614          * @param Title $title
3615          * @param Array $options Array of options to RepoGroup::findFile
3616          * @return Array ( File or false, Title of file )
3617          */
3618         function fetchFileAndTitle( $title, $options = array() ) {
3619                 if ( isset( $options['broken'] ) ) {
3620                         $file = false; // broken thumbnail forced by hook
3621                 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3622                         $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3623                 } else { // get by (name,timestamp)
3624                         $file = wfFindFile( $title, $options );
3625                 }
3626                 $time = $file ? $file->getTimestamp() : false;
3627                 $sha1 = $file ? $file->getSha1() : false;
3628                 # Register the file as a dependency...
3629                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3630                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3631                         # Update fetched file title
3632                         $title = $file->getTitle();
3633                         if ( is_null( $file->getRedirectedTitle() ) ) {
3634                                 # This file was not a redirect, but the title does not match.
3635                                 # Register under the new name because otherwise the link will
3636                                 # get lost.
3637                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3638                         }
3639                 }
3640                 return array( $file, $title );
3641         }
3642
3643         /**
3644          * Transclude an interwiki link.
3645          *
3646          * @param $title Title
3647          * @param $action
3648          *
3649          * @return string
3650          */
3651         function interwikiTransclude( $title, $action ) {
3652                 global $wgEnableScaryTranscluding;
3653
3654                 if ( !$wgEnableScaryTranscluding ) {
3655                         return wfMsgForContent('scarytranscludedisabled');
3656                 }
3657
3658                 $url = $title->getFullUrl( "action=$action" );
3659
3660                 if ( strlen( $url ) > 255 ) {
3661                         return wfMsgForContent( 'scarytranscludetoolong' );
3662                 }
3663                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3664         }
3665
3666         /**
3667          * @param $url string
3668          * @return Mixed|String
3669          */
3670         function fetchScaryTemplateMaybeFromCache( $url ) {
3671                 global $wgTranscludeCacheExpiry;
3672                 $dbr = wfGetDB( DB_SLAVE );
3673                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3674                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3675                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3676                 if ( $obj ) {
3677                         return $obj->tc_contents;
3678                 }
3679
3680                 $text = Http::get( $url );
3681                 if ( !$text ) {
3682                         return wfMsgForContent( 'scarytranscludefailed', $url );
3683                 }
3684
3685                 $dbw = wfGetDB( DB_MASTER );
3686                 $dbw->replace( 'transcache', array('tc_url'), array(
3687                         'tc_url' => $url,
3688                         'tc_time' => $dbw->timestamp( time() ),
3689                         'tc_contents' => $text)
3690                 );
3691                 return $text;
3692         }
3693
3694         /**
3695          * Triple brace replacement -- used for template arguments
3696          * @private
3697          *
3698          * @param $peice array
3699          * @param $frame PPFrame
3700          *
3701          * @return array
3702          */
3703         function argSubstitution( $piece, $frame ) {
3704                 wfProfileIn( __METHOD__ );
3705
3706                 $error = false;
3707                 $parts = $piece['parts'];
3708                 $nameWithSpaces = $frame->expand( $piece['title'] );
3709                 $argName = trim( $nameWithSpaces );
3710                 $object = false;
3711                 $text = $frame->getArgument( $argName );
3712                 if (  $text === false && $parts->getLength() > 0
3713                   && (
3714                         $this->ot['html']
3715                         || $this->ot['pre']
3716                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3717                   )
3718                 ) {
3719                         # No match in frame, use the supplied default
3720                         $object = $parts->item( 0 )->getChildren();
3721                 }
3722                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3723                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3724                         $this->limitationWarn( 'post-expand-template-argument' );
3725                 }
3726
3727                 if ( $text === false && $object === false ) {
3728                         # No match anywhere
3729                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3730                 }
3731                 if ( $error !== false ) {
3732                         $text .= $error;
3733                 }
3734                 if ( $object !== false ) {
3735                         $ret = array( 'object' => $object );
3736                 } else {
3737                         $ret = array( 'text' => $text );
3738                 }
3739
3740                 wfProfileOut( __METHOD__ );
3741                 return $ret;
3742         }
3743
3744         /**
3745          * Return the text to be used for a given extension tag.
3746          * This is the ghost of strip().
3747          *
3748          * @param $params array Associative array of parameters:
3749          *     name       PPNode for the tag name
3750          *     attr       PPNode for unparsed text where tag attributes are thought to be
3751          *     attributes Optional associative array of parsed attributes
3752          *     inner      Contents of extension element
3753          *     noClose    Original text did not have a close tag
3754          * @param $frame PPFrame
3755          *
3756          * @return string
3757          */
3758         function extensionSubstitution( $params, $frame ) {
3759                 $name = $frame->expand( $params['name'] );
3760                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3761                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3762                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3763
3764                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3765                         ( $this->ot['html'] || $this->ot['pre'] );
3766                 if ( $isFunctionTag ) {
3767                         $markerType = 'none';
3768                 } else {
3769                         $markerType = 'general';
3770                 }
3771                 if ( $this->ot['html'] || $isFunctionTag ) {
3772                         $name = strtolower( $name );
3773                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3774                         if ( isset( $params['attributes'] ) ) {
3775                                 $attributes = $attributes + $params['attributes'];
3776                         }
3777
3778                         if ( isset( $this->mTagHooks[$name] ) ) {
3779                                 # Workaround for PHP bug 35229 and similar
3780                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3781                                         throw new MWException( "Tag hook for $name is not callable\n" );
3782                                 }
3783                                 $output = call_user_func_array( $this->mTagHooks[$name],
3784                                         array( $content, $attributes, $this, $frame ) );
3785                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3786                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3787                                 if ( !is_callable( $callback ) ) {
3788                                         throw new MWException( "Tag hook for $name is not callable\n" );
3789                                 }
3790
3791                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3792                         } else {
3793                                 $output = '<span class="error">Invalid tag extension name: ' .
3794                                         htmlspecialchars( $name ) . '</span>';
3795                         }
3796
3797                         if ( is_array( $output ) ) {
3798                                 # Extract flags to local scope (to override $markerType)
3799                                 $flags = $output;
3800                                 $output = $flags[0];
3801                                 unset( $flags[0] );
3802                                 extract( $flags );
3803                         }
3804                 } else {
3805                         if ( is_null( $attrText ) ) {
3806                                 $attrText = '';
3807                         }
3808                         if ( isset( $params['attributes'] ) ) {
3809                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3810                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3811                                                 htmlspecialchars( $attrValue ) . '"';
3812                                 }
3813                         }
3814                         if ( $content === null ) {
3815                                 $output = "<$name$attrText/>";
3816                         } else {
3817                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3818                                 $output = "<$name$attrText>$content$close";
3819                         }
3820                 }
3821
3822                 if ( $markerType === 'none' ) {
3823                         return $output;
3824                 } elseif ( $markerType === 'nowiki' ) {
3825                         $this->mStripState->addNoWiki( $marker, $output );
3826                 } elseif ( $markerType === 'general' ) {
3827                         $this->mStripState->addGeneral( $marker, $output );
3828                 } else {
3829                         throw new MWException( __METHOD__.': invalid marker type' );
3830                 }
3831                 return $marker;
3832         }
3833
3834         /**
3835          * Increment an include size counter
3836          *
3837          * @param $type String: the type of expansion
3838          * @param $size Integer: the size of the text
3839          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3840          */
3841         function incrementIncludeSize( $type, $size ) {
3842                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3843                         return false;
3844                 } else {
3845                         $this->mIncludeSizes[$type] += $size;
3846                         return true;
3847                 }
3848         }
3849
3850         /**
3851          * Increment the expensive function count
3852          *
3853          * @return Boolean: false if the limit has been exceeded
3854          */
3855         function incrementExpensiveFunctionCount() {
3856                 $this->mExpensiveFunctionCount++;
3857                 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3858         }
3859
3860         /**
3861          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3862          * Fills $this->mDoubleUnderscores, returns the modified text
3863          *
3864          * @param $text string
3865          *
3866          * @return string
3867          */
3868         function doDoubleUnderscore( $text ) {
3869                 wfProfileIn( __METHOD__ );
3870
3871                 # The position of __TOC__ needs to be recorded
3872                 $mw = MagicWord::get( 'toc' );
3873                 if ( $mw->match( $text ) ) {
3874                         $this->mShowToc = true;
3875                         $this->mForceTocPosition = true;
3876
3877                         # Set a placeholder. At the end we'll fill it in with the TOC.
3878                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3879
3880                         # Only keep the first one.
3881                         $text = $mw->replace( '', $text );
3882                 }
3883
3884                 # Now match and remove the rest of them
3885                 $mwa = MagicWord::getDoubleUnderscoreArray();
3886                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3887
3888                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3889                         $this->mOutput->mNoGallery = true;
3890                 }
3891                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3892                         $this->mShowToc = false;
3893                 }
3894                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3895                         $this->addTrackingCategory( 'hidden-category-category' );
3896                 }
3897                 # (bug 8068) Allow control over whether robots index a page.
3898                 #
3899                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3900                 # is not desirable, the last one on the page should win.
3901                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3902                         $this->mOutput->setIndexPolicy( 'noindex' );
3903                         $this->addTrackingCategory( 'noindex-category' );
3904                 }
3905                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3906                         $this->mOutput->setIndexPolicy( 'index' );
3907                         $this->addTrackingCategory( 'index-category' );
3908                 }
3909
3910                 # Cache all double underscores in the database
3911                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3912                         $this->mOutput->setProperty( $key, '' );
3913                 }
3914
3915                 wfProfileOut( __METHOD__ );
3916                 return $text;
3917         }
3918
3919         /**
3920          * Add a tracking category, getting the title from a system message,
3921          * or print a debug message if the title is invalid.
3922          *
3923          * @param $msg String: message key
3924          * @return Boolean: whether the addition was successful
3925          */
3926         public function addTrackingCategory( $msg ) {
3927                 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
3928                         wfDebug( __METHOD__.": Not adding tracking category $msg to special page!\n" );
3929                         return false;
3930                 }
3931                 // Important to parse with correct title (bug 31469)
3932                 $cat = wfMessage( $msg )
3933                         ->title( $this->getTitle() )
3934                         ->inContentLanguage()
3935                         ->text();
3936
3937                 # Allow tracking categories to be disabled by setting them to "-"
3938                 if ( $cat === '-' ) {
3939                         return false;
3940                 }
3941
3942                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3943                 if ( $containerCategory ) {
3944                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3945                         return true;
3946                 } else {
3947                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3948                         return false;
3949                 }
3950         }
3951
3952         /**
3953          * This function accomplishes several tasks:
3954          * 1) Auto-number headings if that option is enabled
3955          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3956          * 3) Add a Table of contents on the top for users who have enabled the option
3957          * 4) Auto-anchor headings
3958          *
3959          * It loops through all headlines, collects the necessary data, then splits up the
3960          * string and re-inserts the newly formatted headlines.
3961          *
3962          * @param $text String
3963          * @param $origText String: original, untouched wikitext
3964          * @param $isMain Boolean
3965          * @return mixed|string
3966          * @private
3967          */
3968         function formatHeadings( $text, $origText, $isMain=true ) {
3969                 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3970
3971                 # Inhibit editsection links if requested in the page
3972                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3973                         $maybeShowEditLink = $showEditLink = false;
3974                 } else {
3975                         $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3976                         $showEditLink = $this->mOptions->getEditSection();
3977                 }
3978                 if ( $showEditLink ) {
3979                         $this->mOutput->setEditSectionTokens( true );
3980                 }
3981
3982                 # Get all headlines for numbering them and adding funky stuff like [edit]
3983                 # links - this is for later, but we need the number of headlines right now
3984                 $matches = array();
3985                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3986
3987                 # if there are fewer than 4 headlines in the article, do not show TOC
3988                 # unless it's been explicitly enabled.
3989                 $enoughToc = $this->mShowToc &&
3990                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3991
3992                 # Allow user to stipulate that a page should have a "new section"
3993                 # link added via __NEWSECTIONLINK__
3994                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3995                         $this->mOutput->setNewSection( true );
3996                 }
3997
3998                 # Allow user to remove the "new section"
3999                 # link via __NONEWSECTIONLINK__
4000                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4001                         $this->mOutput->hideNewSection( true );
4002                 }
4003
4004                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4005                 # override above conditions and always show TOC above first header
4006                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4007                         $this->mShowToc = true;
4008                         $enoughToc = true;
4009                 }
4010
4011                 # headline counter
4012                 $headlineCount = 0;
4013                 $numVisible = 0;
4014
4015                 # Ugh .. the TOC should have neat indentation levels which can be
4016                 # passed to the skin functions. These are determined here
4017                 $toc = '';
4018                 $full = '';
4019                 $head = array();
4020                 $sublevelCount = array();
4021                 $levelCount = array();
4022                 $level = 0;
4023                 $prevlevel = 0;
4024                 $toclevel = 0;
4025                 $prevtoclevel = 0;
4026                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4027                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4028                 $oldType = $this->mOutputType;
4029                 $this->setOutputType( self::OT_WIKI );
4030                 $frame = $this->getPreprocessor()->newFrame();
4031                 $root = $this->preprocessToDom( $origText );
4032                 $node = $root->getFirstChild();
4033                 $byteOffset = 0;
4034                 $tocraw = array();
4035                 $refers = array();
4036
4037                 foreach ( $matches[3] as $headline ) {
4038                         $isTemplate = false;
4039                         $titleText = false;
4040                         $sectionIndex = false;
4041                         $numbering = '';
4042                         $markerMatches = array();
4043                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
4044                                 $serial = $markerMatches[1];
4045                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4046                                 $isTemplate = ( $titleText != $baseTitleText );
4047                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
4048                         }
4049
4050                         if ( $toclevel ) {
4051                                 $prevlevel = $level;
4052                         }
4053                         $level = $matches[1][$headlineCount];
4054
4055                         if ( $level > $prevlevel ) {
4056                                 # Increase TOC level
4057                                 $toclevel++;
4058                                 $sublevelCount[$toclevel] = 0;
4059                                 if ( $toclevel<$wgMaxTocLevel ) {
4060                                         $prevtoclevel = $toclevel;
4061                                         $toc .= Linker::tocIndent();
4062                                         $numVisible++;
4063                                 }
4064                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4065                                 # Decrease TOC level, find level to jump to
4066
4067                                 for ( $i = $toclevel; $i > 0; $i-- ) {
4068                                         if ( $levelCount[$i] == $level ) {
4069                                                 # Found last matching level
4070                                                 $toclevel = $i;
4071                                                 break;
4072                                         } elseif ( $levelCount[$i] < $level ) {
4073                                                 # Found first matching level below current level
4074                                                 $toclevel = $i + 1;
4075                                                 break;
4076                                         }
4077                                 }
4078                                 if ( $i == 0 ) {
4079                                         $toclevel = 1;
4080                                 }
4081                                 if ( $toclevel<$wgMaxTocLevel ) {
4082                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
4083                                                 # Unindent only if the previous toc level was shown :p
4084                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4085                                                 $prevtoclevel = $toclevel;
4086                                         } else {
4087                                                 $toc .= Linker::tocLineEnd();
4088                                         }
4089                                 }
4090                         } else {
4091                                 # No change in level, end TOC line
4092                                 if ( $toclevel<$wgMaxTocLevel ) {
4093                                         $toc .= Linker::tocLineEnd();
4094                                 }
4095                         }
4096
4097                         $levelCount[$toclevel] = $level;
4098
4099                         # count number of headlines for each level
4100                         @$sublevelCount[$toclevel]++;
4101                         $dot = 0;
4102                         for( $i = 1; $i <= $toclevel; $i++ ) {
4103                                 if ( !empty( $sublevelCount[$i] ) ) {
4104                                         if ( $dot ) {
4105                                                 $numbering .= '.';
4106                                         }
4107                                         $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4108                                         $dot = 1;
4109                                 }
4110                         }
4111
4112                         # The safe header is a version of the header text safe to use for links
4113
4114                         # Remove link placeholders by the link text.
4115                         #     <!--LINK number-->
4116                         # turns into
4117                         #     link text with suffix
4118                         # Do this before unstrip since link text can contain strip markers
4119                         $safeHeadline = $this->replaceLinkHoldersText( $headline );
4120
4121                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4122                         $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4123
4124                         # Strip out HTML (first regex removes any tag not allowed)
4125                         # Allowed tags are <sup> and <sub> (bug 8393), <i> (bug 26375) and <b> (r105284)
4126                         # We strip any parameter from accepted tags (second regex)
4127                         $tocline = preg_replace(
4128                                 array( '#<(?!/?(sup|sub|i|b)(?: [^>]*)?>).*?'.'>#', '#<(/?(sup|sub|i|b))(?: .*?)?'.'>#' ),
4129                                 array( '',                          '<$1>' ),
4130                                 $safeHeadline
4131                         );
4132                         $tocline = trim( $tocline );
4133
4134                         # For the anchor, strip out HTML-y stuff period
4135                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4136                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4137
4138                         # Save headline for section edit hint before it's escaped
4139                         $headlineHint = $safeHeadline;
4140
4141                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4142                                 # For reverse compatibility, provide an id that's
4143                                 # HTML4-compatible, like we used to.
4144                                 #
4145                                 # It may be worth noting, academically, that it's possible for
4146                                 # the legacy anchor to conflict with a non-legacy headline
4147                                 # anchor on the page.  In this case likely the "correct" thing
4148                                 # would be to either drop the legacy anchors or make sure
4149                                 # they're numbered first.  However, this would require people
4150                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4151                                 # manually, so let's not bother worrying about it.
4152                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4153                                         array( 'noninitial', 'legacy' ) );
4154                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4155
4156                                 if ( $legacyHeadline == $safeHeadline ) {
4157                                         # No reason to have both (in fact, we can't)
4158                                         $legacyHeadline = false;
4159                                 }
4160                         } else {
4161                                 $legacyHeadline = false;
4162                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4163                                         'noninitial' );
4164                         }
4165
4166                         # HTML names must be case-insensitively unique (bug 10721).
4167                         # This does not apply to Unicode characters per
4168                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4169                         # @todo FIXME: We may be changing them depending on the current locale.
4170                         $arrayKey = strtolower( $safeHeadline );
4171                         if ( $legacyHeadline === false ) {
4172                                 $legacyArrayKey = false;
4173                         } else {
4174                                 $legacyArrayKey = strtolower( $legacyHeadline );
4175                         }
4176
4177                         # count how many in assoc. array so we can track dupes in anchors
4178                         if ( isset( $refers[$arrayKey] ) ) {
4179                                 $refers[$arrayKey]++;
4180                         } else {
4181                                 $refers[$arrayKey] = 1;
4182                         }
4183                         if ( isset( $refers[$legacyArrayKey] ) ) {
4184                                 $refers[$legacyArrayKey]++;
4185                         } else {
4186                                 $refers[$legacyArrayKey] = 1;
4187                         }
4188
4189                         # Don't number the heading if it is the only one (looks silly)
4190                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4191                                 # the two are different if the line contains a link
4192                                 $headline = Html::element( 'span', array( 'class' => 'mw-headline-number' ), $numbering ) . ' ' . $headline;
4193                         }
4194
4195                         # Create the anchor for linking from the TOC to the section
4196                         $anchor = $safeHeadline;
4197                         $legacyAnchor = $legacyHeadline;
4198                         if ( $refers[$arrayKey] > 1 ) {
4199                                 $anchor .= '_' . $refers[$arrayKey];
4200                         }
4201                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4202                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4203                         }
4204                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4205                                 $toc .= Linker::tocLine( $anchor, $tocline,
4206                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4207                         }
4208
4209                         # Add the section to the section tree
4210                         # Find the DOM node for this header
4211                         while ( $node && !$isTemplate ) {
4212                                 if ( $node->getName() === 'h' ) {
4213                                         $bits = $node->splitHeading();
4214                                         if ( $bits['i'] == $sectionIndex ) {
4215                                                 break;
4216                                         }
4217                                 }
4218                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4219                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4220                                 $node = $node->getNextSibling();
4221                         }
4222                         $tocraw[] = array(
4223                                 'toclevel' => $toclevel,
4224                                 'level' => $level,
4225                                 'line' => $tocline,
4226                                 'number' => $numbering,
4227                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4228                                 'fromtitle' => $titleText,
4229                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4230                                 'anchor' => $anchor,
4231                         );
4232
4233                         # give headline the correct <h#> tag
4234                         if ( $maybeShowEditLink && $sectionIndex !== false ) {
4235                                 // Output edit section links as markers with styles that can be customized by skins
4236                                 if ( $isTemplate ) {
4237                                         # Put a T flag in the section identifier, to indicate to extractSections()
4238                                         # that sections inside <includeonly> should be counted.
4239                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4240                                 } else {
4241                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4242                                 }
4243                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4244                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4245                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4246                                 // so we don't have to worry about a user trying to input one of these markers directly.
4247                                 // We use a page and section attribute to stop the language converter from converting these important bits
4248                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4249                                 // be able to convert that piece of data.
4250                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4251                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4252                                 if ( isset($editlinkArgs[2]) ) {
4253                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4254                                 } else {
4255                                         $editlink .= '/>';
4256                                 }
4257                         } else {
4258                                 $editlink = '';
4259                         }
4260                         $head[$headlineCount] = Linker::makeHeadline( $level,
4261                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4262                                 $editlink, $legacyAnchor );
4263
4264                         $headlineCount++;
4265                 }
4266
4267                 $this->setOutputType( $oldType );
4268
4269                 # Never ever show TOC if no headers
4270                 if ( $numVisible < 1 ) {
4271                         $enoughToc = false;
4272                 }
4273
4274                 if ( $enoughToc ) {
4275                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4276                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4277                         }
4278                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4279                         $this->mOutput->setTOCHTML( $toc );
4280                 }
4281
4282                 if ( $isMain ) {
4283                         $this->mOutput->setSections( $tocraw );
4284                 }
4285
4286                 # split up and insert constructed headlines
4287                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4288                 $i = 0;
4289
4290                 // build an array of document sections
4291                 $sections = array();
4292                 foreach ( $blocks as $block ) {
4293                         // $head is zero-based, sections aren't.
4294                         if ( empty( $head[$i - 1] ) ) {
4295                                 $sections[$i] = $block;
4296                         } else {
4297                                 $sections[$i] = $head[$i - 1] . $block;
4298                         }
4299
4300                         /**
4301                          * Send a hook, one per section.
4302                          * The idea here is to be able to make section-level DIVs, but to do so in a
4303                          * lower-impact, more correct way than r50769
4304                          *
4305                          * $this : caller
4306                          * $section : the section number
4307                          * &$sectionContent : ref to the content of the section
4308                          * $showEditLinks : boolean describing whether this section has an edit link
4309                          */
4310                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4311
4312                         $i++;
4313                 }
4314
4315                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4316                         // append the TOC at the beginning
4317                         // Top anchor now in skin
4318                         $sections[0] = $sections[0] . $toc . "\n";
4319                 }
4320
4321                 $full .= join( '', $sections );
4322
4323                 if ( $this->mForceTocPosition ) {
4324                         return str_replace( '<!--MWTOC-->', $toc, $full );
4325                 } else {
4326                         return $full;
4327                 }
4328         }
4329
4330         /**
4331          * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4332          * conversion, substitting signatures, {{subst:}} templates, etc.
4333          *
4334          * @param $text String: the text to transform
4335          * @param $title Title: the Title object for the current article
4336          * @param $user User: the User object describing the current user
4337          * @param $options ParserOptions: parsing options
4338          * @param $clearState Boolean: whether to clear the parser state first
4339          * @return String: the altered wiki markup
4340          */
4341         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4342                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4343                 $this->setUser( $user );
4344
4345                 $pairs = array(
4346                         "\r\n" => "\n",
4347                 );
4348                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4349                 if( $options->getPreSaveTransform() ) {
4350                         $text = $this->pstPass2( $text, $user );
4351                 }
4352                 $text = $this->mStripState->unstripBoth( $text );
4353
4354                 $this->setUser( null ); #Reset
4355
4356                 return $text;
4357         }
4358
4359         /**
4360          * Pre-save transform helper function
4361          * @private
4362          *
4363          * @param $text string
4364          * @param $user User
4365          *
4366          * @return string
4367          */
4368         function pstPass2( $text, $user ) {
4369                 global $wgContLang, $wgLocaltimezone;
4370
4371                 # Note: This is the timestamp saved as hardcoded wikitext to
4372                 # the database, we use $wgContLang here in order to give
4373                 # everyone the same signature and use the default one rather
4374                 # than the one selected in each user's preferences.
4375                 # (see also bug 12815)
4376                 $ts = $this->mOptions->getTimestamp();
4377                 if ( isset( $wgLocaltimezone ) ) {
4378                         $tz = $wgLocaltimezone;
4379                 } else {
4380                         $tz = date_default_timezone_get();
4381                 }
4382
4383                 $unixts = wfTimestamp( TS_UNIX, $ts );
4384                 $oldtz = date_default_timezone_get();
4385                 date_default_timezone_set( $tz );
4386                 $ts = date( 'YmdHis', $unixts );
4387                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4388
4389                 # Allow translation of timezones through wiki. date() can return
4390                 # whatever crap the system uses, localised or not, so we cannot
4391                 # ship premade translations.
4392                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4393                 $msg = wfMessage( $key )->inContentLanguage();
4394                 if ( $msg->exists() ) {
4395                         $tzMsg = $msg->text();
4396                 }
4397
4398                 date_default_timezone_set( $oldtz );
4399
4400                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4401
4402                 # Variable replacement
4403                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4404                 $text = $this->replaceVariables( $text );
4405
4406                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4407                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4408
4409                 # Signatures
4410                 $sigText = $this->getUserSig( $user );
4411                 $text = strtr( $text, array(
4412                         '~~~~~' => $d,
4413                         '~~~~' => "$sigText $d",
4414                         '~~~' => $sigText
4415                 ) );
4416
4417                 # Context links: [[|name]] and [[name (context)|]]
4418                 $tc = '[' . Title::legalChars() . ']';
4419                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4420
4421                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";           # [[ns:page (context)|]]
4422                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";           # [[ns:page（context）|]]
4423                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/"; # [[ns:page (context), context|]]
4424                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4425
4426                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4427                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4428                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4429                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4430
4431                 $t = $this->mTitle->getText();
4432                 $m = array();
4433                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4434                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4435                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4436                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4437                 } else {
4438                         # if there's no context, don't bother duplicating the title
4439                         $text = preg_replace( $p2, '[[\\1]]', $text );
4440                 }
4441
4442                 # Trim trailing whitespace
4443                 $text = rtrim( $text );
4444
4445                 return $text;
4446         }
4447
4448         /**
4449          * Fetch the user's signature text, if any, and normalize to
4450          * validated, ready-to-insert wikitext.
4451          * If you have pre-fetched the nickname or the fancySig option, you can
4452          * specify them here to save a database query.
4453          * Do not reuse this parser instance after calling getUserSig(),
4454          * as it may have changed if it's the $wgParser.
4455          *
4456          * @param $user User
4457          * @param $nickname String|bool nickname to use or false to use user's default nickname
4458          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4459          *                  or null to use default value
4460          * @return string
4461          */
4462         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4463                 global $wgMaxSigChars;
4464
4465                 $username = $user->getName();
4466
4467                 # If not given, retrieve from the user object.
4468                 if ( $nickname === false )
4469                         $nickname = $user->getOption( 'nickname' );
4470
4471                 if ( is_null( $fancySig ) ) {
4472                         $fancySig = $user->getBoolOption( 'fancysig' );
4473                 }
4474
4475                 $nickname = $nickname == null ? $username : $nickname;
4476
4477                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4478                         $nickname = $username;
4479                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4480                 } elseif ( $fancySig !== false ) {
4481                         # Sig. might contain markup; validate this
4482                         if ( $this->validateSig( $nickname ) !== false ) {
4483                                 # Validated; clean up (if needed) and return it
4484                                 return $this->cleanSig( $nickname, true );
4485                         } else {
4486                                 # Failed to validate; fall back to the default
4487                                 $nickname = $username;
4488                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4489                         }
4490                 }
4491
4492                 # Make sure nickname doesnt get a sig in a sig
4493                 $nickname = self::cleanSigInSig( $nickname );
4494
4495                 # If we're still here, make it a link to the user page
4496                 $userText = wfEscapeWikiText( $username );
4497                 $nickText = wfEscapeWikiText( $nickname );
4498                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4499
4500                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4501         }
4502
4503         /**
4504          * Check that the user's signature contains no bad XML
4505          *
4506          * @param $text String
4507          * @return mixed An expanded string, or false if invalid.
4508          */
4509         function validateSig( $text ) {
4510                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4511         }
4512
4513         /**
4514          * Clean up signature text
4515          *
4516          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4517          * 2) Substitute all transclusions
4518          *
4519          * @param $text String
4520          * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4521          * @return String: signature text
4522          */
4523         public function cleanSig( $text, $parsing = false ) {
4524                 if ( !$parsing ) {
4525                         global $wgTitle;
4526                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4527                 }
4528
4529                 # Option to disable this feature
4530                 if ( !$this->mOptions->getCleanSignatures() ) {
4531                         return $text;
4532                 }
4533
4534                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4535                 #  => Move this logic to braceSubstitution()
4536                 $substWord = MagicWord::get( 'subst' );
4537                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4538                 $substText = '{{' . $substWord->getSynonym( 0 );
4539
4540                 $text = preg_replace( $substRegex, $substText, $text );
4541                 $text = self::cleanSigInSig( $text );
4542                 $dom = $this->preprocessToDom( $text );
4543                 $frame = $this->getPreprocessor()->newFrame();
4544                 $text = $frame->expand( $dom );
4545
4546                 if ( !$parsing ) {
4547                         $text = $this->mStripState->unstripBoth( $text );
4548                 }
4549
4550                 return $text;
4551         }
4552
4553         /**
4554          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4555          *
4556          * @param $text String
4557          * @return String: signature text with /~{3,5}/ removed
4558          */
4559         public static function cleanSigInSig( $text ) {
4560                 $text = preg_replace( '/~{3,5}/', '', $text );
4561                 return $text;
4562         }
4563
4564         /**
4565          * Set up some variables which are usually set up in parse()
4566          * so that an external function can call some class members with confidence
4567          *
4568          * @param $title Title|null
4569          * @param $options ParserOptions
4570          * @param $outputType
4571          * @param $clearState bool
4572          */
4573         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4574                 $this->startParse( $title, $options, $outputType, $clearState );
4575         }
4576
4577         /**
4578          * @param $title Title|null
4579          * @param $options ParserOptions
4580          * @param $outputType
4581          * @param $clearState bool
4582          */
4583         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4584                 $this->setTitle( $title );
4585                 $this->mOptions = $options;
4586                 $this->setOutputType( $outputType );
4587                 if ( $clearState ) {
4588                         $this->clearState();
4589                 }
4590         }
4591
4592         /**
4593          * Wrapper for preprocess()
4594          *
4595          * @param $text String: the text to preprocess
4596          * @param $options ParserOptions: options
4597          * @param $title Title object or null to use $wgTitle
4598          * @return String
4599          */
4600         public function transformMsg( $text, $options, $title = null ) {
4601                 static $executing = false;
4602
4603                 # Guard against infinite recursion
4604                 if ( $executing ) {
4605                         return $text;
4606                 }
4607                 $executing = true;
4608
4609                 wfProfileIn( __METHOD__ );
4610                 if ( !$title ) {
4611                         global $wgTitle;
4612                         $title = $wgTitle;
4613                 }
4614                 if ( !$title ) {
4615                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4616                         # Create a ghost title in such case
4617                         $title = Title::newFromText( 'Dwimmerlaik' );
4618                 }
4619                 $text = $this->preprocess( $text, $title, $options );
4620
4621                 $executing = false;
4622                 wfProfileOut( __METHOD__ );
4623                 return $text;
4624         }
4625
4626         /**
4627          * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4628          * The callback should have the following form:
4629          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4630          *
4631          * Transform and return $text. Use $parser for any required context, e.g. use
4632          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4633          *
4634          * Hooks may return extended information by returning an array, of which the
4635          * first numbered element (index 0) must be the return string, and all other
4636          * entries are extracted into local variables within an internal function
4637          * in the Parser class.
4638          *
4639          * This interface (introduced r61913) appears to be undocumented, but
4640          * 'markerName' is used by some core tag hooks to override which strip
4641          * array their results are placed in. **Use great caution if attempting
4642          * this interface, as it is not documented and injudicious use could smash
4643          * private variables.**
4644          *
4645          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4646          * @param $callback Mixed: the callback function (and object) to use for the tag
4647          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4648          */
4649         public function setHook( $tag, $callback ) {
4650                 $tag = strtolower( $tag );
4651                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4652                         throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4653                 }
4654                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4655                 $this->mTagHooks[$tag] = $callback;
4656                 if ( !in_array( $tag, $this->mStripList ) ) {
4657                         $this->mStripList[] = $tag;
4658                 }
4659
4660                 return $oldVal;
4661         }
4662
4663         /**
4664          * As setHook(), but letting the contents be parsed.
4665          *
4666          * Transparent tag hooks are like regular XML-style tag hooks, except they
4667          * operate late in the transformation sequence, on HTML instead of wikitext.
4668          *
4669          * This is probably obsoleted by things dealing with parser frames?
4670          * The only extension currently using it is geoserver.
4671          *
4672          * @since 1.10
4673          * @todo better document or deprecate this
4674          *
4675          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4676          * @param $callback Mixed: the callback function (and object) to use for the tag
4677          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4678          */
4679         function setTransparentTagHook( $tag, $callback ) {
4680                 $tag = strtolower( $tag );
4681                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4682                         throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4683                 }
4684                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4685                 $this->mTransparentTagHooks[$tag] = $callback;
4686
4687                 return $oldVal;
4688         }
4689
4690         /**
4691          * Remove all tag hooks
4692          */
4693         function clearTagHooks() {
4694                 $this->mTagHooks = array();
4695                 $this->mFunctionTagHooks = array();
4696                 $this->mStripList = $this->mDefaultStripList;
4697         }
4698
4699         /**
4700          * Create a function, e.g. {{sum:1|2|3}}
4701          * The callback function should have the form:
4702          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4703          *
4704          * Or with SFH_OBJECT_ARGS:
4705          *    function myParserFunction( $parser, $frame, $args ) { ... }
4706          *
4707          * The callback may either return the text result of the function, or an array with the text
4708          * in element 0, and a number of flags in the other elements. The names of the flags are
4709          * specified in the keys. Valid flags are:
4710          *   found                     The text returned is valid, stop processing the template. This
4711          *                             is on by default.
4712          *   nowiki                    Wiki markup in the return value should be escaped
4713          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4714          *
4715          * @param $id String: The magic word ID
4716          * @param $callback Mixed: the callback function (and object) to use
4717          * @param $flags Integer: a combination of the following flags:
4718          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4719          *
4720          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4721          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4722          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4723          *     the arguments, and to control the way they are expanded.
4724          *
4725          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4726          *     arguments, for instance:
4727          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4728          *
4729          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4730          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4731          *     working if/when this is changed.
4732          *
4733          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4734          *     expansion.
4735          *
4736          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4737          *     about the methods available in PPFrame and PPNode.
4738          *
4739          * @return string|callback The old callback function for this name, if any
4740          */
4741         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4742                 global $wgContLang;
4743
4744                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4745                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4746
4747                 # Add to function cache
4748                 $mw = MagicWord::get( $id );
4749                 if ( !$mw )
4750                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4751
4752                 $synonyms = $mw->getSynonyms();
4753                 $sensitive = intval( $mw->isCaseSensitive() );
4754
4755                 foreach ( $synonyms as $syn ) {
4756                         # Case
4757                         if ( !$sensitive ) {
4758                                 $syn = $wgContLang->lc( $syn );
4759                         }
4760                         # Add leading hash
4761                         if ( !( $flags & SFH_NO_HASH ) ) {
4762                                 $syn = '#' . $syn;
4763                         }
4764                         # Remove trailing colon
4765                         if ( substr( $syn, -1, 1 ) === ':' ) {
4766                                 $syn = substr( $syn, 0, -1 );
4767                         }
4768                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4769                 }
4770                 return $oldVal;
4771         }
4772
4773         /**
4774          * Get all registered function hook identifiers
4775          *
4776          * @return Array
4777          */
4778         function getFunctionHooks() {
4779                 return array_keys( $this->mFunctionHooks );
4780         }
4781
4782         /**
4783          * Create a tag function, e.g. "<test>some stuff</test>".
4784          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4785          * Unlike parser functions, their content is not preprocessed.
4786          * @return null
4787          */
4788         function setFunctionTagHook( $tag, $callback, $flags ) {
4789                 $tag = strtolower( $tag );
4790                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4791                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4792                         $this->mFunctionTagHooks[$tag] : null;
4793                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4794
4795                 if ( !in_array( $tag, $this->mStripList ) ) {
4796                         $this->mStripList[] = $tag;
4797                 }
4798
4799                 return $old;
4800         }
4801
4802         /**
4803          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4804          * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4805          * Placeholders created in Skin::makeLinkObj()
4806          *
4807          * @param $text string
4808          * @param $options int
4809          *
4810          * @return array of link CSS classes, indexed by PDBK.
4811          */
4812         function replaceLinkHolders( &$text, $options = 0 ) {
4813                 return $this->mLinkHolders->replace( $text );
4814         }
4815
4816         /**
4817          * Replace "<!--LINK-->" link placeholders with plain text of links
4818          * (not HTML-formatted).
4819          *
4820          * @param $text String
4821          * @return String
4822          */
4823         function replaceLinkHoldersText( $text ) {
4824                 return $this->mLinkHolders->replaceText( $text );
4825         }
4826
4827         /**
4828          * Renders an image gallery from a text with one line per image.
4829          * text labels may be given by using |-style alternative text. E.g.
4830          *   Image:one.jpg|The number "1"
4831          *   Image:tree.jpg|A tree
4832          * given as text will return the HTML of a gallery with two images,
4833          * labeled 'The number "1"' and
4834          * 'A tree'.
4835          *
4836          * @param string $text
4837          * @param array $params
4838          * @return string HTML
4839          */
4840         function renderImageGallery( $text, $params ) {
4841                 $ig = new ImageGallery();
4842                 $ig->setContextTitle( $this->mTitle );
4843                 $ig->setShowBytes( false );
4844                 $ig->setShowFilename( false );
4845                 $ig->setParser( $this );
4846                 $ig->setHideBadImages();
4847                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4848
4849                 if ( isset( $params['showfilename'] ) ) {
4850                         $ig->setShowFilename( true );
4851                 } else {
4852                         $ig->setShowFilename( false );
4853                 }
4854                 if ( isset( $params['caption'] ) ) {
4855                         $caption = $params['caption'];
4856                         $caption = htmlspecialchars( $caption );
4857                         $caption = $this->replaceInternalLinks( $caption );
4858                         $ig->setCaptionHtml( $caption );
4859                 }
4860                 if ( isset( $params['perrow'] ) ) {
4861                         $ig->setPerRow( $params['perrow'] );
4862                 }
4863                 if ( isset( $params['widths'] ) ) {
4864                         $ig->setWidths( $params['widths'] );
4865                 }
4866                 if ( isset( $params['heights'] ) ) {
4867                         $ig->setHeights( $params['heights'] );
4868                 }
4869
4870                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4871
4872                 $lines = StringUtils::explode( "\n", $text );
4873                 foreach ( $lines as $line ) {
4874                         # match lines like these:
4875                         # Image:someimage.jpg|This is some image
4876                         $matches = array();
4877                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4878                         # Skip empty lines
4879                         if ( count( $matches ) == 0 ) {
4880                                 continue;
4881                         }
4882
4883                         if ( strpos( $matches[0], '%' ) !== false ) {
4884                                 $matches[1] = rawurldecode( $matches[1] );
4885                         }
4886                         $title = Title::newFromText( $matches[1], NS_FILE );
4887                         if ( is_null( $title ) ) {
4888                                 # Bogus title. Ignore these so we don't bomb out later.
4889                                 continue;
4890                         }
4891
4892                         $label = '';
4893                         $alt = '';
4894                         $link = '';
4895                         if ( isset( $matches[3] ) ) {
4896                                 // look for an |alt= definition while trying not to break existing
4897                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4898                                 // is defined for images in galleries
4899
4900                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4901                                 $parameterMatches = StringUtils::explode('|', $matches[3]);
4902                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4903                                 $magicWordLink = MagicWord::get( 'img_link' );
4904
4905                                 foreach ( $parameterMatches as $parameterMatch ) {
4906                                         if ( $match = $magicWordAlt->matchVariableStartToEnd( $parameterMatch ) ) {
4907                                                 $alt = $this->stripAltText( $match, false );
4908                                         }
4909                                         elseif( $match = $magicWordLink->matchVariableStartToEnd( $parameterMatch ) ){
4910                                                 $link = strip_tags($this->replaceLinkHoldersText($match));
4911                                                 $chars = self::EXT_LINK_URL_CLASS;
4912                                                 $prots = $this->mUrlProtocols;
4913                                                 //check to see if link matches an absolute url, if not then it must be a wiki link.
4914                                                 if(!preg_match( "/^($prots)$chars+$/u", $link)){
4915                                                         $localLinkTitle = Title::newFromText($link);
4916                                                         $link = $localLinkTitle->getLocalURL();
4917                                                 }
4918                                         }
4919                                         else {
4920                                                 // concatenate all other pipes
4921                                                 $label .= '|' . $parameterMatch;
4922                                         }
4923                                 }
4924                                 // remove the first pipe
4925                                 $label = substr( $label, 1 );
4926                         }
4927
4928                         $ig->add( $title, $label, $alt ,$link);
4929                 }
4930                 return $ig->toHTML();
4931         }
4932
4933         /**
4934          * @param $handler
4935          * @return array
4936          */
4937         function getImageParams( $handler ) {
4938                 if ( $handler ) {
4939                         $handlerClass = get_class( $handler );
4940                 } else {
4941                         $handlerClass = '';
4942                 }
4943                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4944                         # Initialise static lists
4945                         static $internalParamNames = array(
4946                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4947                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4948                                         'bottom', 'text-bottom' ),
4949                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4950                                         'upright', 'border', 'link', 'alt' ),
4951                         );
4952                         static $internalParamMap;
4953                         if ( !$internalParamMap ) {
4954                                 $internalParamMap = array();
4955                                 foreach ( $internalParamNames as $type => $names ) {
4956                                         foreach ( $names as $name ) {
4957                                                 $magicName = str_replace( '-', '_', "img_$name" );
4958                                                 $internalParamMap[$magicName] = array( $type, $name );
4959                                         }
4960                                 }
4961                         }
4962
4963                         # Add handler params
4964                         $paramMap = $internalParamMap;
4965                         if ( $handler ) {
4966                                 $handlerParamMap = $handler->getParamMap();
4967                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4968                                         $paramMap[$magic] = array( 'handler', $paramName );
4969                                 }
4970                         }
4971                         $this->mImageParams[$handlerClass] = $paramMap;
4972                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4973                 }
4974                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4975         }
4976
4977         /**
4978          * Parse image options text and use it to make an image
4979          *
4980          * @param $title Title
4981          * @param $options String
4982          * @param $holders LinkHolderArray|bool
4983          * @return string HTML
4984          */
4985         function makeImage( $title, $options, $holders = false ) {
4986                 # Check if the options text is of the form "options|alt text"
4987                 # Options are:
4988                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4989                 #  * left       no resizing, just left align. label is used for alt= only
4990                 #  * right      same, but right aligned
4991                 #  * none       same, but not aligned
4992                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4993                 #  * center     center the image
4994                 #  * frame      Keep original image size, no magnify-button.
4995                 #  * framed     Same as "frame"
4996                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4997                 #  * upright    reduce width for upright images, rounded to full __0 px
4998                 #  * border     draw a 1px border around the image
4999                 #  * alt        Text for HTML alt attribute (defaults to empty)
5000                 #  * link       Set the target of the image link. Can be external, interwiki, or local
5001                 # vertical-align values (no % or length right now):
5002                 #  * baseline
5003                 #  * sub
5004                 #  * super
5005                 #  * top
5006                 #  * text-top
5007                 #  * middle
5008                 #  * bottom
5009                 #  * text-bottom
5010
5011                 $parts = StringUtils::explode( "|", $options );
5012
5013                 # Give extensions a chance to select the file revision for us
5014                 $options = array();
5015                 $descQuery = false;
5016                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5017                         array( $this, $title, &$options, &$descQuery ) );
5018                 # Fetch and register the file (file title may be different via hooks)
5019                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5020
5021                 # Get parameter map
5022                 $handler = $file ? $file->getHandler() : false;
5023
5024                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5025
5026                 if ( !$file ) {
5027                         $this->addTrackingCategory( 'broken-file-category' );
5028                 }
5029
5030                 # Process the input parameters
5031                 $caption = '';
5032                 $params = array( 'frame' => array(), 'handler' => array(),
5033                         'horizAlign' => array(), 'vertAlign' => array() );
5034                 foreach ( $parts as $part ) {
5035                         $part = trim( $part );
5036                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5037                         $validated = false;
5038                         if ( isset( $paramMap[$magicName] ) ) {
5039                                 list( $type, $paramName ) = $paramMap[$magicName];
5040
5041                                 # Special case; width and height come in one variable together
5042                                 if ( $type === 'handler' && $paramName === 'width' ) {
5043                                         $m = array();
5044                                         # (bug 13500) In both cases (width/height and width only),
5045                                         # permit trailing "px" for backward compatibility.
5046                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5047                                                 $width = intval( $m[1] );
5048                                                 $height = intval( $m[2] );
5049                                                 if ( $handler->validateParam( 'width', $width ) ) {
5050                                                         $params[$type]['width'] = $width;
5051                                                         $validated = true;
5052                                                 }
5053                                                 if ( $handler->validateParam( 'height', $height ) ) {
5054                                                         $params[$type]['height'] = $height;
5055                                                         $validated = true;
5056                                                 }
5057                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5058                                                 $width = intval( $value );
5059                                                 if ( $handler->validateParam( 'width', $width ) ) {
5060                                                         $params[$type]['width'] = $width;
5061                                                         $validated = true;
5062                                                 }
5063                                         } # else no validation -- bug 13436
5064                                 } else {
5065                                         if ( $type === 'handler' ) {
5066                                                 # Validate handler parameter
5067                                                 $validated = $handler->validateParam( $paramName, $value );
5068                                         } else {
5069                                                 # Validate internal parameters
5070                                                 switch( $paramName ) {
5071                                                 case 'manualthumb':
5072                                                 case 'alt':
5073                                                         # @todo FIXME: Possibly check validity here for
5074                                                         # manualthumb? downstream behavior seems odd with
5075                                                         # missing manual thumbs.
5076                                                         $validated = true;
5077                                                         $value = $this->stripAltText( $value, $holders );
5078                                                         break;
5079                                                 case 'link':
5080                                                         $chars = self::EXT_LINK_URL_CLASS;
5081                                                         $prots = $this->mUrlProtocols;
5082                                                         if ( $value === '' ) {
5083                                                                 $paramName = 'no-link';
5084                                                                 $value = true;
5085                                                                 $validated = true;
5086                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
5087                                                                 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
5088                                                                         $paramName = 'link-url';
5089                                                                         $this->mOutput->addExternalLink( $value );
5090                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
5091                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5092                                                                         }
5093                                                                         $validated = true;
5094                                                                 }
5095                                                         } else {
5096                                                                 $linkTitle = Title::newFromText( $value );
5097                                                                 if ( $linkTitle ) {
5098                                                                         $paramName = 'link-title';
5099                                                                         $value = $linkTitle;
5100                                                                         $this->mOutput->addLink( $linkTitle );
5101                                                                         $validated = true;
5102                                                                 }
5103                                                         }
5104                                                         break;
5105                                                 default:
5106                                                         # Most other things appear to be empty or numeric...
5107                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
5108                                                 }
5109                                         }
5110
5111                                         if ( $validated ) {
5112                                                 $params[$type][$paramName] = $value;
5113                                         }
5114                                 }
5115                         }
5116                         if ( !$validated ) {
5117                                 $caption = $part;
5118                         }
5119                 }
5120
5121                 # Process alignment parameters
5122                 if ( $params['horizAlign'] ) {
5123                         $params['frame']['align'] = key( $params['horizAlign'] );
5124                 }
5125                 if ( $params['vertAlign'] ) {
5126                         $params['frame']['valign'] = key( $params['vertAlign'] );
5127                 }
5128
5129                 $params['frame']['caption'] = $caption;
5130
5131                 # Will the image be presented in a frame, with the caption below?
5132                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5133                                                  isset( $params['frame']['framed'] ) ||
5134                                                  isset( $params['frame']['thumbnail'] ) ||
5135                                                  isset( $params['frame']['manualthumb'] );
5136
5137                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5138                 # came to also set the caption, ordinary text after the image -- which
5139                 # makes no sense, because that just repeats the text multiple times in
5140                 # screen readers.  It *also* came to set the title attribute.
5141                 #
5142                 # Now that we have an alt attribute, we should not set the alt text to
5143                 # equal the caption: that's worse than useless, it just repeats the
5144                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5145                 # use the unnamed parameter for alt text as well, just for the time be-
5146                 # ing, if the unnamed param is set and the alt param is not.
5147                 #
5148                 # For the future, we need to figure out if we want to tweak this more,
5149                 # e.g., introducing a title= parameter for the title; ignoring the un-
5150                 # named parameter entirely for images without a caption; adding an ex-
5151                 # plicit caption= parameter and preserving the old magic unnamed para-
5152                 # meter for BC; ...
5153                 if ( $imageIsFramed ) { # Framed image
5154                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5155                                 # No caption or alt text, add the filename as the alt text so
5156                                 # that screen readers at least get some description of the image
5157                                 $params['frame']['alt'] = $title->getText();
5158                         }
5159                         # Do not set $params['frame']['title'] because tooltips don't make sense
5160                         # for framed images
5161                 } else { # Inline image
5162                         if ( !isset( $params['frame']['alt'] ) ) {
5163                                 # No alt text, use the "caption" for the alt text
5164                                 if ( $caption !== '') {
5165                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5166                                 } else {
5167                                         # No caption, fall back to using the filename for the
5168                                         # alt text
5169                                         $params['frame']['alt'] = $title->getText();
5170                                 }
5171                         }
5172                         # Use the "caption" for the tooltip text
5173                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5174                 }
5175
5176                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5177
5178                 # Linker does the rest
5179                 $time = isset( $options['time'] ) ? $options['time'] : false;
5180                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5181                         $time, $descQuery, $this->mOptions->getThumbSize() );
5182
5183                 # Give the handler a chance to modify the parser object
5184                 if ( $handler ) {
5185                         $handler->parserTransformHook( $this, $file );
5186                 }
5187
5188                 return $ret;
5189         }
5190
5191         /**
5192          * @param $caption
5193          * @param $holders LinkHolderArray
5194          * @return mixed|String
5195          */
5196         protected function stripAltText( $caption, $holders ) {
5197                 # Strip bad stuff out of the title (tooltip).  We can't just use
5198                 # replaceLinkHoldersText() here, because if this function is called
5199                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5200                 if ( $holders ) {
5201                         $tooltip = $holders->replaceText( $caption );
5202                 } else {
5203                         $tooltip = $this->replaceLinkHoldersText( $caption );
5204                 }
5205
5206                 # make sure there are no placeholders in thumbnail attributes
5207                 # that are later expanded to html- so expand them now and
5208                 # remove the tags
5209                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5210                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5211
5212                 return $tooltip;
5213         }
5214
5215         /**
5216          * Set a flag in the output object indicating that the content is dynamic and
5217          * shouldn't be cached.
5218          */
5219         function disableCache() {
5220                 wfDebug( "Parser output marked as uncacheable.\n" );
5221                 if ( !$this->mOutput ) {
5222                         throw new MWException( __METHOD__ .
5223                                 " can only be called when actually parsing something" );
5224                 }
5225                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5226                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5227         }
5228
5229         /**
5230          * Callback from the Sanitizer for expanding items found in HTML attribute
5231          * values, so they can be safely tested and escaped.
5232          *
5233          * @param $text String
5234          * @param $frame PPFrame
5235          * @return String
5236          */
5237         function attributeStripCallback( &$text, $frame = false ) {
5238                 $text = $this->replaceVariables( $text, $frame );
5239                 $text = $this->mStripState->unstripBoth( $text );
5240                 return $text;
5241         }
5242
5243         /**
5244          * Accessor
5245          *
5246          * @return array
5247          */
5248         function getTags() {
5249                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ), array_keys( $this->mFunctionTagHooks ) );
5250         }
5251
5252         /**
5253          * Replace transparent tags in $text with the values given by the callbacks.
5254          *
5255          * Transparent tag hooks are like regular XML-style tag hooks, except they
5256          * operate late in the transformation sequence, on HTML instead of wikitext.
5257          *
5258          * @param $text string
5259          *
5260          * @return string
5261          */
5262         function replaceTransparentTags( $text ) {
5263                 $matches = array();
5264                 $elements = array_keys( $this->mTransparentTagHooks );
5265                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5266                 $replacements = array();
5267
5268                 foreach ( $matches as $marker => $data ) {
5269                         list( $element, $content, $params, $tag ) = $data;
5270                         $tagName = strtolower( $element );
5271                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5272                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5273                         } else {
5274                                 $output = $tag;
5275                         }
5276                         $replacements[$marker] = $output;
5277                 }
5278                 return strtr( $text, $replacements );
5279         }
5280
5281         /**
5282          * Break wikitext input into sections, and either pull or replace
5283          * some particular section's text.
5284          *
5285          * External callers should use the getSection and replaceSection methods.
5286          *
5287          * @param $text String: Page wikitext
5288          * @param $section String: a section identifier string of the form:
5289          *   "<flag1> - <flag2> - ... - <section number>"
5290          *
5291          * Currently the only recognised flag is "T", which means the target section number
5292          * was derived during a template inclusion parse, in other words this is a template
5293          * section edit link. If no flags are given, it was an ordinary section edit link.
5294          * This flag is required to avoid a section numbering mismatch when a section is
5295          * enclosed by "<includeonly>" (bug 6563).
5296          *
5297          * The section number 0 pulls the text before the first heading; other numbers will
5298          * pull the given section along with its lower-level subsections. If the section is
5299          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5300          *
5301          * Section 0 is always considered to exist, even if it only contains the empty
5302          * string. If $text is the empty string and section 0 is replaced, $newText is
5303          * returned.
5304          *
5305          * @param $mode String: one of "get" or "replace"
5306          * @param $newText String: replacement text for section data.
5307          * @return String: for "get", the extracted section text.
5308          *                 for "replace", the whole page with the section replaced.
5309          */
5310         private function extractSections( $text, $section, $mode, $newText='' ) {
5311                 global $wgTitle; # not generally used but removes an ugly failure mode
5312                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5313                 $outText = '';
5314                 $frame = $this->getPreprocessor()->newFrame();
5315
5316                 # Process section extraction flags
5317                 $flags = 0;
5318                 $sectionParts = explode( '-', $section );
5319                 $sectionIndex = array_pop( $sectionParts );
5320                 foreach ( $sectionParts as $part ) {
5321                         if ( $part === 'T' ) {
5322                                 $flags |= self::PTD_FOR_INCLUSION;
5323                         }
5324                 }
5325
5326                 # Check for empty input
5327                 if ( strval( $text ) === '' ) {
5328                         # Only sections 0 and T-0 exist in an empty document
5329                         if ( $sectionIndex == 0 ) {
5330                                 if ( $mode === 'get' ) {
5331                                         return '';
5332                                 } else {
5333                                         return $newText;
5334                                 }
5335                         } else {
5336                                 if ( $mode === 'get' ) {
5337                                         return $newText;
5338                                 } else {
5339                                         return $text;
5340                                 }
5341                         }
5342                 }
5343
5344                 # Preprocess the text
5345                 $root = $this->preprocessToDom( $text, $flags );
5346
5347                 # <h> nodes indicate section breaks
5348                 # They can only occur at the top level, so we can find them by iterating the root's children
5349                 $node = $root->getFirstChild();
5350
5351                 # Find the target section
5352                 if ( $sectionIndex == 0 ) {
5353                         # Section zero doesn't nest, level=big
5354                         $targetLevel = 1000;
5355                 } else {
5356                         while ( $node ) {
5357                                 if ( $node->getName() === 'h' ) {
5358                                         $bits = $node->splitHeading();
5359                                         if ( $bits['i'] == $sectionIndex ) {
5360                                                 $targetLevel = $bits['level'];
5361                                                 break;
5362                                         }
5363                                 }
5364                                 if ( $mode === 'replace' ) {
5365                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5366                                 }
5367                                 $node = $node->getNextSibling();
5368                         }
5369                 }
5370
5371                 if ( !$node ) {
5372                         # Not found
5373                         if ( $mode === 'get' ) {
5374                                 return $newText;
5375                         } else {
5376                                 return $text;
5377                         }
5378                 }
5379
5380                 # Find the end of the section, including nested sections
5381                 do {
5382                         if ( $node->getName() === 'h' ) {
5383                                 $bits = $node->splitHeading();
5384                                 $curLevel = $bits['level'];
5385                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5386                                         break;
5387                                 }
5388                         }
5389                         if ( $mode === 'get' ) {
5390                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5391                         }
5392                         $node = $node->getNextSibling();
5393                 } while ( $node );
5394
5395                 # Write out the remainder (in replace mode only)
5396                 if ( $mode === 'replace' ) {
5397                         # Output the replacement text
5398                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5399                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5400                         # Only add trailing whitespace if there is newText
5401                         if ( $newText != "" ) {
5402                                 $outText .= $newText . "\n\n";
5403                         }
5404
5405                         while ( $node ) {
5406                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5407                                 $node = $node->getNextSibling();
5408                         }
5409                 }
5410
5411                 if ( is_string( $outText ) ) {
5412                         # Re-insert stripped tags
5413                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5414                 }
5415
5416                 return $outText;
5417         }
5418
5419         /**
5420          * This function returns the text of a section, specified by a number ($section).
5421          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5422          * the first section before any such heading (section 0).
5423          *
5424          * If a section contains subsections, these are also returned.
5425          *
5426          * @param $text String: text to look in
5427          * @param $section String: section identifier
5428          * @param $deftext String: default to return if section is not found
5429          * @return string text of the requested section
5430          */
5431         public function getSection( $text, $section, $deftext='' ) {
5432                 return $this->extractSections( $text, $section, "get", $deftext );
5433         }
5434
5435         /**
5436          * This function returns $oldtext after the content of the section
5437          * specified by $section has been replaced with $text. If the target
5438          * section does not exist, $oldtext is returned unchanged.
5439          *
5440          * @param $oldtext String: former text of the article
5441          * @param $section int section identifier
5442          * @param $text String: replacing text
5443          * @return String: modified text
5444          */
5445         public function replaceSection( $oldtext, $section, $text ) {
5446                 return $this->extractSections( $oldtext, $section, "replace", $text );
5447         }
5448
5449         /**
5450          * Get the ID of the revision we are parsing
5451          *
5452          * @return Mixed: integer or null
5453          */
5454         function getRevisionId() {
5455                 return $this->mRevisionId;
5456         }
5457
5458         /**
5459          * Get the revision object for $this->mRevisionId
5460          *
5461          * @return Revision|null either a Revision object or null
5462          */
5463         protected function getRevisionObject() {
5464                 if ( !is_null( $this->mRevisionObject ) ) {
5465                         return $this->mRevisionObject;
5466                 }
5467                 if ( is_null( $this->mRevisionId ) ) {
5468                         return null;
5469                 }
5470
5471                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5472                 return $this->mRevisionObject;
5473         }
5474
5475         /**
5476          * Get the timestamp associated with the current revision, adjusted for
5477          * the default server-local timestamp
5478          */
5479         function getRevisionTimestamp() {
5480                 if ( is_null( $this->mRevisionTimestamp ) ) {
5481                         wfProfileIn( __METHOD__ );
5482
5483                         global $wgContLang;
5484
5485                         $revObject = $this->getRevisionObject();
5486                         $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5487
5488                         # The cryptic '' timezone parameter tells to use the site-default
5489                         # timezone offset instead of the user settings.
5490                         #
5491                         # Since this value will be saved into the parser cache, served
5492                         # to other users, and potentially even used inside links and such,
5493                         # it needs to be consistent for all visitors.
5494                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5495
5496                         wfProfileOut( __METHOD__ );
5497                 }
5498                 return $this->mRevisionTimestamp;
5499         }
5500
5501         /**
5502          * Get the name of the user that edited the last revision
5503          *
5504          * @return String: user name
5505          */
5506         function getRevisionUser() {
5507                 if( is_null( $this->mRevisionUser ) ) {
5508                         $revObject = $this->getRevisionObject();
5509
5510                         # if this template is subst: the revision id will be blank,
5511                         # so just use the current user's name
5512                         if( $revObject ) {
5513                                 $this->mRevisionUser = $revObject->getUserText();
5514                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5515                                 $this->mRevisionUser = $this->getUser()->getName();
5516                         }
5517                 }
5518                 return $this->mRevisionUser;
5519         }
5520
5521         /**
5522          * Mutator for $mDefaultSort
5523          *
5524          * @param $sort string New value
5525          */
5526         public function setDefaultSort( $sort ) {
5527                 $this->mDefaultSort = $sort;
5528                 $this->mOutput->setProperty( 'defaultsort', $sort );
5529         }
5530
5531         /**
5532          * Accessor for $mDefaultSort
5533          * Will use the empty string if none is set.
5534          *
5535          * This value is treated as a prefix, so the
5536          * empty string is equivalent to sorting by
5537          * page name.
5538          *
5539          * @return string
5540          */
5541         public function getDefaultSort() {
5542                 if ( $this->mDefaultSort !== false ) {
5543                         return $this->mDefaultSort;
5544                 } else {
5545                         return '';
5546                 }
5547         }
5548
5549         /**
5550          * Accessor for $mDefaultSort
5551          * Unlike getDefaultSort(), will return false if none is set
5552          *
5553          * @return string or false
5554          */
5555         public function getCustomDefaultSort() {
5556                 return $this->mDefaultSort;
5557         }
5558
5559         /**
5560          * Try to guess the section anchor name based on a wikitext fragment
5561          * presumably extracted from a heading, for example "Header" from
5562          * "== Header ==".
5563          *
5564          * @param $text string
5565          *
5566          * @return string
5567          */
5568         public function guessSectionNameFromWikiText( $text ) {
5569                 # Strip out wikitext links(they break the anchor)
5570                 $text = $this->stripSectionName( $text );
5571                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5572                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5573         }
5574
5575         /**
5576          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5577          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5578          * as something other than UTF-8 (apparently?), resulting in breakage.
5579          *
5580          * @param $text String: The section name
5581          * @return string An anchor
5582          */
5583         public function guessLegacySectionNameFromWikiText( $text ) {
5584                 # Strip out wikitext links(they break the anchor)
5585                 $text = $this->stripSectionName( $text );
5586                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5587                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5588         }
5589
5590         /**
5591          * Strips a text string of wikitext for use in a section anchor
5592          *
5593          * Accepts a text string and then removes all wikitext from the
5594          * string and leaves only the resultant text (i.e. the result of
5595          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5596          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5597          * to create valid section anchors by mimicing the output of the
5598          * parser when headings are parsed.
5599          *
5600          * @param $text String: text string to be stripped of wikitext
5601          * for use in a Section anchor
5602          * @return string Filtered text string
5603          */
5604         public function stripSectionName( $text ) {
5605                 # Strip internal link markup
5606                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5607                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5608
5609                 # Strip external link markup
5610                 # @todo FIXME: Not tolerant to blank link text
5611                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5612                 # on how many empty links there are on the page - need to figure that out.
5613                 $text = preg_replace( '/\[(?:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5614
5615                 # Parse wikitext quotes (italics & bold)
5616                 $text = $this->doQuotes( $text );
5617
5618                 # Strip HTML tags
5619                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5620                 return $text;
5621         }
5622
5623         /**
5624          * strip/replaceVariables/unstrip for preprocessor regression testing
5625          *
5626          * @param $text string
5627          * @param $title Title
5628          * @param $options ParserOptions
5629          * @param $outputType int
5630          *
5631          * @return string
5632          */
5633         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5634                 $this->startParse( $title, $options, $outputType, true );
5635
5636                 $text = $this->replaceVariables( $text );
5637                 $text = $this->mStripState->unstripBoth( $text );
5638                 $text = Sanitizer::removeHTMLtags( $text );
5639                 return $text;
5640         }
5641
5642         /**
5643          * @param $text string
5644          * @param $title Title
5645          * @param $options ParserOptions
5646          * @return string
5647          */
5648         function testPst( $text, Title $title, ParserOptions $options ) {
5649                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5650         }
5651
5652         /**
5653          * @param $text
5654          * @param $title Title
5655          * @param $options ParserOptions
5656          * @return string
5657          */
5658         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5659                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5660         }
5661
5662         /**
5663          * Call a callback function on all regions of the given text that are not
5664          * inside strip markers, and replace those regions with the return value
5665          * of the callback. For example, with input:
5666          *
5667          *  aaa<MARKER>bbb
5668          *
5669          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5670          * two strings will be replaced with the value returned by the callback in
5671          * each case.
5672          *
5673          * @param $s string
5674          * @param $callback
5675          *
5676          * @return string
5677          */
5678         function markerSkipCallback( $s, $callback ) {
5679                 $i = 0;
5680                 $out = '';
5681                 while ( $i < strlen( $s ) ) {
5682                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5683                         if ( $markerStart === false ) {
5684                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5685                                 break;
5686                         } else {
5687                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5688                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5689                                 if ( $markerEnd === false ) {
5690                                         $out .= substr( $s, $markerStart );
5691                                         break;
5692                                 } else {
5693                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5694                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5695                                         $i = $markerEnd;
5696                                 }
5697                         }
5698                 }
5699                 return $out;
5700         }
5701
5702         /**
5703          * Remove any strip markers found in the given text.
5704          *
5705          * @param $text Input string
5706          * @return string
5707          */
5708         function killMarkers( $text ) {
5709                 return $this->mStripState->killMarkers( $text );
5710         }
5711
5712         /**
5713          * Save the parser state required to convert the given half-parsed text to
5714          * HTML. "Half-parsed" in this context means the output of
5715          * recursiveTagParse() or internalParse(). This output has strip markers
5716          * from replaceVariables (extensionSubstitution() etc.), and link
5717          * placeholders from replaceLinkHolders().
5718          *
5719          * Returns an array which can be serialized and stored persistently. This
5720          * array can later be loaded into another parser instance with
5721          * unserializeHalfParsedText(). The text can then be safely incorporated into
5722          * the return value of a parser hook.
5723          *
5724          * @param $text string
5725          *
5726          * @return array
5727          */
5728         function serializeHalfParsedText( $text ) {
5729                 wfProfileIn( __METHOD__ );
5730                 $data = array(
5731                         'text' => $text,
5732                         'version' => self::HALF_PARSED_VERSION,
5733                         'stripState' => $this->mStripState->getSubState( $text ),
5734                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5735                 );
5736                 wfProfileOut( __METHOD__ );
5737                 return $data;
5738         }
5739
5740         /**
5741          * Load the parser state given in the $data array, which is assumed to
5742          * have been generated by serializeHalfParsedText(). The text contents is
5743          * extracted from the array, and its markers are transformed into markers
5744          * appropriate for the current Parser instance. This transformed text is
5745          * returned, and can be safely included in the return value of a parser
5746          * hook.
5747          *
5748          * If the $data array has been stored persistently, the caller should first
5749          * check whether it is still valid, by calling isValidHalfParsedText().
5750          *
5751          * @param $data array Serialized data
5752          * @return String
5753          */
5754         function unserializeHalfParsedText( $data ) {
5755                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5756                         throw new MWException( __METHOD__.': invalid version' );
5757                 }
5758
5759                 # First, extract the strip state.
5760                 $texts = array( $data['text'] );
5761                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5762
5763                 # Now renumber links
5764                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5765
5766                 # Should be good to go.
5767                 return $texts[0];
5768         }
5769
5770         /**
5771          * Returns true if the given array, presumed to be generated by
5772          * serializeHalfParsedText(), is compatible with the current version of the
5773          * parser.
5774          *
5775          * @param $data Array
5776          *
5777          * @return bool
5778          */
5779         function isValidHalfParsedText( $data ) {
5780                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5781         }
5782 }