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  * There are seven main entry points into the Parser class:
  35  *
  36  * - Parser::parse()
  37  *     produces HTML output
  38  * - Parser::preSaveTransform().
  39  *     produces altered wiki markup.
  40  * - Parser::preprocess()
  41  *     removes HTML comments and expands templates
  42  * - Parser::cleanSig() and Parser::cleanSigInSig()
  43  *     Cleans a signature before saving it to preferences
  44  * - Parser::getSection()
  45  *     Return the content of a section from an article for section editing
  46  * - Parser::replaceSection()
  47  *     Replaces a section by number inside an article
  48  * - Parser::getPreloadText()
  49  *     Removes <noinclude> sections, and <includeonly> tags.
  50  *
  51  * Globals used:
  52  *    object: $wgContLang
  53  *
  54  * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
  55  *
  56  * @par Settings:
  57  * $wgLocaltimezone
  58  * $wgNamespacesWithSubpages
  59  *
  60  * @par Settings only within ParserOptions:
  61  * $wgAllowExternalImages
  62  * $wgAllowSpecialInclusion
  63  * $wgInterwikiMagic
  64  * $wgMaxArticleSize
  65  * $wgUseDynamicDates
  66  *
  67  * @ingroup Parser
  68  */
  69 class Parser {
  70         /**
  71          * Update this version number when the ParserOutput format
  72          * changes in an incompatible way, so the parser cache
  73          * can automatically discard old data.
  74          */
  75         const VERSION = '1.6.4';
  76
  77         /**
  78          * Update this version number when the output of serialiseHalfParsedText()
  79          * changes in an incompatible way
  80          */
  81         const HALF_PARSED_VERSION = 2;
  82
  83         # Flags for Parser::setFunctionHook
  84         # Also available as global constants from Defines.php
  85         const SFH_NO_HASH = 1;
  86         const SFH_OBJECT_ARGS = 2;
  87
  88         # Constants needed for external link processing
  89         # Everything except bracket, space, or control characters
  90         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  91         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  92         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  93         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  94                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
  95
  96         # State constants for the definition list colon extraction
  97         const COLON_STATE_TEXT = 0;
  98         const COLON_STATE_TAG = 1;
  99         const COLON_STATE_TAGSTART = 2;
 100         const COLON_STATE_CLOSETAG = 3;
 101         const COLON_STATE_TAGSLASH = 4;
 102         const COLON_STATE_COMMENT = 5;
 103         const COLON_STATE_COMMENTDASH = 6;
 104         const COLON_STATE_COMMENTDASHDASH = 7;
 105
 106         # Flags for preprocessToDom
 107         const PTD_FOR_INCLUSION = 1;
 108
 109         # Allowed values for $this->mOutputType
 110         # Parameter to startExternalParse().
 111         const OT_HTML = 1; # like parse()
 112         const OT_WIKI = 2; # like preSaveTransform()
 113         const OT_PREPROCESS = 3; # like preprocess()
 114         const OT_MSG = 3;
 115         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
 116
 117         # Marker Suffix needs to be accessible staticly.
 118         const MARKER_SUFFIX = "-QINU\x7f";
 119
 120         # Persistent:
 121         var $mTagHooks = array();
 122         var $mTransparentTagHooks = array();
 123         var $mFunctionHooks = array();
 124         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 125         var $mFunctionTagHooks = array();
 126         var $mStripList  = array();
 127         var $mDefaultStripList  = array();
 128         var $mVarCache = array();
 129         var $mImageParams = array();
 130         var $mImageParamsMagicArray = array();
 131         var $mMarkerIndex = 0;
 132         var $mFirstCall = true;
 133
 134         # Initialised by initialiseVariables()
 135
 136         /**
 137          * @var MagicWordArray
 138          */
 139         var $mVariables;
 140
 141         /**
 142          * @var MagicWordArray
 143          */
 144         var $mSubstWords;
 145         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 146
 147         # Cleared with clearState():
 148         /**
 149          * @var ParserOutput
 150          */
 151         var $mOutput;
 152         var $mAutonumber, $mDTopen;
 153
 154         /**
 155          * @var StripState
 156          */
 157         var $mStripState;
 158
 159         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 160         /**
 161          * @var LinkHolderArray
 162          */
 163         var $mLinkHolders;
 164
 165         var $mLinkID;
 166         var $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
 167         var $mDefaultSort;
 168         var $mTplExpandCache; # empty-frame expansion cache
 169         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 170         var $mExpensiveFunctionCount; # number of expensive parser function calls
 171         var $mShowToc, $mForceTocPosition;
 172
 173         /**
 174          * @var User
 175          */
 176         var $mUser; # User object; only used when doing pre-save transform
 177
 178         # Temporary
 179         # These are variables reset at least once per parse regardless of $clearState
 180
 181         /**
 182          * @var ParserOptions
 183          */
 184         var $mOptions;
 185
 186         /**
 187          * @var Title
 188          */
 189         var $mTitle;        # Title context, used for self-link rendering and similar things
 190         var $mOutputType;   # Output type, one of the OT_xxx constants
 191         var $ot;            # Shortcut alias, see setOutputType()
 192         var $mRevisionObject; # The revision object of the specified revision ID
 193         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 194         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 195         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 196         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 197
 198         /**
 199          * @var string
 200          */
 201         var $mUniqPrefix;
 202
 203         /**
 204          * Constructor
 205          *
 206          * @param $conf array
 207          */
 208         public function __construct( $conf = array() ) {
 209                 $this->mConf = $conf;
 210                 $this->mUrlProtocols = wfUrlProtocols();
 211                 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')'.
 212                         self::EXT_LINK_URL_CLASS.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 213                 if ( isset( $conf['preprocessorClass'] ) ) {
 214                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 215                 } elseif ( defined( 'MW_COMPILED' ) ) {
 216                         # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
 217                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 218                 } elseif ( extension_loaded( 'domxml' ) ) {
 219                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 220                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 221                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 222                 } elseif ( extension_loaded( 'dom' ) ) {
 223                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 224                 } else {
 225                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 226                 }
 227                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 228         }
 229
 230         /**
 231          * Reduce memory usage to reduce the impact of circular references
 232          */
 233         function __destruct() {
 234                 if ( isset( $this->mLinkHolders ) ) {
 235                         unset( $this->mLinkHolders );
 236                 }
 237                 foreach ( $this as $name => $value ) {
 238                         unset( $this->$name );
 239                 }
 240         }
 241
 242         /**
 243          * Do various kinds of initialisation on the first call of the parser
 244          */
 245         function firstCallInit() {
 246                 if ( !$this->mFirstCall ) {
 247                         return;
 248                 }
 249                 $this->mFirstCall = false;
 250
 251                 wfProfileIn( __METHOD__ );
 252
 253                 CoreParserFunctions::register( $this );
 254                 CoreTagHooks::register( $this );
 255                 $this->initialiseVariables();
 256
 257                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 258                 wfProfileOut( __METHOD__ );
 259         }
 260
 261         /**
 262          * Clear Parser state
 263          *
 264          * @private
 265          */
 266         function clearState() {
 267                 wfProfileIn( __METHOD__ );
 268                 if ( $this->mFirstCall ) {
 269                         $this->firstCallInit();
 270                 }
 271                 $this->mOutput = new ParserOutput;
 272                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 273                 $this->mAutonumber = 0;
 274                 $this->mLastSection = '';
 275                 $this->mDTopen = false;
 276                 $this->mIncludeCount = array();
 277                 $this->mArgStack = false;
 278                 $this->mInPre = false;
 279                 $this->mLinkHolders = new LinkHolderArray( $this );
 280                 $this->mLinkID = 0;
 281                 $this->mRevisionObject = $this->mRevisionTimestamp =
 282                         $this->mRevisionId = $this->mRevisionUser = null;
 283                 $this->mVarCache = array();
 284                 $this->mUser = null;
 285
 286                 /**
 287                  * Prefix for temporary replacement strings for the multipass parser.
 288                  * \x07 should never appear in input as it's disallowed in XML.
 289                  * Using it at the front also gives us a little extra robustness
 290                  * since it shouldn't match when butted up against identifier-like
 291                  * string constructs.
 292                  *
 293                  * Must not consist of all title characters, or else it will change
 294                  * the behaviour of <nowiki> in a link.
 295                  */
 296                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 297                 $this->mStripState = new StripState( $this->mUniqPrefix );
 298
 299
 300                 # Clear these on every parse, bug 4549
 301                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 302
 303                 $this->mShowToc = true;
 304                 $this->mForceTocPosition = false;
 305                 $this->mIncludeSizes = array(
 306                         'post-expand' => 0,
 307                         'arg' => 0,
 308                 );
 309                 $this->mPPNodeCount = 0;
 310                 $this->mGeneratedPPNodeCount = 0;
 311                 $this->mHighestExpansionDepth = 0;
 312                 $this->mDefaultSort = false;
 313                 $this->mHeadings = array();
 314                 $this->mDoubleUnderscores = array();
 315                 $this->mExpensiveFunctionCount = 0;
 316
 317                 # Fix cloning
 318                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 319                         $this->mPreprocessor = null;
 320                 }
 321
 322                 wfRunHooks( 'ParserClearState', array( &$this ) );
 323                 wfProfileOut( __METHOD__ );
 324         }
 325
 326         /**
 327          * Convert wikitext to HTML
 328          * Do not call this function recursively.
 329          *
 330          * @param $text String: text we want to parse
 331          * @param $title Title object
 332          * @param $options ParserOptions
 333          * @param $linestart boolean
 334          * @param $clearState boolean
 335          * @param $revid Int: number to pass in {{REVISIONID}}
 336          * @return ParserOutput a ParserOutput
 337          */
 338         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 339                 /**
 340                  * First pass--just handle <nowiki> sections, pass the rest off
 341                  * to internalParse() which does all the real work.
 342                  */
 343
 344                 global $wgUseTidy, $wgAlwaysUseTidy;
 345                 $fname = __METHOD__.'-' . wfGetCaller();
 346                 wfProfileIn( __METHOD__ );
 347                 wfProfileIn( $fname );
 348
 349                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 350
 351                 # Remove the strip marker tag prefix from the input, if present.
 352                 if ( $clearState ) {
 353                         $text = str_replace( $this->mUniqPrefix, '', $text );
 354                 }
 355
 356                 $oldRevisionId = $this->mRevisionId;
 357                 $oldRevisionObject = $this->mRevisionObject;
 358                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 359                 $oldRevisionUser = $this->mRevisionUser;
 360                 if ( $revid !== null ) {
 361                         $this->mRevisionId = $revid;
 362                         $this->mRevisionObject = null;
 363                         $this->mRevisionTimestamp = null;
 364                         $this->mRevisionUser = null;
 365                 }
 366
 367                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 368                 # No more strip!
 369                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 370                 $text = $this->internalParse( $text );
 371                 wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
 372
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 376                 $fixtags = array(
 377                         # french spaces, last one Guillemet-left
 378                         # only if there is something before the space
 379                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
 380                         # french spaces, Guillemet-right
 381                         '/(\\302\\253) /' => '\\1&#160;',
 382                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 383                 );
 384                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 385
 386                 $text = $this->doBlockLevels( $text, $linestart );
 387
 388                 $this->replaceLinkHolders( $text );
 389
 390                 /**
 391                  * The input doesn't get language converted if
 392                  * a) It's disabled
 393                  * b) Content isn't converted
 394                  * c) It's a conversion table
 395                  * d) it is an interface message (which is in the user language)
 396                  */
 397                 if ( !( $options->getDisableContentConversion()
 398                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) ) )
 399                 {
 400                         # Run convert unconditionally in 1.18-compatible mode
 401                         global $wgBug34832TransitionalRollback;
 402                         if ( $wgBug34832TransitionalRollback || !$this->mOptions->getInterfaceMessage() ) {
 403                                 # The position of the convert() call should not be changed. it
 404                                 # assumes that the links are all replaced and the only thing left
 405                                 # is the <nowiki> mark.
 406                                 $text = $this->getConverterLanguage()->convert( $text );
 407                         }
 408                 }
 409
 410                 /**
 411                  * A converted title will be provided in the output object if title and
 412                  * content conversion are enabled, the article text does not contain
 413                  * a conversion-suppressing double-underscore tag, and no
 414                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 415                  * automatic link conversion.
 416                  */
 417                 if ( !( $options->getDisableTitleConversion()
 418                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 419                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 420                                 || $this->mOutput->getDisplayTitle() !== false ) )
 421                 {
 422                         $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
 423                         if ( $convruletitle ) {
 424                                 $this->mOutput->setTitleText( $convruletitle );
 425                         } else {
 426                                 $titleText = $this->getConverterLanguage()->convertTitle( $title );
 427                                 $this->mOutput->setTitleText( $titleText );
 428                         }
 429                 }
 430
 431                 $text = $this->mStripState->unstripNoWiki( $text );
 432
 433                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 434
 435                 $text = $this->replaceTransparentTags( $text );
 436                 $text = $this->mStripState->unstripGeneral( $text );
 437
 438                 $text = Sanitizer::normalizeCharReferences( $text );
 439
 440                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 441                         $text = MWTidy::tidy( $text );
 442                 } else {
 443                         # attempt to sanitize at least some nesting problems
 444                         # (bug #2702 and quite a few others)
 445                         $tidyregs = array(
 446                                 # ''Something [http://www.cool.com cool''] -->
 447                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 448                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 449                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 450                                 # fix up an anchor inside another anchor, only
 451                                 # at least for a single single nested link (bug 3695)
 452                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 453                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 454                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 455                                 # contains a div, so fix it up here; replace
 456                                 # div with escaped text
 457                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 458                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 459                                 # remove empty italic or bold tag pairs, some
 460                                 # introduced by rules above
 461                                 '/<([bi])><\/\\1>/' => '',
 462                         );
 463
 464                         $text = preg_replace(
 465                                 array_keys( $tidyregs ),
 466                                 array_values( $tidyregs ),
 467                                 $text );
 468                 }
 469
 470                 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
 471                         $this->limitationWarn( 'expensive-parserfunction',
 472                                 $this->mExpensiveFunctionCount,
 473                                 $this->mOptions->getExpensiveParserFunctionLimit()
 474                         );
 475                 }
 476
 477                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 478
 479                 # Information on include size limits, for the benefit of users who try to skirt them
 480                 if ( $this->mOptions->getEnableLimitReport() ) {
 481                         $max = $this->mOptions->getMaxIncludeSize();
 482                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/{$this->mOptions->getExpensiveParserFunctionLimit()}\n";
 483                         $limitReport =
 484                                 "NewPP limit report\n" .
 485                                 "Preprocessor visited node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 486                                 "Preprocessor generated node count: " .
 487                                         "{$this->mGeneratedPPNodeCount}/{$this->mOptions->getMaxGeneratedPPNodeCount()}\n" .
 488                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 489                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 490                                 "Highest expansion depth: {$this->mHighestExpansionDepth}/{$this->mOptions->getMaxPPExpandDepth()}\n".
 491                                 $PFreport;
 492                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 493                         $text .= "\n<!-- \n$limitReport-->\n";
 494                 }
 495                 $this->mOutput->setText( $text );
 496
 497                 $this->mRevisionId = $oldRevisionId;
 498                 $this->mRevisionObject = $oldRevisionObject;
 499                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 500                 $this->mRevisionUser = $oldRevisionUser;
 501                 wfProfileOut( $fname );
 502                 wfProfileOut( __METHOD__ );
 503
 504                 return $this->mOutput;
 505         }
 506
 507         /**
 508          * Recursive parser entry point that can be called from an extension tag
 509          * hook.
 510          *
 511          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 512          *
 513          * @param $text String: text extension wants to have parsed
 514          * @param $frame PPFrame: The frame to use for expanding any template variables
 515          *
 516          * @return string
 517          */
 518         function recursiveTagParse( $text, $frame=false ) {
 519                 wfProfileIn( __METHOD__ );
 520                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 521                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 522                 $text = $this->internalParse( $text, false, $frame );
 523                 wfProfileOut( __METHOD__ );
 524                 return $text;
 525         }
 526
 527         /**
 528          * Expand templates and variables in the text, producing valid, static wikitext.
 529          * Also removes comments.
 530          * @return mixed|string
 531          */
 532         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 533                 wfProfileIn( __METHOD__ );
 534                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 535                 if ( $revid !== null ) {
 536                         $this->mRevisionId = $revid;
 537                 }
 538                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 539                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 540                 $text = $this->replaceVariables( $text );
 541                 $text = $this->mStripState->unstripBoth( $text );
 542                 wfProfileOut( __METHOD__ );
 543                 return $text;
 544         }
 545
 546         /**
 547          * Recursive parser entry point that can be called from an extension tag
 548          * hook.
 549          *
 550          * @param $text String: text to be expanded
 551          * @param $frame PPFrame: The frame to use for expanding any template variables
 552          * @return String
 553          * @since 1.19
 554          */
 555         public function recursivePreprocess( $text, $frame = false ) {
 556                 wfProfileIn( __METHOD__ );
 557                 $text = $this->replaceVariables( $text, $frame );
 558                 $text = $this->mStripState->unstripBoth( $text );
 559                 wfProfileOut( __METHOD__ );
 560                 return $text;
 561         }
 562
 563         /**
 564          * Process the wikitext for the "?preload=" feature. (bug 5210)
 565          *
 566          * "<noinclude>", "<includeonly>" etc. are parsed as for template
 567          * transclusion, comments, templates, arguments, tags hooks and parser
 568          * functions are untouched.
 569          *
 570          * @param $text String
 571          * @param $title Title
 572          * @param $options ParserOptions
 573          * @return String
 574          */
 575         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 576                 # Parser (re)initialisation
 577                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 578
 579                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 580                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 581                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 582                 $text = $this->mStripState->unstripBoth( $text );
 583                 return $text;
 584         }
 585
 586         /**
 587          * Get a random string
 588          *
 589          * @return string
 590          */
 591         static public function getRandomString() {
 592                 return wfRandomString( 16 );
 593         }
 594
 595         /**
 596          * Set the current user.
 597          * Should only be used when doing pre-save transform.
 598          *
 599          * @param $user Mixed: User object or null (to reset)
 600          */
 601         function setUser( $user ) {
 602                 $this->mUser = $user;
 603         }
 604
 605         /**
 606          * Accessor for mUniqPrefix.
 607          *
 608          * @return String
 609          */
 610         public function uniqPrefix() {
 611                 if ( !isset( $this->mUniqPrefix ) ) {
 612                         # @todo FIXME: This is probably *horribly wrong*
 613                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 614                         # if this is called for a parser cache hit, the parser may not
 615                         # have ever been initialized in the first place.
 616                         # Not really sure what the heck is supposed to be going on here.
 617                         return '';
 618                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 619                 }
 620                 return $this->mUniqPrefix;
 621         }
 622
 623         /**
 624          * Set the context title
 625          *
 626          * @param $t Title
 627          */
 628         function setTitle( $t ) {
 629                 if ( !$t || $t instanceof FakeTitle ) {
 630                         $t = Title::newFromText( 'NO TITLE' );
 631                 }
 632
 633                 if ( strval( $t->getFragment() ) !== '' ) {
 634                         # Strip the fragment to avoid various odd effects
 635                         $this->mTitle = clone $t;
 636                         $this->mTitle->setFragment( '' );
 637                 } else {
 638                         $this->mTitle = $t;
 639                 }
 640         }
 641
 642         /**
 643          * Accessor for the Title object
 644          *
 645          * @return Title object
 646          */
 647         function getTitle() {
 648                 return $this->mTitle;
 649         }
 650
 651         /**
 652          * Accessor/mutator for the Title object
 653          *
 654          * @param $x Title object or null to just get the current one
 655          * @return Title object
 656          */
 657         function Title( $x = null ) {
 658                 return wfSetVar( $this->mTitle, $x );
 659         }
 660
 661         /**
 662          * Set the output type
 663          *
 664          * @param $ot Integer: new value
 665          */
 666         function setOutputType( $ot ) {
 667                 $this->mOutputType = $ot;
 668                 # Shortcut alias
 669                 $this->ot = array(
 670                         'html' => $ot == self::OT_HTML,
 671                         'wiki' => $ot == self::OT_WIKI,
 672                         'pre' => $ot == self::OT_PREPROCESS,
 673                         'plain' => $ot == self::OT_PLAIN,
 674                 );
 675         }
 676
 677         /**
 678          * Accessor/mutator for the output type
 679          *
 680          * @param $x int|null New value or null to just get the current one
 681          * @return Integer
 682          */
 683         function OutputType( $x = null ) {
 684                 return wfSetVar( $this->mOutputType, $x );
 685         }
 686
 687         /**
 688          * Get the ParserOutput object
 689          *
 690          * @return ParserOutput object
 691          */
 692         function getOutput() {
 693                 return $this->mOutput;
 694         }
 695
 696         /**
 697          * Get the ParserOptions object
 698          *
 699          * @return ParserOptions object
 700          */
 701         function getOptions() {
 702                 return $this->mOptions;
 703         }
 704
 705         /**
 706          * Accessor/mutator for the ParserOptions object
 707          *
 708          * @param $x ParserOptions New value or null to just get the current one
 709          * @return ParserOptions Current ParserOptions object
 710          */
 711         function Options( $x = null ) {
 712                 return wfSetVar( $this->mOptions, $x );
 713         }
 714
 715         /**
 716          * @return int
 717          */
 718         function nextLinkID() {
 719                 return $this->mLinkID++;
 720         }
 721
 722         /**
 723          * @param $id int
 724          */
 725         function setLinkID( $id ) {
 726                 $this->mLinkID = $id;
 727         }
 728
 729         /**
 730          * Get a language object for use in parser functions such as {{FORMATNUM:}}
 731          * @return Language
 732          */
 733         function getFunctionLang() {
 734                 return $this->getTargetLanguage();
 735         }
 736
 737         /**
 738          * Get the target language for the content being parsed. This is usually the
 739          * language that the content is in.
 740          *
 741          * @since 1.19
 742          *
 743          * @return Language|null
 744          */
 745         public function getTargetLanguage() {
 746                 $target = $this->mOptions->getTargetLanguage();
 747
 748                 if ( $target !== null ) {
 749                         return $target;
 750                 } elseif( $this->mOptions->getInterfaceMessage() ) {
 751                         return $this->mOptions->getUserLangObj();
 752                 } elseif( is_null( $this->mTitle ) ) {
 753                         throw new MWException( __METHOD__ . ': $this->mTitle is null' );
 754                 }
 755
 756                 return $this->mTitle->getPageLanguage();
 757         }
 758
 759         /**
 760          * Get the language object for language conversion
 761          */
 762         function getConverterLanguage() {
 763                 global $wgBug34832TransitionalRollback, $wgContLang;
 764                 if ( $wgBug34832TransitionalRollback ) {
 765                         return $wgContLang;
 766                 } else {
 767                         return $this->getTargetLanguage();
 768                 }
 769         }
 770
 771         /**
 772          * Get a User object either from $this->mUser, if set, or from the
 773          * ParserOptions object otherwise
 774          *
 775          * @return User object
 776          */
 777         function getUser() {
 778                 if ( !is_null( $this->mUser ) ) {
 779                         return $this->mUser;
 780                 }
 781                 return $this->mOptions->getUser();
 782         }
 783
 784         /**
 785          * Get a preprocessor object
 786          *
 787          * @return Preprocessor instance
 788          */
 789         function getPreprocessor() {
 790                 if ( !isset( $this->mPreprocessor ) ) {
 791                         $class = $this->mPreprocessorClass;
 792                         $this->mPreprocessor = new $class( $this );
 793                 }
 794                 return $this->mPreprocessor;
 795         }
 796
 797         /**
 798          * Replaces all occurrences of HTML-style comments and the given tags
 799          * in the text with a random marker and returns the next text. The output
 800          * parameter $matches will be an associative array filled with data in
 801          * the form:
 802          *
 803          * @code
 804          *   'UNIQ-xxxxx' => array(
 805          *     'element',
 806          *     'tag content',
 807          *     array( 'param' => 'x' ),
 808          *     '<element param="x">tag content</element>' ) )
 809          * @endcode
 810          *
 811          * @param $elements array list of element names. Comments are always extracted.
 812          * @param $text string Source text string.
 813          * @param $matches array Out parameter, Array: extracted tags
 814          * @param $uniq_prefix string
 815          * @return String: stripped text
 816          */
 817         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 818                 static $n = 1;
 819                 $stripped = '';
 820                 $matches = array();
 821
 822                 $taglist = implode( '|', $elements );
 823                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 824
 825                 while ( $text != '' ) {
 826                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 827                         $stripped .= $p[0];
 828                         if ( count( $p ) < 5 ) {
 829                                 break;
 830                         }
 831                         if ( count( $p ) > 5 ) {
 832                                 # comment
 833                                 $element    = $p[4];
 834                                 $attributes = '';
 835                                 $close      = '';
 836                                 $inside     = $p[5];
 837                         } else {
 838                                 # tag
 839                                 $element    = $p[1];
 840                                 $attributes = $p[2];
 841                                 $close      = $p[3];
 842                                 $inside     = $p[4];
 843                         }
 844
 845                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 846                         $stripped .= $marker;
 847
 848                         if ( $close === '/>' ) {
 849                                 # Empty element tag, <tag />
 850                                 $content = null;
 851                                 $text = $inside;
 852                                 $tail = null;
 853                         } else {
 854                                 if ( $element === '!--' ) {
 855                                         $end = '/(-->)/';
 856                                 } else {
 857                                         $end = "/(<\\/$element\\s*>)/i";
 858                                 }
 859                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 860                                 $content = $q[0];
 861                                 if ( count( $q ) < 3 ) {
 862                                         # No end tag -- let it run out to the end of the text.
 863                                         $tail = '';
 864                                         $text = '';
 865                                 } else {
 866                                         $tail = $q[1];
 867                                         $text = $q[2];
 868                                 }
 869                         }
 870
 871                         $matches[$marker] = array( $element,
 872                                 $content,
 873                                 Sanitizer::decodeTagAttributes( $attributes ),
 874                                 "<$element$attributes$close$content$tail" );
 875                 }
 876                 return $stripped;
 877         }
 878
 879         /**
 880          * Get a list of strippable XML-like elements
 881          *
 882          * @return array
 883          */
 884         function getStripList() {
 885                 return $this->mStripList;
 886         }
 887
 888         /**
 889          * Add an item to the strip state
 890          * Returns the unique tag which must be inserted into the stripped text
 891          * The tag will be replaced with the original text in unstrip()
 892          *
 893          * @param $text string
 894          *
 895          * @return string
 896          */
 897         function insertStripItem( $text ) {
 898                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 899                 $this->mMarkerIndex++;
 900                 $this->mStripState->addGeneral( $rnd, $text );
 901                 return $rnd;
 902         }
 903
 904         /**
 905          * parse the wiki syntax used to render tables
 906          *
 907          * @private
 908          * @return string
 909          */
 910         function doTableStuff( $text ) {
 911                 wfProfileIn( __METHOD__ );
 912
 913                 $lines = StringUtils::explode( "\n", $text );
 914                 $out = '';
 915                 $td_history = array(); # Is currently a td tag open?
 916                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 917                 $tr_history = array(); # Is currently a tr tag open?
 918                 $tr_attributes = array(); # history of tr attributes
 919                 $has_opened_tr = array(); # Did this table open a <tr> element?
 920                 $indent_level = 0; # indent level of the table
 921
 922                 foreach ( $lines as $outLine ) {
 923                         $line = trim( $outLine );
 924
 925                         if ( $line === '' ) { # empty line, go to next line
 926                                 $out .= $outLine."\n";
 927                                 continue;
 928                         }
 929
 930                         $first_character = $line[0];
 931                         $matches = array();
 932
 933                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 934                                 # First check if we are starting a new table
 935                                 $indent_level = strlen( $matches[1] );
 936
 937                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 938                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 939
 940                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 941                                 array_push( $td_history , false );
 942                                 array_push( $last_tag_history , '' );
 943                                 array_push( $tr_history , false );
 944                                 array_push( $tr_attributes , '' );
 945                                 array_push( $has_opened_tr , false );
 946                         } elseif ( count( $td_history ) == 0 ) {
 947                                 # Don't do any of the following
 948                                 $out .= $outLine."\n";
 949                                 continue;
 950                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 951                                 # We are ending a table
 952                                 $line = '</table>' . substr( $line , 2 );
 953                                 $last_tag = array_pop( $last_tag_history );
 954
 955                                 if ( !array_pop( $has_opened_tr ) ) {
 956                                         $line = "<tr><td></td></tr>{$line}";
 957                                 }
 958
 959                                 if ( array_pop( $tr_history ) ) {
 960                                         $line = "</tr>{$line}";
 961                                 }
 962
 963                                 if ( array_pop( $td_history ) ) {
 964                                         $line = "</{$last_tag}>{$line}";
 965                                 }
 966                                 array_pop( $tr_attributes );
 967                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 968                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 969                                 # Now we have a table row
 970                                 $line = preg_replace( '#^\|-+#', '', $line );
 971
 972                                 # Whats after the tag is now only attributes
 973                                 $attributes = $this->mStripState->unstripBoth( $line );
 974                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 975                                 array_pop( $tr_attributes );
 976                                 array_push( $tr_attributes, $attributes );
 977
 978                                 $line = '';
 979                                 $last_tag = array_pop( $last_tag_history );
 980                                 array_pop( $has_opened_tr );
 981                                 array_push( $has_opened_tr , true );
 982
 983                                 if ( array_pop( $tr_history ) ) {
 984                                         $line = '</tr>';
 985                                 }
 986
 987                                 if ( array_pop( $td_history ) ) {
 988                                         $line = "</{$last_tag}>{$line}";
 989                                 }
 990
 991                                 $outLine = $line;
 992                                 array_push( $tr_history , false );
 993                                 array_push( $td_history , false );
 994                                 array_push( $last_tag_history , '' );
 995                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 996                                 # This might be cell elements, td, th or captions
 997                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 998                                         $first_character = '+';
 999                                         $line = substr( $line , 1 );
1000                                 }
1001
1002                                 $line = substr( $line , 1 );
1003
1004                                 if ( $first_character === '!' ) {
1005                                         $line = str_replace( '!!' , '||' , $line );
1006                                 }
1007
1008                                 # Split up multiple cells on the same line.
1009                                 # FIXME : This can result in improper nesting of tags processed
1010                                 # by earlier parser steps, but should avoid splitting up eg
1011                                 # attribute values containing literal "||".
1012                                 $cells = StringUtils::explodeMarkup( '||' , $line );
1013
1014                                 $outLine = '';
1015
1016                                 # Loop through each table cell
1017                                 foreach ( $cells as $cell ) {
1018                                         $previous = '';
1019                                         if ( $first_character !== '+' ) {
1020                                                 $tr_after = array_pop( $tr_attributes );
1021                                                 if ( !array_pop( $tr_history ) ) {
1022                                                         $previous = "<tr{$tr_after}>\n";
1023                                                 }
1024                                                 array_push( $tr_history , true );
1025                                                 array_push( $tr_attributes , '' );
1026                                                 array_pop( $has_opened_tr );
1027                                                 array_push( $has_opened_tr , true );
1028                                         }
1029
1030                                         $last_tag = array_pop( $last_tag_history );
1031
1032                                         if ( array_pop( $td_history ) ) {
1033                                                 $previous = "</{$last_tag}>\n{$previous}";
1034                                         }
1035
1036                                         if ( $first_character === '|' ) {
1037                                                 $last_tag = 'td';
1038                                         } elseif ( $first_character === '!' ) {
1039                                                 $last_tag = 'th';
1040                                         } elseif ( $first_character === '+' ) {
1041                                                 $last_tag = 'caption';
1042                                         } else {
1043                                                 $last_tag = '';
1044                                         }
1045
1046                                         array_push( $last_tag_history , $last_tag );
1047
1048                                         # A cell could contain both parameters and data
1049                                         $cell_data = explode( '|' , $cell , 2 );
1050
1051                                         # Bug 553: Note that a '|' inside an invalid link should not
1052                                         # be mistaken as delimiting cell parameters
1053                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
1054                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
1055                                         } elseif ( count( $cell_data ) == 1 ) {
1056                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1057                                         } else {
1058                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1059                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
1060                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1061                                         }
1062
1063                                         $outLine .= $cell;
1064                                         array_push( $td_history , true );
1065                                 }
1066                         }
1067                         $out .= $outLine . "\n";
1068                 }
1069
1070                 # Closing open td, tr && table
1071                 while ( count( $td_history ) > 0 ) {
1072                         if ( array_pop( $td_history ) ) {
1073                                 $out .= "</td>\n";
1074                         }
1075                         if ( array_pop( $tr_history ) ) {
1076                                 $out .= "</tr>\n";
1077                         }
1078                         if ( !array_pop( $has_opened_tr ) ) {
1079                                 $out .= "<tr><td></td></tr>\n" ;
1080                         }
1081
1082                         $out .= "</table>\n";
1083                 }
1084
1085                 # Remove trailing line-ending (b/c)
1086                 if ( substr( $out, -1 ) === "\n" ) {
1087                         $out = substr( $out, 0, -1 );
1088                 }
1089
1090                 # special case: don't return empty table
1091                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1092                         $out = '';
1093                 }
1094
1095                 wfProfileOut( __METHOD__ );
1096
1097                 return $out;
1098         }
1099
1100         /**
1101          * Helper function for parse() that transforms wiki markup into
1102          * HTML. Only called for $mOutputType == self::OT_HTML.
1103          *
1104          * @private
1105          *
1106          * @param $text string
1107          * @param $isMain bool
1108          * @param $frame bool
1109          *
1110          * @return string
1111          */
1112         function internalParse( $text, $isMain = true, $frame = false ) {
1113                 wfProfileIn( __METHOD__ );
1114
1115                 $origText = $text;
1116
1117                 # Hook to suspend the parser in this state
1118                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1119                         wfProfileOut( __METHOD__ );
1120                         return $text ;
1121                 }
1122
1123                 # if $frame is provided, then use $frame for replacing any variables
1124                 if ( $frame ) {
1125                         # use frame depth to infer how include/noinclude tags should be handled
1126                         # depth=0 means this is the top-level document; otherwise it's an included document
1127                         if ( !$frame->depth ) {
1128                                 $flag = 0;
1129                         } else {
1130                                 $flag = Parser::PTD_FOR_INCLUSION;
1131                         }
1132                         $dom = $this->preprocessToDom( $text, $flag );
1133                         $text = $frame->expand( $dom );
1134                 } else {
1135                         # if $frame is not provided, then use old-style replaceVariables
1136                         $text = $this->replaceVariables( $text );
1137                 }
1138
1139                 wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1140                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1141                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1142
1143                 # Tables need to come after variable replacement for things to work
1144                 # properly; putting them before other transformations should keep
1145                 # exciting things like link expansions from showing up in surprising
1146                 # places.
1147                 $text = $this->doTableStuff( $text );
1148
1149                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1150
1151                 $text = $this->doDoubleUnderscore( $text );
1152
1153                 $text = $this->doHeadings( $text );
1154                 if ( $this->mOptions->getUseDynamicDates() ) {
1155                         $df = DateFormatter::getInstance();
1156                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1157                 }
1158                 $text = $this->replaceInternalLinks( $text );
1159                 $text = $this->doAllQuotes( $text );
1160                 $text = $this->replaceExternalLinks( $text );
1161
1162                 # replaceInternalLinks may sometimes leave behind
1163                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1164                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1165
1166                 $text = $this->doMagicLinks( $text );
1167                 $text = $this->formatHeadings( $text, $origText, $isMain );
1168
1169                 wfProfileOut( __METHOD__ );
1170                 return $text;
1171         }
1172
1173         /**
1174          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1175          * magic external links.
1176          *
1177          * DML
1178          * @private
1179          *
1180          * @param $text string
1181          *
1182          * @return string
1183          */
1184         function doMagicLinks( $text ) {
1185                 wfProfileIn( __METHOD__ );
1186                 $prots = wfUrlProtocolsWithoutProtRel();
1187                 $urlChar = self::EXT_LINK_URL_CLASS;
1188                 $text = preg_replace_callback(
1189                         '!(?:                           # Start cases
1190                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1191                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1192                                 (\\b(?i:$prots)$urlChar+) |  # m[3]: Free external links" . '
1193                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1194                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1195                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1196                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1197                                         [0-9Xx]                 # check digit
1198                                         \b)
1199                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1200                 wfProfileOut( __METHOD__ );
1201                 return $text;
1202         }
1203
1204         /**
1205          * @throws MWException
1206          * @param $m array
1207          * @return HTML|string
1208          */
1209         function magicLinkCallback( $m ) {
1210                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1211                         # Skip anchor
1212                         return $m[0];
1213                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1214                         # Skip HTML element
1215                         return $m[0];
1216                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1217                         # Free external link
1218                         return $this->makeFreeExternalLink( $m[0] );
1219                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1220                         # RFC or PMID
1221                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1222                                 $keyword = 'RFC';
1223                                 $urlmsg = 'rfcurl';
1224                                 $CssClass = 'mw-magiclink-rfc';
1225                                 $id = $m[4];
1226                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1227                                 $keyword = 'PMID';
1228                                 $urlmsg = 'pubmedurl';
1229                                 $CssClass = 'mw-magiclink-pmid';
1230                                 $id = $m[4];
1231                         } else {
1232                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1233                                         substr( $m[0], 0, 20 ) . '"' );
1234                         }
1235                         $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1236                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1237                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1238                         # ISBN
1239                         $isbn = $m[5];
1240                         $num = strtr( $isbn, array(
1241                                 '-' => '',
1242                                 ' ' => '',
1243                                 'x' => 'X',
1244                         ));
1245                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1246                         return'<a href="' .
1247                                 htmlspecialchars( $titleObj->getLocalUrl() ) .
1248                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1249                 } else {
1250                         return $m[0];
1251                 }
1252         }
1253
1254         /**
1255          * Make a free external link, given a user-supplied URL
1256          *
1257          * @param $url string
1258          *
1259          * @return string HTML
1260          * @private
1261          */
1262         function makeFreeExternalLink( $url ) {
1263                 wfProfileIn( __METHOD__ );
1264
1265                 $trail = '';
1266
1267                 # The characters '<' and '>' (which were escaped by
1268                 # removeHTMLtags()) should not be included in
1269                 # URLs, per RFC 2396.
1270                 $m2 = array();
1271                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1272                         $trail = substr( $url, $m2[0][1] ) . $trail;
1273                         $url = substr( $url, 0, $m2[0][1] );
1274                 }
1275
1276                 # Move trailing punctuation to $trail
1277                 $sep = ',;\.:!?';
1278                 # If there is no left bracket, then consider right brackets fair game too
1279                 if ( strpos( $url, '(' ) === false ) {
1280                         $sep .= ')';
1281                 }
1282
1283                 $numSepChars = strspn( strrev( $url ), $sep );
1284                 if ( $numSepChars ) {
1285                         $trail = substr( $url, -$numSepChars ) . $trail;
1286                         $url = substr( $url, 0, -$numSepChars );
1287                 }
1288
1289                 $url = Sanitizer::cleanUrl( $url );
1290
1291                 # Is this an external image?
1292                 $text = $this->maybeMakeExternalImage( $url );
1293                 if ( $text === false ) {
1294                         # Not an image, make a link
1295                         $text = Linker::makeExternalLink( $url,
1296                                 $this->getConverterLanguage()->markNoConversion($url), true, 'free',
1297                                 $this->getExternalLinkAttribs( $url ) );
1298                         # Register it in the output object...
1299                         # Replace unnecessary URL escape codes with their equivalent characters
1300                         $pasteurized = self::replaceUnusualEscapes( $url );
1301                         $this->mOutput->addExternalLink( $pasteurized );
1302                 }
1303                 wfProfileOut( __METHOD__ );
1304                 return $text . $trail;
1305         }
1306
1307
1308         /**
1309          * Parse headers and return html
1310          *
1311          * @private
1312          *
1313          * @param $text string
1314          *
1315          * @return string
1316          */
1317         function doHeadings( $text ) {
1318                 wfProfileIn( __METHOD__ );
1319                 for ( $i = 6; $i >= 1; --$i ) {
1320                         $h = str_repeat( '=', $i );
1321                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1322                           "<h$i>\\1</h$i>", $text );
1323                 }
1324                 wfProfileOut( __METHOD__ );
1325                 return $text;
1326         }
1327
1328         /**
1329          * Replace single quotes with HTML markup
1330          * @private
1331          *
1332          * @param $text string
1333          *
1334          * @return string the altered text
1335          */
1336         function doAllQuotes( $text ) {
1337                 wfProfileIn( __METHOD__ );
1338                 $outtext = '';
1339                 $lines = StringUtils::explode( "\n", $text );
1340                 foreach ( $lines as $line ) {
1341                         $outtext .= $this->doQuotes( $line ) . "\n";
1342                 }
1343                 $outtext = substr( $outtext, 0,-1 );
1344                 wfProfileOut( __METHOD__ );
1345                 return $outtext;
1346         }
1347
1348         /**
1349          * Helper function for doAllQuotes()
1350          *
1351          * @param $text string
1352          *
1353          * @return string
1354          */
1355         public function doQuotes( $text ) {
1356                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1357                 if ( count( $arr ) == 1 ) {
1358                         return $text;
1359                 } else {
1360                         # First, do some preliminary work. This may shift some apostrophes from
1361                         # being mark-up to being text. It also counts the number of occurrences
1362                         # of bold and italics mark-ups.
1363                         $numbold = 0;
1364                         $numitalics = 0;
1365                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1366                                 if ( ( $i % 2 ) == 1 ) {
1367                                         # If there are ever four apostrophes, assume the first is supposed to
1368                                         # be text, and the remaining three constitute mark-up for bold text.
1369                                         if ( strlen( $arr[$i] ) == 4 ) {
1370                                                 $arr[$i-1] .= "'";
1371                                                 $arr[$i] = "'''";
1372                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1373                                                 # If there are more than 5 apostrophes in a row, assume they're all
1374                                                 # text except for the last 5.
1375                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1376                                                 $arr[$i] = "'''''";
1377                                         }
1378                                         # Count the number of occurrences of bold and italics mark-ups.
1379                                         # We are not counting sequences of five apostrophes.
1380                                         if ( strlen( $arr[$i] ) == 2 ) {
1381                                                 $numitalics++;
1382                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1383                                                 $numbold++;
1384                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1385                                                 $numitalics++;
1386                                                 $numbold++;
1387                                         }
1388                                 }
1389                         }
1390
1391                         # If there is an odd number of both bold and italics, it is likely
1392                         # that one of the bold ones was meant to be an apostrophe followed
1393                         # by italics. Which one we cannot know for certain, but it is more
1394                         # likely to be one that has a single-letter word before it.
1395                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1396                                 $i = 0;
1397                                 $firstsingleletterword = -1;
1398                                 $firstmultiletterword = -1;
1399                                 $firstspace = -1;
1400                                 foreach ( $arr as $r ) {
1401                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1402                                                 $x1 = substr( $arr[$i-1], -1 );
1403                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1404                                                 if ( $x1 === ' ' ) {
1405                                                         if ( $firstspace == -1 ) {
1406                                                                 $firstspace = $i;
1407                                                         }
1408                                                 } elseif ( $x2 === ' ') {
1409                                                         if ( $firstsingleletterword == -1 ) {
1410                                                                 $firstsingleletterword = $i;
1411                                                         }
1412                                                 } else {
1413                                                         if ( $firstmultiletterword == -1 ) {
1414                                                                 $firstmultiletterword = $i;
1415                                                         }
1416                                                 }
1417                                         }
1418                                         $i++;
1419                                 }
1420
1421                                 # If there is a single-letter word, use it!
1422                                 if ( $firstsingleletterword > -1 ) {
1423                                         $arr[$firstsingleletterword] = "''";
1424                                         $arr[$firstsingleletterword-1] .= "'";
1425                                 } elseif ( $firstmultiletterword > -1 ) {
1426                                         # If not, but there's a multi-letter word, use that one.
1427                                         $arr[$firstmultiletterword] = "''";
1428                                         $arr[$firstmultiletterword-1] .= "'";
1429                                 } elseif ( $firstspace > -1 ) {
1430                                         # ... otherwise use the first one that has neither.
1431                                         # (notice that it is possible for all three to be -1 if, for example,
1432                                         # there is only one pentuple-apostrophe in the line)
1433                                         $arr[$firstspace] = "''";
1434                                         $arr[$firstspace-1] .= "'";
1435                                 }
1436                         }
1437
1438                         # Now let's actually convert our apostrophic mush to HTML!
1439                         $output = '';
1440                         $buffer = '';
1441                         $state = '';
1442                         $i = 0;
1443                         foreach ( $arr as $r ) {
1444                                 if ( ( $i % 2 ) == 0 ) {
1445                                         if ( $state === 'both' ) {
1446                                                 $buffer .= $r;
1447                                         } else {
1448                                                 $output .= $r;
1449                                         }
1450                                 } else {
1451                                         if ( strlen( $r ) == 2 ) {
1452                                                 if ( $state === 'i' ) {
1453                                                         $output .= '</i>'; $state = '';
1454                                                 } elseif ( $state === 'bi' ) {
1455                                                         $output .= '</i>'; $state = 'b';
1456                                                 } elseif ( $state === 'ib' ) {
1457                                                         $output .= '</b></i><b>'; $state = 'b';
1458                                                 } elseif ( $state === 'both' ) {
1459                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1460                                                 } else { # $state can be 'b' or ''
1461                                                         $output .= '<i>'; $state .= 'i';
1462                                                 }
1463                                         } elseif ( strlen( $r ) == 3 ) {
1464                                                 if ( $state === 'b' ) {
1465                                                         $output .= '</b>'; $state = '';
1466                                                 } elseif ( $state === 'bi' ) {
1467                                                         $output .= '</i></b><i>'; $state = 'i';
1468                                                 } elseif ( $state === 'ib' ) {
1469                                                         $output .= '</b>'; $state = 'i';
1470                                                 } elseif ( $state === 'both' ) {
1471                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1472                                                 } else { # $state can be 'i' or ''
1473                                                         $output .= '<b>'; $state .= 'b';
1474                                                 }
1475                                         } elseif ( strlen( $r ) == 5 ) {
1476                                                 if ( $state === 'b' ) {
1477                                                         $output .= '</b><i>'; $state = 'i';
1478                                                 } elseif ( $state === 'i' ) {
1479                                                         $output .= '</i><b>'; $state = 'b';
1480                                                 } elseif ( $state === 'bi' ) {
1481                                                         $output .= '</i></b>'; $state = '';
1482                                                 } elseif ( $state === 'ib' ) {
1483                                                         $output .= '</b></i>'; $state = '';
1484                                                 } elseif ( $state === 'both' ) {
1485                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1486                                                 } else { # ($state == '')
1487                                                         $buffer = ''; $state = 'both';
1488                                                 }
1489                                         }
1490                                 }
1491                                 $i++;
1492                         }
1493                         # Now close all remaining tags.  Notice that the order is important.
1494                         if ( $state === 'b' || $state === 'ib' ) {
1495                                 $output .= '</b>';
1496                         }
1497                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1498                                 $output .= '</i>';
1499                         }
1500                         if ( $state === 'bi' ) {
1501                                 $output .= '</b>';
1502                         }
1503                         # There might be lonely ''''', so make sure we have a buffer
1504                         if ( $state === 'both' && $buffer ) {
1505                                 $output .= '<b><i>'.$buffer.'</i></b>';
1506                         }
1507                         return $output;
1508                 }
1509         }
1510
1511         /**
1512          * Replace external links (REL)
1513          *
1514          * Note: this is all very hackish and the order of execution matters a lot.
1515          * Make sure to run maintenance/parserTests.php if you change this code.
1516          *
1517          * @private
1518          *
1519          * @param $text string
1520          *
1521          * @return string
1522          */
1523         function replaceExternalLinks( $text ) {
1524                 wfProfileIn( __METHOD__ );
1525
1526                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1527                 if ( $bits === false ) {
1528                         throw new MWException( "PCRE needs to be compiled with --enable-unicode-properties in order for MediaWiki to function" );
1529                 }
1530                 $s = array_shift( $bits );
1531
1532                 $i = 0;
1533                 while ( $i<count( $bits ) ) {
1534                         $url = $bits[$i++];
1535                         $protocol = $bits[$i++];
1536                         $text = $bits[$i++];
1537                         $trail = $bits[$i++];
1538
1539                         # The characters '<' and '>' (which were escaped by
1540                         # removeHTMLtags()) should not be included in
1541                         # URLs, per RFC 2396.
1542                         $m2 = array();
1543                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1544                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1545                                 $url = substr( $url, 0, $m2[0][1] );
1546                         }
1547
1548                         # If the link text is an image URL, replace it with an <img> tag
1549                         # This happened by accident in the original parser, but some people used it extensively
1550                         $img = $this->maybeMakeExternalImage( $text );
1551                         if ( $img !== false ) {
1552                                 $text = $img;
1553                         }
1554
1555                         $dtrail = '';
1556
1557                         # Set linktype for CSS - if URL==text, link is essentially free
1558                         $linktype = ( $text === $url ) ? 'free' : 'text';
1559
1560                         # No link text, e.g. [http://domain.tld/some.link]
1561                         if ( $text == '' ) {
1562                                 # Autonumber
1563                                 $langObj = $this->getTargetLanguage();
1564                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1565                                 $linktype = 'autonumber';
1566                         } else {
1567                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1568                                 # Check for trail
1569                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1570                         }
1571
1572                         $text = $this->getConverterLanguage()->markNoConversion( $text );
1573
1574                         $url = Sanitizer::cleanUrl( $url );
1575
1576                         # Use the encoded URL
1577                         # This means that users can paste URLs directly into the text
1578                         # Funny characters like ö aren't valid in URLs anyway
1579                         # This was changed in August 2004
1580                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1581                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1582
1583                         # Register link in the output object.
1584                         # Replace unnecessary URL escape codes with the referenced character
1585                         # This prevents spammers from hiding links from the filters
1586                         $pasteurized = self::replaceUnusualEscapes( $url );
1587                         $this->mOutput->addExternalLink( $pasteurized );
1588                 }
1589
1590                 wfProfileOut( __METHOD__ );
1591                 return $s;
1592         }
1593
1594         /**
1595          * Get an associative array of additional HTML attributes appropriate for a
1596          * particular external link.  This currently may include rel => nofollow
1597          * (depending on configuration, namespace, and the URL's domain) and/or a
1598          * target attribute (depending on configuration).
1599          *
1600          * @param $url String|bool optional URL, to extract the domain from for rel =>
1601          *   nofollow if appropriate
1602          * @return Array associative array of HTML attributes
1603          */
1604         function getExternalLinkAttribs( $url = false ) {
1605                 $attribs = array();
1606                 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1607                 $ns = $this->mTitle->getNamespace();
1608                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1609                                 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1610                 {
1611                         $attribs['rel'] = 'nofollow';
1612                 }
1613                 if ( $this->mOptions->getExternalLinkTarget() ) {
1614                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1615                 }
1616                 return $attribs;
1617         }
1618
1619         /**
1620          * Replace unusual URL escape codes with their equivalent characters
1621          *
1622          * @param $url String
1623          * @return String
1624          *
1625          * @todo  This can merge genuinely required bits in the path or query string,
1626          *        breaking legit URLs. A proper fix would treat the various parts of
1627          *        the URL differently; as a workaround, just use the output for
1628          *        statistical records, not for actual linking/output.
1629          */
1630         static function replaceUnusualEscapes( $url ) {
1631                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1632                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1633         }
1634
1635         /**
1636          * Callback function used in replaceUnusualEscapes().
1637          * Replaces unusual URL escape codes with their equivalent character
1638          *
1639          * @param $matches array
1640          *
1641          * @return string
1642          */
1643         private static function replaceUnusualEscapesCallback( $matches ) {
1644                 $char = urldecode( $matches[0] );
1645                 $ord = ord( $char );
1646                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1647                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1648                         # No, shouldn't be escaped
1649                         return $char;
1650                 } else {
1651                         # Yes, leave it escaped
1652                         return $matches[0];
1653                 }
1654         }
1655
1656         /**
1657          * make an image if it's allowed, either through the global
1658          * option, through the exception, or through the on-wiki whitelist
1659          * @private
1660          *
1661          * $param $url string
1662          *
1663          * @return string
1664          */
1665         function maybeMakeExternalImage( $url ) {
1666                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1667                 $imagesexception = !empty( $imagesfrom );
1668                 $text = false;
1669                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1670                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1671                         $imagematch = false;
1672                         foreach ( $imagesfrom as $match ) {
1673                                 if ( strpos( $url, $match ) === 0 ) {
1674                                         $imagematch = true;
1675                                         break;
1676                                 }
1677                         }
1678                 } elseif ( $imagesexception ) {
1679                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1680                 } else {
1681                         $imagematch = false;
1682                 }
1683                 if ( $this->mOptions->getAllowExternalImages()
1684                          || ( $imagesexception && $imagematch ) ) {
1685                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1686                                 # Image found
1687                                 $text = Linker::makeExternalImage( $url );
1688                         }
1689                 }
1690                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1691                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1692                         $whitelist = explode( "\n", wfMessage( 'external_image_whitelist' )->inContentLanguage()->text() );
1693                         foreach ( $whitelist as $entry ) {
1694                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1695                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1696                                         continue;
1697                                 }
1698                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1699                                         # Image matches a whitelist entry
1700                                         $text = Linker::makeExternalImage( $url );
1701                                         break;
1702                                 }
1703                         }
1704                 }
1705                 return $text;
1706         }
1707
1708         /**
1709          * Process [[ ]] wikilinks
1710          *
1711          * @param $s string
1712          *
1713          * @return String: processed text
1714          *
1715          * @private
1716          */
1717         function replaceInternalLinks( $s ) {
1718                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1719                 return $s;
1720         }
1721
1722         /**
1723          * Process [[ ]] wikilinks (RIL)
1724          * @return LinkHolderArray
1725          *
1726          * @private
1727          */
1728         function replaceInternalLinks2( &$s ) {
1729                 wfProfileIn( __METHOD__ );
1730
1731                 wfProfileIn( __METHOD__.'-setup' );
1732                 static $tc = FALSE, $e1, $e1_img;
1733                 # the % is needed to support urlencoded titles as well
1734                 if ( !$tc ) {
1735                         $tc = Title::legalChars() . '#%';
1736                         # Match a link having the form [[namespace:link|alternate]]trail
1737                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1738                         # Match cases where there is no "]]", which might still be images
1739                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1740                 }
1741
1742                 $holders = new LinkHolderArray( $this );
1743
1744                 # split the entire text string on occurrences of [[
1745                 $a = StringUtils::explode( '[[', ' ' . $s );
1746                 # get the first element (all text up to first [[), and remove the space we added
1747                 $s = $a->current();
1748                 $a->next();
1749                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1750                 $s = substr( $s, 1 );
1751
1752                 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1753                 $e2 = null;
1754                 if ( $useLinkPrefixExtension ) {
1755                         # Match the end of a line for a word that's not followed by whitespace,
1756                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1757                         $e2 = wfMessage( 'linkprefix' )->inContentLanguage()->text();
1758                 }
1759
1760                 if ( is_null( $this->mTitle ) ) {
1761                         wfProfileOut( __METHOD__.'-setup' );
1762                         wfProfileOut( __METHOD__ );
1763                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1764                 }
1765                 $nottalk = !$this->mTitle->isTalkPage();
1766
1767                 if ( $useLinkPrefixExtension ) {
1768                         $m = array();
1769                         if ( preg_match( $e2, $s, $m ) ) {
1770                                 $first_prefix = $m[2];
1771                         } else {
1772                                 $first_prefix = false;
1773                         }
1774                 } else {
1775                         $prefix = '';
1776                 }
1777
1778                 if ( $this->getConverterLanguage()->hasVariants() ) {
1779                         $selflink = $this->getConverterLanguage()->autoConvertToAllVariants(
1780                                 $this->mTitle->getPrefixedText() );
1781                 } else {
1782                         $selflink = array( $this->mTitle->getPrefixedText() );
1783                 }
1784                 $useSubpages = $this->areSubpagesAllowed();
1785                 wfProfileOut( __METHOD__.'-setup' );
1786
1787                 # Loop for each link
1788                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1789                         # Check for excessive memory usage
1790                         if ( $holders->isBig() ) {
1791                                 # Too big
1792                                 # Do the existence check, replace the link holders and clear the array
1793                                 $holders->replace( $s );
1794                                 $holders->clear();
1795                         }
1796
1797                         if ( $useLinkPrefixExtension ) {
1798                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1799                                 if ( preg_match( $e2, $s, $m ) ) {
1800                                         $prefix = $m[2];
1801                                         $s = $m[1];
1802                                 } else {
1803                                         $prefix='';
1804                                 }
1805                                 # first link
1806                                 if ( $first_prefix ) {
1807                                         $prefix = $first_prefix;
1808                                         $first_prefix = false;
1809                                 }
1810                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1811                         }
1812
1813                         $might_be_img = false;
1814
1815                         wfProfileIn( __METHOD__."-e1" );
1816                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1817                                 $text = $m[2];
1818                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1819                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1820                                 # the real problem is with the $e1 regex
1821                                 # See bug 1300.
1822                                 #
1823                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1824                                 # and no image is in sight. See bug 2095.
1825                                 #
1826                                 if ( $text !== '' &&
1827                                         substr( $m[3], 0, 1 ) === ']' &&
1828                                         strpos( $text, '[' ) !== false
1829                                 )
1830                                 {
1831                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1832                                         $m[3] = substr( $m[3], 1 );
1833                                 }
1834                                 # fix up urlencoded title texts
1835                                 if ( strpos( $m[1], '%' ) !== false ) {
1836                                         # Should anchors '#' also be rejected?
1837                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1838                                 }
1839                                 $trail = $m[3];
1840                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1841                                 $might_be_img = true;
1842                                 $text = $m[2];
1843                                 if ( strpos( $m[1], '%' ) !== false ) {
1844                                         $m[1] = rawurldecode( $m[1] );
1845                                 }
1846                                 $trail = "";
1847                         } else { # Invalid form; output directly
1848                                 $s .= $prefix . '[[' . $line ;
1849                                 wfProfileOut( __METHOD__."-e1" );
1850                                 continue;
1851                         }
1852                         wfProfileOut( __METHOD__."-e1" );
1853                         wfProfileIn( __METHOD__."-misc" );
1854
1855                         # Don't allow internal links to pages containing
1856                         # PROTO: where PROTO is a valid URL protocol; these
1857                         # should be external links.
1858                         if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $m[1] ) ) {
1859                                 $s .= $prefix . '[[' . $line ;
1860                                 wfProfileOut( __METHOD__."-misc" );
1861                                 continue;
1862                         }
1863
1864                         # Make subpage if necessary
1865                         if ( $useSubpages ) {
1866                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1867                         } else {
1868                                 $link = $m[1];
1869                         }
1870
1871                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1872                         if ( !$noforce ) {
1873                                 # Strip off leading ':'
1874                                 $link = substr( $link, 1 );
1875                         }
1876
1877                         wfProfileOut( __METHOD__."-misc" );
1878                         wfProfileIn( __METHOD__."-title" );
1879                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1880                         if ( $nt === null ) {
1881                                 $s .= $prefix . '[[' . $line;
1882                                 wfProfileOut( __METHOD__."-title" );
1883                                 continue;
1884                         }
1885
1886                         $ns = $nt->getNamespace();
1887                         $iw = $nt->getInterWiki();
1888                         wfProfileOut( __METHOD__."-title" );
1889
1890                         if ( $might_be_img ) { # if this is actually an invalid link
1891                                 wfProfileIn( __METHOD__."-might_be_img" );
1892                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1893                                         $found = false;
1894                                         while ( true ) {
1895                                                 # look at the next 'line' to see if we can close it there
1896                                                 $a->next();
1897                                                 $next_line = $a->current();
1898                                                 if ( $next_line === false || $next_line === null ) {
1899                                                         break;
1900                                                 }
1901                                                 $m = explode( ']]', $next_line, 3 );
1902                                                 if ( count( $m ) == 3 ) {
1903                                                         # the first ]] closes the inner link, the second the image
1904                                                         $found = true;
1905                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1906                                                         $trail = $m[2];
1907                                                         break;
1908                                                 } elseif ( count( $m ) == 2 ) {
1909                                                         # if there's exactly one ]] that's fine, we'll keep looking
1910                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1911                                                 } else {
1912                                                         # if $next_line is invalid too, we need look no further
1913                                                         $text .= '[[' . $next_line;
1914                                                         break;
1915                                                 }
1916                                         }
1917                                         if ( !$found ) {
1918                                                 # we couldn't find the end of this imageLink, so output it raw
1919                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1920                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1921                                                 $s .= "{$prefix}[[$link|$text";
1922                                                 # note: no $trail, because without an end, there *is* no trail
1923                                                 wfProfileOut( __METHOD__."-might_be_img" );
1924                                                 continue;
1925                                         }
1926                                 } else { # it's not an image, so output it raw
1927                                         $s .= "{$prefix}[[$link|$text";
1928                                         # note: no $trail, because without an end, there *is* no trail
1929                                         wfProfileOut( __METHOD__."-might_be_img" );
1930                                         continue;
1931                                 }
1932                                 wfProfileOut( __METHOD__."-might_be_img" );
1933                         }
1934
1935                         $wasblank = ( $text  == '' );
1936                         if ( $wasblank ) {
1937                                 $text = $link;
1938                         } else {
1939                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1940                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1941                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1942                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1943                                 $text = $this->doQuotes( $text );
1944                         }
1945
1946                         # Link not escaped by : , create the various objects
1947                         if ( $noforce ) {
1948                                 # Interwikis
1949                                 wfProfileIn( __METHOD__."-interwiki" );
1950                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && Language::fetchLanguageName( $iw, null, 'mw' ) ) {
1951                                         // FIXME: the above check prevents links to sites with identifiers that are not language codes
1952                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1953                                         $s = rtrim( $s . $prefix );
1954                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1955                                         wfProfileOut( __METHOD__."-interwiki" );
1956                                         continue;
1957                                 }
1958                                 wfProfileOut( __METHOD__."-interwiki" );
1959
1960                                 if ( $ns == NS_FILE ) {
1961                                         wfProfileIn( __METHOD__."-image" );
1962                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1963                                                 if ( $wasblank ) {
1964                                                         # if no parameters were passed, $text
1965                                                         # becomes something like "File:Foo.png",
1966                                                         # which we don't want to pass on to the
1967                                                         # image generator
1968                                                         $text = '';
1969                                                 } else {
1970                                                         # recursively parse links inside the image caption
1971                                                         # actually, this will parse them in any other parameters, too,
1972                                                         # but it might be hard to fix that, and it doesn't matter ATM
1973                                                         $text = $this->replaceExternalLinks( $text );
1974                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1975                                                 }
1976                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1977                                                 $s .= $prefix . $this->armorLinks(
1978                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1979                                         } else {
1980                                                 $s .= $prefix . $trail;
1981                                         }
1982                                         wfProfileOut( __METHOD__."-image" );
1983                                         continue;
1984                                 }
1985
1986                                 if ( $ns == NS_CATEGORY ) {
1987                                         wfProfileIn( __METHOD__."-category" );
1988                                         $s = rtrim( $s . "\n" ); # bug 87
1989
1990                                         if ( $wasblank ) {
1991                                                 $sortkey = $this->getDefaultSort();
1992                                         } else {
1993                                                 $sortkey = $text;
1994                                         }
1995                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1996                                         $sortkey = str_replace( "\n", '', $sortkey );
1997                                         $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
1998                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1999
2000                                         /**
2001                                          * Strip the whitespace Category links produce, see bug 87
2002                                          * @todo We might want to use trim($tmp, "\n") here.
2003                                          */
2004                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2005
2006                                         wfProfileOut( __METHOD__."-category" );
2007                                         continue;
2008                                 }
2009                         }
2010
2011                         # Self-link checking
2012                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2013                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2014                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2015                                         continue;
2016                                 }
2017                         }
2018
2019                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2020                         # @todo FIXME: Should do batch file existence checks, see comment below
2021                         if ( $ns == NS_MEDIA ) {
2022                                 wfProfileIn( __METHOD__."-media" );
2023                                 # Give extensions a chance to select the file revision for us
2024                                 $options = array();
2025                                 $descQuery = false;
2026                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2027                                         array( $this, $nt, &$options, &$descQuery ) );
2028                                 # Fetch and register the file (file title may be different via hooks)
2029                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2030                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2031                                 $s .= $prefix . $this->armorLinks(
2032                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2033                                 wfProfileOut( __METHOD__."-media" );
2034                                 continue;
2035                         }
2036
2037                         wfProfileIn( __METHOD__."-always_known" );
2038                         # Some titles, such as valid special pages or files in foreign repos, should
2039                         # be shown as bluelinks even though they're not included in the page table
2040                         #
2041                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2042                         # batch file existence checks for NS_FILE and NS_MEDIA
2043                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2044                                 $this->mOutput->addLink( $nt );
2045                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2046                         } else {
2047                                 # Links will be added to the output link list after checking
2048                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2049                         }
2050                         wfProfileOut( __METHOD__."-always_known" );
2051                 }
2052                 wfProfileOut( __METHOD__ );
2053                 return $holders;
2054         }
2055
2056         /**
2057          * Render a forced-blue link inline; protect against double expansion of
2058          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2059          * Since this little disaster has to split off the trail text to avoid
2060          * breaking URLs in the following text without breaking trails on the
2061          * wiki links, it's been made into a horrible function.
2062          *
2063          * @param $nt Title
2064          * @param $text String
2065          * @param $query Array or String
2066          * @param $trail String
2067          * @param $prefix String
2068          * @return String: HTML-wikitext mix oh yuck
2069          */
2070         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2071                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2072
2073                 if ( is_string( $query ) ) {
2074                         $query = wfCgiToArray( $query );
2075                 }
2076                 if ( $text == '' ) {
2077                         $text = htmlspecialchars( $nt->getPrefixedText() );
2078                 }
2079
2080                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2081
2082                 return $this->armorLinks( $link ) . $trail;
2083         }
2084
2085         /**
2086          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2087          * going to go through further parsing steps before inline URL expansion.
2088          *
2089          * Not needed quite as much as it used to be since free links are a bit
2090          * more sensible these days. But bracketed links are still an issue.
2091          *
2092          * @param $text String: more-or-less HTML
2093          * @return String: less-or-more HTML with NOPARSE bits
2094          */
2095         function armorLinks( $text ) {
2096                 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2097                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2098         }
2099
2100         /**
2101          * Return true if subpage links should be expanded on this page.
2102          * @return Boolean
2103          */
2104         function areSubpagesAllowed() {
2105                 # Some namespaces don't allow subpages
2106                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2107         }
2108
2109         /**
2110          * Handle link to subpage if necessary
2111          *
2112          * @param $target String: the source of the link
2113          * @param &$text String: the link text, modified as necessary
2114          * @return string the full name of the link
2115          * @private
2116          */
2117         function maybeDoSubpageLink( $target, &$text ) {
2118                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2119         }
2120
2121         /**#@+
2122          * Used by doBlockLevels()
2123          * @private
2124          *
2125          * @return string
2126          */
2127         function closeParagraph() {
2128                 $result = '';
2129                 if ( $this->mLastSection != '' ) {
2130                         $result = '</' . $this->mLastSection  . ">\n";
2131                 }
2132                 $this->mInPre = false;
2133                 $this->mLastSection = '';
2134                 return $result;
2135         }
2136
2137         /**
2138          * getCommon() returns the length of the longest common substring
2139          * of both arguments, starting at the beginning of both.
2140          * @private
2141          *
2142          * @param $st1 string
2143          * @param $st2 string
2144          *
2145          * @return int
2146          */
2147         function getCommon( $st1, $st2 ) {
2148                 $fl = strlen( $st1 );
2149                 $shorter = strlen( $st2 );
2150                 if ( $fl < $shorter ) {
2151                         $shorter = $fl;
2152                 }
2153
2154                 for ( $i = 0; $i < $shorter; ++$i ) {
2155                         if ( $st1[$i] != $st2[$i] ) {
2156                                 break;
2157                         }
2158                 }
2159                 return $i;
2160         }
2161
2162         /**
2163          * These next three functions open, continue, and close the list
2164          * element appropriate to the prefix character passed into them.
2165          * @private
2166          *
2167          * @param $char string
2168          *
2169          * @return string
2170          */
2171         function openList( $char ) {
2172                 $result = $this->closeParagraph();
2173
2174                 if ( '*' === $char ) {
2175                         $result .= '<ul><li>';
2176                 } elseif ( '#' === $char ) {
2177                         $result .= '<ol><li>';
2178                 } elseif ( ':' === $char ) {
2179                         $result .= '<dl><dd>';
2180                 } elseif ( ';' === $char ) {
2181                         $result .= '<dl><dt>';
2182                         $this->mDTopen = true;
2183                 } else {
2184                         $result = '<!-- ERR 1 -->';
2185                 }
2186
2187                 return $result;
2188         }
2189
2190         /**
2191          * TODO: document
2192          * @param $char String
2193          * @private
2194          *
2195          * @return string
2196          */
2197         function nextItem( $char ) {
2198                 if ( '*' === $char || '#' === $char ) {
2199                         return '</li><li>';
2200                 } elseif ( ':' === $char || ';' === $char ) {
2201                         $close = '</dd>';
2202                         if ( $this->mDTopen ) {
2203                                 $close = '</dt>';
2204                         }
2205                         if ( ';' === $char ) {
2206                                 $this->mDTopen = true;
2207                                 return $close . '<dt>';
2208                         } else {
2209                                 $this->mDTopen = false;
2210                                 return $close . '<dd>';
2211                         }
2212                 }
2213                 return '<!-- ERR 2 -->';
2214         }
2215
2216         /**
2217          * TODO: document
2218          * @param $char String
2219          * @private
2220          *
2221          * @return string
2222          */
2223         function closeList( $char ) {
2224                 if ( '*' === $char ) {
2225                         $text = '</li></ul>';
2226                 } elseif ( '#' === $char ) {
2227                         $text = '</li></ol>';
2228                 } elseif ( ':' === $char ) {
2229                         if ( $this->mDTopen ) {
2230                                 $this->mDTopen = false;
2231                                 $text = '</dt></dl>';
2232                         } else {
2233                                 $text = '</dd></dl>';
2234                         }
2235                 } else {
2236                         return '<!-- ERR 3 -->';
2237                 }
2238                 return $text."\n";
2239         }
2240         /**#@-*/
2241
2242         /**
2243          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2244          *
2245          * @param $text String
2246          * @param $linestart Boolean: whether or not this is at the start of a line.
2247          * @private
2248          * @return string the lists rendered as HTML
2249          */
2250         function doBlockLevels( $text, $linestart ) {
2251                 wfProfileIn( __METHOD__ );
2252
2253                 # Parsing through the text line by line.  The main thing
2254                 # happening here is handling of block-level elements p, pre,
2255                 # and making lists from lines starting with * # : etc.
2256                 #
2257                 $textLines = StringUtils::explode( "\n", $text );
2258
2259                 $lastPrefix = $output = '';
2260                 $this->mDTopen = $inBlockElem = false;
2261                 $prefixLength = 0;
2262                 $paragraphStack = false;
2263
2264                 foreach ( $textLines as $oLine ) {
2265                         # Fix up $linestart
2266                         if ( !$linestart ) {
2267                                 $output .= $oLine;
2268                                 $linestart = true;
2269                                 continue;
2270                         }
2271                         # * = ul
2272                         # # = ol
2273                         # ; = dt
2274                         # : = dd
2275
2276                         $lastPrefixLength = strlen( $lastPrefix );
2277                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2278                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2279                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2280                         if ( !$this->mInPre ) {
2281                                 # Multiple prefixes may abut each other for nested lists.
2282                                 $prefixLength = strspn( $oLine, '*#:;' );
2283                                 $prefix = substr( $oLine, 0, $prefixLength );
2284
2285                                 # eh?
2286                                 # ; and : are both from definition-lists, so they're equivalent
2287                                 #  for the purposes of determining whether or not we need to open/close
2288                                 #  elements.
2289                                 $prefix2 = str_replace( ';', ':', $prefix );
2290                                 $t = substr( $oLine, $prefixLength );
2291                                 $this->mInPre = (bool)$preOpenMatch;
2292                         } else {
2293                                 # Don't interpret any other prefixes in preformatted text
2294                                 $prefixLength = 0;
2295                                 $prefix = $prefix2 = '';
2296                                 $t = $oLine;
2297                         }
2298
2299                         # List generation
2300                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2301                                 # Same as the last item, so no need to deal with nesting or opening stuff
2302                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2303                                 $paragraphStack = false;
2304
2305                                 if ( substr( $prefix, -1 ) === ';') {
2306                                         # The one nasty exception: definition lists work like this:
2307                                         # ; title : definition text
2308                                         # So we check for : in the remainder text to split up the
2309                                         # title and definition, without b0rking links.
2310                                         $term = $t2 = '';
2311                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2312                                                 $t = $t2;
2313                                                 $output .= $term . $this->nextItem( ':' );
2314                                         }
2315                                 }
2316                         } elseif ( $prefixLength || $lastPrefixLength ) {
2317                                 # We need to open or close prefixes, or both.
2318
2319                                 # Either open or close a level...
2320                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2321                                 $paragraphStack = false;
2322
2323                                 # Close all the prefixes which aren't shared.
2324                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2325                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2326                                         --$lastPrefixLength;
2327                                 }
2328
2329                                 # Continue the current prefix if appropriate.
2330                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2331                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2332                                 }
2333
2334                                 # Open prefixes where appropriate.
2335                                 while ( $prefixLength > $commonPrefixLength ) {
2336                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2337                                         $output .= $this->openList( $char );
2338
2339                                         if ( ';' === $char ) {
2340                                                 # @todo FIXME: This is dupe of code above
2341                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2342                                                         $t = $t2;
2343                                                         $output .= $term . $this->nextItem( ':' );
2344                                                 }
2345                                         }
2346                                         ++$commonPrefixLength;
2347                                 }
2348                                 $lastPrefix = $prefix2;
2349                         }
2350
2351                         # If we have no prefixes, go to paragraph mode.
2352                         if ( 0 == $prefixLength ) {
2353                                 wfProfileIn( __METHOD__."-paragraph" );
2354                                 # No prefix (not in list)--go to paragraph mode
2355                                 # XXX: use a stack for nestable elements like span, table and div
2356                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2357                                 $closematch = preg_match(
2358                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2359                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2360                                 if ( $openmatch or $closematch ) {
2361                                         $paragraphStack = false;
2362                                         # TODO bug 5718: paragraph closed
2363                                         $output .= $this->closeParagraph();
2364                                         if ( $preOpenMatch and !$preCloseMatch ) {
2365                                                 $this->mInPre = true;
2366                                         }
2367                                         $inBlockElem = !$closematch;
2368                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2369                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2370                                                 # pre
2371                                                 if ( $this->mLastSection !== 'pre' ) {
2372                                                         $paragraphStack = false;
2373                                                         $output .= $this->closeParagraph().'<pre>';
2374                                                         $this->mLastSection = 'pre';
2375                                                 }
2376                                                 $t = substr( $t, 1 );
2377                                         } else {
2378                                                 # paragraph
2379                                                 if ( trim( $t ) === '' ) {
2380                                                         if ( $paragraphStack ) {
2381                                                                 $output .= $paragraphStack.'<br />';
2382                                                                 $paragraphStack = false;
2383                                                                 $this->mLastSection = 'p';
2384                                                         } else {
2385                                                                 if ( $this->mLastSection !== 'p' ) {
2386                                                                         $output .= $this->closeParagraph();
2387                                                                         $this->mLastSection = '';
2388                                                                         $paragraphStack = '<p>';
2389                                                                 } else {
2390                                                                         $paragraphStack = '</p><p>';
2391                                                                 }
2392                                                         }
2393                                                 } else {
2394                                                         if ( $paragraphStack ) {
2395                                                                 $output .= $paragraphStack;
2396                                                                 $paragraphStack = false;
2397                                                                 $this->mLastSection = 'p';
2398                                                         } elseif ( $this->mLastSection !== 'p' ) {
2399                                                                 $output .= $this->closeParagraph().'<p>';
2400                                                                 $this->mLastSection = 'p';
2401                                                         }
2402                                                 }
2403                                         }
2404                                 }
2405                                 wfProfileOut( __METHOD__."-paragraph" );
2406                         }
2407                         # somewhere above we forget to get out of pre block (bug 785)
2408                         if ( $preCloseMatch && $this->mInPre ) {
2409                                 $this->mInPre = false;
2410                         }
2411                         if ( $paragraphStack === false ) {
2412                                 $output .= $t."\n";
2413                         }
2414                 }
2415                 while ( $prefixLength ) {
2416                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2417                         --$prefixLength;
2418                 }
2419                 if ( $this->mLastSection != '' ) {
2420                         $output .= '</' . $this->mLastSection . '>';
2421                         $this->mLastSection = '';
2422                 }
2423
2424                 wfProfileOut( __METHOD__ );
2425                 return $output;
2426         }
2427
2428         /**
2429          * Split up a string on ':', ignoring any occurrences inside tags
2430          * to prevent illegal overlapping.
2431          *
2432          * @param $str String the string to split
2433          * @param &$before String set to everything before the ':'
2434          * @param &$after String set to everything after the ':'
2435          * @return String the position of the ':', or false if none found
2436          */
2437         function findColonNoLinks( $str, &$before, &$after ) {
2438                 wfProfileIn( __METHOD__ );
2439
2440                 $pos = strpos( $str, ':' );
2441                 if ( $pos === false ) {
2442                         # Nothing to find!
2443                         wfProfileOut( __METHOD__ );
2444                         return false;
2445                 }
2446
2447                 $lt = strpos( $str, '<' );
2448                 if ( $lt === false || $lt > $pos ) {
2449                         # Easy; no tag nesting to worry about
2450                         $before = substr( $str, 0, $pos );
2451                         $after = substr( $str, $pos+1 );
2452                         wfProfileOut( __METHOD__ );
2453                         return $pos;
2454                 }
2455
2456                 # Ugly state machine to walk through avoiding tags.
2457                 $state = self::COLON_STATE_TEXT;
2458                 $stack = 0;
2459                 $len = strlen( $str );
2460                 for( $i = 0; $i < $len; $i++ ) {
2461                         $c = $str[$i];
2462
2463                         switch( $state ) {
2464                         # (Using the number is a performance hack for common cases)
2465                         case 0: # self::COLON_STATE_TEXT:
2466                                 switch( $c ) {
2467                                 case "<":
2468                                         # Could be either a <start> tag or an </end> tag
2469                                         $state = self::COLON_STATE_TAGSTART;
2470                                         break;
2471                                 case ":":
2472                                         if ( $stack == 0 ) {
2473                                                 # We found it!
2474                                                 $before = substr( $str, 0, $i );
2475                                                 $after = substr( $str, $i + 1 );
2476                                                 wfProfileOut( __METHOD__ );
2477                                                 return $i;
2478                                         }
2479                                         # Embedded in a tag; don't break it.
2480                                         break;
2481                                 default:
2482                                         # Skip ahead looking for something interesting
2483                                         $colon = strpos( $str, ':', $i );
2484                                         if ( $colon === false ) {
2485                                                 # Nothing else interesting
2486                                                 wfProfileOut( __METHOD__ );
2487                                                 return false;
2488                                         }
2489                                         $lt = strpos( $str, '<', $i );
2490                                         if ( $stack === 0 ) {
2491                                                 if ( $lt === false || $colon < $lt ) {
2492                                                         # We found it!
2493                                                         $before = substr( $str, 0, $colon );
2494                                                         $after = substr( $str, $colon + 1 );
2495                                                         wfProfileOut( __METHOD__ );
2496                                                         return $i;
2497                                                 }
2498                                         }
2499                                         if ( $lt === false ) {
2500                                                 # Nothing else interesting to find; abort!
2501                                                 # We're nested, but there's no close tags left. Abort!
2502                                                 break 2;
2503                                         }
2504                                         # Skip ahead to next tag start
2505                                         $i = $lt;
2506                                         $state = self::COLON_STATE_TAGSTART;
2507                                 }
2508                                 break;
2509                         case 1: # self::COLON_STATE_TAG:
2510                                 # In a <tag>
2511                                 switch( $c ) {
2512                                 case ">":
2513                                         $stack++;
2514                                         $state = self::COLON_STATE_TEXT;
2515                                         break;
2516                                 case "/":
2517                                         # Slash may be followed by >?
2518                                         $state = self::COLON_STATE_TAGSLASH;
2519                                         break;
2520                                 default:
2521                                         # ignore
2522                                 }
2523                                 break;
2524                         case 2: # self::COLON_STATE_TAGSTART:
2525                                 switch( $c ) {
2526                                 case "/":
2527                                         $state = self::COLON_STATE_CLOSETAG;
2528                                         break;
2529                                 case "!":
2530                                         $state = self::COLON_STATE_COMMENT;
2531                                         break;
2532                                 case ">":
2533                                         # Illegal early close? This shouldn't happen D:
2534                                         $state = self::COLON_STATE_TEXT;
2535                                         break;
2536                                 default:
2537                                         $state = self::COLON_STATE_TAG;
2538                                 }
2539                                 break;
2540                         case 3: # self::COLON_STATE_CLOSETAG:
2541                                 # In a </tag>
2542                                 if ( $c === ">" ) {
2543                                         $stack--;
2544                                         if ( $stack < 0 ) {
2545                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2546                                                 wfProfileOut( __METHOD__ );
2547                                                 return false;
2548                                         }
2549                                         $state = self::COLON_STATE_TEXT;
2550                                 }
2551                                 break;
2552                         case self::COLON_STATE_TAGSLASH:
2553                                 if ( $c === ">" ) {
2554                                         # Yes, a self-closed tag <blah/>
2555                                         $state = self::COLON_STATE_TEXT;
2556                                 } else {
2557                                         # Probably we're jumping the gun, and this is an attribute
2558                                         $state = self::COLON_STATE_TAG;
2559                                 }
2560                                 break;
2561                         case 5: # self::COLON_STATE_COMMENT:
2562                                 if ( $c === "-" ) {
2563                                         $state = self::COLON_STATE_COMMENTDASH;
2564                                 }
2565                                 break;
2566                         case self::COLON_STATE_COMMENTDASH:
2567                                 if ( $c === "-" ) {
2568                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2569                                 } else {
2570                                         $state = self::COLON_STATE_COMMENT;
2571                                 }
2572                                 break;
2573                         case self::COLON_STATE_COMMENTDASHDASH:
2574                                 if ( $c === ">" ) {
2575                                         $state = self::COLON_STATE_TEXT;
2576                                 } else {
2577                                         $state = self::COLON_STATE_COMMENT;
2578                                 }
2579                                 break;
2580                         default:
2581                                 throw new MWException( "State machine error in " . __METHOD__ );
2582                         }
2583                 }
2584                 if ( $stack > 0 ) {
2585                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2586                         wfProfileOut( __METHOD__ );
2587                         return false;
2588                 }
2589                 wfProfileOut( __METHOD__ );
2590                 return false;
2591         }
2592
2593         /**
2594          * Return value of a magic variable (like PAGENAME)
2595          *
2596          * @private
2597          *
2598          * @param $index integer
2599          * @param $frame PPFrame
2600          *
2601          * @return string
2602          */
2603         function getVariableValue( $index, $frame = false ) {
2604                 global $wgContLang, $wgSitename, $wgServer;
2605                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2606
2607                 if ( is_null( $this->mTitle ) ) {
2608                         // If no title set, bad things are going to happen
2609                         // later. Title should always be set since this
2610                         // should only be called in the middle of a parse
2611                         // operation (but the unit-tests do funky stuff)
2612                         throw new MWException( __METHOD__ . ' Should only be '
2613                                 . ' called while parsing (no title set)' );
2614                 }
2615
2616                 /**
2617                  * Some of these require message or data lookups and can be
2618                  * expensive to check many times.
2619                  */
2620                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2621                         if ( isset( $this->mVarCache[$index] ) ) {
2622                                 return $this->mVarCache[$index];
2623                         }
2624                 }
2625
2626                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2627                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2628
2629                 # Use the time zone
2630                 global $wgLocaltimezone;
2631                 if ( isset( $wgLocaltimezone ) ) {
2632                         $oldtz = date_default_timezone_get();
2633                         date_default_timezone_set( $wgLocaltimezone );
2634                 }
2635
2636                 $localTimestamp = date( 'YmdHis', $ts );
2637                 $localMonth = date( 'm', $ts );
2638                 $localMonth1 = date( 'n', $ts );
2639                 $localMonthName = date( 'n', $ts );
2640                 $localDay = date( 'j', $ts );
2641                 $localDay2 = date( 'd', $ts );
2642                 $localDayOfWeek = date( 'w', $ts );
2643                 $localWeek = date( 'W', $ts );
2644                 $localYear = date( 'Y', $ts );
2645                 $localHour = date( 'H', $ts );
2646                 if ( isset( $wgLocaltimezone ) ) {
2647                         date_default_timezone_set( $oldtz );
2648                 }
2649
2650                 $pageLang = $this->getFunctionLang();
2651
2652                 switch ( $index ) {
2653                         case 'currentmonth':
2654                                 $value = $pageLang->formatNum( gmdate( 'm', $ts ) );
2655                                 break;
2656                         case 'currentmonth1':
2657                                 $value = $pageLang->formatNum( gmdate( 'n', $ts ) );
2658                                 break;
2659                         case 'currentmonthname':
2660                                 $value = $pageLang->getMonthName( gmdate( 'n', $ts ) );
2661                                 break;
2662                         case 'currentmonthnamegen':
2663                                 $value = $pageLang->getMonthNameGen( gmdate( 'n', $ts ) );
2664                                 break;
2665                         case 'currentmonthabbrev':
2666                                 $value = $pageLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2667                                 break;
2668                         case 'currentday':
2669                                 $value = $pageLang->formatNum( gmdate( 'j', $ts ) );
2670                                 break;
2671                         case 'currentday2':
2672                                 $value = $pageLang->formatNum( gmdate( 'd', $ts ) );
2673                                 break;
2674                         case 'localmonth':
2675                                 $value = $pageLang->formatNum( $localMonth );
2676                                 break;
2677                         case 'localmonth1':
2678                                 $value = $pageLang->formatNum( $localMonth1 );
2679                                 break;
2680                         case 'localmonthname':
2681                                 $value = $pageLang->getMonthName( $localMonthName );
2682                                 break;
2683                         case 'localmonthnamegen':
2684                                 $value = $pageLang->getMonthNameGen( $localMonthName );
2685                                 break;
2686                         case 'localmonthabbrev':
2687                                 $value = $pageLang->getMonthAbbreviation( $localMonthName );
2688                                 break;
2689                         case 'localday':
2690                                 $value = $pageLang->formatNum( $localDay );
2691                                 break;
2692                         case 'localday2':
2693                                 $value = $pageLang->formatNum( $localDay2 );
2694                                 break;
2695                         case 'pagename':
2696                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2697                                 break;
2698                         case 'pagenamee':
2699                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2700                                 break;
2701                         case 'fullpagename':
2702                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2703                                 break;
2704                         case 'fullpagenamee':
2705                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2706                                 break;
2707                         case 'subpagename':
2708                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2709                                 break;
2710                         case 'subpagenamee':
2711                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2712                                 break;
2713                         case 'basepagename':
2714                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2715                                 break;
2716                         case 'basepagenamee':
2717                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2718                                 break;
2719                         case 'talkpagename':
2720                                 if ( $this->mTitle->canTalk() ) {
2721                                         $talkPage = $this->mTitle->getTalkPage();
2722                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2723                                 } else {
2724                                         $value = '';
2725                                 }
2726                                 break;
2727                         case 'talkpagenamee':
2728                                 if ( $this->mTitle->canTalk() ) {
2729                                         $talkPage = $this->mTitle->getTalkPage();
2730                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2731                                 } else {
2732                                         $value = '';
2733                                 }
2734                                 break;
2735                         case 'subjectpagename':
2736                                 $subjPage = $this->mTitle->getSubjectPage();
2737                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2738                                 break;
2739                         case 'subjectpagenamee':
2740                                 $subjPage = $this->mTitle->getSubjectPage();
2741                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2742                                 break;
2743                         case 'pageid': // requested in bug 23427
2744                                 $pageid = $this->getTitle()->getArticleId();
2745                                 if( $pageid == 0 ) {
2746                                         # 0 means the page doesn't exist in the database,
2747                                         # which means the user is previewing a new page.
2748                                         # The vary-revision flag must be set, because the magic word
2749                                         # will have a different value once the page is saved.
2750                                         $this->mOutput->setFlag( 'vary-revision' );
2751                                         wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2752                                 }
2753                                 $value = $pageid ? $pageid : null;
2754                                 break;
2755                         case 'revisionid':
2756                                 # Let the edit saving system know we should parse the page
2757                                 # *after* a revision ID has been assigned.
2758                                 $this->mOutput->setFlag( 'vary-revision' );
2759                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2760                                 $value = $this->mRevisionId;
2761                                 break;
2762                         case 'revisionday':
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__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2767                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2768                                 break;
2769                         case 'revisionday2':
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__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2774                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2775                                 break;
2776                         case 'revisionmonth':
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__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2781                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2782                                 break;
2783                         case 'revisionmonth1':
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__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2788                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2789                                 break;
2790                         case 'revisionyear':
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__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2795                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2796                                 break;
2797                         case 'revisiontimestamp':
2798                                 # Let the edit saving system know we should parse the page
2799                                 # *after* a revision ID has been assigned. This is for null edits.
2800                                 $this->mOutput->setFlag( 'vary-revision' );
2801                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2802                                 $value = $this->getRevisionTimestamp();
2803                                 break;
2804                         case 'revisionuser':
2805                                 # Let the edit saving system know we should parse the page
2806                                 # *after* a revision ID has been assigned. This is for null edits.
2807                                 $this->mOutput->setFlag( 'vary-revision' );
2808                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2809                                 $value = $this->getRevisionUser();
2810                                 break;
2811                         case 'namespace':
2812                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2813                                 break;
2814                         case 'namespacee':
2815                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2816                                 break;
2817                         case 'namespacenumber':
2818                                 $value = $this->mTitle->getNamespace();
2819                                 break;
2820                         case 'talkspace':
2821                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2822                                 break;
2823                         case 'talkspacee':
2824                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2825                                 break;
2826                         case 'subjectspace':
2827                                 $value = $this->mTitle->getSubjectNsText();
2828                                 break;
2829                         case 'subjectspacee':
2830                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2831                                 break;
2832                         case 'currentdayname':
2833                                 $value = $pageLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2834                                 break;
2835                         case 'currentyear':
2836                                 $value = $pageLang->formatNum( gmdate( 'Y', $ts ), true );
2837                                 break;
2838                         case 'currenttime':
2839                                 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2840                                 break;
2841                         case 'currenthour':
2842                                 $value = $pageLang->formatNum( gmdate( 'H', $ts ), true );
2843                                 break;
2844                         case 'currentweek':
2845                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2846                                 # int to remove the padding
2847                                 $value = $pageLang->formatNum( (int)gmdate( 'W', $ts ) );
2848                                 break;
2849                         case 'currentdow':
2850                                 $value = $pageLang->formatNum( gmdate( 'w', $ts ) );
2851                                 break;
2852                         case 'localdayname':
2853                                 $value = $pageLang->getWeekdayName( $localDayOfWeek + 1 );
2854                                 break;
2855                         case 'localyear':
2856                                 $value = $pageLang->formatNum( $localYear, true );
2857                                 break;
2858                         case 'localtime':
2859                                 $value = $pageLang->time( $localTimestamp, false, false );
2860                                 break;
2861                         case 'localhour':
2862                                 $value = $pageLang->formatNum( $localHour, true );
2863                                 break;
2864                         case 'localweek':
2865                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2866                                 # int to remove the padding
2867                                 $value = $pageLang->formatNum( (int)$localWeek );
2868                                 break;
2869                         case 'localdow':
2870                                 $value = $pageLang->formatNum( $localDayOfWeek );
2871                                 break;
2872                         case 'numberofarticles':
2873                                 $value = $pageLang->formatNum( SiteStats::articles() );
2874                                 break;
2875                         case 'numberoffiles':
2876                                 $value = $pageLang->formatNum( SiteStats::images() );
2877                                 break;
2878                         case 'numberofusers':
2879                                 $value = $pageLang->formatNum( SiteStats::users() );
2880                                 break;
2881                         case 'numberofactiveusers':
2882                                 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2883                                 break;
2884                         case 'numberofpages':
2885                                 $value = $pageLang->formatNum( SiteStats::pages() );
2886                                 break;
2887                         case 'numberofadmins':
2888                                 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2889                                 break;
2890                         case 'numberofedits':
2891                                 $value = $pageLang->formatNum( SiteStats::edits() );
2892                                 break;
2893                         case 'numberofviews':
2894                                 global $wgDisableCounters;
2895                                 $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
2896                                 break;
2897                         case 'currenttimestamp':
2898                                 $value = wfTimestamp( TS_MW, $ts );
2899                                 break;
2900                         case 'localtimestamp':
2901                                 $value = $localTimestamp;
2902                                 break;
2903                         case 'currentversion':
2904                                 $value = SpecialVersion::getVersion();
2905                                 break;
2906                         case 'articlepath':
2907                                 return $wgArticlePath;
2908                         case 'sitename':
2909                                 return $wgSitename;
2910                         case 'server':
2911                                 return $wgServer;
2912                         case 'servername':
2913                                 $serverParts = wfParseUrl( $wgServer );
2914                                 return $serverParts && isset( $serverParts['host'] ) ? $serverParts['host'] : $wgServer;
2915                         case 'scriptpath':
2916                                 return $wgScriptPath;
2917                         case 'stylepath':
2918                                 return $wgStylePath;
2919                         case 'directionmark':
2920                                 return $pageLang->getDirMark();
2921                         case 'contentlanguage':
2922                                 global $wgLanguageCode;
2923                                 return $wgLanguageCode;
2924                         default:
2925                                 $ret = null;
2926                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2927                                         return $ret;
2928                                 } else {
2929                                         return null;
2930                                 }
2931                 }
2932
2933                 if ( $index ) {
2934                         $this->mVarCache[$index] = $value;
2935                 }
2936
2937                 return $value;
2938         }
2939
2940         /**
2941          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2942          *
2943          * @private
2944          */
2945         function initialiseVariables() {
2946                 wfProfileIn( __METHOD__ );
2947                 $variableIDs = MagicWord::getVariableIDs();
2948                 $substIDs = MagicWord::getSubstIDs();
2949
2950                 $this->mVariables = new MagicWordArray( $variableIDs );
2951                 $this->mSubstWords = new MagicWordArray( $substIDs );
2952                 wfProfileOut( __METHOD__ );
2953         }
2954
2955         /**
2956          * Preprocess some wikitext and return the document tree.
2957          * This is the ghost of replace_variables().
2958          *
2959          * @param $text String: The text to parse
2960          * @param $flags Integer: bitwise combination of:
2961          *          self::PTD_FOR_INCLUSION    Handle "<noinclude>" and "<includeonly>" as if the text is being
2962          *                                     included. Default is to assume a direct page view.
2963          *
2964          * The generated DOM tree must depend only on the input text and the flags.
2965          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2966          *
2967          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2968          * change in the DOM tree for a given text, must be passed through the section identifier
2969          * in the section edit link and thus back to extractSections().
2970          *
2971          * The output of this function is currently only cached in process memory, but a persistent
2972          * cache may be implemented at a later date which takes further advantage of these strict
2973          * dependency requirements.
2974          *
2975          * @private
2976          *
2977          * @return PPNode
2978          */
2979         function preprocessToDom( $text, $flags = 0 ) {
2980                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2981                 return $dom;
2982         }
2983
2984         /**
2985          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2986          *
2987          * @param $s string
2988          *
2989          * @return array
2990          */
2991         public static function splitWhitespace( $s ) {
2992                 $ltrimmed = ltrim( $s );
2993                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2994                 $trimmed = rtrim( $ltrimmed );
2995                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2996                 if ( $diff > 0 ) {
2997                         $w2 = substr( $ltrimmed, -$diff );
2998                 } else {
2999                         $w2 = '';
3000                 }
3001                 return array( $w1, $trimmed, $w2 );
3002         }
3003
3004         /**
3005          * Replace magic variables, templates, and template arguments
3006          * with the appropriate text. Templates are substituted recursively,
3007          * taking care to avoid infinite loops.
3008          *
3009          * Note that the substitution depends on value of $mOutputType:
3010          *  self::OT_WIKI: only {{subst:}} templates
3011          *  self::OT_PREPROCESS: templates but not extension tags
3012          *  self::OT_HTML: all templates and extension tags
3013          *
3014          * @param $text String the text to transform
3015          * @param $frame PPFrame Object describing the arguments passed to the template.
3016          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
3017          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
3018          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
3019          * @private
3020          *
3021          * @return string
3022          */
3023         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3024                 # Is there any text? Also, Prevent too big inclusions!
3025                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3026                         return $text;
3027                 }
3028                 wfProfileIn( __METHOD__ );
3029
3030                 if ( $frame === false ) {
3031                         $frame = $this->getPreprocessor()->newFrame();
3032                 } elseif ( !( $frame instanceof PPFrame ) ) {
3033                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
3034                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3035                 }
3036
3037                 $dom = $this->preprocessToDom( $text );
3038                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3039                 $text = $frame->expand( $dom, $flags );
3040
3041                 wfProfileOut( __METHOD__ );
3042                 return $text;
3043         }
3044
3045         /**
3046          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3047          *
3048          * @param $args array
3049          *
3050          * @return array
3051          */
3052         static function createAssocArgs( $args ) {
3053                 $assocArgs = array();
3054                 $index = 1;
3055                 foreach ( $args as $arg ) {
3056                         $eqpos = strpos( $arg, '=' );
3057                         if ( $eqpos === false ) {
3058                                 $assocArgs[$index++] = $arg;
3059                         } else {
3060                                 $name = trim( substr( $arg, 0, $eqpos ) );
3061                                 $value = trim( substr( $arg, $eqpos+1 ) );
3062                                 if ( $value === false ) {
3063                                         $value = '';
3064                                 }
3065                                 if ( $name !== false ) {
3066                                         $assocArgs[$name] = $value;
3067                                 }
3068                         }
3069                 }
3070
3071                 return $assocArgs;
3072         }
3073
3074         /**
3075          * Warn the user when a parser limitation is reached
3076          * Will warn at most once the user per limitation type
3077          *
3078          * @param $limitationType String: should be one of:
3079          *   'expensive-parserfunction' (corresponding messages:
3080          *       'expensive-parserfunction-warning',
3081          *       'expensive-parserfunction-category')
3082          *   'post-expand-template-argument' (corresponding messages:
3083          *       'post-expand-template-argument-warning',
3084          *       'post-expand-template-argument-category')
3085          *   'post-expand-template-inclusion' (corresponding messages:
3086          *       'post-expand-template-inclusion-warning',
3087          *       'post-expand-template-inclusion-category')
3088          * @param $current int|null Current value
3089          * @param $max int|null Maximum allowed, when an explicit limit has been
3090          *       exceeded, provide the values (optional)
3091          */
3092         function limitationWarn( $limitationType, $current = '', $max = '' ) {
3093                 # does no harm if $current and $max are present but are unnecessary for the message
3094                 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3095                         ->inContentLanguage()->escaped();
3096                 $this->mOutput->addWarning( $warning );
3097                 $this->addTrackingCategory( "$limitationType-category" );
3098         }
3099
3100         /**
3101          * Return the text of a template, after recursively
3102          * replacing any variables or templates within the template.
3103          *
3104          * @param $piece Array: the parts of the template
3105          *  $piece['title']: the title, i.e. the part before the |
3106          *  $piece['parts']: the parameter array
3107          *  $piece['lineStart']: whether the brace was at the start of a line
3108          * @param $frame PPFrame The current frame, contains template arguments
3109          * @return String: the text of the template
3110          * @private
3111          */
3112         function braceSubstitution( $piece, $frame ) {
3113                 global $wgContLang;
3114                 wfProfileIn( __METHOD__ );
3115                 wfProfileIn( __METHOD__.'-setup' );
3116
3117                 # Flags
3118                 $found = false;             # $text has been filled
3119                 $nowiki = false;            # wiki markup in $text should be escaped
3120                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3121                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3122                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3123                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3124
3125                 # Title object, where $text came from
3126                 $title = false;
3127
3128                 # $part1 is the bit before the first |, and must contain only title characters.
3129                 # Various prefixes will be stripped from it later.
3130                 $titleWithSpaces = $frame->expand( $piece['title'] );
3131                 $part1 = trim( $titleWithSpaces );
3132                 $titleText = false;
3133
3134                 # Original title text preserved for various purposes
3135                 $originalTitle = $part1;
3136
3137                 # $args is a list of argument nodes, starting from index 0, not including $part1
3138                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3139                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3140                 wfProfileOut( __METHOD__.'-setup' );
3141
3142                 $titleProfileIn = null; // profile templates
3143
3144                 # SUBST
3145                 wfProfileIn( __METHOD__.'-modifiers' );
3146                 if ( !$found ) {
3147
3148                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3149
3150                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3151                         # Decide whether to expand template or keep wikitext as-is.
3152                         if ( $this->ot['wiki'] ) {
3153                                 if ( $substMatch === false ) {
3154                                         $literal = true;  # literal when in PST with no prefix
3155                                 } else {
3156                                         $literal = false; # expand when in PST with subst: or safesubst:
3157                                 }
3158                         } else {
3159                                 if ( $substMatch == 'subst' ) {
3160                                         $literal = true;  # literal when not in PST with plain subst:
3161                                 } else {
3162                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3163                                 }
3164                         }
3165                         if ( $literal ) {
3166                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3167                                 $isLocalObj = true;
3168                                 $found = true;
3169                         }
3170                 }
3171
3172                 # Variables
3173                 if ( !$found && $args->getLength() == 0 ) {
3174                         $id = $this->mVariables->matchStartToEnd( $part1 );
3175                         if ( $id !== false ) {
3176                                 $text = $this->getVariableValue( $id, $frame );
3177                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3178                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3179                                 }
3180                                 $found = true;
3181                         }
3182                 }
3183
3184                 # MSG, MSGNW and RAW
3185                 if ( !$found ) {
3186                         # Check for MSGNW:
3187                         $mwMsgnw = MagicWord::get( 'msgnw' );
3188                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3189                                 $nowiki = true;
3190                         } else {
3191                                 # Remove obsolete MSG:
3192                                 $mwMsg = MagicWord::get( 'msg' );
3193                                 $mwMsg->matchStartAndRemove( $part1 );
3194                         }
3195
3196                         # Check for RAW:
3197                         $mwRaw = MagicWord::get( 'raw' );
3198                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3199                                 $forceRawInterwiki = true;
3200                         }
3201                 }
3202                 wfProfileOut( __METHOD__.'-modifiers' );
3203
3204                 # Parser functions
3205                 if ( !$found ) {
3206                         wfProfileIn( __METHOD__ . '-pfunc' );
3207
3208                         $colonPos = strpos( $part1, ':' );
3209                         if ( $colonPos !== false ) {
3210                                 # Case sensitive functions
3211                                 $function = substr( $part1, 0, $colonPos );
3212                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3213                                         $function = $this->mFunctionSynonyms[1][$function];
3214                                 } else {
3215                                         # Case insensitive functions
3216                                         $function = $wgContLang->lc( $function );
3217                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3218                                                 $function = $this->mFunctionSynonyms[0][$function];
3219                                         } else {
3220                                                 $function = false;
3221                                         }
3222                                 }
3223                                 if ( $function ) {
3224                                         wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3225                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3226                                         $initialArgs = array( &$this );
3227                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3228                                         if ( $flags & SFH_OBJECT_ARGS ) {
3229                                                 # Add a frame parameter, and pass the arguments as an array
3230                                                 $allArgs = $initialArgs;
3231                                                 $allArgs[] = $frame;
3232                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3233                                                         $funcArgs[] = $args->item( $i );
3234                                                 }
3235                                                 $allArgs[] = $funcArgs;
3236                                         } else {
3237                                                 # Convert arguments to plain text
3238                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3239                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3240                                                 }
3241                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3242                                         }
3243
3244                                         # Workaround for PHP bug 35229 and similar
3245                                         if ( !is_callable( $callback ) ) {
3246                                                 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3247                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3248                                                 wfProfileOut( __METHOD__ );
3249                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3250                                         }
3251                                         $result = call_user_func_array( $callback, $allArgs );
3252                                         $found = true;
3253                                         $noparse = true;
3254                                         $preprocessFlags = 0;
3255
3256                                         if ( is_array( $result ) ) {
3257                                                 if ( isset( $result[0] ) ) {
3258                                                         $text = $result[0];
3259                                                         unset( $result[0] );
3260                                                 }
3261
3262                                                 # Extract flags into the local scope
3263                                                 # This allows callers to set flags such as nowiki, found, etc.
3264                                                 extract( $result );
3265                                         } else {
3266                                                 $text = $result;
3267                                         }
3268                                         if ( !$noparse ) {
3269                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3270                                                 $isChildObj = true;
3271                                         }
3272                                         wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3273                                 }
3274                         }
3275                         wfProfileOut( __METHOD__ . '-pfunc' );
3276                 }
3277
3278                 # Finish mangling title and then check for loops.
3279                 # Set $title to a Title object and $titleText to the PDBK
3280                 if ( !$found ) {
3281                         $ns = NS_TEMPLATE;
3282                         # Split the title into page and subpage
3283                         $subpage = '';
3284                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3285                         if ( $subpage !== '' ) {
3286                                 $ns = $this->mTitle->getNamespace();
3287                         }
3288                         $title = Title::newFromText( $part1, $ns );
3289                         if ( $title ) {
3290                                 $titleText = $title->getPrefixedText();
3291                                 # Check for language variants if the template is not found
3292                                 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3293                                         $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3294                                 }
3295                                 # Do recursion depth check
3296                                 $limit = $this->mOptions->getMaxTemplateDepth();
3297                                 if ( $frame->depth >= $limit ) {
3298                                         $found = true;
3299                                         $text = '<span class="error">'
3300                                                 . wfMessage( 'parser-template-recursion-depth-warning' )
3301                                                         ->numParams( $limit )->inContentLanguage()->text()
3302                                                 . '</span>';
3303                                 }
3304                         }
3305                 }
3306
3307                 # Load from database
3308                 if ( !$found && $title ) {
3309                         if ( !Profiler::instance()->isPersistent() ) {
3310                                 # Too many unique items can kill profiling DBs/collectors
3311                                 $titleProfileIn = __METHOD__ . "-title-" . $title->getDBKey();
3312                                 wfProfileIn( $titleProfileIn ); // template in
3313                         }
3314                         wfProfileIn( __METHOD__ . '-loadtpl' );
3315                         if ( !$title->isExternal() ) {
3316                                 if ( $title->isSpecialPage()
3317                                         && $this->mOptions->getAllowSpecialInclusion()
3318                                         && $this->ot['html'] )
3319                                 {
3320                                         // Pass the template arguments as URL parameters.
3321                                         // "uselang" will have no effect since the Language object
3322                                         // is forced to the one defined in ParserOptions.
3323                                         $pageArgs = array();
3324                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3325                                                 $bits = $args->item( $i )->splitArg();
3326                                                 if ( strval( $bits['index'] ) === '' ) {
3327                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3328                                                         $value = trim( $frame->expand( $bits['value'] ) );
3329                                                         $pageArgs[$name] = $value;
3330                                                 }
3331                                         }
3332
3333                                         // Create a new context to execute the special page
3334                                         $context = new RequestContext;
3335                                         $context->setTitle( $title );
3336                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3337                                         $context->setUser( $this->getUser() );
3338                                         $context->setLanguage( $this->mOptions->getUserLangObj() );
3339                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3340                                         if ( $ret ) {
3341                                                 $text = $context->getOutput()->getHTML();
3342                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3343                                                 $found = true;
3344                                                 $isHTML = true;
3345                                                 $this->disableCache();
3346                                         }
3347                                 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3348                                         $found = false; # access denied
3349                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3350                                 } else {
3351                                         list( $text, $title ) = $this->getTemplateDom( $title );
3352                                         if ( $text !== false ) {
3353                                                 $found = true;
3354                                                 $isChildObj = true;
3355                                         }
3356                                 }
3357
3358                                 # If the title is valid but undisplayable, make a link to it
3359                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3360                                         $text = "[[:$titleText]]";
3361                                         $found = true;
3362                                 }
3363                         } elseif ( $title->isTrans() ) {
3364                                 # Interwiki transclusion
3365                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3366                                         $text = $this->interwikiTransclude( $title, 'render' );
3367                                         $isHTML = true;
3368                                 } else {
3369                                         $text = $this->interwikiTransclude( $title, 'raw' );
3370                                         # Preprocess it like a template
3371                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3372                                         $isChildObj = true;
3373                                 }
3374                                 $found = true;
3375                         }
3376
3377                         # Do infinite loop check
3378                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3379                         if ( !$frame->loopCheck( $title ) ) {
3380                                 $found = true;
3381                                 $text = '<span class="error">'
3382                                         . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3383                                         . '</span>';
3384                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3385                         }
3386                         wfProfileOut( __METHOD__ . '-loadtpl' );
3387                 }
3388
3389                 # If we haven't found text to substitute by now, we're done
3390                 # Recover the source wikitext and return it
3391                 if ( !$found ) {
3392                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3393                         if ( $titleProfileIn ) {
3394                                 wfProfileOut( $titleProfileIn ); // template out
3395                         }
3396                         wfProfileOut( __METHOD__ );
3397                         return array( 'object' => $text );
3398                 }
3399
3400                 # Expand DOM-style return values in a child frame
3401                 if ( $isChildObj ) {
3402                         # Clean up argument array
3403                         $newFrame = $frame->newChild( $args, $title );
3404
3405                         if ( $nowiki ) {
3406                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3407                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3408                                 # Expansion is eligible for the empty-frame cache
3409                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3410                                         $text = $this->mTplExpandCache[$titleText];
3411                                 } else {
3412                                         $text = $newFrame->expand( $text );
3413                                         $this->mTplExpandCache[$titleText] = $text;
3414                                 }
3415                         } else {
3416                                 # Uncached expansion
3417                                 $text = $newFrame->expand( $text );
3418                         }
3419                 }
3420                 if ( $isLocalObj && $nowiki ) {
3421                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3422                         $isLocalObj = false;
3423                 }
3424
3425                 if ( $titleProfileIn ) {
3426                         wfProfileOut( $titleProfileIn ); // template out
3427                 }
3428
3429                 # Replace raw HTML by a placeholder
3430                 if ( $isHTML ) {
3431                         $text = $this->insertStripItem( $text );
3432                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3433                         # Escape nowiki-style return values
3434                         $text = wfEscapeWikiText( $text );
3435                 } elseif ( is_string( $text )
3436                         && !$piece['lineStart']
3437                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3438                 {
3439                         # Bug 529: if the template begins with a table or block-level
3440                         # element, it should be treated as beginning a new line.
3441                         # This behaviour is somewhat controversial.
3442                         $text = "\n" . $text;
3443                 }
3444
3445                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3446                         # Error, oversize inclusion
3447                         if ( $titleText !== false ) {
3448                                 # Make a working, properly escaped link if possible (bug 23588)
3449                                 $text = "[[:$titleText]]";
3450                         } else {
3451                                 # This will probably not be a working link, but at least it may
3452                                 # provide some hint of where the problem is
3453                                 preg_replace( '/^:/', '', $originalTitle );
3454                                 $text = "[[:$originalTitle]]";
3455                         }
3456                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3457                         $this->limitationWarn( 'post-expand-template-inclusion' );
3458                 }
3459
3460                 if ( $isLocalObj ) {
3461                         $ret = array( 'object' => $text );
3462                 } else {
3463                         $ret = array( 'text' => $text );
3464                 }
3465
3466                 wfProfileOut( __METHOD__ );
3467                 return $ret;
3468         }
3469
3470         /**
3471          * Get the semi-parsed DOM representation of a template with a given title,
3472          * and its redirect destination title. Cached.
3473          *
3474          * @param $title Title
3475          *
3476          * @return array
3477          */
3478         function getTemplateDom( $title ) {
3479                 $cacheTitle = $title;
3480                 $titleText = $title->getPrefixedDBkey();
3481
3482                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3483                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3484                         $title = Title::makeTitle( $ns, $dbk );
3485                         $titleText = $title->getPrefixedDBkey();
3486                 }
3487                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3488                         return array( $this->mTplDomCache[$titleText], $title );
3489                 }
3490
3491                 # Cache miss, go to the database
3492                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3493
3494                 if ( $text === false ) {
3495                         $this->mTplDomCache[$titleText] = false;
3496                         return array( false, $title );
3497                 }
3498
3499                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3500                 $this->mTplDomCache[ $titleText ] = $dom;
3501
3502                 if ( !$title->equals( $cacheTitle ) ) {
3503                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3504                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3505                 }
3506
3507                 return array( $dom, $title );
3508         }
3509
3510         /**
3511          * Fetch the unparsed text of a template and register a reference to it.
3512          * @param Title $title
3513          * @return Array ( string or false, Title )
3514          */
3515         function fetchTemplateAndTitle( $title ) {
3516                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3517                 $stuff = call_user_func( $templateCb, $title, $this );
3518                 $text = $stuff['text'];
3519                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3520                 if ( isset( $stuff['deps'] ) ) {
3521                         foreach ( $stuff['deps'] as $dep ) {
3522                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3523                         }
3524                 }
3525                 return array( $text, $finalTitle );
3526         }
3527
3528         /**
3529          * Fetch the unparsed text of a template and register a reference to it.
3530          * @param Title $title
3531          * @return mixed string or false
3532          */
3533         function fetchTemplate( $title ) {
3534                 $rv = $this->fetchTemplateAndTitle( $title );
3535                 return $rv[0];
3536         }
3537
3538         /**
3539          * Static function to get a template
3540          * Can be overridden via ParserOptions::setTemplateCallback().
3541          *
3542          * @param $title  Title
3543          * @param $parser Parser
3544          *
3545          * @return array
3546          */
3547         static function statelessFetchTemplate( $title, $parser = false ) {
3548                 $text = $skip = false;
3549                 $finalTitle = $title;
3550                 $deps = array();
3551
3552                 # Loop to fetch the article, with up to 1 redirect
3553                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3554                         # Give extensions a chance to select the revision instead
3555                         $id = false; # Assume current
3556                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3557                                 array( $parser, $title, &$skip, &$id ) );
3558
3559                         if ( $skip ) {
3560                                 $text = false;
3561                                 $deps[] = array(
3562                                         'title'         => $title,
3563                                         'page_id'       => $title->getArticleID(),
3564                                         'rev_id'        => null
3565                                 );
3566                                 break;
3567                         }
3568                         # Get the revision
3569                         $rev = $id
3570                                 ? Revision::newFromId( $id )
3571                                 : Revision::newFromTitle( $title, false, Revision::READ_NORMAL );
3572                         $rev_id = $rev ? $rev->getId() : 0;
3573                         # If there is no current revision, there is no page
3574                         if ( $id === false && !$rev ) {
3575                                 $linkCache = LinkCache::singleton();
3576                                 $linkCache->addBadLinkObj( $title );
3577                         }
3578
3579                         $deps[] = array(
3580                                 'title'         => $title,
3581                                 'page_id'       => $title->getArticleID(),
3582                                 'rev_id'        => $rev_id );
3583                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3584                                 # We fetched a rev from a different title; register it too...
3585                                 $deps[] = array(
3586                                         'title'         => $rev->getTitle(),
3587                                         'page_id'       => $rev->getPage(),
3588                                         'rev_id'        => $rev_id );
3589                         }
3590
3591                         if ( $rev ) {
3592                                 $content = $rev->getContent();
3593                                 $text = $content->getWikitextForTransclusion();
3594
3595                                 if ( $text === false || $text === null ) {
3596                                         $text = false;
3597                                         break;
3598                                 }
3599                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3600                                 global $wgContLang;
3601                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3602                                 if ( !$message->exists() ) {
3603                                         $text = false;
3604                                         break;
3605                                 }
3606                                 $content = $message->content();
3607                                 $text = $message->plain();
3608                         } else {
3609                                 break;
3610                         }
3611                         if ( !$content ) {
3612                                 break;
3613                         }
3614                         # Redirect?
3615                         $finalTitle = $title;
3616                         $title = $content->getRedirectTarget();
3617                 }
3618                 return array(
3619                         'text' => $text,
3620                         'finalTitle' => $finalTitle,
3621                         'deps' => $deps );
3622         }
3623
3624         /**
3625          * Fetch a file and its title and register a reference to it.
3626          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3627          * @param Title $title
3628          * @param Array $options Array of options to RepoGroup::findFile
3629          * @return File|bool
3630          */
3631         function fetchFile( $title, $options = array() ) {
3632                 $res = $this->fetchFileAndTitle( $title, $options );
3633                 return $res[0];
3634         }
3635
3636         /**
3637          * Fetch a file and its title and register a reference to it.
3638          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3639          * @param Title $title
3640          * @param Array $options Array of options to RepoGroup::findFile
3641          * @return Array ( File or false, Title of file )
3642          */
3643         function fetchFileAndTitle( $title, $options = array() ) {
3644                 if ( isset( $options['broken'] ) ) {
3645                         $file = false; // broken thumbnail forced by hook
3646                 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3647                         $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3648                 } else { // get by (name,timestamp)
3649                         $file = wfFindFile( $title, $options );
3650                 }
3651                 $time = $file ? $file->getTimestamp() : false;
3652                 $sha1 = $file ? $file->getSha1() : false;
3653                 # Register the file as a dependency...
3654                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3655                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3656                         # Update fetched file title
3657                         $title = $file->getTitle();
3658                         if ( is_null( $file->getRedirectedTitle() ) ) {
3659                                 # This file was not a redirect, but the title does not match.
3660                                 # Register under the new name because otherwise the link will
3661                                 # get lost.
3662                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3663                         }
3664                 }
3665                 return array( $file, $title );
3666         }
3667
3668         /**
3669          * Transclude an interwiki link.
3670          *
3671          * @param $title Title
3672          * @param $action
3673          *
3674          * @return string
3675          */
3676         function interwikiTransclude( $title, $action ) {
3677                 global $wgEnableScaryTranscluding;
3678
3679                 if ( !$wgEnableScaryTranscluding ) {
3680                         return wfMessage('scarytranscludedisabled')->inContentLanguage()->text();
3681                 }
3682
3683                 $url = $title->getFullUrl( "action=$action" );
3684
3685                 if ( strlen( $url ) > 255 ) {
3686                         return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3687                 }
3688                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3689         }
3690
3691         /**
3692          * @param $url string
3693          * @return Mixed|String
3694          */
3695         function fetchScaryTemplateMaybeFromCache( $url ) {
3696                 global $wgTranscludeCacheExpiry;
3697                 $dbr = wfGetDB( DB_SLAVE );
3698                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3699                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3700                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3701                 if ( $obj ) {
3702                         return $obj->tc_contents;
3703                 }
3704
3705                 $text = Http::get( $url );
3706                 if ( !$text ) {
3707                         return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3708                 }
3709
3710                 $dbw = wfGetDB( DB_MASTER );
3711                 $dbw->replace( 'transcache', array('tc_url'), array(
3712                         'tc_url' => $url,
3713                         'tc_time' => $dbw->timestamp( time() ),
3714                         'tc_contents' => $text)
3715                 );
3716                 return $text;
3717         }
3718
3719         /**
3720          * Triple brace replacement -- used for template arguments
3721          * @private
3722          *
3723          * @param $piece array
3724          * @param $frame PPFrame
3725          *
3726          * @return array
3727          */
3728         function argSubstitution( $piece, $frame ) {
3729                 wfProfileIn( __METHOD__ );
3730
3731                 $error = false;
3732                 $parts = $piece['parts'];
3733                 $nameWithSpaces = $frame->expand( $piece['title'] );
3734                 $argName = trim( $nameWithSpaces );
3735                 $object = false;
3736                 $text = $frame->getArgument( $argName );
3737                 if (  $text === false && $parts->getLength() > 0
3738                   && (
3739                         $this->ot['html']
3740                         || $this->ot['pre']
3741                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3742                   )
3743                 ) {
3744                         # No match in frame, use the supplied default
3745                         $object = $parts->item( 0 )->getChildren();
3746                 }
3747                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3748                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3749                         $this->limitationWarn( 'post-expand-template-argument' );
3750                 }
3751
3752                 if ( $text === false && $object === false ) {
3753                         # No match anywhere
3754                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3755                 }
3756                 if ( $error !== false ) {
3757                         $text .= $error;
3758                 }
3759                 if ( $object !== false ) {
3760                         $ret = array( 'object' => $object );
3761                 } else {
3762                         $ret = array( 'text' => $text );
3763                 }
3764
3765                 wfProfileOut( __METHOD__ );
3766                 return $ret;
3767         }
3768
3769         /**
3770          * Return the text to be used for a given extension tag.
3771          * This is the ghost of strip().
3772          *
3773          * @param $params array Associative array of parameters:
3774          *     name       PPNode for the tag name
3775          *     attr       PPNode for unparsed text where tag attributes are thought to be
3776          *     attributes Optional associative array of parsed attributes
3777          *     inner      Contents of extension element
3778          *     noClose    Original text did not have a close tag
3779          * @param $frame PPFrame
3780          *
3781          * @return string
3782          */
3783         function extensionSubstitution( $params, $frame ) {
3784                 $name = $frame->expand( $params['name'] );
3785                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3786                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3787                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3788
3789                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3790                         ( $this->ot['html'] || $this->ot['pre'] );
3791                 if ( $isFunctionTag ) {
3792                         $markerType = 'none';
3793                 } else {
3794                         $markerType = 'general';
3795                 }
3796                 if ( $this->ot['html'] || $isFunctionTag ) {
3797                         $name = strtolower( $name );
3798                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3799                         if ( isset( $params['attributes'] ) ) {
3800                                 $attributes = $attributes + $params['attributes'];
3801                         }
3802
3803                         if ( isset( $this->mTagHooks[$name] ) ) {
3804                                 # Workaround for PHP bug 35229 and similar
3805                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3806                                         throw new MWException( "Tag hook for $name is not callable\n" );
3807                                 }
3808                                 $output = call_user_func_array( $this->mTagHooks[$name],
3809                                         array( $content, $attributes, $this, $frame ) );
3810                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3811                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3812                                 if ( !is_callable( $callback ) ) {
3813                                         throw new MWException( "Tag hook for $name is not callable\n" );
3814                                 }
3815
3816                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3817                         } else {
3818                                 $output = '<span class="error">Invalid tag extension name: ' .
3819                                         htmlspecialchars( $name ) . '</span>';
3820                         }
3821
3822                         if ( is_array( $output ) ) {
3823                                 # Extract flags to local scope (to override $markerType)
3824                                 $flags = $output;
3825                                 $output = $flags[0];
3826                                 unset( $flags[0] );
3827                                 extract( $flags );
3828                         }
3829                 } else {
3830                         if ( is_null( $attrText ) ) {
3831                                 $attrText = '';
3832                         }
3833                         if ( isset( $params['attributes'] ) ) {
3834                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3835                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3836                                                 htmlspecialchars( $attrValue ) . '"';
3837                                 }
3838                         }
3839                         if ( $content === null ) {
3840                                 $output = "<$name$attrText/>";
3841                         } else {
3842                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3843                                 $output = "<$name$attrText>$content$close";
3844                         }
3845                 }
3846
3847                 if ( $markerType === 'none' ) {
3848                         return $output;
3849                 } elseif ( $markerType === 'nowiki' ) {
3850                         $this->mStripState->addNoWiki( $marker, $output );
3851                 } elseif ( $markerType === 'general' ) {
3852                         $this->mStripState->addGeneral( $marker, $output );
3853                 } else {
3854                         throw new MWException( __METHOD__.': invalid marker type' );
3855                 }
3856                 return $marker;
3857         }
3858
3859         /**
3860          * Increment an include size counter
3861          *
3862          * @param $type String: the type of expansion
3863          * @param $size Integer: the size of the text
3864          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3865          */
3866         function incrementIncludeSize( $type, $size ) {
3867                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3868                         return false;
3869                 } else {
3870                         $this->mIncludeSizes[$type] += $size;
3871                         return true;
3872                 }
3873         }
3874
3875         /**
3876          * Increment the expensive function count
3877          *
3878          * @return Boolean: false if the limit has been exceeded
3879          */
3880         function incrementExpensiveFunctionCount() {
3881                 $this->mExpensiveFunctionCount++;
3882                 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3883         }
3884
3885         /**
3886          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3887          * Fills $this->mDoubleUnderscores, returns the modified text
3888          *
3889          * @param $text string
3890          *
3891          * @return string
3892          */
3893         function doDoubleUnderscore( $text ) {
3894                 wfProfileIn( __METHOD__ );
3895
3896                 # The position of __TOC__ needs to be recorded
3897                 $mw = MagicWord::get( 'toc' );
3898                 if ( $mw->match( $text ) ) {
3899                         $this->mShowToc = true;
3900                         $this->mForceTocPosition = true;
3901
3902                         # Set a placeholder. At the end we'll fill it in with the TOC.
3903                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3904
3905                         # Only keep the first one.
3906                         $text = $mw->replace( '', $text );
3907                 }
3908
3909                 # Now match and remove the rest of them
3910                 $mwa = MagicWord::getDoubleUnderscoreArray();
3911                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3912
3913                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3914                         $this->mOutput->mNoGallery = true;
3915                 }
3916                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3917                         $this->mShowToc = false;
3918                 }
3919                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3920                         $this->addTrackingCategory( 'hidden-category-category' );
3921                 }
3922                 # (bug 8068) Allow control over whether robots index a page.
3923                 #
3924                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3925                 # is not desirable, the last one on the page should win.
3926                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3927                         $this->mOutput->setIndexPolicy( 'noindex' );
3928                         $this->addTrackingCategory( 'noindex-category' );
3929                 }
3930                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3931                         $this->mOutput->setIndexPolicy( 'index' );
3932                         $this->addTrackingCategory( 'index-category' );
3933                 }
3934
3935                 # Cache all double underscores in the database
3936                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3937                         $this->mOutput->setProperty( $key, '' );
3938                 }
3939
3940                 wfProfileOut( __METHOD__ );
3941                 return $text;
3942         }
3943
3944         /**
3945          * Add a tracking category, getting the title from a system message,
3946          * or print a debug message if the title is invalid.
3947          *
3948          * @param $msg String: message key
3949          * @return Boolean: whether the addition was successful
3950          */
3951         public function addTrackingCategory( $msg ) {
3952                 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
3953                         wfDebug( __METHOD__.": Not adding tracking category $msg to special page!\n" );
3954                         return false;
3955                 }
3956                 // Important to parse with correct title (bug 31469)
3957                 $cat = wfMessage( $msg )
3958                         ->title( $this->getTitle() )
3959                         ->inContentLanguage()
3960                         ->text();
3961
3962                 # Allow tracking categories to be disabled by setting them to "-"
3963                 if ( $cat === '-' ) {
3964                         return false;
3965                 }
3966
3967                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3968                 if ( $containerCategory ) {
3969                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3970                         return true;
3971                 } else {
3972                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3973                         return false;
3974                 }
3975         }
3976
3977         /**
3978          * This function accomplishes several tasks:
3979          * 1) Auto-number headings if that option is enabled
3980          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3981          * 3) Add a Table of contents on the top for users who have enabled the option
3982          * 4) Auto-anchor headings
3983          *
3984          * It loops through all headlines, collects the necessary data, then splits up the
3985          * string and re-inserts the newly formatted headlines.
3986          *
3987          * @param $text String
3988          * @param $origText String: original, untouched wikitext
3989          * @param $isMain Boolean
3990          * @return mixed|string
3991          * @private
3992          */
3993         function formatHeadings( $text, $origText, $isMain=true ) {
3994                 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3995
3996                 # Inhibit editsection links if requested in the page
3997                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3998                         $maybeShowEditLink = $showEditLink = false;
3999                 } else {
4000                         $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4001                         $showEditLink = $this->mOptions->getEditSection();
4002                 }
4003                 if ( $showEditLink ) {
4004                         $this->mOutput->setEditSectionTokens( true );
4005                 }
4006
4007                 # Get all headlines for numbering them and adding funky stuff like [edit]
4008                 # links - this is for later, but we need the number of headlines right now
4009                 $matches = array();
4010                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
4011
4012                 # if there are fewer than 4 headlines in the article, do not show TOC
4013                 # unless it's been explicitly enabled.
4014                 $enoughToc = $this->mShowToc &&
4015                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4016
4017                 # Allow user to stipulate that a page should have a "new section"
4018                 # link added via __NEWSECTIONLINK__
4019                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4020                         $this->mOutput->setNewSection( true );
4021                 }
4022
4023                 # Allow user to remove the "new section"
4024                 # link via __NONEWSECTIONLINK__
4025                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4026                         $this->mOutput->hideNewSection( true );
4027                 }
4028
4029                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4030                 # override above conditions and always show TOC above first header
4031                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4032                         $this->mShowToc = true;
4033                         $enoughToc = true;
4034                 }
4035
4036                 # headline counter
4037                 $headlineCount = 0;
4038                 $numVisible = 0;
4039
4040                 # Ugh .. the TOC should have neat indentation levels which can be
4041                 # passed to the skin functions. These are determined here
4042                 $toc = '';
4043                 $full = '';
4044                 $head = array();
4045                 $sublevelCount = array();
4046                 $levelCount = array();
4047                 $level = 0;
4048                 $prevlevel = 0;
4049                 $toclevel = 0;
4050                 $prevtoclevel = 0;
4051                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4052                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4053                 $oldType = $this->mOutputType;
4054                 $this->setOutputType( self::OT_WIKI );
4055                 $frame = $this->getPreprocessor()->newFrame();
4056                 $root = $this->preprocessToDom( $origText );
4057                 $node = $root->getFirstChild();
4058                 $byteOffset = 0;
4059                 $tocraw = array();
4060                 $refers = array();
4061
4062                 foreach ( $matches[3] as $headline ) {
4063                         $isTemplate = false;
4064                         $titleText = false;
4065                         $sectionIndex = false;
4066                         $numbering = '';
4067                         $markerMatches = array();
4068                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
4069                                 $serial = $markerMatches[1];
4070                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4071                                 $isTemplate = ( $titleText != $baseTitleText );
4072                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
4073                         }
4074
4075                         if ( $toclevel ) {
4076                                 $prevlevel = $level;
4077                         }
4078                         $level = $matches[1][$headlineCount];
4079
4080                         if ( $level > $prevlevel ) {
4081                                 # Increase TOC level
4082                                 $toclevel++;
4083                                 $sublevelCount[$toclevel] = 0;
4084                                 if ( $toclevel<$wgMaxTocLevel ) {
4085                                         $prevtoclevel = $toclevel;
4086                                         $toc .= Linker::tocIndent();
4087                                         $numVisible++;
4088                                 }
4089                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4090                                 # Decrease TOC level, find level to jump to
4091
4092                                 for ( $i = $toclevel; $i > 0; $i-- ) {
4093                                         if ( $levelCount[$i] == $level ) {
4094                                                 # Found last matching level
4095                                                 $toclevel = $i;
4096                                                 break;
4097                                         } elseif ( $levelCount[$i] < $level ) {
4098                                                 # Found first matching level below current level
4099                                                 $toclevel = $i + 1;
4100                                                 break;
4101                                         }
4102                                 }
4103                                 if ( $i == 0 ) {
4104                                         $toclevel = 1;
4105                                 }
4106                                 if ( $toclevel<$wgMaxTocLevel ) {
4107                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
4108                                                 # Unindent only if the previous toc level was shown :p
4109                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4110                                                 $prevtoclevel = $toclevel;
4111                                         } else {
4112                                                 $toc .= Linker::tocLineEnd();
4113                                         }
4114                                 }
4115                         } else {
4116                                 # No change in level, end TOC line
4117                                 if ( $toclevel<$wgMaxTocLevel ) {
4118                                         $toc .= Linker::tocLineEnd();
4119                                 }
4120                         }
4121
4122                         $levelCount[$toclevel] = $level;
4123
4124                         # count number of headlines for each level
4125                         @$sublevelCount[$toclevel]++;
4126                         $dot = 0;
4127                         for( $i = 1; $i <= $toclevel; $i++ ) {
4128                                 if ( !empty( $sublevelCount[$i] ) ) {
4129                                         if ( $dot ) {
4130                                                 $numbering .= '.';
4131                                         }
4132                                         $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4133                                         $dot = 1;
4134                                 }
4135                         }
4136
4137                         # The safe header is a version of the header text safe to use for links
4138
4139                         # Remove link placeholders by the link text.
4140                         #     <!--LINK number-->
4141                         # turns into
4142                         #     link text with suffix
4143                         # Do this before unstrip since link text can contain strip markers
4144                         $safeHeadline = $this->replaceLinkHoldersText( $headline );
4145
4146                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4147                         $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4148
4149                         # Strip out HTML (first regex removes any tag not allowed)
4150                         # Allowed tags are <sup> and <sub> (bug 8393), <i> (bug 26375) and <b> (r105284)
4151                         # We strip any parameter from accepted tags (second regex)
4152                         $tocline = preg_replace(
4153                                 array( '#<(?!/?(sup|sub|i|b)(?: [^>]*)?>).*?'.'>#', '#<(/?(sup|sub|i|b))(?: .*?)?'.'>#' ),
4154                                 array( '',                          '<$1>' ),
4155                                 $safeHeadline
4156                         );
4157                         $tocline = trim( $tocline );
4158
4159                         # For the anchor, strip out HTML-y stuff period
4160                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4161                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4162
4163                         # Save headline for section edit hint before it's escaped
4164                         $headlineHint = $safeHeadline;
4165
4166                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4167                                 # For reverse compatibility, provide an id that's
4168                                 # HTML4-compatible, like we used to.
4169                                 #
4170                                 # It may be worth noting, academically, that it's possible for
4171                                 # the legacy anchor to conflict with a non-legacy headline
4172                                 # anchor on the page.  In this case likely the "correct" thing
4173                                 # would be to either drop the legacy anchors or make sure
4174                                 # they're numbered first.  However, this would require people
4175                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4176                                 # manually, so let's not bother worrying about it.
4177                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4178                                         array( 'noninitial', 'legacy' ) );
4179                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4180
4181                                 if ( $legacyHeadline == $safeHeadline ) {
4182                                         # No reason to have both (in fact, we can't)
4183                                         $legacyHeadline = false;
4184                                 }
4185                         } else {
4186                                 $legacyHeadline = false;
4187                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4188                                         'noninitial' );
4189                         }
4190
4191                         # HTML names must be case-insensitively unique (bug 10721).
4192                         # This does not apply to Unicode characters per
4193                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4194                         # @todo FIXME: We may be changing them depending on the current locale.
4195                         $arrayKey = strtolower( $safeHeadline );
4196                         if ( $legacyHeadline === false ) {
4197                                 $legacyArrayKey = false;
4198                         } else {
4199                                 $legacyArrayKey = strtolower( $legacyHeadline );
4200                         }
4201
4202                         # count how many in assoc. array so we can track dupes in anchors
4203                         if ( isset( $refers[$arrayKey] ) ) {
4204                                 $refers[$arrayKey]++;
4205                         } else {
4206                                 $refers[$arrayKey] = 1;
4207                         }
4208                         if ( isset( $refers[$legacyArrayKey] ) ) {
4209                                 $refers[$legacyArrayKey]++;
4210                         } else {
4211                                 $refers[$legacyArrayKey] = 1;
4212                         }
4213
4214                         # Don't number the heading if it is the only one (looks silly)
4215                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4216                                 # the two are different if the line contains a link
4217                                 $headline = Html::element( 'span', array( 'class' => 'mw-headline-number' ), $numbering ) . ' ' . $headline;
4218                         }
4219
4220                         # Create the anchor for linking from the TOC to the section
4221                         $anchor = $safeHeadline;
4222                         $legacyAnchor = $legacyHeadline;
4223                         if ( $refers[$arrayKey] > 1 ) {
4224                                 $anchor .= '_' . $refers[$arrayKey];
4225                         }
4226                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4227                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4228                         }
4229                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4230                                 $toc .= Linker::tocLine( $anchor, $tocline,
4231                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4232                         }
4233
4234                         # Add the section to the section tree
4235                         # Find the DOM node for this header
4236                         while ( $node && !$isTemplate ) {
4237                                 if ( $node->getName() === 'h' ) {
4238                                         $bits = $node->splitHeading();
4239                                         if ( $bits['i'] == $sectionIndex ) {
4240                                                 break;
4241                                         }
4242                                 }
4243                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4244                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4245                                 $node = $node->getNextSibling();
4246                         }
4247                         $tocraw[] = array(
4248                                 'toclevel' => $toclevel,
4249                                 'level' => $level,
4250                                 'line' => $tocline,
4251                                 'number' => $numbering,
4252                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4253                                 'fromtitle' => $titleText,
4254                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4255                                 'anchor' => $anchor,
4256                         );
4257
4258                         # give headline the correct <h#> tag
4259                         if ( $maybeShowEditLink && $sectionIndex !== false ) {
4260                                 // Output edit section links as markers with styles that can be customized by skins
4261                                 if ( $isTemplate ) {
4262                                         # Put a T flag in the section identifier, to indicate to extractSections()
4263                                         # that sections inside <includeonly> should be counted.
4264                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4265                                 } else {
4266                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4267                                 }
4268                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4269                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4270                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4271                                 // so we don't have to worry about a user trying to input one of these markers directly.
4272                                 // We use a page and section attribute to stop the language converter from converting these important bits
4273                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4274                                 // be able to convert that piece of data.
4275                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4276                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4277                                 if ( isset($editlinkArgs[2]) ) {
4278                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4279                                 } else {
4280                                         $editlink .= '/>';
4281                                 }
4282                         } else {
4283                                 $editlink = '';
4284                         }
4285                         $head[$headlineCount] = Linker::makeHeadline( $level,
4286                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4287                                 $editlink, $legacyAnchor );
4288
4289                         $headlineCount++;
4290                 }
4291
4292                 $this->setOutputType( $oldType );
4293
4294                 # Never ever show TOC if no headers
4295                 if ( $numVisible < 1 ) {
4296                         $enoughToc = false;
4297                 }
4298
4299                 if ( $enoughToc ) {
4300                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4301                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4302                         }
4303                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4304                         $this->mOutput->setTOCHTML( $toc );
4305                 }
4306
4307                 if ( $isMain ) {
4308                         $this->mOutput->setSections( $tocraw );
4309                 }
4310
4311                 # split up and insert constructed headlines
4312                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4313                 $i = 0;
4314
4315                 // build an array of document sections
4316                 $sections = array();
4317                 foreach ( $blocks as $block ) {
4318                         // $head is zero-based, sections aren't.
4319                         if ( empty( $head[$i - 1] ) ) {
4320                                 $sections[$i] = $block;
4321                         } else {
4322                                 $sections[$i] = $head[$i - 1] . $block;
4323                         }
4324
4325                         /**
4326                          * Send a hook, one per section.
4327                          * The idea here is to be able to make section-level DIVs, but to do so in a
4328                          * lower-impact, more correct way than r50769
4329                          *
4330                          * $this : caller
4331                          * $section : the section number
4332                          * &$sectionContent : ref to the content of the section
4333                          * $showEditLinks : boolean describing whether this section has an edit link
4334                          */
4335                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4336
4337                         $i++;
4338                 }
4339
4340                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4341                         // append the TOC at the beginning
4342                         // Top anchor now in skin
4343                         $sections[0] = $sections[0] . $toc . "\n";
4344                 }
4345
4346                 $full .= join( '', $sections );
4347
4348                 if ( $this->mForceTocPosition ) {
4349                         return str_replace( '<!--MWTOC-->', $toc, $full );
4350                 } else {
4351                         return $full;
4352                 }
4353         }
4354
4355         /**
4356          * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4357          * conversion, substitting signatures, {{subst:}} templates, etc.
4358          *
4359          * @param $text String: the text to transform
4360          * @param $title Title: the Title object for the current article
4361          * @param $user User: the User object describing the current user
4362          * @param $options ParserOptions: parsing options
4363          * @param $clearState Boolean: whether to clear the parser state first
4364          * @return String: the altered wiki markup
4365          */
4366         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4367                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4368                 $this->setUser( $user );
4369
4370                 $pairs = array(
4371                         "\r\n" => "\n",
4372                 );
4373                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4374                 if( $options->getPreSaveTransform() ) {
4375                         $text = $this->pstPass2( $text, $user );
4376                 }
4377                 $text = $this->mStripState->unstripBoth( $text );
4378
4379                 $this->setUser( null ); #Reset
4380
4381                 return $text;
4382         }
4383
4384         /**
4385          * Pre-save transform helper function
4386          * @private
4387          *
4388          * @param $text string
4389          * @param $user User
4390          *
4391          * @return string
4392          */
4393         function pstPass2( $text, $user ) {
4394                 global $wgContLang, $wgLocaltimezone;
4395
4396                 # Note: This is the timestamp saved as hardcoded wikitext to
4397                 # the database, we use $wgContLang here in order to give
4398                 # everyone the same signature and use the default one rather
4399                 # than the one selected in each user's preferences.
4400                 # (see also bug 12815)
4401                 $ts = $this->mOptions->getTimestamp();
4402                 if ( isset( $wgLocaltimezone ) ) {
4403                         $tz = $wgLocaltimezone;
4404                 } else {
4405                         $tz = date_default_timezone_get();
4406                 }
4407
4408                 $unixts = wfTimestamp( TS_UNIX, $ts );
4409                 $oldtz = date_default_timezone_get();
4410                 date_default_timezone_set( $tz );
4411                 $ts = date( 'YmdHis', $unixts );
4412                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4413
4414                 # Allow translation of timezones through wiki. date() can return
4415                 # whatever crap the system uses, localised or not, so we cannot
4416                 # ship premade translations.
4417                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4418                 $msg = wfMessage( $key )->inContentLanguage();
4419                 if ( $msg->exists() ) {
4420                         $tzMsg = $msg->text();
4421                 }
4422
4423                 date_default_timezone_set( $oldtz );
4424
4425                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4426
4427                 # Variable replacement
4428                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4429                 $text = $this->replaceVariables( $text );
4430
4431                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4432                 # which may corrupt this parser instance via its wfMessage()->text() call-
4433
4434                 # Signatures
4435                 $sigText = $this->getUserSig( $user );
4436                 $text = strtr( $text, array(
4437                         '~~~~~' => $d,
4438                         '~~~~' => "$sigText $d",
4439                         '~~~' => $sigText
4440                 ) );
4441
4442                 # Context links: [[|name]] and [[name (context)|]]
4443                 $tc = '[' . Title::legalChars() . ']';
4444                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4445
4446                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";           # [[ns:page (context)|]]
4447                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";           # [[ns:page（context）|]]
4448                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/"; # [[ns:page (context), context|]]
4449                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4450
4451                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4452                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4453                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4454                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4455
4456                 $t = $this->mTitle->getText();
4457                 $m = array();
4458                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4459                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4460                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4461                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4462                 } else {
4463                         # if there's no context, don't bother duplicating the title
4464                         $text = preg_replace( $p2, '[[\\1]]', $text );
4465                 }
4466
4467                 # Trim trailing whitespace
4468                 $text = rtrim( $text );
4469
4470                 return $text;
4471         }
4472
4473         /**
4474          * Fetch the user's signature text, if any, and normalize to
4475          * validated, ready-to-insert wikitext.
4476          * If you have pre-fetched the nickname or the fancySig option, you can
4477          * specify them here to save a database query.
4478          * Do not reuse this parser instance after calling getUserSig(),
4479          * as it may have changed if it's the $wgParser.
4480          *
4481          * @param $user User
4482          * @param $nickname String|bool nickname to use or false to use user's default nickname
4483          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4484          *                  or null to use default value
4485          * @return string
4486          */
4487         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4488                 global $wgMaxSigChars;
4489
4490                 $username = $user->getName();
4491
4492                 # If not given, retrieve from the user object.
4493                 if ( $nickname === false )
4494                         $nickname = $user->getOption( 'nickname' );
4495
4496                 if ( is_null( $fancySig ) ) {
4497                         $fancySig = $user->getBoolOption( 'fancysig' );
4498                 }
4499
4500                 $nickname = $nickname == null ? $username : $nickname;
4501
4502                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4503                         $nickname = $username;
4504                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4505                 } elseif ( $fancySig !== false ) {
4506                         # Sig. might contain markup; validate this
4507                         if ( $this->validateSig( $nickname ) !== false ) {
4508                                 # Validated; clean up (if needed) and return it
4509                                 return $this->cleanSig( $nickname, true );
4510                         } else {
4511                                 # Failed to validate; fall back to the default
4512                                 $nickname = $username;
4513                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4514                         }
4515                 }
4516
4517                 # Make sure nickname doesnt get a sig in a sig
4518                 $nickname = self::cleanSigInSig( $nickname );
4519
4520                 # If we're still here, make it a link to the user page
4521                 $userText = wfEscapeWikiText( $username );
4522                 $nickText = wfEscapeWikiText( $nickname );
4523                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4524
4525                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4526         }
4527
4528         /**
4529          * Check that the user's signature contains no bad XML
4530          *
4531          * @param $text String
4532          * @return mixed An expanded string, or false if invalid.
4533          */
4534         function validateSig( $text ) {
4535                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4536         }
4537
4538         /**
4539          * Clean up signature text
4540          *
4541          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4542          * 2) Substitute all transclusions
4543          *
4544          * @param $text String
4545          * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4546          * @return String: signature text
4547          */
4548         public function cleanSig( $text, $parsing = false ) {
4549                 if ( !$parsing ) {
4550                         global $wgTitle;
4551                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4552                 }
4553
4554                 # Option to disable this feature
4555                 if ( !$this->mOptions->getCleanSignatures() ) {
4556                         return $text;
4557                 }
4558
4559                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4560                 #  => Move this logic to braceSubstitution()
4561                 $substWord = MagicWord::get( 'subst' );
4562                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4563                 $substText = '{{' . $substWord->getSynonym( 0 );
4564
4565                 $text = preg_replace( $substRegex, $substText, $text );
4566                 $text = self::cleanSigInSig( $text );
4567                 $dom = $this->preprocessToDom( $text );
4568                 $frame = $this->getPreprocessor()->newFrame();
4569                 $text = $frame->expand( $dom );
4570
4571                 if ( !$parsing ) {
4572                         $text = $this->mStripState->unstripBoth( $text );
4573                 }
4574
4575                 return $text;
4576         }
4577
4578         /**
4579          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4580          *
4581          * @param $text String
4582          * @return String: signature text with /~{3,5}/ removed
4583          */
4584         public static function cleanSigInSig( $text ) {
4585                 $text = preg_replace( '/~{3,5}/', '', $text );
4586                 return $text;
4587         }
4588
4589         /**
4590          * Set up some variables which are usually set up in parse()
4591          * so that an external function can call some class members with confidence
4592          *
4593          * @param $title Title|null
4594          * @param $options ParserOptions
4595          * @param $outputType
4596          * @param $clearState bool
4597          */
4598         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4599                 $this->startParse( $title, $options, $outputType, $clearState );
4600         }
4601
4602         /**
4603          * @param $title Title|null
4604          * @param $options ParserOptions
4605          * @param $outputType
4606          * @param $clearState bool
4607          */
4608         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4609                 $this->setTitle( $title );
4610                 $this->mOptions = $options;
4611                 $this->setOutputType( $outputType );
4612                 if ( $clearState ) {
4613                         $this->clearState();
4614                 }
4615         }
4616
4617         /**
4618          * Wrapper for preprocess()
4619          *
4620          * @param $text String: the text to preprocess
4621          * @param $options ParserOptions: options
4622          * @param $title Title object or null to use $wgTitle
4623          * @return String
4624          */
4625         public function transformMsg( $text, $options, $title = null ) {
4626                 static $executing = false;
4627
4628                 # Guard against infinite recursion
4629                 if ( $executing ) {
4630                         return $text;
4631                 }
4632                 $executing = true;
4633
4634                 wfProfileIn( __METHOD__ );
4635                 if ( !$title ) {
4636                         global $wgTitle;
4637                         $title = $wgTitle;
4638                 }
4639                 if ( !$title ) {
4640                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4641                         # Create a ghost title in such case
4642                         $title = Title::newFromText( 'Dwimmerlaik' );
4643                 }
4644                 $text = $this->preprocess( $text, $title, $options );
4645
4646                 $executing = false;
4647                 wfProfileOut( __METHOD__ );
4648                 return $text;
4649         }
4650
4651         /**
4652          * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4653          * The callback should have the following form:
4654          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4655          *
4656          * Transform and return $text. Use $parser for any required context, e.g. use
4657          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4658          *
4659          * Hooks may return extended information by returning an array, of which the
4660          * first numbered element (index 0) must be the return string, and all other
4661          * entries are extracted into local variables within an internal function
4662          * in the Parser class.
4663          *
4664          * This interface (introduced r61913) appears to be undocumented, but
4665          * 'markerName' is used by some core tag hooks to override which strip
4666          * array their results are placed in. **Use great caution if attempting
4667          * this interface, as it is not documented and injudicious use could smash
4668          * private variables.**
4669          *
4670          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4671          * @param $callback Mixed: the callback function (and object) to use for the tag
4672          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4673          */
4674         public function setHook( $tag, $callback ) {
4675                 $tag = strtolower( $tag );
4676                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4677                         throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4678                 }
4679                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4680                 $this->mTagHooks[$tag] = $callback;
4681                 if ( !in_array( $tag, $this->mStripList ) ) {
4682                         $this->mStripList[] = $tag;
4683                 }
4684
4685                 return $oldVal;
4686         }
4687
4688         /**
4689          * As setHook(), but letting the contents be parsed.
4690          *
4691          * Transparent tag hooks are like regular XML-style tag hooks, except they
4692          * operate late in the transformation sequence, on HTML instead of wikitext.
4693          *
4694          * This is probably obsoleted by things dealing with parser frames?
4695          * The only extension currently using it is geoserver.
4696          *
4697          * @since 1.10
4698          * @todo better document or deprecate this
4699          *
4700          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4701          * @param $callback Mixed: the callback function (and object) to use for the tag
4702          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4703          */
4704         function setTransparentTagHook( $tag, $callback ) {
4705                 $tag = strtolower( $tag );
4706                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4707                         throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4708                 }
4709                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4710                 $this->mTransparentTagHooks[$tag] = $callback;
4711
4712                 return $oldVal;
4713         }
4714
4715         /**
4716          * Remove all tag hooks
4717          */
4718         function clearTagHooks() {
4719                 $this->mTagHooks = array();
4720                 $this->mFunctionTagHooks = array();
4721                 $this->mStripList = $this->mDefaultStripList;
4722         }
4723
4724         /**
4725          * Create a function, e.g. {{sum:1|2|3}}
4726          * The callback function should have the form:
4727          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4728          *
4729          * Or with SFH_OBJECT_ARGS:
4730          *    function myParserFunction( $parser, $frame, $args ) { ... }
4731          *
4732          * The callback may either return the text result of the function, or an array with the text
4733          * in element 0, and a number of flags in the other elements. The names of the flags are
4734          * specified in the keys. Valid flags are:
4735          *   found                     The text returned is valid, stop processing the template. This
4736          *                             is on by default.
4737          *   nowiki                    Wiki markup in the return value should be escaped
4738          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4739          *
4740          * @param $id String: The magic word ID
4741          * @param $callback Mixed: the callback function (and object) to use
4742          * @param $flags Integer: a combination of the following flags:
4743          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4744          *
4745          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4746          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4747          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4748          *     the arguments, and to control the way they are expanded.
4749          *
4750          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4751          *     arguments, for instance:
4752          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4753          *
4754          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4755          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4756          *     working if/when this is changed.
4757          *
4758          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4759          *     expansion.
4760          *
4761          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4762          *     about the methods available in PPFrame and PPNode.
4763          *
4764          * @return string|callback The old callback function for this name, if any
4765          */
4766         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4767                 global $wgContLang;
4768
4769                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4770                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4771
4772                 # Add to function cache
4773                 $mw = MagicWord::get( $id );
4774                 if ( !$mw )
4775                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4776
4777                 $synonyms = $mw->getSynonyms();
4778                 $sensitive = intval( $mw->isCaseSensitive() );
4779
4780                 foreach ( $synonyms as $syn ) {
4781                         # Case
4782                         if ( !$sensitive ) {
4783                                 $syn = $wgContLang->lc( $syn );
4784                         }
4785                         # Add leading hash
4786                         if ( !( $flags & SFH_NO_HASH ) ) {
4787                                 $syn = '#' . $syn;
4788                         }
4789                         # Remove trailing colon
4790                         if ( substr( $syn, -1, 1 ) === ':' ) {
4791                                 $syn = substr( $syn, 0, -1 );
4792                         }
4793                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4794                 }
4795                 return $oldVal;
4796         }
4797
4798         /**
4799          * Get all registered function hook identifiers
4800          *
4801          * @return Array
4802          */
4803         function getFunctionHooks() {
4804                 return array_keys( $this->mFunctionHooks );
4805         }
4806
4807         /**
4808          * Create a tag function, e.g. "<test>some stuff</test>".
4809          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4810          * Unlike parser functions, their content is not preprocessed.
4811          * @return null
4812          */
4813         function setFunctionTagHook( $tag, $callback, $flags ) {
4814                 $tag = strtolower( $tag );
4815                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4816                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4817                         $this->mFunctionTagHooks[$tag] : null;
4818                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4819
4820                 if ( !in_array( $tag, $this->mStripList ) ) {
4821                         $this->mStripList[] = $tag;
4822                 }
4823
4824                 return $old;
4825         }
4826
4827         /**
4828          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4829          * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4830          * Placeholders created in Skin::makeLinkObj()
4831          *
4832          * @param $text string
4833          * @param $options int
4834          *
4835          * @return array of link CSS classes, indexed by PDBK.
4836          */
4837         function replaceLinkHolders( &$text, $options = 0 ) {
4838                 return $this->mLinkHolders->replace( $text );
4839         }
4840
4841         /**
4842          * Replace "<!--LINK-->" link placeholders with plain text of links
4843          * (not HTML-formatted).
4844          *
4845          * @param $text String
4846          * @return String
4847          */
4848         function replaceLinkHoldersText( $text ) {
4849                 return $this->mLinkHolders->replaceText( $text );
4850         }
4851
4852         /**
4853          * Renders an image gallery from a text with one line per image.
4854          * text labels may be given by using |-style alternative text. E.g.
4855          *   Image:one.jpg|The number "1"
4856          *   Image:tree.jpg|A tree
4857          * given as text will return the HTML of a gallery with two images,
4858          * labeled 'The number "1"' and
4859          * 'A tree'.
4860          *
4861          * @param string $text
4862          * @param array $params
4863          * @return string HTML
4864          */
4865         function renderImageGallery( $text, $params ) {
4866                 $ig = new ImageGallery();
4867                 $ig->setContextTitle( $this->mTitle );
4868                 $ig->setShowBytes( false );
4869                 $ig->setShowFilename( false );
4870                 $ig->setParser( $this );
4871                 $ig->setHideBadImages();
4872                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4873
4874                 if ( isset( $params['showfilename'] ) ) {
4875                         $ig->setShowFilename( true );
4876                 } else {
4877                         $ig->setShowFilename( false );
4878                 }
4879                 if ( isset( $params['caption'] ) ) {
4880                         $caption = $params['caption'];
4881                         $caption = htmlspecialchars( $caption );
4882                         $caption = $this->replaceInternalLinks( $caption );
4883                         $ig->setCaptionHtml( $caption );
4884                 }
4885                 if ( isset( $params['perrow'] ) ) {
4886                         $ig->setPerRow( $params['perrow'] );
4887                 }
4888                 if ( isset( $params['widths'] ) ) {
4889                         $ig->setWidths( $params['widths'] );
4890                 }
4891                 if ( isset( $params['heights'] ) ) {
4892                         $ig->setHeights( $params['heights'] );
4893                 }
4894
4895                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4896
4897                 $lines = StringUtils::explode( "\n", $text );
4898                 foreach ( $lines as $line ) {
4899                         # match lines like these:
4900                         # Image:someimage.jpg|This is some image
4901                         $matches = array();
4902                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4903                         # Skip empty lines
4904                         if ( count( $matches ) == 0 ) {
4905                                 continue;
4906                         }
4907
4908                         if ( strpos( $matches[0], '%' ) !== false ) {
4909                                 $matches[1] = rawurldecode( $matches[1] );
4910                         }
4911                         $title = Title::newFromText( $matches[1], NS_FILE );
4912                         if ( is_null( $title ) ) {
4913                                 # Bogus title. Ignore these so we don't bomb out later.
4914                                 continue;
4915                         }
4916
4917                         $label = '';
4918                         $alt = '';
4919                         $link = '';
4920                         if ( isset( $matches[3] ) ) {
4921                                 // look for an |alt= definition while trying not to break existing
4922                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4923                                 // is defined for images in galleries
4924
4925                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4926                                 $parameterMatches = StringUtils::explode('|', $matches[3]);
4927                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4928                                 $magicWordLink = MagicWord::get( 'img_link' );
4929
4930                                 foreach ( $parameterMatches as $parameterMatch ) {
4931                                         if ( $match = $magicWordAlt->matchVariableStartToEnd( $parameterMatch ) ) {
4932                                                 $alt = $this->stripAltText( $match, false );
4933                                         }
4934                                         elseif( $match = $magicWordLink->matchVariableStartToEnd( $parameterMatch ) ){
4935                                                 $link = strip_tags($this->replaceLinkHoldersText($match));
4936                                                 $chars = self::EXT_LINK_URL_CLASS;
4937                                                 $prots = $this->mUrlProtocols;
4938                                                 //check to see if link matches an absolute url, if not then it must be a wiki link.
4939                                                 if(!preg_match( "/^($prots)$chars+$/u", $link)){
4940                                                         $localLinkTitle = Title::newFromText($link);
4941                                                         $link = $localLinkTitle->getLocalURL();
4942                                                 }
4943                                         }
4944                                         else {
4945                                                 // concatenate all other pipes
4946                                                 $label .= '|' . $parameterMatch;
4947                                         }
4948                                 }
4949                                 // remove the first pipe
4950                                 $label = substr( $label, 1 );
4951                         }
4952
4953                         $ig->add( $title, $label, $alt ,$link);
4954                 }
4955                 return $ig->toHTML();
4956         }
4957
4958         /**
4959          * @param $handler
4960          * @return array
4961          */
4962         function getImageParams( $handler ) {
4963                 if ( $handler ) {
4964                         $handlerClass = get_class( $handler );
4965                 } else {
4966                         $handlerClass = '';
4967                 }
4968                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4969                         # Initialise static lists
4970                         static $internalParamNames = array(
4971                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4972                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4973                                         'bottom', 'text-bottom' ),
4974                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4975                                         'upright', 'border', 'link', 'alt', 'class' ),
4976                         );
4977                         static $internalParamMap;
4978                         if ( !$internalParamMap ) {
4979                                 $internalParamMap = array();
4980                                 foreach ( $internalParamNames as $type => $names ) {
4981                                         foreach ( $names as $name ) {
4982                                                 $magicName = str_replace( '-', '_', "img_$name" );
4983                                                 $internalParamMap[$magicName] = array( $type, $name );
4984                                         }
4985                                 }
4986                         }
4987
4988                         # Add handler params
4989                         $paramMap = $internalParamMap;
4990                         if ( $handler ) {
4991                                 $handlerParamMap = $handler->getParamMap();
4992                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4993                                         $paramMap[$magic] = array( 'handler', $paramName );
4994                                 }
4995                         }
4996                         $this->mImageParams[$handlerClass] = $paramMap;
4997                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4998                 }
4999                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
5000         }
5001
5002         /**
5003          * Parse image options text and use it to make an image
5004          *
5005          * @param $title Title
5006          * @param $options String
5007          * @param $holders LinkHolderArray|bool
5008          * @return string HTML
5009          */
5010         function makeImage( $title, $options, $holders = false ) {
5011                 # Check if the options text is of the form "options|alt text"
5012                 # Options are:
5013                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
5014                 #  * left       no resizing, just left align. label is used for alt= only
5015                 #  * right      same, but right aligned
5016                 #  * none       same, but not aligned
5017                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
5018                 #  * center     center the image
5019                 #  * frame      Keep original image size, no magnify-button.
5020                 #  * framed     Same as "frame"
5021                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
5022                 #  * upright    reduce width for upright images, rounded to full __0 px
5023                 #  * border     draw a 1px border around the image
5024                 #  * alt        Text for HTML alt attribute (defaults to empty)
5025                 #  * class      Set a class for img node
5026                 #  * link       Set the target of the image link. Can be external, interwiki, or local
5027                 # vertical-align values (no % or length right now):
5028                 #  * baseline
5029                 #  * sub
5030                 #  * super
5031                 #  * top
5032                 #  * text-top
5033                 #  * middle
5034                 #  * bottom
5035                 #  * text-bottom
5036
5037                 $parts = StringUtils::explode( "|", $options );
5038
5039                 # Give extensions a chance to select the file revision for us
5040                 $options = array();
5041                 $descQuery = false;
5042                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5043                         array( $this, $title, &$options, &$descQuery ) );
5044                 # Fetch and register the file (file title may be different via hooks)
5045                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5046
5047                 # Get parameter map
5048                 $handler = $file ? $file->getHandler() : false;
5049
5050                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5051
5052                 if ( !$file ) {
5053                         $this->addTrackingCategory( 'broken-file-category' );
5054                 }
5055
5056                 # Process the input parameters
5057                 $caption = '';
5058                 $params = array( 'frame' => array(), 'handler' => array(),
5059                         'horizAlign' => array(), 'vertAlign' => array() );
5060                 foreach ( $parts as $part ) {
5061                         $part = trim( $part );
5062                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5063                         $validated = false;
5064                         if ( isset( $paramMap[$magicName] ) ) {
5065                                 list( $type, $paramName ) = $paramMap[$magicName];
5066
5067                                 # Special case; width and height come in one variable together
5068                                 if ( $type === 'handler' && $paramName === 'width' ) {
5069                                         $parsedWidthParam = $this->parseWidthParam( $value );
5070                                         if( isset( $parsedWidthParam['width'] ) ) {
5071                                                 $width = $parsedWidthParam['width'];
5072                                                 if ( $handler->validateParam( 'width', $width ) ) {
5073                                                         $params[$type]['width'] = $width;
5074                                                         $validated = true;
5075                                                 }
5076                                         }
5077                                         if( isset( $parsedWidthParam['height'] ) ) {
5078                                                 $height = $parsedWidthParam['height'];
5079                                                 if ( $handler->validateParam( 'height', $height ) ) {
5080                                                         $params[$type]['height'] = $height;
5081                                                         $validated = true;
5082                                                 }
5083                                         }
5084                                         # else no validation -- bug 13436
5085                                 } else {
5086                                         if ( $type === 'handler' ) {
5087                                                 # Validate handler parameter
5088                                                 $validated = $handler->validateParam( $paramName, $value );
5089                                         } else {
5090                                                 # Validate internal parameters
5091                                                 switch( $paramName ) {
5092                                                 case 'manualthumb':
5093                                                 case 'alt':
5094                                                 case 'class':
5095                                                         # @todo FIXME: Possibly check validity here for
5096                                                         # manualthumb? downstream behavior seems odd with
5097                                                         # missing manual thumbs.
5098                                                         $validated = true;
5099                                                         $value = $this->stripAltText( $value, $holders );
5100                                                         break;
5101                                                 case 'link':
5102                                                         $chars = self::EXT_LINK_URL_CLASS;
5103                                                         $prots = $this->mUrlProtocols;
5104                                                         if ( $value === '' ) {
5105                                                                 $paramName = 'no-link';
5106                                                                 $value = true;
5107                                                                 $validated = true;
5108                                                         } elseif ( preg_match( "/^(?i)$prots/", $value ) ) {
5109                                                                 if ( preg_match( "/^((?i)$prots)$chars+$/u", $value, $m ) ) {
5110                                                                         $paramName = 'link-url';
5111                                                                         $this->mOutput->addExternalLink( $value );
5112                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
5113                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5114                                                                         }
5115                                                                         $validated = true;
5116                                                                 }
5117                                                         } else {
5118                                                                 $linkTitle = Title::newFromText( $value );
5119                                                                 if ( $linkTitle ) {
5120                                                                         $paramName = 'link-title';
5121                                                                         $value = $linkTitle;
5122                                                                         $this->mOutput->addLink( $linkTitle );
5123                                                                         $validated = true;
5124                                                                 }
5125                                                         }
5126                                                         break;
5127                                                 default:
5128                                                         # Most other things appear to be empty or numeric...
5129                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
5130                                                 }
5131                                         }
5132
5133                                         if ( $validated ) {
5134                                                 $params[$type][$paramName] = $value;
5135                                         }
5136                                 }
5137                         }
5138                         if ( !$validated ) {
5139                                 $caption = $part;
5140                         }
5141                 }
5142
5143                 # Process alignment parameters
5144                 if ( $params['horizAlign'] ) {
5145                         $params['frame']['align'] = key( $params['horizAlign'] );
5146                 }
5147                 if ( $params['vertAlign'] ) {
5148                         $params['frame']['valign'] = key( $params['vertAlign'] );
5149                 }
5150
5151                 $params['frame']['caption'] = $caption;
5152
5153                 # Will the image be presented in a frame, with the caption below?
5154                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5155                                                  isset( $params['frame']['framed'] ) ||
5156                                                  isset( $params['frame']['thumbnail'] ) ||
5157                                                  isset( $params['frame']['manualthumb'] );
5158
5159                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5160                 # came to also set the caption, ordinary text after the image -- which
5161                 # makes no sense, because that just repeats the text multiple times in
5162                 # screen readers.  It *also* came to set the title attribute.
5163                 #
5164                 # Now that we have an alt attribute, we should not set the alt text to
5165                 # equal the caption: that's worse than useless, it just repeats the
5166                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5167                 # use the unnamed parameter for alt text as well, just for the time be-
5168                 # ing, if the unnamed param is set and the alt param is not.
5169                 #
5170                 # For the future, we need to figure out if we want to tweak this more,
5171                 # e.g., introducing a title= parameter for the title; ignoring the un-
5172                 # named parameter entirely for images without a caption; adding an ex-
5173                 # plicit caption= parameter and preserving the old magic unnamed para-
5174                 # meter for BC; ...
5175                 if ( $imageIsFramed ) { # Framed image
5176                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5177                                 # No caption or alt text, add the filename as the alt text so
5178                                 # that screen readers at least get some description of the image
5179                                 $params['frame']['alt'] = $title->getText();
5180                         }
5181                         # Do not set $params['frame']['title'] because tooltips don't make sense
5182                         # for framed images
5183                 } else { # Inline image
5184                         if ( !isset( $params['frame']['alt'] ) ) {
5185                                 # No alt text, use the "caption" for the alt text
5186                                 if ( $caption !== '') {
5187                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5188                                 } else {
5189                                         # No caption, fall back to using the filename for the
5190                                         # alt text
5191                                         $params['frame']['alt'] = $title->getText();
5192                                 }
5193                         }
5194                         # Use the "caption" for the tooltip text
5195                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5196                 }
5197
5198                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5199
5200                 # Linker does the rest
5201                 $time = isset( $options['time'] ) ? $options['time'] : false;
5202                 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5203                         $time, $descQuery, $this->mOptions->getThumbSize() );
5204
5205                 # Give the handler a chance to modify the parser object
5206                 if ( $handler ) {
5207                         $handler->parserTransformHook( $this, $file );
5208                 }
5209
5210                 return $ret;
5211         }
5212
5213         /**
5214          * @param $caption
5215          * @param $holders LinkHolderArray
5216          * @return mixed|String
5217          */
5218         protected function stripAltText( $caption, $holders ) {
5219                 # Strip bad stuff out of the title (tooltip).  We can't just use
5220                 # replaceLinkHoldersText() here, because if this function is called
5221                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5222                 if ( $holders ) {
5223                         $tooltip = $holders->replaceText( $caption );
5224                 } else {
5225                         $tooltip = $this->replaceLinkHoldersText( $caption );
5226                 }
5227
5228                 # make sure there are no placeholders in thumbnail attributes
5229                 # that are later expanded to html- so expand them now and
5230                 # remove the tags
5231                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5232                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5233
5234                 return $tooltip;
5235         }
5236
5237         /**
5238          * Set a flag in the output object indicating that the content is dynamic and
5239          * shouldn't be cached.
5240          */
5241         function disableCache() {
5242                 wfDebug( "Parser output marked as uncacheable.\n" );
5243                 if ( !$this->mOutput ) {
5244                         throw new MWException( __METHOD__ .
5245                                 " can only be called when actually parsing something" );
5246                 }
5247                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5248                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5249         }
5250
5251         /**
5252          * Callback from the Sanitizer for expanding items found in HTML attribute
5253          * values, so they can be safely tested and escaped.
5254          *
5255          * @param $text String
5256          * @param $frame PPFrame
5257          * @return String
5258          */
5259         function attributeStripCallback( &$text, $frame = false ) {
5260                 $text = $this->replaceVariables( $text, $frame );
5261                 $text = $this->mStripState->unstripBoth( $text );
5262                 return $text;
5263         }
5264
5265         /**
5266          * Accessor
5267          *
5268          * @return array
5269          */
5270         function getTags() {
5271                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ), array_keys( $this->mFunctionTagHooks ) );
5272         }
5273
5274         /**
5275          * Replace transparent tags in $text with the values given by the callbacks.
5276          *
5277          * Transparent tag hooks are like regular XML-style tag hooks, except they
5278          * operate late in the transformation sequence, on HTML instead of wikitext.
5279          *
5280          * @param $text string
5281          *
5282          * @return string
5283          */
5284         function replaceTransparentTags( $text ) {
5285                 $matches = array();
5286                 $elements = array_keys( $this->mTransparentTagHooks );
5287                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5288                 $replacements = array();
5289
5290                 foreach ( $matches as $marker => $data ) {
5291                         list( $element, $content, $params, $tag ) = $data;
5292                         $tagName = strtolower( $element );
5293                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5294                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5295                         } else {
5296                                 $output = $tag;
5297                         }
5298                         $replacements[$marker] = $output;
5299                 }
5300                 return strtr( $text, $replacements );
5301         }
5302
5303         /**
5304          * Break wikitext input into sections, and either pull or replace
5305          * some particular section's text.
5306          *
5307          * External callers should use the getSection and replaceSection methods.
5308          *
5309          * @param $text String: Page wikitext
5310          * @param $section String: a section identifier string of the form:
5311          *   "<flag1> - <flag2> - ... - <section number>"
5312          *
5313          * Currently the only recognised flag is "T", which means the target section number
5314          * was derived during a template inclusion parse, in other words this is a template
5315          * section edit link. If no flags are given, it was an ordinary section edit link.
5316          * This flag is required to avoid a section numbering mismatch when a section is
5317          * enclosed by "<includeonly>" (bug 6563).
5318          *
5319          * The section number 0 pulls the text before the first heading; other numbers will
5320          * pull the given section along with its lower-level subsections. If the section is
5321          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5322          *
5323          * Section 0 is always considered to exist, even if it only contains the empty
5324          * string. If $text is the empty string and section 0 is replaced, $newText is
5325          * returned.
5326          *
5327          * @param $mode String: one of "get" or "replace"
5328          * @param $newText String: replacement text for section data.
5329          * @return String: for "get", the extracted section text.
5330          *                 for "replace", the whole page with the section replaced.
5331          */
5332         private function extractSections( $text, $section, $mode, $newText='' ) {
5333                 global $wgTitle; # not generally used but removes an ugly failure mode
5334                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5335                 $outText = '';
5336                 $frame = $this->getPreprocessor()->newFrame();
5337
5338                 # Process section extraction flags
5339                 $flags = 0;
5340                 $sectionParts = explode( '-', $section );
5341                 $sectionIndex = array_pop( $sectionParts );
5342                 foreach ( $sectionParts as $part ) {
5343                         if ( $part === 'T' ) {
5344                                 $flags |= self::PTD_FOR_INCLUSION;
5345                         }
5346                 }
5347
5348                 # Check for empty input
5349                 if ( strval( $text ) === '' ) {
5350                         # Only sections 0 and T-0 exist in an empty document
5351                         if ( $sectionIndex == 0 ) {
5352                                 if ( $mode === 'get' ) {
5353                                         return '';
5354                                 } else {
5355                                         return $newText;
5356                                 }
5357                         } else {
5358                                 if ( $mode === 'get' ) {
5359                                         return $newText;
5360                                 } else {
5361                                         return $text;
5362                                 }
5363                         }
5364                 }
5365
5366                 # Preprocess the text
5367                 $root = $this->preprocessToDom( $text, $flags );
5368
5369                 # <h> nodes indicate section breaks
5370                 # They can only occur at the top level, so we can find them by iterating the root's children
5371                 $node = $root->getFirstChild();
5372
5373                 # Find the target section
5374                 if ( $sectionIndex == 0 ) {
5375                         # Section zero doesn't nest, level=big
5376                         $targetLevel = 1000;
5377                 } else {
5378                         while ( $node ) {
5379                                 if ( $node->getName() === 'h' ) {
5380                                         $bits = $node->splitHeading();
5381                                         if ( $bits['i'] == $sectionIndex ) {
5382                                                 $targetLevel = $bits['level'];
5383                                                 break;
5384                                         }
5385                                 }
5386                                 if ( $mode === 'replace' ) {
5387                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5388                                 }
5389                                 $node = $node->getNextSibling();
5390                         }
5391                 }
5392
5393                 if ( !$node ) {
5394                         # Not found
5395                         if ( $mode === 'get' ) {
5396                                 return $newText;
5397                         } else {
5398                                 return $text;
5399                         }
5400                 }
5401
5402                 # Find the end of the section, including nested sections
5403                 do {
5404                         if ( $node->getName() === 'h' ) {
5405                                 $bits = $node->splitHeading();
5406                                 $curLevel = $bits['level'];
5407                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5408                                         break;
5409                                 }
5410                         }
5411                         if ( $mode === 'get' ) {
5412                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5413                         }
5414                         $node = $node->getNextSibling();
5415                 } while ( $node );
5416
5417                 # Write out the remainder (in replace mode only)
5418                 if ( $mode === 'replace' ) {
5419                         # Output the replacement text
5420                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5421                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5422                         # Only add trailing whitespace if there is newText
5423                         if ( $newText != "" ) {
5424                                 $outText .= $newText . "\n\n";
5425                         }
5426
5427                         while ( $node ) {
5428                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5429                                 $node = $node->getNextSibling();
5430                         }
5431                 }
5432
5433                 if ( is_string( $outText ) ) {
5434                         # Re-insert stripped tags
5435                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5436                 }
5437
5438                 return $outText;
5439         }
5440
5441         /**
5442          * This function returns the text of a section, specified by a number ($section).
5443          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5444          * the first section before any such heading (section 0).
5445          *
5446          * If a section contains subsections, these are also returned.
5447          *
5448          * @param $text String: text to look in
5449          * @param $section String: section identifier
5450          * @param $deftext String: default to return if section is not found
5451          * @return string text of the requested section
5452          */
5453         public function getSection( $text, $section, $deftext='' ) {
5454                 return $this->extractSections( $text, $section, "get", $deftext );
5455         }
5456
5457         /**
5458          * This function returns $oldtext after the content of the section
5459          * specified by $section has been replaced with $text. If the target
5460          * section does not exist, $oldtext is returned unchanged.
5461          *
5462          * @param $oldtext String: former text of the article
5463          * @param $section int section identifier
5464          * @param $text String: replacing text
5465          * @return String: modified text
5466          */
5467         public function replaceSection( $oldtext, $section, $text ) {
5468                 return $this->extractSections( $oldtext, $section, "replace", $text );
5469         }
5470
5471         /**
5472          * Get the ID of the revision we are parsing
5473          *
5474          * @return Mixed: integer or null
5475          */
5476         function getRevisionId() {
5477                 return $this->mRevisionId;
5478         }
5479
5480         /**
5481          * Get the revision object for $this->mRevisionId
5482          *
5483          * @return Revision|null either a Revision object or null
5484          */
5485         protected function getRevisionObject() {
5486                 if ( !is_null( $this->mRevisionObject ) ) {
5487                         return $this->mRevisionObject;
5488                 }
5489                 if ( is_null( $this->mRevisionId ) ) {
5490                         return null;
5491                 }
5492
5493                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5494                 return $this->mRevisionObject;
5495         }
5496
5497         /**
5498          * Get the timestamp associated with the current revision, adjusted for
5499          * the default server-local timestamp
5500          */
5501         function getRevisionTimestamp() {
5502                 if ( is_null( $this->mRevisionTimestamp ) ) {
5503                         wfProfileIn( __METHOD__ );
5504
5505                         global $wgContLang;
5506
5507                         $revObject = $this->getRevisionObject();
5508                         $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5509
5510                         # The cryptic '' timezone parameter tells to use the site-default
5511                         # timezone offset instead of the user settings.
5512                         #
5513                         # Since this value will be saved into the parser cache, served
5514                         # to other users, and potentially even used inside links and such,
5515                         # it needs to be consistent for all visitors.
5516                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5517
5518                         wfProfileOut( __METHOD__ );
5519                 }
5520                 return $this->mRevisionTimestamp;
5521         }
5522
5523         /**
5524          * Get the name of the user that edited the last revision
5525          *
5526          * @return String: user name
5527          */
5528         function getRevisionUser() {
5529                 if( is_null( $this->mRevisionUser ) ) {
5530                         $revObject = $this->getRevisionObject();
5531
5532                         # if this template is subst: the revision id will be blank,
5533                         # so just use the current user's name
5534                         if( $revObject ) {
5535                                 $this->mRevisionUser = $revObject->getUserText();
5536                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5537                                 $this->mRevisionUser = $this->getUser()->getName();
5538                         }
5539                 }
5540                 return $this->mRevisionUser;
5541         }
5542
5543         /**
5544          * Mutator for $mDefaultSort
5545          *
5546          * @param $sort string New value
5547          */
5548         public function setDefaultSort( $sort ) {
5549                 $this->mDefaultSort = $sort;
5550                 $this->mOutput->setProperty( 'defaultsort', $sort );
5551         }
5552
5553         /**
5554          * Accessor for $mDefaultSort
5555          * Will use the empty string if none is set.
5556          *
5557          * This value is treated as a prefix, so the
5558          * empty string is equivalent to sorting by
5559          * page name.
5560          *
5561          * @return string
5562          */
5563         public function getDefaultSort() {
5564                 if ( $this->mDefaultSort !== false ) {
5565                         return $this->mDefaultSort;
5566                 } else {
5567                         return '';
5568                 }
5569         }
5570
5571         /**
5572          * Accessor for $mDefaultSort
5573          * Unlike getDefaultSort(), will return false if none is set
5574          *
5575          * @return string or false
5576          */
5577         public function getCustomDefaultSort() {
5578                 return $this->mDefaultSort;
5579         }
5580
5581         /**
5582          * Try to guess the section anchor name based on a wikitext fragment
5583          * presumably extracted from a heading, for example "Header" from
5584          * "== Header ==".
5585          *
5586          * @param $text string
5587          *
5588          * @return string
5589          */
5590         public function guessSectionNameFromWikiText( $text ) {
5591                 # Strip out wikitext links(they break the anchor)
5592                 $text = $this->stripSectionName( $text );
5593                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5594                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5595         }
5596
5597         /**
5598          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5599          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5600          * as something other than UTF-8 (apparently?), resulting in breakage.
5601          *
5602          * @param $text String: The section name
5603          * @return string An anchor
5604          */
5605         public function guessLegacySectionNameFromWikiText( $text ) {
5606                 # Strip out wikitext links(they break the anchor)
5607                 $text = $this->stripSectionName( $text );
5608                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5609                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5610         }
5611
5612         /**
5613          * Strips a text string of wikitext for use in a section anchor
5614          *
5615          * Accepts a text string and then removes all wikitext from the
5616          * string and leaves only the resultant text (i.e. the result of
5617          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5618          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5619          * to create valid section anchors by mimicing the output of the
5620          * parser when headings are parsed.
5621          *
5622          * @param $text String: text string to be stripped of wikitext
5623          * for use in a Section anchor
5624          * @return string Filtered text string
5625          */
5626         public function stripSectionName( $text ) {
5627                 # Strip internal link markup
5628                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5629                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5630
5631                 # Strip external link markup
5632                 # @todo FIXME: Not tolerant to blank link text
5633                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5634                 # on how many empty links there are on the page - need to figure that out.
5635                 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5636
5637                 # Parse wikitext quotes (italics & bold)
5638                 $text = $this->doQuotes( $text );
5639
5640                 # Strip HTML tags
5641                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5642                 return $text;
5643         }
5644
5645         /**
5646          * strip/replaceVariables/unstrip for preprocessor regression testing
5647          *
5648          * @param $text string
5649          * @param $title Title
5650          * @param $options ParserOptions
5651          * @param $outputType int
5652          *
5653          * @return string
5654          */
5655         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5656                 $this->startParse( $title, $options, $outputType, true );
5657
5658                 $text = $this->replaceVariables( $text );
5659                 $text = $this->mStripState->unstripBoth( $text );
5660                 $text = Sanitizer::removeHTMLtags( $text );
5661                 return $text;
5662         }
5663
5664         /**
5665          * @param $text string
5666          * @param $title Title
5667          * @param $options ParserOptions
5668          * @return string
5669          */
5670         function testPst( $text, Title $title, ParserOptions $options ) {
5671                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5672         }
5673
5674         /**
5675          * @param $text
5676          * @param $title Title
5677          * @param $options ParserOptions
5678          * @return string
5679          */
5680         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5681                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5682         }
5683
5684         /**
5685          * Call a callback function on all regions of the given text that are not
5686          * inside strip markers, and replace those regions with the return value
5687          * of the callback. For example, with input:
5688          *
5689          *  aaa<MARKER>bbb
5690          *
5691          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5692          * two strings will be replaced with the value returned by the callback in
5693          * each case.
5694          *
5695          * @param $s string
5696          * @param $callback
5697          *
5698          * @return string
5699          */
5700         function markerSkipCallback( $s, $callback ) {
5701                 $i = 0;
5702                 $out = '';
5703                 while ( $i < strlen( $s ) ) {
5704                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5705                         if ( $markerStart === false ) {
5706                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5707                                 break;
5708                         } else {
5709                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5710                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5711                                 if ( $markerEnd === false ) {
5712                                         $out .= substr( $s, $markerStart );
5713                                         break;
5714                                 } else {
5715                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5716                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5717                                         $i = $markerEnd;
5718                                 }
5719                         }
5720                 }
5721                 return $out;
5722         }
5723
5724         /**
5725          * Remove any strip markers found in the given text.
5726          *
5727          * @param $text Input string
5728          * @return string
5729          */
5730         function killMarkers( $text ) {
5731                 return $this->mStripState->killMarkers( $text );
5732         }
5733
5734         /**
5735          * Save the parser state required to convert the given half-parsed text to
5736          * HTML. "Half-parsed" in this context means the output of
5737          * recursiveTagParse() or internalParse(). This output has strip markers
5738          * from replaceVariables (extensionSubstitution() etc.), and link
5739          * placeholders from replaceLinkHolders().
5740          *
5741          * Returns an array which can be serialized and stored persistently. This
5742          * array can later be loaded into another parser instance with
5743          * unserializeHalfParsedText(). The text can then be safely incorporated into
5744          * the return value of a parser hook.
5745          *
5746          * @param $text string
5747          *
5748          * @return array
5749          */
5750         function serializeHalfParsedText( $text ) {
5751                 wfProfileIn( __METHOD__ );
5752                 $data = array(
5753                         'text' => $text,
5754                         'version' => self::HALF_PARSED_VERSION,
5755                         'stripState' => $this->mStripState->getSubState( $text ),
5756                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5757                 );
5758                 wfProfileOut( __METHOD__ );
5759                 return $data;
5760         }
5761
5762         /**
5763          * Load the parser state given in the $data array, which is assumed to
5764          * have been generated by serializeHalfParsedText(). The text contents is
5765          * extracted from the array, and its markers are transformed into markers
5766          * appropriate for the current Parser instance. This transformed text is
5767          * returned, and can be safely included in the return value of a parser
5768          * hook.
5769          *
5770          * If the $data array has been stored persistently, the caller should first
5771          * check whether it is still valid, by calling isValidHalfParsedText().
5772          *
5773          * @param $data array Serialized data
5774          * @return String
5775          */
5776         function unserializeHalfParsedText( $data ) {
5777                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5778                         throw new MWException( __METHOD__.': invalid version' );
5779                 }
5780
5781                 # First, extract the strip state.
5782                 $texts = array( $data['text'] );
5783                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5784
5785                 # Now renumber links
5786                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5787
5788                 # Should be good to go.
5789                 return $texts[0];
5790         }
5791
5792         /**
5793          * Returns true if the given array, presumed to be generated by
5794          * serializeHalfParsedText(), is compatible with the current version of the
5795          * parser.
5796          *
5797          * @param $data Array
5798          *
5799          * @return bool
5800          */
5801         function isValidHalfParsedText( $data ) {
5802                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5803         }
5804
5805         /**
5806          * Parsed a width param of imagelink like 300px or 200x300px
5807          *
5808          * @param $value String
5809          *
5810          * @return array
5811          * @since 1.20
5812          */
5813         public function parseWidthParam( $value ) {
5814                 $parsedWidthParam = array();
5815                 if( $value === '' ) {
5816                         return $parsedWidthParam;
5817                 }
5818                 $m = array();
5819                 # (bug 13500) In both cases (width/height and width only),
5820                 # permit trailing "px" for backward compatibility.
5821                 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5822                         $width = intval( $m[1] );
5823                         $height = intval( $m[2] );
5824                         $parsedWidthParam['width'] = $width;
5825                         $parsedWidthParam['height'] = $height;
5826                 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5827                         $width = intval( $value );
5828                         $parsedWidthParam['width'] = $width;
5829                 }
5830                 return $parsedWidthParam;
5831         }
5832 }