includes/parser/Preprocessor_DOM.php

   1 <?php
   2 /**
   3  * Preprocessor using PHP's dom extension
   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  * @ingroup Parser
  26  * @codingStandardsIgnoreStart
  27  */
  28 class Preprocessor_DOM implements Preprocessor {
  29         // @codingStandardsIgnoreEnd
  30
  31         /**
  32          * @var Parser
  33          */
  34         public $parser;
  35
  36         public $memoryLimit;
  37
  38         const CACHE_VERSION = 1;
  39
  40         public function __construct( $parser ) {
  41                 $this->parser = $parser;
  42                 $mem = ini_get( 'memory_limit' );
  43                 $this->memoryLimit = false;
  44                 if ( strval( $mem ) !== '' && $mem != -1 ) {
  45                         if ( preg_match( '/^\d+$/', $mem ) ) {
  46                                 $this->memoryLimit = $mem;
  47                         } elseif ( preg_match( '/^(\d+)M$/i', $mem, $m ) ) {
  48                                 $this->memoryLimit = $m[1] * 1048576;
  49                         }
  50                 }
  51         }
  52
  53         /**
  54          * @return PPFrame_DOM
  55          */
  56         public function newFrame() {
  57                 return new PPFrame_DOM( $this );
  58         }
  59
  60         /**
  61          * @param array $args
  62          * @return PPCustomFrame_DOM
  63          */
  64         public function newCustomFrame( $args ) {
  65                 return new PPCustomFrame_DOM( $this, $args );
  66         }
  67
  68         /**
  69          * @param array $values
  70          * @return PPNode_DOM
  71          * @throws MWException
  72          */
  73         public function newPartNodeArray( $values ) {
  74                 // NOTE: DOM manipulation is slower than building & parsing XML! (or so Tim sais)
  75                 $xml = "<list>";
  76
  77                 foreach ( $values as $k => $val ) {
  78                         if ( is_int( $k ) ) {
  79                                 $xml .= "<part><name index=\"$k\"/><value>"
  80                                         . htmlspecialchars( $val ) . "</value></part>";
  81                         } else {
  82                                 $xml .= "<part><name>" . htmlspecialchars( $k )
  83                                         . "</name>=<value>" . htmlspecialchars( $val ) . "</value></part>";
  84                         }
  85                 }
  86
  87                 $xml .= "</list>";
  88
  89                 $dom = new DOMDocument();
  90                 MediaWiki\suppressWarnings();
  91                 $result = $dom->loadXML( $xml );
  92                 MediaWiki\restoreWarnings();
  93                 if ( !$result ) {
  94                         // Try running the XML through UtfNormal to get rid of invalid characters
  95                         $xml = UtfNormal\Validator::cleanUp( $xml );
  96                         // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
  97                         // don't barf when the XML is >256 levels deep
  98                         $result = $dom->loadXML( $xml, 1 << 19 );
  99                 }
 100
 101                 if ( !$result ) {
 102                         throw new MWException( 'Parameters passed to ' . __METHOD__ . ' result in invalid XML' );
 103                 }
 104
 105                 $root = $dom->documentElement;
 106                 $node = new PPNode_DOM( $root->childNodes );
 107                 return $node;
 108         }
 109
 110         /**
 111          * @throws MWException
 112          * @return bool
 113          */
 114         public function memCheck() {
 115                 if ( $this->memoryLimit === false ) {
 116                         return true;
 117                 }
 118                 $usage = memory_get_usage();
 119                 if ( $usage > $this->memoryLimit * 0.9 ) {
 120                         $limit = intval( $this->memoryLimit * 0.9 / 1048576 + 0.5 );
 121                         throw new MWException( "Preprocessor hit 90% memory limit ($limit MB)" );
 122                 }
 123                 return $usage <= $this->memoryLimit * 0.8;
 124         }
 125
 126         /**
 127          * Preprocess some wikitext and return the document tree.
 128          * This is the ghost of Parser::replace_variables().
 129          *
 130          * @param string $text The text to parse
 131          * @param int $flags Bitwise combination of:
 132          *     Parser::PTD_FOR_INCLUSION  Handle "<noinclude>" and "<includeonly>"
 133          *                                as if the text is being included. Default
 134          *                                is to assume a direct page view.
 135          *
 136          * The generated DOM tree must depend only on the input text and the flags.
 137          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
 138          *
 139          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
 140          * change in the DOM tree for a given text, must be passed through the section identifier
 141          * in the section edit link and thus back to extractSections().
 142          *
 143          * The output of this function is currently only cached in process memory, but a persistent
 144          * cache may be implemented at a later date which takes further advantage of these strict
 145          * dependency requirements.
 146          *
 147          * @throws MWException
 148          * @return PPNode_DOM
 149          */
 150         public function preprocessToObj( $text, $flags = 0 ) {
 151                 global $wgMemc, $wgPreprocessorCacheThreshold;
 152
 153                 $xml = false;
 154                 $cacheable = ( $wgPreprocessorCacheThreshold !== false
 155                         && strlen( $text ) > $wgPreprocessorCacheThreshold );
 156                 if ( $cacheable ) {
 157                         $cacheKey = wfMemcKey( 'preprocess-xml', md5( $text ), $flags );
 158                         $cacheValue = $wgMemc->get( $cacheKey );
 159                         if ( $cacheValue ) {
 160                                 $version = substr( $cacheValue, 0, 8 );
 161                                 if ( intval( $version ) == self::CACHE_VERSION ) {
 162                                         $xml = substr( $cacheValue, 8 );
 163                                         // From the cache
 164                                         wfDebugLog( "Preprocessor", "Loaded preprocessor XML from memcached (key $cacheKey)" );
 165                                 }
 166                         }
 167                         if ( $xml === false ) {
 168                                 $xml = $this->preprocessToXml( $text, $flags );
 169                                 $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . $xml;
 170                                 $wgMemc->set( $cacheKey, $cacheValue, 86400 );
 171                                 wfDebugLog( "Preprocessor", "Saved preprocessor XML to memcached (key $cacheKey)" );
 172                         }
 173                 } else {
 174                         $xml = $this->preprocessToXml( $text, $flags );
 175                 }
 176
 177                 // Fail if the number of elements exceeds acceptable limits
 178                 // Do not attempt to generate the DOM
 179                 $this->parser->mGeneratedPPNodeCount += substr_count( $xml, '<' );
 180                 $max = $this->parser->mOptions->getMaxGeneratedPPNodeCount();
 181                 if ( $this->parser->mGeneratedPPNodeCount > $max ) {
 182                         if ( $cacheable ) {
 183                         }
 184                         throw new MWException( __METHOD__ . ': generated node count limit exceeded' );
 185                 }
 186
 187                 $dom = new DOMDocument;
 188                 MediaWiki\suppressWarnings();
 189                 $result = $dom->loadXML( $xml );
 190                 MediaWiki\restoreWarnings();
 191                 if ( !$result ) {
 192                         // Try running the XML through UtfNormal to get rid of invalid characters
 193                         $xml = UtfNormal\Validator::cleanUp( $xml );
 194                         // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
 195                         // don't barf when the XML is >256 levels deep.
 196                         $result = $dom->loadXML( $xml, 1 << 19 );
 197                 }
 198                 if ( $result ) {
 199                         $obj = new PPNode_DOM( $dom->documentElement );
 200                 }
 201
 202                 if ( $cacheable ) {
 203                 }
 204
 205                 if ( !$result ) {
 206                         throw new MWException( __METHOD__ . ' generated invalid XML' );
 207                 }
 208                 return $obj;
 209         }
 210
 211         /**
 212          * @param string $text
 213          * @param int $flags
 214          * @return string
 215          */
 216         public function preprocessToXml( $text, $flags = 0 ) {
 217                 $rules = array(
 218                         '{' => array(
 219                                 'end' => '}',
 220                                 'names' => array(
 221                                         2 => 'template',
 222                                         3 => 'tplarg',
 223                                 ),
 224                                 'min' => 2,
 225                                 'max' => 3,
 226                         ),
 227                         '[' => array(
 228                                 'end' => ']',
 229                                 'names' => array( 2 => null ),
 230                                 'min' => 2,
 231                                 'max' => 2,
 232                         )
 233                 );
 234
 235                 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
 236
 237                 $xmlishElements = $this->parser->getStripList();
 238                 $enableOnlyinclude = false;
 239                 if ( $forInclusion ) {
 240                         $ignoredTags = array( 'includeonly', '/includeonly' );
 241                         $ignoredElements = array( 'noinclude' );
 242                         $xmlishElements[] = 'noinclude';
 243                         if ( strpos( $text, '<onlyinclude>' ) !== false
 244                                 && strpos( $text, '</onlyinclude>' ) !== false
 245                         ) {
 246                                 $enableOnlyinclude = true;
 247                         }
 248                 } else {
 249                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
 250                         $ignoredElements = array( 'includeonly' );
 251                         $xmlishElements[] = 'includeonly';
 252                 }
 253                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
 254
 255                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
 256                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
 257
 258                 $stack = new PPDStack;
 259
 260                 $searchBase = "[{<\n"; # }
 261                 // For fast reverse searches
 262                 $revText = strrev( $text );
 263                 $lengthText = strlen( $text );
 264
 265                 // Input pointer, starts out pointing to a pseudo-newline before the start
 266                 $i = 0;
 267                 // Current accumulator
 268                 $accum =& $stack->getAccum();
 269                 $accum = '<root>';
 270                 // True to find equals signs in arguments
 271                 $findEquals = false;
 272                 // True to take notice of pipe characters
 273                 $findPipe = false;
 274                 $headingIndex = 1;
 275                 // True if $i is inside a possible heading
 276                 $inHeading = false;
 277                 // True if there are no more greater-than (>) signs right of $i
 278                 $noMoreGT = false;
 279                 // True to ignore all input up to the next <onlyinclude>
 280                 $findOnlyinclude = $enableOnlyinclude;
 281                 // Do a line-start run without outputting an LF character
 282                 $fakeLineStart = true;
 283
 284                 while ( true ) {
 285                         // $this->memCheck();
 286
 287                         if ( $findOnlyinclude ) {
 288                                 // Ignore all input up to the next <onlyinclude>
 289                                 $startPos = strpos( $text, '<onlyinclude>', $i );
 290                                 if ( $startPos === false ) {
 291                                         // Ignored section runs to the end
 292                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
 293                                         break;
 294                                 }
 295                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
 296                                 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
 297                                 $i = $tagEndPos;
 298                                 $findOnlyinclude = false;
 299                         }
 300
 301                         if ( $fakeLineStart ) {
 302                                 $found = 'line-start';
 303                                 $curChar = '';
 304                         } else {
 305                                 # Find next opening brace, closing brace or pipe
 306                                 $search = $searchBase;
 307                                 if ( $stack->top === false ) {
 308                                         $currentClosing = '';
 309                                 } else {
 310                                         $currentClosing = $stack->top->close;
 311                                         $search .= $currentClosing;
 312                                 }
 313                                 if ( $findPipe ) {
 314                                         $search .= '|';
 315                                 }
 316                                 if ( $findEquals ) {
 317                                         // First equals will be for the template
 318                                         $search .= '=';
 319                                 }
 320                                 $rule = null;
 321                                 # Output literal section, advance input counter
 322                                 $literalLength = strcspn( $text, $search, $i );
 323                                 if ( $literalLength > 0 ) {
 324                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
 325                                         $i += $literalLength;
 326                                 }
 327                                 if ( $i >= $lengthText ) {
 328                                         if ( $currentClosing == "\n" ) {
 329                                                 // Do a past-the-end run to finish off the heading
 330                                                 $curChar = '';
 331                                                 $found = 'line-end';
 332                                         } else {
 333                                                 # All done
 334                                                 break;
 335                                         }
 336                                 } else {
 337                                         $curChar = $text[$i];
 338                                         if ( $curChar == '|' ) {
 339                                                 $found = 'pipe';
 340                                         } elseif ( $curChar == '=' ) {
 341                                                 $found = 'equals';
 342                                         } elseif ( $curChar == '<' ) {
 343                                                 $found = 'angle';
 344                                         } elseif ( $curChar == "\n" ) {
 345                                                 if ( $inHeading ) {
 346                                                         $found = 'line-end';
 347                                                 } else {
 348                                                         $found = 'line-start';
 349                                                 }
 350                                         } elseif ( $curChar == $currentClosing ) {
 351                                                 $found = 'close';
 352                                         } elseif ( isset( $rules[$curChar] ) ) {
 353                                                 $found = 'open';
 354                                                 $rule = $rules[$curChar];
 355                                         } else {
 356                                                 # Some versions of PHP have a strcspn which stops on null characters
 357                                                 # Ignore and continue
 358                                                 ++$i;
 359                                                 continue;
 360                                         }
 361                                 }
 362                         }
 363
 364                         if ( $found == 'angle' ) {
 365                                 $matches = false;
 366                                 // Handle </onlyinclude>
 367                                 if ( $enableOnlyinclude
 368                                         && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>'
 369                                 ) {
 370                                         $findOnlyinclude = true;
 371                                         continue;
 372                                 }
 373
 374                                 // Determine element name
 375                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
 376                                         // Element name missing or not listed
 377                                         $accum .= '&lt;';
 378                                         ++$i;
 379                                         continue;
 380                                 }
 381                                 // Handle comments
 382                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
 383
 384                                         // To avoid leaving blank lines, when a sequence of
 385                                         // space-separated comments is both preceded and followed by
 386                                         // a newline (ignoring spaces), then
 387                                         // trim leading and trailing spaces and the trailing newline.
 388
 389                                         // Find the end
 390                                         $endPos = strpos( $text, '-->', $i + 4 );
 391                                         if ( $endPos === false ) {
 392                                                 // Unclosed comment in input, runs to end
 393                                                 $inner = substr( $text, $i );
 394                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
 395                                                 $i = $lengthText;
 396                                         } else {
 397                                                 // Search backwards for leading whitespace
 398                                                 $wsStart = $i ? ( $i - strspn( $revText, " \t", $lengthText - $i ) ) : 0;
 399
 400                                                 // Search forwards for trailing whitespace
 401                                                 // $wsEnd will be the position of the last space (or the '>' if there's none)
 402                                                 $wsEnd = $endPos + 2 + strspn( $text, " \t", $endPos + 3 );
 403
 404                                                 // Keep looking forward as long as we're finding more
 405                                                 // comments.
 406                                                 $comments = array( array( $wsStart, $wsEnd ) );
 407                                                 while ( substr( $text, $wsEnd + 1, 4 ) == '<!--' ) {
 408                                                         $c = strpos( $text, '-->', $wsEnd + 4 );
 409                                                         if ( $c === false ) {
 410                                                                 break;
 411                                                         }
 412                                                         $c = $c + 2 + strspn( $text, " \t", $c + 3 );
 413                                                         $comments[] = array( $wsEnd + 1, $c );
 414                                                         $wsEnd = $c;
 415                                                 }
 416
 417                                                 // Eat the line if possible
 418                                                 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
 419                                                 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
 420                                                 // it's a possible beneficial b/c break.
 421                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
 422                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n"
 423                                                 ) {
 424                                                         // Remove leading whitespace from the end of the accumulator
 425                                                         // Sanity check first though
 426                                                         $wsLength = $i - $wsStart;
 427                                                         if ( $wsLength > 0
 428                                                                 && strspn( $accum, " \t", -$wsLength ) === $wsLength
 429                                                         ) {
 430                                                                 $accum = substr( $accum, 0, -$wsLength );
 431                                                         }
 432
 433                                                         // Dump all but the last comment to the accumulator
 434                                                         foreach ( $comments as $j => $com ) {
 435                                                                 $startPos = $com[0];
 436                                                                 $endPos = $com[1] + 1;
 437                                                                 if ( $j == ( count( $comments ) - 1 ) ) {
 438                                                                         break;
 439                                                                 }
 440                                                                 $inner = substr( $text, $startPos, $endPos - $startPos );
 441                                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
 442                                                         }
 443
 444                                                         // Do a line-start run next time to look for headings after the comment
 445                                                         $fakeLineStart = true;
 446                                                 } else {
 447                                                         // No line to eat, just take the comment itself
 448                                                         $startPos = $i;
 449                                                         $endPos += 2;
 450                                                 }
 451
 452                                                 if ( $stack->top ) {
 453                                                         $part = $stack->top->getCurrentPart();
 454                                                         if ( !( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) ) {
 455                                                                 $part->visualEnd = $wsStart;
 456                                                         }
 457                                                         // Else comments abutting, no change in visual end
 458                                                         $part->commentEnd = $endPos;
 459                                                 }
 460                                                 $i = $endPos + 1;
 461                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
 462                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
 463                                         }
 464                                         continue;
 465                                 }
 466                                 $name = $matches[1];
 467                                 $lowerName = strtolower( $name );
 468                                 $attrStart = $i + strlen( $name ) + 1;
 469
 470                                 // Find end of tag
 471                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
 472                                 if ( $tagEndPos === false ) {
 473                                         // Infinite backtrack
 474                                         // Disable tag search to prevent worst-case O(N^2) performance
 475                                         $noMoreGT = true;
 476                                         $accum .= '&lt;';
 477                                         ++$i;
 478                                         continue;
 479                                 }
 480
 481                                 // Handle ignored tags
 482                                 if ( in_array( $lowerName, $ignoredTags ) ) {
 483                                         $accum .= '<ignore>'
 484                                                 . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) )
 485                                                 . '</ignore>';
 486                                         $i = $tagEndPos + 1;
 487                                         continue;
 488                                 }
 489
 490                                 $tagStartPos = $i;
 491                                 if ( $text[$tagEndPos - 1] == '/' ) {
 492                                         $attrEnd = $tagEndPos - 1;
 493                                         $inner = null;
 494                                         $i = $tagEndPos + 1;
 495                                         $close = '';
 496                                 } else {
 497                                         $attrEnd = $tagEndPos;
 498                                         // Find closing tag
 499                                         if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
 500                                                         $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 )
 501                                         ) {
 502                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
 503                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
 504                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
 505                                         } else {
 506                                                 // No end tag -- let it run out to the end of the text.
 507                                                 $inner = substr( $text, $tagEndPos + 1 );
 508                                                 $i = $lengthText;
 509                                                 $close = '';
 510                                         }
 511                                 }
 512                                 // <includeonly> and <noinclude> just become <ignore> tags
 513                                 if ( in_array( $lowerName, $ignoredElements ) ) {
 514                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
 515                                                 . '</ignore>';
 516                                         continue;
 517                                 }
 518
 519                                 $accum .= '<ext>';
 520                                 if ( $attrEnd <= $attrStart ) {
 521                                         $attr = '';
 522                                 } else {
 523                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
 524                                 }
 525                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
 526                                         // Note that the attr element contains the whitespace between name and attribute,
 527                                         // this is necessary for precise reconstruction during pre-save transform.
 528                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
 529                                 if ( $inner !== null ) {
 530                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
 531                                 }
 532                                 $accum .= $close . '</ext>';
 533                         } elseif ( $found == 'line-start' ) {
 534                                 // Is this the start of a heading?
 535                                 // Line break belongs before the heading element in any case
 536                                 if ( $fakeLineStart ) {
 537                                         $fakeLineStart = false;
 538                                 } else {
 539                                         $accum .= $curChar;
 540                                         $i++;
 541                                 }
 542
 543                                 $count = strspn( $text, '=', $i, 6 );
 544                                 if ( $count == 1 && $findEquals ) {
 545                                         // DWIM: This looks kind of like a name/value separator.
 546                                         // Let's let the equals handler have it and break the
 547                                         // potential heading. This is heuristic, but AFAICT the
 548                                         // methods for completely correct disambiguation are very
 549                                         // complex.
 550                                 } elseif ( $count > 0 ) {
 551                                         $piece = array(
 552                                                 'open' => "\n",
 553                                                 'close' => "\n",
 554                                                 'parts' => array( new PPDPart( str_repeat( '=', $count ) ) ),
 555                                                 'startPos' => $i,
 556                                                 'count' => $count );
 557                                         $stack->push( $piece );
 558                                         $accum =& $stack->getAccum();
 559                                         $flags = $stack->getFlags();
 560                                         extract( $flags );
 561                                         $i += $count;
 562                                 }
 563                         } elseif ( $found == 'line-end' ) {
 564                                 $piece = $stack->top;
 565                                 // A heading must be open, otherwise \n wouldn't have been in the search list
 566                                 assert( '$piece->open == "\n"' );
 567                                 $part = $piece->getCurrentPart();
 568                                 // Search back through the input to see if it has a proper close.
 569                                 // Do this using the reversed string since the other solutions
 570                                 // (end anchor, etc.) are inefficient.
 571                                 $wsLength = strspn( $revText, " \t", $lengthText - $i );
 572                                 $searchStart = $i - $wsLength;
 573                                 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
 574                                         // Comment found at line end
 575                                         // Search for equals signs before the comment
 576                                         $searchStart = $part->visualEnd;
 577                                         $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
 578                                 }
 579                                 $count = $piece->count;
 580                                 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
 581                                 if ( $equalsLength > 0 ) {
 582                                         if ( $searchStart - $equalsLength == $piece->startPos ) {
 583                                                 // This is just a single string of equals signs on its own line
 584                                                 // Replicate the doHeadings behavior /={count}(.+)={count}/
 585                                                 // First find out how many equals signs there really are (don't stop at 6)
 586                                                 $count = $equalsLength;
 587                                                 if ( $count < 3 ) {
 588                                                         $count = 0;
 589                                                 } else {
 590                                                         $count = min( 6, intval( ( $count - 1 ) / 2 ) );
 591                                                 }
 592                                         } else {
 593                                                 $count = min( $equalsLength, $count );
 594                                         }
 595                                         if ( $count > 0 ) {
 596                                                 // Normal match, output <h>
 597                                                 $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
 598                                                 $headingIndex++;
 599                                         } else {
 600                                                 // Single equals sign on its own line, count=0
 601                                                 $element = $accum;
 602                                         }
 603                                 } else {
 604                                         // No match, no <h>, just pass down the inner text
 605                                         $element = $accum;
 606                                 }
 607                                 // Unwind the stack
 608                                 $stack->pop();
 609                                 $accum =& $stack->getAccum();
 610                                 $flags = $stack->getFlags();
 611                                 extract( $flags );
 612
 613                                 // Append the result to the enclosing accumulator
 614                                 $accum .= $element;
 615                                 // Note that we do NOT increment the input pointer.
 616                                 // This is because the closing linebreak could be the opening linebreak of
 617                                 // another heading. Infinite loops are avoided because the next iteration MUST
 618                                 // hit the heading open case above, which unconditionally increments the
 619                                 // input pointer.
 620                         } elseif ( $found == 'open' ) {
 621                                 # count opening brace characters
 622                                 $count = strspn( $text, $curChar, $i );
 623
 624                                 # we need to add to stack only if opening brace count is enough for one of the rules
 625                                 if ( $count >= $rule['min'] ) {
 626                                         # Add it to the stack
 627                                         $piece = array(
 628                                                 'open' => $curChar,
 629                                                 'close' => $rule['end'],
 630                                                 'count' => $count,
 631                                                 'lineStart' => ( $i > 0 && $text[$i - 1] == "\n" ),
 632                                         );
 633
 634                                         $stack->push( $piece );
 635                                         $accum =& $stack->getAccum();
 636                                         $flags = $stack->getFlags();
 637                                         extract( $flags );
 638                                 } else {
 639                                         # Add literal brace(s)
 640                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
 641                                 }
 642                                 $i += $count;
 643                         } elseif ( $found == 'close' ) {
 644                                 $piece = $stack->top;
 645                                 # lets check if there are enough characters for closing brace
 646                                 $maxCount = $piece->count;
 647                                 $count = strspn( $text, $curChar, $i, $maxCount );
 648
 649                                 # check for maximum matching characters (if there are 5 closing
 650                                 # characters, we will probably need only 3 - depending on the rules)
 651                                 $rule = $rules[$piece->open];
 652                                 if ( $count > $rule['max'] ) {
 653                                         # The specified maximum exists in the callback array, unless the caller
 654                                         # has made an error
 655                                         $matchingCount = $rule['max'];
 656                                 } else {
 657                                         # Count is less than the maximum
 658                                         # Skip any gaps in the callback array to find the true largest match
 659                                         # Need to use array_key_exists not isset because the callback can be null
 660                                         $matchingCount = $count;
 661                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
 662                                                 --$matchingCount;
 663                                         }
 664                                 }
 665
 666                                 if ( $matchingCount <= 0 ) {
 667                                         # No matching element found in callback array
 668                                         # Output a literal closing brace and continue
 669                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
 670                                         $i += $count;
 671                                         continue;
 672                                 }
 673                                 $name = $rule['names'][$matchingCount];
 674                                 if ( $name === null ) {
 675                                         // No element, just literal text
 676                                         $element = $piece->breakSyntax( $matchingCount ) . str_repeat( $rule['end'], $matchingCount );
 677                                 } else {
 678                                         # Create XML element
 679                                         # Note: $parts is already XML, does not need to be encoded further
 680                                         $parts = $piece->parts;
 681                                         $title = $parts[0]->out;
 682                                         unset( $parts[0] );
 683
 684                                         # The invocation is at the start of the line if lineStart is set in
 685                                         # the stack, and all opening brackets are used up.
 686                                         if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
 687                                                 $attr = ' lineStart="1"';
 688                                         } else {
 689                                                 $attr = '';
 690                                         }
 691
 692                                         $element = "<$name$attr>";
 693                                         $element .= "<title>$title</title>";
 694                                         $argIndex = 1;
 695                                         foreach ( $parts as $part ) {
 696                                                 if ( isset( $part->eqpos ) ) {
 697                                                         $argName = substr( $part->out, 0, $part->eqpos );
 698                                                         $argValue = substr( $part->out, $part->eqpos + 1 );
 699                                                         $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
 700                                                 } else {
 701                                                         $element .= "<part><name index=\"$argIndex\" /><value>{$part->out}</value></part>";
 702                                                         $argIndex++;
 703                                                 }
 704                                         }
 705                                         $element .= "</$name>";
 706                                 }
 707
 708                                 # Advance input pointer
 709                                 $i += $matchingCount;
 710
 711                                 # Unwind the stack
 712                                 $stack->pop();
 713                                 $accum =& $stack->getAccum();
 714
 715                                 # Re-add the old stack element if it still has unmatched opening characters remaining
 716                                 if ( $matchingCount < $piece->count ) {
 717                                         $piece->parts = array( new PPDPart );
 718                                         $piece->count -= $matchingCount;
 719                                         # do we still qualify for any callback with remaining count?
 720                                         $min = $rules[$piece->open]['min'];
 721                                         if ( $piece->count >= $min ) {
 722                                                 $stack->push( $piece );
 723                                                 $accum =& $stack->getAccum();
 724                                         } else {
 725                                                 $accum .= str_repeat( $piece->open, $piece->count );
 726                                         }
 727                                 }
 728                                 $flags = $stack->getFlags();
 729                                 extract( $flags );
 730
 731                                 # Add XML element to the enclosing accumulator
 732                                 $accum .= $element;
 733                         } elseif ( $found == 'pipe' ) {
 734                                 $findEquals = true; // shortcut for getFlags()
 735                                 $stack->addPart();
 736                                 $accum =& $stack->getAccum();
 737                                 ++$i;
 738                         } elseif ( $found == 'equals' ) {
 739                                 $findEquals = false; // shortcut for getFlags()
 740                                 $stack->getCurrentPart()->eqpos = strlen( $accum );
 741                                 $accum .= '=';
 742                                 ++$i;
 743                         }
 744                 }
 745
 746                 # Output any remaining unclosed brackets
 747                 foreach ( $stack->stack as $piece ) {
 748                         $stack->rootAccum .= $piece->breakSyntax();
 749                 }
 750                 $stack->rootAccum .= '</root>';
 751                 $xml = $stack->rootAccum;
 752
 753                 return $xml;
 754         }
 755 }
 756
 757 /**
 758  * Stack class to help Preprocessor::preprocessToObj()
 759  * @ingroup Parser
 760  */
 761 class PPDStack {
 762         public $stack, $rootAccum;
 763
 764         /**
 765          * @var PPDStack
 766          */
 767         public $top;
 768         public $out;
 769         public $elementClass = 'PPDStackElement';
 770
 771         public static $false = false;
 772
 773         public function __construct() {
 774                 $this->stack = array();
 775                 $this->top = false;
 776                 $this->rootAccum = '';
 777                 $this->accum =& $this->rootAccum;
 778         }
 779
 780         /**
 781          * @return int
 782          */
 783         public function count() {
 784                 return count( $this->stack );
 785         }
 786
 787         public function &getAccum() {
 788                 return $this->accum;
 789         }
 790
 791         public function getCurrentPart() {
 792                 if ( $this->top === false ) {
 793                         return false;
 794                 } else {
 795                         return $this->top->getCurrentPart();
 796                 }
 797         }
 798
 799         public function push( $data ) {
 800                 if ( $data instanceof $this->elementClass ) {
 801                         $this->stack[] = $data;
 802                 } else {
 803                         $class = $this->elementClass;
 804                         $this->stack[] = new $class( $data );
 805                 }
 806                 $this->top = $this->stack[count( $this->stack ) - 1];
 807                 $this->accum =& $this->top->getAccum();
 808         }
 809
 810         public function pop() {
 811                 if ( !count( $this->stack ) ) {
 812                         throw new MWException( __METHOD__ . ': no elements remaining' );
 813                 }
 814                 $temp = array_pop( $this->stack );
 815
 816                 if ( count( $this->stack ) ) {
 817                         $this->top = $this->stack[count( $this->stack ) - 1];
 818                         $this->accum =& $this->top->getAccum();
 819                 } else {
 820                         $this->top = self::$false;
 821                         $this->accum =& $this->rootAccum;
 822                 }
 823                 return $temp;
 824         }
 825
 826         public function addPart( $s = '' ) {
 827                 $this->top->addPart( $s );
 828                 $this->accum =& $this->top->getAccum();
 829         }
 830
 831         /**
 832          * @return array
 833          */
 834         public function getFlags() {
 835                 if ( !count( $this->stack ) ) {
 836                         return array(
 837                                 'findEquals' => false,
 838                                 'findPipe' => false,
 839                                 'inHeading' => false,
 840                         );
 841                 } else {
 842                         return $this->top->getFlags();
 843                 }
 844         }
 845 }
 846
 847 /**
 848  * @ingroup Parser
 849  */
 850 class PPDStackElement {
 851         /**
 852          * @var string Opening character (\n for heading)
 853          */
 854         public $open;
 855
 856         /**
 857          * @var string Matching closing character
 858          */
 859         public $close;
 860
 861         /**
 862          * @var int Number of opening characters found (number of "=" for heading)
 863          */
 864         public $count;
 865
 866         /**
 867          * @var PPDPart[] Array of PPDPart objects describing pipe-separated parts.
 868          */
 869         public $parts;
 870
 871         /**
 872          * @var bool True if the open char appeared at the start of the input line.
 873          *  Not set for headings.
 874          */
 875         public $lineStart;
 876
 877         public $partClass = 'PPDPart';
 878
 879         public function __construct( $data = array() ) {
 880                 $class = $this->partClass;
 881                 $this->parts = array( new $class );
 882
 883                 foreach ( $data as $name => $value ) {
 884                         $this->$name = $value;
 885                 }
 886         }
 887
 888         public function &getAccum() {
 889                 return $this->parts[count( $this->parts ) - 1]->out;
 890         }
 891
 892         public function addPart( $s = '' ) {
 893                 $class = $this->partClass;
 894                 $this->parts[] = new $class( $s );
 895         }
 896
 897         public function getCurrentPart() {
 898                 return $this->parts[count( $this->parts ) - 1];
 899         }
 900
 901         /**
 902          * @return array
 903          */
 904         public function getFlags() {
 905                 $partCount = count( $this->parts );
 906                 $findPipe = $this->open != "\n" && $this->open != '[';
 907                 return array(
 908                         'findPipe' => $findPipe,
 909                         'findEquals' => $findPipe && $partCount > 1 && !isset( $this->parts[$partCount - 1]->eqpos ),
 910                         'inHeading' => $this->open == "\n",
 911                 );
 912         }
 913
 914         /**
 915          * Get the output string that would result if the close is not found.
 916          *
 917          * @param bool|int $openingCount
 918          * @return string
 919          */
 920         public function breakSyntax( $openingCount = false ) {
 921                 if ( $this->open == "\n" ) {
 922                         $s = $this->parts[0]->out;
 923                 } else {
 924                         if ( $openingCount === false ) {
 925                                 $openingCount = $this->count;
 926                         }
 927                         $s = str_repeat( $this->open, $openingCount );
 928                         $first = true;
 929                         foreach ( $this->parts as $part ) {
 930                                 if ( $first ) {
 931                                         $first = false;
 932                                 } else {
 933                                         $s .= '|';
 934                                 }
 935                                 $s .= $part->out;
 936                         }
 937                 }
 938                 return $s;
 939         }
 940 }
 941
 942 /**
 943  * @ingroup Parser
 944  */
 945 class PPDPart {
 946         /**
 947          * @var string Output accumulator string
 948          */
 949         public $out;
 950
 951         // Optional member variables:
 952         //   eqpos        Position of equals sign in output accumulator
 953         //   commentEnd   Past-the-end input pointer for the last comment encountered
 954         //   visualEnd    Past-the-end input pointer for the end of the accumulator minus comments
 955
 956         public function __construct( $out = '' ) {
 957                 $this->out = $out;
 958         }
 959 }
 960
 961 /**
 962  * An expansion frame, used as a context to expand the result of preprocessToObj()
 963  * @ingroup Parser
 964  * @codingStandardsIgnoreStart
 965  */
 966 class PPFrame_DOM implements PPFrame {
 967         // @codingStandardsIgnoreEnd
 968
 969         /**
 970          * @var Preprocessor
 971          */
 972         public $preprocessor;
 973
 974         /**
 975          * @var Parser
 976          */
 977         public $parser;
 978
 979         /**
 980          * @var Title
 981          */
 982         public $title;
 983         public $titleCache;
 984
 985         /**
 986          * Hashtable listing templates which are disallowed for expansion in this frame,
 987          * having been encountered previously in parent frames.
 988          */
 989         public $loopCheckHash;
 990
 991         /**
 992          * Recursion depth of this frame, top = 0
 993          * Note that this is NOT the same as expansion depth in expand()
 994          */
 995         public $depth;
 996
 997         private $volatile = false;
 998         private $ttl = null;
 999
1000         /**
1001          * @var array
1002          */
1003         protected $childExpansionCache;
1004
1005         /**
1006          * Construct a new preprocessor frame.
1007          * @param Preprocessor $preprocessor The parent preprocessor
1008          */
1009         public function __construct( $preprocessor ) {
1010                 $this->preprocessor = $preprocessor;
1011                 $this->parser = $preprocessor->parser;
1012                 $this->title = $this->parser->mTitle;
1013                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
1014                 $this->loopCheckHash = array();
1015                 $this->depth = 0;
1016                 $this->childExpansionCache = array();
1017         }
1018
1019         /**
1020          * Create a new child frame
1021          * $args is optionally a multi-root PPNode or array containing the template arguments
1022          *
1023          * @param bool|array $args
1024          * @param Title|bool $title
1025          * @param int $indexOffset
1026          * @return PPTemplateFrame_DOM
1027          */
1028         public function newChild( $args = false, $title = false, $indexOffset = 0 ) {
1029                 $namedArgs = array();
1030                 $numberedArgs = array();
1031                 if ( $title === false ) {
1032                         $title = $this->title;
1033                 }
1034                 if ( $args !== false ) {
1035                         $xpath = false;
1036                         if ( $args instanceof PPNode ) {
1037                                 $args = $args->node;
1038                         }
1039                         foreach ( $args as $arg ) {
1040                                 if ( $arg instanceof PPNode ) {
1041                                         $arg = $arg->node;
1042                                 }
1043                                 if ( !$xpath || $xpath->document !== $arg->ownerDocument ) {
1044                                         $xpath = new DOMXPath( $arg->ownerDocument );
1045                                 }
1046
1047                                 $nameNodes = $xpath->query( 'name', $arg );
1048                                 $value = $xpath->query( 'value', $arg );
1049                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
1050                                         // Numbered parameter
1051                                         $index = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
1052                                         $index = $index - $indexOffset;
1053                                         if ( isset( $namedArgs[$index] ) || isset( $numberedArgs[$index] ) ) {
1054                                                 $this->parser->getOutput()->addWarning( wfMessage( 'duplicate-args-warning',
1055                                                         wfEscapeWikiText( $this->title ),
1056                                                         wfEscapeWikiText( $title ),
1057                                                         wfEscapeWikiText( $index ) )->text() );
1058                                                 $this->parser->addTrackingCategory( 'duplicate-args-category' );
1059                                         }
1060                                         $numberedArgs[$index] = $value->item( 0 );
1061                                         unset( $namedArgs[$index] );
1062                                 } else {
1063                                         // Named parameter
1064                                         $name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS ) );
1065                                         if ( isset( $namedArgs[$name] ) || isset( $numberedArgs[$name] ) ) {
1066                                                 $this->parser->getOutput()->addWarning( wfMessage( 'duplicate-args-warning',
1067                                                         wfEscapeWikiText( $this->title ),
1068                                                         wfEscapeWikiText( $title ),
1069                                                         wfEscapeWikiText( $name ) )->text() );
1070                                                 $this->parser->addTrackingCategory( 'duplicate-args-category' );
1071                                         }
1072                                         $namedArgs[$name] = $value->item( 0 );
1073                                         unset( $numberedArgs[$name] );
1074                                 }
1075                         }
1076                 }
1077                 return new PPTemplateFrame_DOM( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
1078         }
1079
1080         /**
1081          * @throws MWException
1082          * @param string|int $key
1083          * @param string|PPNode_DOM|DOMDocument $root
1084          * @param int $flags
1085          * @return string
1086          */
1087         public function cachedExpand( $key, $root, $flags = 0 ) {
1088                 // we don't have a parent, so we don't have a cache
1089                 return $this->expand( $root, $flags );
1090         }
1091
1092         /**
1093          * @throws MWException
1094          * @param string|PPNode_DOM|DOMDocument $root
1095          * @param int $flags
1096          * @return string
1097          */
1098         public function expand( $root, $flags = 0 ) {
1099                 static $expansionDepth = 0;
1100                 if ( is_string( $root ) ) {
1101                         return $root;
1102                 }
1103
1104                 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) {
1105                         $this->parser->limitationWarn( 'node-count-exceeded',
1106                                 $this->parser->mPPNodeCount,
1107                                 $this->parser->mOptions->getMaxPPNodeCount()
1108                         );
1109                         return '<span class="error">Node-count limit exceeded</span>';
1110                 }
1111
1112                 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
1113                         $this->parser->limitationWarn( 'expansion-depth-exceeded',
1114                                 $expansionDepth,
1115                                 $this->parser->mOptions->getMaxPPExpandDepth()
1116                         );
1117                         return '<span class="error">Expansion depth limit exceeded</span>';
1118                 }
1119                 ++$expansionDepth;
1120                 if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
1121                         $this->parser->mHighestExpansionDepth = $expansionDepth;
1122                 }
1123
1124                 if ( $root instanceof PPNode_DOM ) {
1125                         $root = $root->node;
1126                 }
1127                 if ( $root instanceof DOMDocument ) {
1128                         $root = $root->documentElement;
1129                 }
1130
1131                 $outStack = array( '', '' );
1132                 $iteratorStack = array( false, $root );
1133                 $indexStack = array( 0, 0 );
1134
1135                 while ( count( $iteratorStack ) > 1 ) {
1136                         $level = count( $outStack ) - 1;
1137                         $iteratorNode =& $iteratorStack[$level];
1138                         $out =& $outStack[$level];
1139                         $index =& $indexStack[$level];
1140
1141                         if ( $iteratorNode instanceof PPNode_DOM ) {
1142                                 $iteratorNode = $iteratorNode->node;
1143                         }
1144
1145                         if ( is_array( $iteratorNode ) ) {
1146                                 if ( $index >= count( $iteratorNode ) ) {
1147                                         // All done with this iterator
1148                                         $iteratorStack[$level] = false;
1149                                         $contextNode = false;
1150                                 } else {
1151                                         $contextNode = $iteratorNode[$index];
1152                                         $index++;
1153                                 }
1154                         } elseif ( $iteratorNode instanceof DOMNodeList ) {
1155                                 if ( $index >= $iteratorNode->length ) {
1156                                         // All done with this iterator
1157                                         $iteratorStack[$level] = false;
1158                                         $contextNode = false;
1159                                 } else {
1160                                         $contextNode = $iteratorNode->item( $index );
1161                                         $index++;
1162                                 }
1163                         } else {
1164                                 // Copy to $contextNode and then delete from iterator stack,
1165                                 // because this is not an iterator but we do have to execute it once
1166                                 $contextNode = $iteratorStack[$level];
1167                                 $iteratorStack[$level] = false;
1168                         }
1169
1170                         if ( $contextNode instanceof PPNode_DOM ) {
1171                                 $contextNode = $contextNode->node;
1172                         }
1173
1174                         $newIterator = false;
1175
1176                         if ( $contextNode === false ) {
1177                                 // nothing to do
1178                         } elseif ( is_string( $contextNode ) ) {
1179                                 $out .= $contextNode;
1180                         } elseif ( is_array( $contextNode ) || $contextNode instanceof DOMNodeList ) {
1181                                 $newIterator = $contextNode;
1182                         } elseif ( $contextNode instanceof DOMNode ) {
1183                                 if ( $contextNode->nodeType == XML_TEXT_NODE ) {
1184                                         $out .= $contextNode->nodeValue;
1185                                 } elseif ( $contextNode->nodeName == 'template' ) {
1186                                         # Double-brace expansion
1187                                         $xpath = new DOMXPath( $contextNode->ownerDocument );
1188                                         $titles = $xpath->query( 'title', $contextNode );
1189                                         $title = $titles->item( 0 );
1190                                         $parts = $xpath->query( 'part', $contextNode );
1191                                         if ( $flags & PPFrame::NO_TEMPLATES ) {
1192                                                 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $title, $parts );
1193                                         } else {
1194                                                 $lineStart = $contextNode->getAttribute( 'lineStart' );
1195                                                 $params = array(
1196                                                         'title' => new PPNode_DOM( $title ),
1197                                                         'parts' => new PPNode_DOM( $parts ),
1198                                                         'lineStart' => $lineStart );
1199                                                 $ret = $this->parser->braceSubstitution( $params, $this );
1200                                                 if ( isset( $ret['object'] ) ) {
1201                                                         $newIterator = $ret['object'];
1202                                                 } else {
1203                                                         $out .= $ret['text'];
1204                                                 }
1205                                         }
1206                                 } elseif ( $contextNode->nodeName == 'tplarg' ) {
1207                                         # Triple-brace expansion
1208                                         $xpath = new DOMXPath( $contextNode->ownerDocument );
1209                                         $titles = $xpath->query( 'title', $contextNode );
1210                                         $title = $titles->item( 0 );
1211                                         $parts = $xpath->query( 'part', $contextNode );
1212                                         if ( $flags & PPFrame::NO_ARGS ) {
1213                                                 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $title, $parts );
1214                                         } else {
1215                                                 $params = array(
1216                                                         'title' => new PPNode_DOM( $title ),
1217                                                         'parts' => new PPNode_DOM( $parts ) );
1218                                                 $ret = $this->parser->argSubstitution( $params, $this );
1219                                                 if ( isset( $ret['object'] ) ) {
1220                                                         $newIterator = $ret['object'];
1221                                                 } else {
1222                                                         $out .= $ret['text'];
1223                                                 }
1224                                         }
1225                                 } elseif ( $contextNode->nodeName == 'comment' ) {
1226                                         # HTML-style comment
1227                                         # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1228                                         # Not in RECOVER_COMMENTS mode (msgnw) though.
1229                                         if ( ( $this->parser->ot['html']
1230                                                 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
1231                                                 || ( $flags & PPFrame::STRIP_COMMENTS )
1232                                                 ) && !( $flags & PPFrame::RECOVER_COMMENTS )
1233                                         ) {
1234                                                 $out .= '';
1235                                         } elseif ( $this->parser->ot['wiki'] && !( $flags & PPFrame::RECOVER_COMMENTS ) ) {
1236                                                 # Add a strip marker in PST mode so that pstPass2() can
1237                                                 # run some old-fashioned regexes on the result.
1238                                                 # Not in RECOVER_COMMENTS mode (extractSections) though.
1239                                                 $out .= $this->parser->insertStripItem( $contextNode->textContent );
1240                                         } else {
1241                                                 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1242                                                 $out .= $contextNode->textContent;
1243                                         }
1244                                 } elseif ( $contextNode->nodeName == 'ignore' ) {
1245                                         # Output suppression used by <includeonly> etc.
1246                                         # OT_WIKI will only respect <ignore> in substed templates.
1247                                         # The other output types respect it unless NO_IGNORE is set.
1248                                         # extractSections() sets NO_IGNORE and so never respects it.
1249                                         if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] )
1250                                                 || ( $flags & PPFrame::NO_IGNORE )
1251                                         ) {
1252                                                 $out .= $contextNode->textContent;
1253                                         } else {
1254                                                 $out .= '';
1255                                         }
1256                                 } elseif ( $contextNode->nodeName == 'ext' ) {
1257                                         # Extension tag
1258                                         $xpath = new DOMXPath( $contextNode->ownerDocument );
1259                                         $names = $xpath->query( 'name', $contextNode );
1260                                         $attrs = $xpath->query( 'attr', $contextNode );
1261                                         $inners = $xpath->query( 'inner', $contextNode );
1262                                         $closes = $xpath->query( 'close', $contextNode );
1263                                         if ( $flags & PPFrame::NO_TAGS ) {
1264                                                 $s = '<' . $this->expand( $names->item( 0 ), $flags );
1265                                                 if ( $attrs->length > 0 ) {
1266                                                         $s .= $this->expand( $attrs->item( 0 ), $flags );
1267                                                 }
1268                                                 if ( $inners->length > 0 ) {
1269                                                         $s .= '>' . $this->expand( $inners->item( 0 ), $flags );
1270                                                         if ( $closes->length > 0 ) {
1271                                                                 $s .= $this->expand( $closes->item( 0 ), $flags );
1272                                                         }
1273                                                 } else {
1274                                                         $s .= '/>';
1275                                                 }
1276                                                 $out .= $s;
1277                                         } else {
1278                                                 $params = array(
1279                                                         'name' => new PPNode_DOM( $names->item( 0 ) ),
1280                                                         'attr' => $attrs->length > 0 ? new PPNode_DOM( $attrs->item( 0 ) ) : null,
1281                                                         'inner' => $inners->length > 0 ? new PPNode_DOM( $inners->item( 0 ) ) : null,
1282                                                         'close' => $closes->length > 0 ? new PPNode_DOM( $closes->item( 0 ) ) : null,
1283                                                 );
1284                                                 $out .= $this->parser->extensionSubstitution( $params, $this );
1285                                         }
1286                                 } elseif ( $contextNode->nodeName == 'h' ) {
1287                                         # Heading
1288                                         $s = $this->expand( $contextNode->childNodes, $flags );
1289
1290                                         # Insert a heading marker only for <h> children of <root>
1291                                         # This is to stop extractSections from going over multiple tree levels
1292                                         if ( $contextNode->parentNode->nodeName == 'root' && $this->parser->ot['html'] ) {
1293                                                 # Insert heading index marker
1294                                                 $headingIndex = $contextNode->getAttribute( 'i' );
1295                                                 $titleText = $this->title->getPrefixedDBkey();
1296                                                 $this->parser->mHeadings[] = array( $titleText, $headingIndex );
1297                                                 $serial = count( $this->parser->mHeadings ) - 1;
1298                                                 $marker = Parser::MARKER_PREFIX . "-h-$serial-" . Parser::MARKER_SUFFIX;
1299                                                 $count = $contextNode->getAttribute( 'level' );
1300                                                 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
1301                                                 $this->parser->mStripState->addGeneral( $marker, '' );
1302                                         }
1303                                         $out .= $s;
1304                                 } else {
1305                                         # Generic recursive expansion
1306                                         $newIterator = $contextNode->childNodes;
1307                                 }
1308                         } else {
1309                                 throw new MWException( __METHOD__ . ': Invalid parameter type' );
1310                         }
1311
1312                         if ( $newIterator !== false ) {
1313                                 if ( $newIterator instanceof PPNode_DOM ) {
1314                                         $newIterator = $newIterator->node;
1315                                 }
1316                                 $outStack[] = '';
1317                                 $iteratorStack[] = $newIterator;
1318                                 $indexStack[] = 0;
1319                         } elseif ( $iteratorStack[$level] === false ) {
1320                                 // Return accumulated value to parent
1321                                 // With tail recursion
1322                                 while ( $iteratorStack[$level] === false && $level > 0 ) {
1323                                         $outStack[$level - 1] .= $out;
1324                                         array_pop( $outStack );
1325                                         array_pop( $iteratorStack );
1326                                         array_pop( $indexStack );
1327                                         $level--;
1328                                 }
1329                         }
1330                 }
1331                 --$expansionDepth;
1332                 return $outStack[0];
1333         }
1334
1335         /**
1336          * @param string $sep
1337          * @param int $flags
1338          * @param string|PPNode_DOM|DOMDocument $args,...
1339          * @return string
1340          */
1341         public function implodeWithFlags( $sep, $flags /*, ... */ ) {
1342                 $args = array_slice( func_get_args(), 2 );
1343
1344                 $first = true;
1345                 $s = '';
1346                 foreach ( $args as $root ) {
1347                         if ( $root instanceof PPNode_DOM ) {
1348                                 $root = $root->node;
1349                         }
1350                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1351                                 $root = array( $root );
1352                         }
1353                         foreach ( $root as $node ) {
1354                                 if ( $first ) {
1355                                         $first = false;
1356                                 } else {
1357                                         $s .= $sep;
1358                                 }
1359                                 $s .= $this->expand( $node, $flags );
1360                         }
1361                 }
1362                 return $s;
1363         }
1364
1365         /**
1366          * Implode with no flags specified
1367          * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1368          *
1369          * @param string $sep
1370          * @param string|PPNode_DOM|DOMDocument $args,...
1371          * @return string
1372          */
1373         public function implode( $sep /*, ... */ ) {
1374                 $args = array_slice( func_get_args(), 1 );
1375
1376                 $first = true;
1377                 $s = '';
1378                 foreach ( $args as $root ) {
1379                         if ( $root instanceof PPNode_DOM ) {
1380                                 $root = $root->node;
1381                         }
1382                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1383                                 $root = array( $root );
1384                         }
1385                         foreach ( $root as $node ) {
1386                                 if ( $first ) {
1387                                         $first = false;
1388                                 } else {
1389                                         $s .= $sep;
1390                                 }
1391                                 $s .= $this->expand( $node );
1392                         }
1393                 }
1394                 return $s;
1395         }
1396
1397         /**
1398          * Makes an object that, when expand()ed, will be the same as one obtained
1399          * with implode()
1400          *
1401          * @param string $sep
1402          * @param string|PPNode_DOM|DOMDocument $args,...
1403          * @return array
1404          */
1405         public function virtualImplode( $sep /*, ... */ ) {
1406                 $args = array_slice( func_get_args(), 1 );
1407                 $out = array();
1408                 $first = true;
1409
1410                 foreach ( $args as $root ) {
1411                         if ( $root instanceof PPNode_DOM ) {
1412                                 $root = $root->node;
1413                         }
1414                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1415                                 $root = array( $root );
1416                         }
1417                         foreach ( $root as $node ) {
1418                                 if ( $first ) {
1419                                         $first = false;
1420                                 } else {
1421                                         $out[] = $sep;
1422                                 }
1423                                 $out[] = $node;
1424                         }
1425                 }
1426                 return $out;
1427         }
1428
1429         /**
1430          * Virtual implode with brackets
1431          * @param string $start
1432          * @param string $sep
1433          * @param string $end
1434          * @param string|PPNode_DOM|DOMDocument $args,...
1435          * @return array
1436          */
1437         public function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1438                 $args = array_slice( func_get_args(), 3 );
1439                 $out = array( $start );
1440                 $first = true;
1441
1442                 foreach ( $args as $root ) {
1443                         if ( $root instanceof PPNode_DOM ) {
1444                                 $root = $root->node;
1445                         }
1446                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1447                                 $root = array( $root );
1448                         }
1449                         foreach ( $root as $node ) {
1450                                 if ( $first ) {
1451                                         $first = false;
1452                                 } else {
1453                                         $out[] = $sep;
1454                                 }
1455                                 $out[] = $node;
1456                         }
1457                 }
1458                 $out[] = $end;
1459                 return $out;
1460         }
1461
1462         public function __toString() {
1463                 return 'frame{}';
1464         }
1465
1466         public function getPDBK( $level = false ) {
1467                 if ( $level === false ) {
1468                         return $this->title->getPrefixedDBkey();
1469                 } else {
1470                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1471                 }
1472         }
1473
1474         /**
1475          * @return array
1476          */
1477         public function getArguments() {
1478                 return array();
1479         }
1480
1481         /**
1482          * @return array
1483          */
1484         public function getNumberedArguments() {
1485                 return array();
1486         }
1487
1488         /**
1489          * @return array
1490          */
1491         public function getNamedArguments() {
1492                 return array();
1493         }
1494
1495         /**
1496          * Returns true if there are no arguments in this frame
1497          *
1498          * @return bool
1499          */
1500         public function isEmpty() {
1501                 return true;
1502         }
1503
1504         public function getArgument( $name ) {
1505                 return false;
1506         }
1507
1508         /**
1509          * Returns true if the infinite loop check is OK, false if a loop is detected
1510          *
1511          * @param Title $title
1512          * @return bool
1513          */
1514         public function loopCheck( $title ) {
1515                 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1516         }
1517
1518         /**
1519          * Return true if the frame is a template frame
1520          *
1521          * @return bool
1522          */
1523         public function isTemplate() {
1524                 return false;
1525         }
1526
1527         /**
1528          * Get a title of frame
1529          *
1530          * @return Title
1531          */
1532         public function getTitle() {
1533                 return $this->title;
1534         }
1535
1536         /**
1537          * Set the volatile flag
1538          *
1539          * @param bool $flag
1540          */
1541         public function setVolatile( $flag = true ) {
1542                 $this->volatile = $flag;
1543         }
1544
1545         /**
1546          * Get the volatile flag
1547          *
1548          * @return bool
1549          */
1550         public function isVolatile() {
1551                 return $this->volatile;
1552         }
1553
1554         /**
1555          * Set the TTL
1556          *
1557          * @param int $ttl
1558          */
1559         public function setTTL( $ttl ) {
1560                 if ( $ttl !== null && ( $this->ttl === null || $ttl < $this->ttl ) ) {
1561                         $this->ttl = $ttl;
1562                 }
1563         }
1564
1565         /**
1566          * Get the TTL
1567          *
1568          * @return int|null
1569          */
1570         public function getTTL() {
1571                 return $this->ttl;
1572         }
1573 }
1574
1575 /**
1576  * Expansion frame with template arguments
1577  * @ingroup Parser
1578  * @codingStandardsIgnoreStart
1579  */
1580 class PPTemplateFrame_DOM extends PPFrame_DOM {
1581         // @codingStandardsIgnoreEnd
1582
1583         public $numberedArgs, $namedArgs;
1584
1585         /**
1586          * @var PPFrame_DOM
1587          */
1588         public $parent;
1589         public $numberedExpansionCache, $namedExpansionCache;
1590
1591         /**
1592          * @param Preprocessor $preprocessor
1593          * @param bool|PPFrame_DOM $parent
1594          * @param array $numberedArgs
1595          * @param array $namedArgs
1596          * @param bool|Title $title
1597          */
1598         public function __construct( $preprocessor, $parent = false, $numberedArgs = array(),
1599                 $namedArgs = array(), $title = false
1600         ) {
1601                 parent::__construct( $preprocessor );
1602
1603                 $this->parent = $parent;
1604                 $this->numberedArgs = $numberedArgs;
1605                 $this->namedArgs = $namedArgs;
1606                 $this->title = $title;
1607                 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1608                 $this->titleCache = $parent->titleCache;
1609                 $this->titleCache[] = $pdbk;
1610                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1611                 if ( $pdbk !== false ) {
1612                         $this->loopCheckHash[$pdbk] = true;
1613                 }
1614                 $this->depth = $parent->depth + 1;
1615                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1616         }
1617
1618         public function __toString() {
1619                 $s = 'tplframe{';
1620                 $first = true;
1621                 $args = $this->numberedArgs + $this->namedArgs;
1622                 foreach ( $args as $name => $value ) {
1623                         if ( $first ) {
1624                                 $first = false;
1625                         } else {
1626                                 $s .= ', ';
1627                         }
1628                         $s .= "\"$name\":\"" .
1629                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
1630                 }
1631                 $s .= '}';
1632                 return $s;
1633         }
1634
1635         /**
1636          * @throws MWException
1637          * @param string|int $key
1638          * @param string|PPNode_DOM|DOMDocument $root
1639          * @param int $flags
1640          * @return string
1641          */
1642         public function cachedExpand( $key, $root, $flags = 0 ) {
1643                 if ( isset( $this->parent->childExpansionCache[$key] ) ) {
1644                         return $this->parent->childExpansionCache[$key];
1645                 }
1646                 $retval = $this->expand( $root, $flags );
1647                 if ( !$this->isVolatile() ) {
1648                         $this->parent->childExpansionCache[$key] = $retval;
1649                 }
1650                 return $retval;
1651         }
1652
1653         /**
1654          * Returns true if there are no arguments in this frame
1655          *
1656          * @return bool
1657          */
1658         public function isEmpty() {
1659                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1660         }
1661
1662         public function getArguments() {
1663                 $arguments = array();
1664                 foreach ( array_merge(
1665                                 array_keys( $this->numberedArgs ),
1666                                 array_keys( $this->namedArgs ) ) as $key ) {
1667                         $arguments[$key] = $this->getArgument( $key );
1668                 }
1669                 return $arguments;
1670         }
1671
1672         public function getNumberedArguments() {
1673                 $arguments = array();
1674                 foreach ( array_keys( $this->numberedArgs ) as $key ) {
1675                         $arguments[$key] = $this->getArgument( $key );
1676                 }
1677                 return $arguments;
1678         }
1679
1680         public function getNamedArguments() {
1681                 $arguments = array();
1682                 foreach ( array_keys( $this->namedArgs ) as $key ) {
1683                         $arguments[$key] = $this->getArgument( $key );
1684                 }
1685                 return $arguments;
1686         }
1687
1688         public function getNumberedArgument( $index ) {
1689                 if ( !isset( $this->numberedArgs[$index] ) ) {
1690                         return false;
1691                 }
1692                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1693                         # No trimming for unnamed arguments
1694                         $this->numberedExpansionCache[$index] = $this->parent->expand(
1695                                 $this->numberedArgs[$index],
1696                                 PPFrame::STRIP_COMMENTS
1697                         );
1698                 }
1699                 return $this->numberedExpansionCache[$index];
1700         }
1701
1702         public function getNamedArgument( $name ) {
1703                 if ( !isset( $this->namedArgs[$name] ) ) {
1704                         return false;
1705                 }
1706                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1707                         # Trim named arguments post-expand, for backwards compatibility
1708                         $this->namedExpansionCache[$name] = trim(
1709                                 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1710                 }
1711                 return $this->namedExpansionCache[$name];
1712         }
1713
1714         public function getArgument( $name ) {
1715                 $text = $this->getNumberedArgument( $name );
1716                 if ( $text === false ) {
1717                         $text = $this->getNamedArgument( $name );
1718                 }
1719                 return $text;
1720         }
1721
1722         /**
1723          * Return true if the frame is a template frame
1724          *
1725          * @return bool
1726          */
1727         public function isTemplate() {
1728                 return true;
1729         }
1730
1731         public function setVolatile( $flag = true ) {
1732                 parent::setVolatile( $flag );
1733                 $this->parent->setVolatile( $flag );
1734         }
1735
1736         public function setTTL( $ttl ) {
1737                 parent::setTTL( $ttl );
1738                 $this->parent->setTTL( $ttl );
1739         }
1740 }
1741
1742 /**
1743  * Expansion frame with custom arguments
1744  * @ingroup Parser
1745  * @codingStandardsIgnoreStart
1746  */
1747 class PPCustomFrame_DOM extends PPFrame_DOM {
1748         // @codingStandardsIgnoreEnd
1749
1750         public $args;
1751
1752         public function __construct( $preprocessor, $args ) {
1753                 parent::__construct( $preprocessor );
1754                 $this->args = $args;
1755         }
1756
1757         public function __toString() {
1758                 $s = 'cstmframe{';
1759                 $first = true;
1760                 foreach ( $this->args as $name => $value ) {
1761                         if ( $first ) {
1762                                 $first = false;
1763                         } else {
1764                                 $s .= ', ';
1765                         }
1766                         $s .= "\"$name\":\"" .
1767                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1768                 }
1769                 $s .= '}';
1770                 return $s;
1771         }
1772
1773         /**
1774          * @return bool
1775          */
1776         public function isEmpty() {
1777                 return !count( $this->args );
1778         }
1779
1780         public function getArgument( $index ) {
1781                 if ( !isset( $this->args[$index] ) ) {
1782                         return false;
1783                 }
1784                 return $this->args[$index];
1785         }
1786
1787         public function getArguments() {
1788                 return $this->args;
1789         }
1790 }
1791
1792 /**
1793  * @ingroup Parser
1794  * @codingStandardsIgnoreStart
1795  */
1796 class PPNode_DOM implements PPNode {
1797         // @codingStandardsIgnoreEnd
1798
1799         /**
1800          * @var DOMElement
1801          */
1802         public $node;
1803         public $xpath;
1804
1805         public function __construct( $node, $xpath = false ) {
1806                 $this->node = $node;
1807         }
1808
1809         /**
1810          * @return DOMXPath
1811          */
1812         public function getXPath() {
1813                 if ( $this->xpath === null ) {
1814                         $this->xpath = new DOMXPath( $this->node->ownerDocument );
1815                 }
1816                 return $this->xpath;
1817         }
1818
1819         public function __toString() {
1820                 if ( $this->node instanceof DOMNodeList ) {
1821                         $s = '';
1822                         foreach ( $this->node as $node ) {
1823                                 $s .= $node->ownerDocument->saveXML( $node );
1824                         }
1825                 } else {
1826                         $s = $this->node->ownerDocument->saveXML( $this->node );
1827                 }
1828                 return $s;
1829         }
1830
1831         /**
1832          * @return bool|PPNode_DOM
1833          */
1834         public function getChildren() {
1835                 return $this->node->childNodes ? new self( $this->node->childNodes ) : false;
1836         }
1837
1838         /**
1839          * @return bool|PPNode_DOM
1840          */
1841         public function getFirstChild() {
1842                 return $this->node->firstChild ? new self( $this->node->firstChild ) : false;
1843         }
1844
1845         /**
1846          * @return bool|PPNode_DOM
1847          */
1848         public function getNextSibling() {
1849                 return $this->node->nextSibling ? new self( $this->node->nextSibling ) : false;
1850         }
1851
1852         /**
1853          * @param string $type
1854          *
1855          * @return bool|PPNode_DOM
1856          */
1857         public function getChildrenOfType( $type ) {
1858                 return new self( $this->getXPath()->query( $type, $this->node ) );
1859         }
1860
1861         /**
1862          * @return int
1863          */
1864         public function getLength() {
1865                 if ( $this->node instanceof DOMNodeList ) {
1866                         return $this->node->length;
1867                 } else {
1868                         return false;
1869                 }
1870         }
1871
1872         /**
1873          * @param int $i
1874          * @return bool|PPNode_DOM
1875          */
1876         public function item( $i ) {
1877                 $item = $this->node->item( $i );
1878                 return $item ? new self( $item ) : false;
1879         }
1880
1881         /**
1882          * @return string
1883          */
1884         public function getName() {
1885                 if ( $this->node instanceof DOMNodeList ) {
1886                         return '#nodelist';
1887                 } else {
1888                         return $this->node->nodeName;
1889                 }
1890         }
1891
1892         /**
1893          * Split a "<part>" node into an associative array containing:
1894          *  - name          PPNode name
1895          *  - index         String index
1896          *  - value         PPNode value
1897          *
1898          * @throws MWException
1899          * @return array
1900          */
1901         public function splitArg() {
1902                 $xpath = $this->getXPath();
1903                 $names = $xpath->query( 'name', $this->node );
1904                 $values = $xpath->query( 'value', $this->node );
1905                 if ( !$names->length || !$values->length ) {
1906                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1907                 }
1908                 $name = $names->item( 0 );
1909                 $index = $name->getAttribute( 'index' );
1910                 return array(
1911                         'name' => new self( $name ),
1912                         'index' => $index,
1913                         'value' => new self( $values->item( 0 ) ) );
1914         }
1915
1916         /**
1917          * Split an "<ext>" node into an associative array containing name, attr, inner and close
1918          * All values in the resulting array are PPNodes. Inner and close are optional.
1919          *
1920          * @throws MWException
1921          * @return array
1922          */
1923         public function splitExt() {
1924                 $xpath = $this->getXPath();
1925                 $names = $xpath->query( 'name', $this->node );
1926                 $attrs = $xpath->query( 'attr', $this->node );
1927                 $inners = $xpath->query( 'inner', $this->node );
1928                 $closes = $xpath->query( 'close', $this->node );
1929                 if ( !$names->length || !$attrs->length ) {
1930                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1931                 }
1932                 $parts = array(
1933                         'name' => new self( $names->item( 0 ) ),
1934                         'attr' => new self( $attrs->item( 0 ) ) );
1935                 if ( $inners->length ) {
1936                         $parts['inner'] = new self( $inners->item( 0 ) );
1937                 }
1938                 if ( $closes->length ) {
1939                         $parts['close'] = new self( $closes->item( 0 ) );
1940                 }
1941                 return $parts;
1942         }
1943
1944         /**
1945          * Split a "<h>" node
1946          * @throws MWException
1947          * @return array
1948          */
1949         public function splitHeading() {
1950                 if ( $this->getName() !== 'h' ) {
1951                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1952                 }
1953                 return array(
1954                         'i' => $this->node->getAttribute( 'i' ),
1955                         'level' => $this->node->getAttribute( 'level' ),
1956                         'contents' => $this->getChildren()
1957                 );
1958         }
1959 }