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