includes/parser/Preprocessor_DOM.php

   1 <?php
   2 /**
   3  * Preprocessor using PHP's dom extension
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Parser
  22  */
  23
  24 /**
  25  * @ingroup Parser
  26  * @codingStandardsIgnoreStart
  27  */
  28 class Preprocessor_DOM implements Preprocessor {
  29         // @codingStandardsIgnoreEnd
  30
  31         /**
  32          * @var Parser
  33          */
  34         public $parser;
  35
  36         public $memoryLimit;
  37
  38         const CACHE_VERSION = 1;
  39
  40         public function __construct( $parser ) {
  41                 $this->parser = $parser;
  42                 $mem = ini_get( 'memory_limit' );
  43                 $this->memoryLimit = false;
  44                 if ( strval( $mem ) !== '' && $mem != -1 ) {
  45                         if ( preg_match( '/^\d+$/', $mem ) ) {
  46                                 $this->memoryLimit = $mem;
  47                         } elseif ( preg_match( '/^(\d+)M$/i', $mem, $m ) ) {
  48                                 $this->memoryLimit = $m[1] * 1048576;
  49                         }
  50                 }
  51         }
  52
  53         /**
  54          * @return PPFrame_DOM
  55          */
  56         public function newFrame() {
  57                 return new PPFrame_DOM( $this );
  58         }
  59
  60         /**
  61          * @param array $args
  62          * @return PPCustomFrame_DOM
  63          */
  64         public function newCustomFrame( $args ) {
  65                 return new PPCustomFrame_DOM( $this, $args );
  66         }
  67
  68         /**
  69          * @param array $values
  70          * @return PPNode_DOM
  71          * @throws MWException
  72          */
  73         public function newPartNodeArray( $values ) {
  74                 //NOTE: DOM manipulation is slower than building & parsing XML! (or so Tim sais)
  75                 $xml = "<list>";
  76
  77                 foreach ( $values as $k => $val ) {
  78                         if ( is_int( $k ) ) {
  79                                 $xml .= "<part><name index=\"$k\"/><value>"
  80                                         . htmlspecialchars( $val ) . "</value></part>";
  81                         } else {
  82                                 $xml .= "<part><name>" . htmlspecialchars( $k )
  83                                         . "</name>=<value>" . htmlspecialchars( $val ) . "</value></part>";
  84                         }
  85                 }
  86
  87                 $xml .= "</list>";
  88
  89                 wfProfileIn( __METHOD__ . '-loadXML' );
  90                 $dom = new DOMDocument();
  91                 wfSuppressWarnings();
  92                 $result = $dom->loadXML( $xml );
  93                 wfRestoreWarnings();
  94                 if ( !$result ) {
  95                         // Try running the XML through UtfNormal to get rid of invalid characters
  96                         $xml = UtfNormal::cleanUp( $xml );
  97                         // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
  98                         // don't barf when the XML is >256 levels deep
  99                         $result = $dom->loadXML( $xml, 1 << 19 );
 100                 }
 101                 wfProfileOut( __METHOD__ . '-loadXML' );
 102
 103                 if ( !$result ) {
 104                         throw new MWException( 'Parameters passed to ' . __METHOD__ . ' result in invalid XML' );
 105                 }
 106
 107                 $root = $dom->documentElement;
 108                 $node = new PPNode_DOM( $root->childNodes );
 109                 return $node;
 110         }
 111
 112         /**
 113          * @throws MWException
 114          * @return bool
 115          */
 116         public function memCheck() {
 117                 if ( $this->memoryLimit === false ) {
 118                         return true;
 119                 }
 120                 $usage = memory_get_usage();
 121                 if ( $usage > $this->memoryLimit * 0.9 ) {
 122                         $limit = intval( $this->memoryLimit * 0.9 / 1048576 + 0.5 );
 123                         throw new MWException( "Preprocessor hit 90% memory limit ($limit MB)" );
 124                 }
 125                 return $usage <= $this->memoryLimit * 0.8;
 126         }
 127
 128         /**
 129          * Preprocess some wikitext and return the document tree.
 130          * This is the ghost of Parser::replace_variables().
 131          *
 132          * @param string $text The text to parse
 133          * @param int $flags Bitwise combination of:
 134          *     Parser::PTD_FOR_INCLUSION  Handle "<noinclude>" and "<includeonly>"
 135          *                                as if the text is being included. Default
 136          *                                is to assume a direct page view.
 137          *
 138          * The generated DOM tree must depend only on the input text and the flags.
 139          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
 140          *
 141          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
 142          * change in the DOM tree for a given text, must be passed through the section identifier
 143          * in the section edit link and thus back to extractSections().
 144          *
 145          * The output of this function is currently only cached in process memory, but a persistent
 146          * cache may be implemented at a later date which takes further advantage of these strict
 147          * dependency requirements.
 148          *
 149          * @throws MWException
 150          * @return PPNode_DOM
 151          */
 152         public function preprocessToObj( $text, $flags = 0 ) {
 153                 wfProfileIn( __METHOD__ );
 154                 global $wgMemc, $wgPreprocessorCacheThreshold;
 155
 156                 $xml = false;
 157                 $cacheable = ( $wgPreprocessorCacheThreshold !== false
 158                         && strlen( $text ) > $wgPreprocessorCacheThreshold );
 159                 if ( $cacheable ) {
 160                         wfProfileIn( __METHOD__ . '-cacheable' );
 161
 162                         $cacheKey = wfMemcKey( 'preprocess-xml', md5( $text ), $flags );
 163                         $cacheValue = $wgMemc->get( $cacheKey );
 164                         if ( $cacheValue ) {
 165                                 $version = substr( $cacheValue, 0, 8 );
 166                                 if ( intval( $version ) == self::CACHE_VERSION ) {
 167                                         $xml = substr( $cacheValue, 8 );
 168                                         // From the cache
 169                                         wfDebugLog( "Preprocessor", "Loaded preprocessor XML from memcached (key $cacheKey)" );
 170                                 }
 171                         }
 172                         if ( $xml === false ) {
 173                                 wfProfileIn( __METHOD__ . '-cache-miss' );
 174                                 $xml = $this->preprocessToXml( $text, $flags );
 175                                 $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . $xml;
 176                                 $wgMemc->set( $cacheKey, $cacheValue, 86400 );
 177                                 wfProfileOut( __METHOD__ . '-cache-miss' );
 178                                 wfDebugLog( "Preprocessor", "Saved preprocessor XML to memcached (key $cacheKey)" );
 179                         }
 180                 } else {
 181                         $xml = $this->preprocessToXml( $text, $flags );
 182                 }
 183
 184                 // Fail if the number of elements exceeds acceptable limits
 185                 // Do not attempt to generate the DOM
 186                 $this->parser->mGeneratedPPNodeCount += substr_count( $xml, '<' );
 187                 $max = $this->parser->mOptions->getMaxGeneratedPPNodeCount();
 188                 if ( $this->parser->mGeneratedPPNodeCount > $max ) {
 189                         if ( $cacheable ) {
 190                                 wfProfileOut( __METHOD__ . '-cacheable' );
 191                         }
 192                         wfProfileOut( __METHOD__ );
 193                         throw new MWException( __METHOD__ . ': generated node count limit exceeded' );
 194                 }
 195
 196                 wfProfileIn( __METHOD__ . '-loadXML' );
 197                 $dom = new DOMDocument;
 198                 wfSuppressWarnings();
 199                 $result = $dom->loadXML( $xml );
 200                 wfRestoreWarnings();
 201                 if ( !$result ) {
 202                         // Try running the XML through UtfNormal to get rid of invalid characters
 203                         $xml = UtfNormal::cleanUp( $xml );
 204                         // 1 << 19 == XML_PARSE_HUGE, needed so newer versions of libxml2
 205                         // don't barf when the XML is >256 levels deep.
 206                         $result = $dom->loadXML( $xml, 1 << 19 );
 207                 }
 208                 if ( $result ) {
 209                         $obj = new PPNode_DOM( $dom->documentElement );
 210                 }
 211                 wfProfileOut( __METHOD__ . '-loadXML' );
 212
 213                 if ( $cacheable ) {
 214                         wfProfileOut( __METHOD__ . '-cacheable' );
 215                 }
 216
 217                 wfProfileOut( __METHOD__ );
 218
 219                 if ( !$result ) {
 220                         throw new MWException( __METHOD__ . ' generated invalid XML' );
 221                 }
 222                 return $obj;
 223         }
 224
 225         /**
 226          * @param string $text
 227          * @param int $flags
 228          * @return string
 229          */
 230         public function preprocessToXml( $text, $flags = 0 ) {
 231                 wfProfileIn( __METHOD__ );
 232                 $rules = array(
 233                         '{' => array(
 234                                 'end' => '}',
 235                                 'names' => array(
 236                                         2 => 'template',
 237                                         3 => 'tplarg',
 238                                 ),
 239                                 'min' => 2,
 240                                 'max' => 3,
 241                         ),
 242                         '[' => array(
 243                                 'end' => ']',
 244                                 'names' => array( 2 => null ),
 245                                 'min' => 2,
 246                                 'max' => 2,
 247                         )
 248                 );
 249
 250                 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
 251
 252                 $xmlishElements = $this->parser->getStripList();
 253                 $enableOnlyinclude = false;
 254                 if ( $forInclusion ) {
 255                         $ignoredTags = array( 'includeonly', '/includeonly' );
 256                         $ignoredElements = array( 'noinclude' );
 257                         $xmlishElements[] = 'noinclude';
 258                         if ( strpos( $text, '<onlyinclude>' ) !== false
 259                                 && strpos( $text, '</onlyinclude>' ) !== false
 260                         ) {
 261                                 $enableOnlyinclude = true;
 262                         }
 263                 } else {
 264                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
 265                         $ignoredElements = array( 'includeonly' );
 266                         $xmlishElements[] = 'includeonly';
 267                 }
 268                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
 269
 270                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
 271                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
 272
 273                 $stack = new PPDStack;
 274
 275                 $searchBase = "[{<\n"; #}
 276                 // For fast reverse searches
 277                 $revText = strrev( $text );
 278                 $lengthText = strlen( $text );
 279
 280                 // Input pointer, starts out pointing to a pseudo-newline before the start
 281                 $i = 0;
 282                 // Current accumulator
 283                 $accum =& $stack->getAccum();
 284                 $accum = '<root>';
 285                 // True to find equals signs in arguments
 286                 $findEquals = false;
 287                 // True to take notice of pipe characters
 288                 $findPipe = false;
 289                 $headingIndex = 1;
 290                 // True if $i is inside a possible heading
 291                 $inHeading = false;
 292                 // True if there are no more greater-than (>) signs right of $i
 293                 $noMoreGT = false;
 294                 // True to ignore all input up to the next <onlyinclude>
 295                 $findOnlyinclude = $enableOnlyinclude;
 296                 // Do a line-start run without outputting an LF character
 297                 $fakeLineStart = true;
 298
 299                 while ( true ) {
 300                         //$this->memCheck();
 301
 302                         if ( $findOnlyinclude ) {
 303                                 // Ignore all input up to the next <onlyinclude>
 304                                 $startPos = strpos( $text, '<onlyinclude>', $i );
 305                                 if ( $startPos === false ) {
 306                                         // Ignored section runs to the end
 307                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
 308                                         break;
 309                                 }
 310                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
 311                                 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
 312                                 $i = $tagEndPos;
 313                                 $findOnlyinclude = false;
 314                         }
 315
 316                         if ( $fakeLineStart ) {
 317                                 $found = 'line-start';
 318                                 $curChar = '';
 319                         } else {
 320                                 # Find next opening brace, closing brace or pipe
 321                                 $search = $searchBase;
 322                                 if ( $stack->top === false ) {
 323                                         $currentClosing = '';
 324                                 } else {
 325                                         $currentClosing = $stack->top->close;
 326                                         $search .= $currentClosing;
 327                                 }
 328                                 if ( $findPipe ) {
 329                                         $search .= '|';
 330                                 }
 331                                 if ( $findEquals ) {
 332                                         // First equals will be for the template
 333                                         $search .= '=';
 334                                 }
 335                                 $rule = null;
 336                                 # Output literal section, advance input counter
 337                                 $literalLength = strcspn( $text, $search, $i );
 338                                 if ( $literalLength > 0 ) {
 339                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
 340                                         $i += $literalLength;
 341                                 }
 342                                 if ( $i >= $lengthText ) {
 343                                         if ( $currentClosing == "\n" ) {
 344                                                 // Do a past-the-end run to finish off the heading
 345                                                 $curChar = '';
 346                                                 $found = 'line-end';
 347                                         } else {
 348                                                 # All done
 349                                                 break;
 350                                         }
 351                                 } else {
 352                                         $curChar = $text[$i];
 353                                         if ( $curChar == '|' ) {
 354                                                 $found = 'pipe';
 355                                         } elseif ( $curChar == '=' ) {
 356                                                 $found = 'equals';
 357                                         } elseif ( $curChar == '<' ) {
 358                                                 $found = 'angle';
 359                                         } elseif ( $curChar == "\n" ) {
 360                                                 if ( $inHeading ) {
 361                                                         $found = 'line-end';
 362                                                 } else {
 363                                                         $found = 'line-start';
 364                                                 }
 365                                         } elseif ( $curChar == $currentClosing ) {
 366                                                 $found = 'close';
 367                                         } elseif ( isset( $rules[$curChar] ) ) {
 368                                                 $found = 'open';
 369                                                 $rule = $rules[$curChar];
 370                                         } else {
 371                                                 # Some versions of PHP have a strcspn which stops on null characters
 372                                                 # Ignore and continue
 373                                                 ++$i;
 374                                                 continue;
 375                                         }
 376                                 }
 377                         }
 378
 379                         if ( $found == 'angle' ) {
 380                                 $matches = false;
 381                                 // Handle </onlyinclude>
 382                                 if ( $enableOnlyinclude
 383                                         && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>'
 384                                 ) {
 385                                         $findOnlyinclude = true;
 386                                         continue;
 387                                 }
 388
 389                                 // Determine element name
 390                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
 391                                         // Element name missing or not listed
 392                                         $accum .= '&lt;';
 393                                         ++$i;
 394                                         continue;
 395                                 }
 396                                 // Handle comments
 397                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
 398
 399                                         // To avoid leaving blank lines, when a sequence of
 400                                         // space-separated comments is both preceded and followed by
 401                                         // a newline (ignoring spaces), then
 402                                         // trim leading and trailing spaces and the trailing newline.
 403
 404                                         // Find the end
 405                                         $endPos = strpos( $text, '-->', $i + 4 );
 406                                         if ( $endPos === false ) {
 407                                                 // Unclosed comment in input, runs to end
 408                                                 $inner = substr( $text, $i );
 409                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
 410                                                 $i = $lengthText;
 411                                         } else {
 412                                                 // Search backwards for leading whitespace
 413                                                 $wsStart = $i ? ( $i - strspn( $revText, " \t", $lengthText - $i ) ) : 0;
 414
 415                                                 // Search forwards for trailing whitespace
 416                                                 // $wsEnd will be the position of the last space (or the '>' if there's none)
 417                                                 $wsEnd = $endPos + 2 + strspn( $text, " \t", $endPos + 3 );
 418
 419                                                 // Keep looking forward as long as we're finding more
 420                                                 // comments.
 421                                                 $comments = array( array( $wsStart, $wsEnd ) );
 422                                                 while ( substr( $text, $wsEnd + 1, 4 ) == '<!--' ) {
 423                                                         $c = strpos( $text, '-->', $wsEnd + 4 );
 424                                                         if ( $c === false ) {
 425                                                                 break;
 426                                                         }
 427                                                         $c = $c + 2 + strspn( $text, " \t", $c + 3 );
 428                                                         $comments[] = array( $wsEnd + 1, $c );
 429                                                         $wsEnd = $c;
 430                                                 }
 431
 432                                                 // Eat the line if possible
 433                                                 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
 434                                                 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
 435                                                 // it's a possible beneficial b/c break.
 436                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
 437                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n"
 438                                                 ) {
 439                                                         // Remove leading whitespace from the end of the accumulator
 440                                                         // Sanity check first though
 441                                                         $wsLength = $i - $wsStart;
 442                                                         if ( $wsLength > 0
 443                                                                 && strspn( $accum, " \t", -$wsLength ) === $wsLength
 444                                                         ) {
 445                                                                 $accum = substr( $accum, 0, -$wsLength );
 446                                                         }
 447
 448                                                         // Dump all but the last comment to the accumulator
 449                                                         foreach ( $comments as $j => $com ) {
 450                                                                 $startPos = $com[0];
 451                                                                 $endPos = $com[1] + 1;
 452                                                                 if ( $j == ( count( $comments ) - 1 ) ) {
 453                                                                         break;
 454                                                                 }
 455                                                                 $inner = substr( $text, $startPos, $endPos - $startPos );
 456                                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
 457                                                         }
 458
 459                                                         // Do a line-start run next time to look for headings after the comment
 460                                                         $fakeLineStart = true;
 461                                                 } else {
 462                                                         // No line to eat, just take the comment itself
 463                                                         $startPos = $i;
 464                                                         $endPos += 2;
 465                                                 }
 466
 467                                                 if ( $stack->top ) {
 468                                                         $part = $stack->top->getCurrentPart();
 469                                                         if ( !( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) ) {
 470                                                                 $part->visualEnd = $wsStart;
 471                                                         }
 472                                                         // Else comments abutting, no change in visual end
 473                                                         $part->commentEnd = $endPos;
 474                                                 }
 475                                                 $i = $endPos + 1;
 476                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
 477                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
 478                                         }
 479                                         continue;
 480                                 }
 481                                 $name = $matches[1];
 482                                 $lowerName = strtolower( $name );
 483                                 $attrStart = $i + strlen( $name ) + 1;
 484
 485                                 // Find end of tag
 486                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
 487                                 if ( $tagEndPos === false ) {
 488                                         // Infinite backtrack
 489                                         // Disable tag search to prevent worst-case O(N^2) performance
 490                                         $noMoreGT = true;
 491                                         $accum .= '&lt;';
 492                                         ++$i;
 493                                         continue;
 494                                 }
 495
 496                                 // Handle ignored tags
 497                                 if ( in_array( $lowerName, $ignoredTags ) ) {
 498                                         $accum .= '<ignore>'
 499                                                 . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) )
 500                                                 . '</ignore>';
 501                                         $i = $tagEndPos + 1;
 502                                         continue;
 503                                 }
 504
 505                                 $tagStartPos = $i;
 506                                 if ( $text[$tagEndPos - 1] == '/' ) {
 507                                         $attrEnd = $tagEndPos - 1;
 508                                         $inner = null;
 509                                         $i = $tagEndPos + 1;
 510                                         $close = '';
 511                                 } else {
 512                                         $attrEnd = $tagEndPos;
 513                                         // Find closing tag
 514                                         if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
 515                                                         $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 )
 516                                         ) {
 517                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
 518                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
 519                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
 520                                         } else {
 521                                                 // No end tag -- let it run out to the end of the text.
 522                                                 $inner = substr( $text, $tagEndPos + 1 );
 523                                                 $i = $lengthText;
 524                                                 $close = '';
 525                                         }
 526                                 }
 527                                 // <includeonly> and <noinclude> just become <ignore> tags
 528                                 if ( in_array( $lowerName, $ignoredElements ) ) {
 529                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
 530                                                 . '</ignore>';
 531                                         continue;
 532                                 }
 533
 534                                 $accum .= '<ext>';
 535                                 if ( $attrEnd <= $attrStart ) {
 536                                         $attr = '';
 537                                 } else {
 538                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
 539                                 }
 540                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
 541                                         // Note that the attr element contains the whitespace between name and attribute,
 542                                         // this is necessary for precise reconstruction during pre-save transform.
 543                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
 544                                 if ( $inner !== null ) {
 545                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
 546                                 }
 547                                 $accum .= $close . '</ext>';
 548                         } elseif ( $found == 'line-start' ) {
 549                                 // Is this the start of a heading?
 550                                 // Line break belongs before the heading element in any case
 551                                 if ( $fakeLineStart ) {
 552                                         $fakeLineStart = false;
 553                                 } else {
 554                                         $accum .= $curChar;
 555                                         $i++;
 556                                 }
 557
 558                                 $count = strspn( $text, '=', $i, 6 );
 559                                 if ( $count == 1 && $findEquals ) {
 560                                         // DWIM: This looks kind of like a name/value separator.
 561                                         // Let's let the equals handler have it and break the
 562                                         // potential heading. This is heuristic, but AFAICT the
 563                                         // methods for completely correct disambiguation are very
 564                                         // complex.
 565                                 } elseif ( $count > 0 ) {
 566                                         $piece = array(
 567                                                 'open' => "\n",
 568                                                 'close' => "\n",
 569                                                 'parts' => array( new PPDPart( str_repeat( '=', $count ) ) ),
 570                                                 'startPos' => $i,
 571                                                 'count' => $count );
 572                                         $stack->push( $piece );
 573                                         $accum =& $stack->getAccum();
 574                                         $flags = $stack->getFlags();
 575                                         extract( $flags );
 576                                         $i += $count;
 577                                 }
 578                         } elseif ( $found == 'line-end' ) {
 579                                 $piece = $stack->top;
 580                                 // A heading must be open, otherwise \n wouldn't have been in the search list
 581                                 assert( '$piece->open == "\n"' );
 582                                 $part = $piece->getCurrentPart();
 583                                 // Search back through the input to see if it has a proper close.
 584                                 // Do this using the reversed string since the other solutions
 585                                 // (end anchor, etc.) are inefficient.
 586                                 $wsLength = strspn( $revText, " \t", $lengthText - $i );
 587                                 $searchStart = $i - $wsLength;
 588                                 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
 589                                         // Comment found at line end
 590                                         // Search for equals signs before the comment
 591                                         $searchStart = $part->visualEnd;
 592                                         $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
 593                                 }
 594                                 $count = $piece->count;
 595                                 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
 596                                 if ( $equalsLength > 0 ) {
 597                                         if ( $searchStart - $equalsLength == $piece->startPos ) {
 598                                                 // This is just a single string of equals signs on its own line
 599                                                 // Replicate the doHeadings behavior /={count}(.+)={count}/
 600                                                 // First find out how many equals signs there really are (don't stop at 6)
 601                                                 $count = $equalsLength;
 602                                                 if ( $count < 3 ) {
 603                                                         $count = 0;
 604                                                 } else {
 605                                                         $count = min( 6, intval( ( $count - 1 ) / 2 ) );
 606                                                 }
 607                                         } else {
 608                                                 $count = min( $equalsLength, $count );
 609                                         }
 610                                         if ( $count > 0 ) {
 611                                                 // Normal match, output <h>
 612                                                 $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
 613                                                 $headingIndex++;
 614                                         } else {
 615                                                 // Single equals sign on its own line, count=0
 616                                                 $element = $accum;
 617                                         }
 618                                 } else {
 619                                         // No match, no <h>, just pass down the inner text
 620                                         $element = $accum;
 621                                 }
 622                                 // Unwind the stack
 623                                 $stack->pop();
 624                                 $accum =& $stack->getAccum();
 625                                 $flags = $stack->getFlags();
 626                                 extract( $flags );
 627
 628                                 // Append the result to the enclosing accumulator
 629                                 $accum .= $element;
 630                                 // Note that we do NOT increment the input pointer.
 631                                 // This is because the closing linebreak could be the opening linebreak of
 632                                 // another heading. Infinite loops are avoided because the next iteration MUST
 633                                 // hit the heading open case above, which unconditionally increments the
 634                                 // input pointer.
 635                         } elseif ( $found == 'open' ) {
 636                                 # count opening brace characters
 637                                 $count = strspn( $text, $curChar, $i );
 638
 639                                 # we need to add to stack only if opening brace count is enough for one of the rules
 640                                 if ( $count >= $rule['min'] ) {
 641                                         # Add it to the stack
 642                                         $piece = array(
 643                                                 'open' => $curChar,
 644                                                 'close' => $rule['end'],
 645                                                 'count' => $count,
 646                                                 'lineStart' => ( $i > 0 && $text[$i - 1] == "\n" ),
 647                                         );
 648
 649                                         $stack->push( $piece );
 650                                         $accum =& $stack->getAccum();
 651                                         $flags = $stack->getFlags();
 652                                         extract( $flags );
 653                                 } else {
 654                                         # Add literal brace(s)
 655                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
 656                                 }
 657                                 $i += $count;
 658                         } elseif ( $found == 'close' ) {
 659                                 $piece = $stack->top;
 660                                 # lets check if there are enough characters for closing brace
 661                                 $maxCount = $piece->count;
 662                                 $count = strspn( $text, $curChar, $i, $maxCount );
 663
 664                                 # check for maximum matching characters (if there are 5 closing
 665                                 # characters, we will probably need only 3 - depending on the rules)
 666                                 $rule = $rules[$piece->open];
 667                                 if ( $count > $rule['max'] ) {
 668                                         # The specified maximum exists in the callback array, unless the caller
 669                                         # has made an error
 670                                         $matchingCount = $rule['max'];
 671                                 } else {
 672                                         # Count is less than the maximum
 673                                         # Skip any gaps in the callback array to find the true largest match
 674                                         # Need to use array_key_exists not isset because the callback can be null
 675                                         $matchingCount = $count;
 676                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
 677                                                 --$matchingCount;
 678                                         }
 679                                 }
 680
 681                                 if ( $matchingCount <= 0 ) {
 682                                         # No matching element found in callback array
 683                                         # Output a literal closing brace and continue
 684                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
 685                                         $i += $count;
 686                                         continue;
 687                                 }
 688                                 $name = $rule['names'][$matchingCount];
 689                                 if ( $name === null ) {
 690                                         // No element, just literal text
 691                                         $element = $piece->breakSyntax( $matchingCount ) . str_repeat( $rule['end'], $matchingCount );
 692                                 } else {
 693                                         # Create XML element
 694                                         # Note: $parts is already XML, does not need to be encoded further
 695                                         $parts = $piece->parts;
 696                                         $title = $parts[0]->out;
 697                                         unset( $parts[0] );
 698
 699                                         # The invocation is at the start of the line if lineStart is set in
 700                                         # the stack, and all opening brackets are used up.
 701                                         if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
 702                                                 $attr = ' lineStart="1"';
 703                                         } else {
 704                                                 $attr = '';
 705                                         }
 706
 707                                         $element = "<$name$attr>";
 708                                         $element .= "<title>$title</title>";
 709                                         $argIndex = 1;
 710                                         foreach ( $parts as $part ) {
 711                                                 if ( isset( $part->eqpos ) ) {
 712                                                         $argName = substr( $part->out, 0, $part->eqpos );
 713                                                         $argValue = substr( $part->out, $part->eqpos + 1 );
 714                                                         $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
 715                                                 } else {
 716                                                         $element .= "<part><name index=\"$argIndex\" /><value>{$part->out}</value></part>";
 717                                                         $argIndex++;
 718                                                 }
 719                                         }
 720                                         $element .= "</$name>";
 721                                 }
 722
 723                                 # Advance input pointer
 724                                 $i += $matchingCount;
 725
 726                                 # Unwind the stack
 727                                 $stack->pop();
 728                                 $accum =& $stack->getAccum();
 729
 730                                 # Re-add the old stack element if it still has unmatched opening characters remaining
 731                                 if ( $matchingCount < $piece->count ) {
 732                                         $piece->parts = array( new PPDPart );
 733                                         $piece->count -= $matchingCount;
 734                                         # do we still qualify for any callback with remaining count?
 735                                         $min = $rules[$piece->open]['min'];
 736                                         if ( $piece->count >= $min ) {
 737                                                 $stack->push( $piece );
 738                                                 $accum =& $stack->getAccum();
 739                                         } else {
 740                                                 $accum .= str_repeat( $piece->open, $piece->count );
 741                                         }
 742                                 }
 743                                 $flags = $stack->getFlags();
 744                                 extract( $flags );
 745
 746                                 # Add XML element to the enclosing accumulator
 747                                 $accum .= $element;
 748                         } elseif ( $found == 'pipe' ) {
 749                                 $findEquals = true; // shortcut for getFlags()
 750                                 $stack->addPart();
 751                                 $accum =& $stack->getAccum();
 752                                 ++$i;
 753                         } elseif ( $found == 'equals' ) {
 754                                 $findEquals = false; // shortcut for getFlags()
 755                                 $stack->getCurrentPart()->eqpos = strlen( $accum );
 756                                 $accum .= '=';
 757                                 ++$i;
 758                         }
 759                 }
 760
 761                 # Output any remaining unclosed brackets
 762                 foreach ( $stack->stack as $piece ) {
 763                         $stack->rootAccum .= $piece->breakSyntax();
 764                 }
 765                 $stack->rootAccum .= '</root>';
 766                 $xml = $stack->rootAccum;
 767
 768                 wfProfileOut( __METHOD__ );
 769
 770                 return $xml;
 771         }
 772 }
 773
 774 /**
 775  * Stack class to help Preprocessor::preprocessToObj()
 776  * @ingroup Parser
 777  */
 778 class PPDStack {
 779         public $stack, $rootAccum;
 780
 781         /**
 782          * @var PPDStack
 783          */
 784         public $top;
 785         public $out;
 786         public $elementClass = 'PPDStackElement';
 787
 788         public static $false = false;
 789
 790         public function __construct() {
 791                 $this->stack = array();
 792                 $this->top = false;
 793                 $this->rootAccum = '';
 794                 $this->accum =& $this->rootAccum;
 795         }
 796
 797         /**
 798          * @return int
 799          */
 800         public function count() {
 801                 return count( $this->stack );
 802         }
 803
 804         public function &getAccum() {
 805                 return $this->accum;
 806         }
 807
 808         public function getCurrentPart() {
 809                 if ( $this->top === false ) {
 810                         return false;
 811                 } else {
 812                         return $this->top->getCurrentPart();
 813                 }
 814         }
 815
 816         public function push( $data ) {
 817                 if ( $data instanceof $this->elementClass ) {
 818                         $this->stack[] = $data;
 819                 } else {
 820                         $class = $this->elementClass;
 821                         $this->stack[] = new $class( $data );
 822                 }
 823                 $this->top = $this->stack[count( $this->stack ) - 1];
 824                 $this->accum =& $this->top->getAccum();
 825         }
 826
 827         public function pop() {
 828                 if ( !count( $this->stack ) ) {
 829                         throw new MWException( __METHOD__ . ': no elements remaining' );
 830                 }
 831                 $temp = array_pop( $this->stack );
 832
 833                 if ( count( $this->stack ) ) {
 834                         $this->top = $this->stack[count( $this->stack ) - 1];
 835                         $this->accum =& $this->top->getAccum();
 836                 } else {
 837                         $this->top = self::$false;
 838                         $this->accum =& $this->rootAccum;
 839                 }
 840                 return $temp;
 841         }
 842
 843         public function addPart( $s = '' ) {
 844                 $this->top->addPart( $s );
 845                 $this->accum =& $this->top->getAccum();
 846         }
 847
 848         /**
 849          * @return array
 850          */
 851         public function getFlags() {
 852                 if ( !count( $this->stack ) ) {
 853                         return array(
 854                                 'findEquals' => false,
 855                                 'findPipe' => false,
 856                                 'inHeading' => false,
 857                         );
 858                 } else {
 859                         return $this->top->getFlags();
 860                 }
 861         }
 862 }
 863
 864 /**
 865  * @ingroup Parser
 866  */
 867 class PPDStackElement {
 868         public $open,              // Opening character (\n for heading)
 869                 $close,             // Matching closing character
 870                 $count,             // Number of opening characters found (number of "=" for heading)
 871                 $parts,             // Array of PPDPart objects describing pipe-separated parts.
 872                 $lineStart;         // True if the open char appeared at the start of the input line. Not set for headings.
 873
 874         public $partClass = 'PPDPart';
 875
 876         public function __construct( $data = array() ) {
 877                 $class = $this->partClass;
 878                 $this->parts = array( new $class );
 879
 880                 foreach ( $data as $name => $value ) {
 881                         $this->$name = $value;
 882                 }
 883         }
 884
 885         public function &getAccum() {
 886                 return $this->parts[count( $this->parts ) - 1]->out;
 887         }
 888
 889         public function addPart( $s = '' ) {
 890                 $class = $this->partClass;
 891                 $this->parts[] = new $class( $s );
 892         }
 893
 894         public function getCurrentPart() {
 895                 return $this->parts[count( $this->parts ) - 1];
 896         }
 897
 898         /**
 899          * @return array
 900          */
 901         public function getFlags() {
 902                 $partCount = count( $this->parts );
 903                 $findPipe = $this->open != "\n" && $this->open != '[';
 904                 return array(
 905                         'findPipe' => $findPipe,
 906                         'findEquals' => $findPipe && $partCount > 1 && !isset( $this->parts[$partCount - 1]->eqpos ),
 907                         'inHeading' => $this->open == "\n",
 908                 );
 909         }
 910
 911         /**
 912          * Get the output string that would result if the close is not found.
 913          *
 914          * @param bool|int $openingCount
 915          * @return string
 916          */
 917         public function breakSyntax( $openingCount = false ) {
 918                 if ( $this->open == "\n" ) {
 919                         $s = $this->parts[0]->out;
 920                 } else {
 921                         if ( $openingCount === false ) {
 922                                 $openingCount = $this->count;
 923                         }
 924                         $s = str_repeat( $this->open, $openingCount );
 925                         $first = true;
 926                         foreach ( $this->parts as $part ) {
 927                                 if ( $first ) {
 928                                         $first = false;
 929                                 } else {
 930                                         $s .= '|';
 931                                 }
 932                                 $s .= $part->out;
 933                         }
 934                 }
 935                 return $s;
 936         }
 937 }
 938
 939 /**
 940  * @ingroup Parser
 941  */
 942 class PPDPart {
 943         public $out; // Output accumulator string
 944
 945         // Optional member variables:
 946         //   eqpos        Position of equals sign in output accumulator
 947         //   commentEnd   Past-the-end input pointer for the last comment encountered
 948         //   visualEnd    Past-the-end input pointer for the end of the accumulator minus comments
 949
 950         public function __construct( $out = '' ) {
 951                 $this->out = $out;
 952         }
 953 }
 954
 955 /**
 956  * An expansion frame, used as a context to expand the result of preprocessToObj()
 957  * @ingroup Parser
 958  * @codingStandardsIgnoreStart
 959  */
 960 class PPFrame_DOM implements PPFrame {
 961         // @codingStandardsIgnoreEnd
 962
 963         /**
 964          * @var Preprocessor
 965          */
 966         public $preprocessor;
 967
 968         /**
 969          * @var Parser
 970          */
 971         public $parser;
 972
 973         /**
 974          * @var Title
 975          */
 976         public $title;
 977         public $titleCache;
 978
 979         /**
 980          * Hashtable listing templates which are disallowed for expansion in this frame,
 981          * having been encountered previously in parent frames.
 982          */
 983         public $loopCheckHash;
 984
 985         /**
 986          * Recursion depth of this frame, top = 0
 987          * Note that this is NOT the same as expansion depth in expand()
 988          */
 989         public $depth;
 990
 991         private $volatile = false;
 992         private $ttl = null;
 993
 994         /**
 995          * @var array
 996          */
 997         protected $childExpansionCache;
 998
 999         /**
1000          * Construct a new preprocessor frame.
1001          * @param Preprocessor $preprocessor The parent preprocessor
1002          */
1003         public function __construct( $preprocessor ) {
1004                 $this->preprocessor = $preprocessor;
1005                 $this->parser = $preprocessor->parser;
1006                 $this->title = $this->parser->mTitle;
1007                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
1008                 $this->loopCheckHash = array();
1009                 $this->depth = 0;
1010                 $this->childExpansionCache = array();
1011         }
1012
1013         /**
1014          * Create a new child frame
1015          * $args is optionally a multi-root PPNode or array containing the template arguments
1016          *
1017          * @param bool|array $args
1018          * @param Title|bool $title
1019          * @param int $indexOffset
1020          * @return PPTemplateFrame_DOM
1021          */
1022         public function newChild( $args = false, $title = false, $indexOffset = 0 ) {
1023                 $namedArgs = array();
1024                 $numberedArgs = array();
1025                 if ( $title === false ) {
1026                         $title = $this->title;
1027                 }
1028                 if ( $args !== false ) {
1029                         $xpath = false;
1030                         if ( $args instanceof PPNode ) {
1031                                 $args = $args->node;
1032                         }
1033                         foreach ( $args as $arg ) {
1034                                 if ( $arg instanceof PPNode ) {
1035                                         $arg = $arg->node;
1036                                 }
1037                                 if ( !$xpath || $xpath->document !== $arg->ownerDocument ) {
1038                                         $xpath = new DOMXPath( $arg->ownerDocument );
1039                                 }
1040
1041                                 $nameNodes = $xpath->query( 'name', $arg );
1042                                 $value = $xpath->query( 'value', $arg );
1043                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
1044                                         // Numbered parameter
1045                                         $index = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
1046                                         $index = $index - $indexOffset;
1047                                         if ( isset( $namedArgs[$index] ) || isset( $numberedArgs[$index] ) ) {
1048                                                 $this->parser->addTrackingCategory( 'duplicate-args-category' );
1049                                         }
1050                                         $numberedArgs[$index] = $value->item( 0 );
1051                                         unset( $namedArgs[$index] );
1052                                 } else {
1053                                         // Named parameter
1054                                         $name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS ) );
1055                                         if ( isset( $namedArgs[$name] ) || isset( $numberedArgs[$name] ) ) {
1056                                                 $this->parser->addTrackingCategory( 'duplicate-args-category' );
1057                                         }
1058                                         $namedArgs[$name] = $value->item( 0 );
1059                                         unset( $numberedArgs[$name] );
1060                                 }
1061                         }
1062                 }
1063                 return new PPTemplateFrame_DOM( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
1064         }
1065
1066         /**
1067          * @throws MWException
1068          * @param string|int $key
1069          * @param string|PPNode_DOM|DOMDocument $root
1070          * @param int $flags
1071          * @return string
1072          */
1073         public function cachedExpand( $key, $root, $flags = 0 ) {
1074                 // we don't have a parent, so we don't have a cache
1075                 return $this->expand( $root, $flags );
1076         }
1077
1078         /**
1079          * @throws MWException
1080          * @param string|PPNode_DOM|DOMDocument $root
1081          * @param int $flags
1082          * @return string
1083          */
1084         public function expand( $root, $flags = 0 ) {
1085                 static $expansionDepth = 0;
1086                 if ( is_string( $root ) ) {
1087                         return $root;
1088                 }
1089
1090                 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) {
1091                         $this->parser->limitationWarn( 'node-count-exceeded',
1092                                 $this->parser->mPPNodeCount,
1093                                 $this->parser->mOptions->getMaxPPNodeCount()
1094                         );
1095                         return '<span class="error">Node-count limit exceeded</span>';
1096                 }
1097
1098                 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
1099                         $this->parser->limitationWarn( 'expansion-depth-exceeded',
1100                                 $expansionDepth,
1101                                 $this->parser->mOptions->getMaxPPExpandDepth()
1102                         );
1103                         return '<span class="error">Expansion depth limit exceeded</span>';
1104                 }
1105                 wfProfileIn( __METHOD__ );
1106                 ++$expansionDepth;
1107                 if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
1108                         $this->parser->mHighestExpansionDepth = $expansionDepth;
1109                 }
1110
1111                 if ( $root instanceof PPNode_DOM ) {
1112                         $root = $root->node;
1113                 }
1114                 if ( $root instanceof DOMDocument ) {
1115                         $root = $root->documentElement;
1116                 }
1117
1118                 $outStack = array( '', '' );
1119                 $iteratorStack = array( false, $root );
1120                 $indexStack = array( 0, 0 );
1121
1122                 while ( count( $iteratorStack ) > 1 ) {
1123                         $level = count( $outStack ) - 1;
1124                         $iteratorNode =& $iteratorStack[$level];
1125                         $out =& $outStack[$level];
1126                         $index =& $indexStack[$level];
1127
1128                         if ( $iteratorNode instanceof PPNode_DOM ) {
1129                                 $iteratorNode = $iteratorNode->node;
1130                         }
1131
1132                         if ( is_array( $iteratorNode ) ) {
1133                                 if ( $index >= count( $iteratorNode ) ) {
1134                                         // All done with this iterator
1135                                         $iteratorStack[$level] = false;
1136                                         $contextNode = false;
1137                                 } else {
1138                                         $contextNode = $iteratorNode[$index];
1139                                         $index++;
1140                                 }
1141                         } elseif ( $iteratorNode instanceof DOMNodeList ) {
1142                                 if ( $index >= $iteratorNode->length ) {
1143                                         // All done with this iterator
1144                                         $iteratorStack[$level] = false;
1145                                         $contextNode = false;
1146                                 } else {
1147                                         $contextNode = $iteratorNode->item( $index );
1148                                         $index++;
1149                                 }
1150                         } else {
1151                                 // Copy to $contextNode and then delete from iterator stack,
1152                                 // because this is not an iterator but we do have to execute it once
1153                                 $contextNode = $iteratorStack[$level];
1154                                 $iteratorStack[$level] = false;
1155                         }
1156
1157                         if ( $contextNode instanceof PPNode_DOM ) {
1158                                 $contextNode = $contextNode->node;
1159                         }
1160
1161                         $newIterator = false;
1162
1163                         if ( $contextNode === false ) {
1164                                 // nothing to do
1165                         } elseif ( is_string( $contextNode ) ) {
1166                                 $out .= $contextNode;
1167                         } elseif ( is_array( $contextNode ) || $contextNode instanceof DOMNodeList ) {
1168                                 $newIterator = $contextNode;
1169                         } elseif ( $contextNode instanceof DOMNode ) {
1170                                 if ( $contextNode->nodeType == XML_TEXT_NODE ) {
1171                                         $out .= $contextNode->nodeValue;
1172                                 } elseif ( $contextNode->nodeName == 'template' ) {
1173                                         # Double-brace expansion
1174                                         $xpath = new DOMXPath( $contextNode->ownerDocument );
1175                                         $titles = $xpath->query( 'title', $contextNode );
1176                                         $title = $titles->item( 0 );
1177                                         $parts = $xpath->query( 'part', $contextNode );
1178                                         if ( $flags & PPFrame::NO_TEMPLATES ) {
1179                                                 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $title, $parts );
1180                                         } else {
1181                                                 $lineStart = $contextNode->getAttribute( 'lineStart' );
1182                                                 $params = array(
1183                                                         'title' => new PPNode_DOM( $title ),
1184                                                         'parts' => new PPNode_DOM( $parts ),
1185                                                         'lineStart' => $lineStart );
1186                                                 $ret = $this->parser->braceSubstitution( $params, $this );
1187                                                 if ( isset( $ret['object'] ) ) {
1188                                                         $newIterator = $ret['object'];
1189                                                 } else {
1190                                                         $out .= $ret['text'];
1191                                                 }
1192                                         }
1193                                 } elseif ( $contextNode->nodeName == 'tplarg' ) {
1194                                         # Triple-brace expansion
1195                                         $xpath = new DOMXPath( $contextNode->ownerDocument );
1196                                         $titles = $xpath->query( 'title', $contextNode );
1197                                         $title = $titles->item( 0 );
1198                                         $parts = $xpath->query( 'part', $contextNode );
1199                                         if ( $flags & PPFrame::NO_ARGS ) {
1200                                                 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $title, $parts );
1201                                         } else {
1202                                                 $params = array(
1203                                                         'title' => new PPNode_DOM( $title ),
1204                                                         'parts' => new PPNode_DOM( $parts ) );
1205                                                 $ret = $this->parser->argSubstitution( $params, $this );
1206                                                 if ( isset( $ret['object'] ) ) {
1207                                                         $newIterator = $ret['object'];
1208                                                 } else {
1209                                                         $out .= $ret['text'];
1210                                                 }
1211                                         }
1212                                 } elseif ( $contextNode->nodeName == 'comment' ) {
1213                                         # HTML-style comment
1214                                         # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1215                                         if ( $this->parser->ot['html']
1216                                                 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
1217                                                 || ( $flags & PPFrame::STRIP_COMMENTS )
1218                                         ) {
1219                                                 $out .= '';
1220                                         } elseif ( $this->parser->ot['wiki'] && !( $flags & PPFrame::RECOVER_COMMENTS ) ) {
1221                                                 # Add a strip marker in PST mode so that pstPass2() can
1222                                                 # run some old-fashioned regexes on the result.
1223                                                 # Not in RECOVER_COMMENTS mode (extractSections) though.
1224                                                 $out .= $this->parser->insertStripItem( $contextNode->textContent );
1225                                         } else {
1226                                                 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1227                                                 $out .= $contextNode->textContent;
1228                                         }
1229                                 } elseif ( $contextNode->nodeName == 'ignore' ) {
1230                                         # Output suppression used by <includeonly> etc.
1231                                         # OT_WIKI will only respect <ignore> in substed templates.
1232                                         # The other output types respect it unless NO_IGNORE is set.
1233                                         # extractSections() sets NO_IGNORE and so never respects it.
1234                                         if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] )
1235                                                 || ( $flags & PPFrame::NO_IGNORE )
1236                                         ) {
1237                                                 $out .= $contextNode->textContent;
1238                                         } else {
1239                                                 $out .= '';
1240                                         }
1241                                 } elseif ( $contextNode->nodeName == 'ext' ) {
1242                                         # Extension tag
1243                                         $xpath = new DOMXPath( $contextNode->ownerDocument );
1244                                         $names = $xpath->query( 'name', $contextNode );
1245                                         $attrs = $xpath->query( 'attr', $contextNode );
1246                                         $inners = $xpath->query( 'inner', $contextNode );
1247                                         $closes = $xpath->query( 'close', $contextNode );
1248                                         if ( $flags & PPFrame::NO_TAGS ) {
1249                                                 $s = '<' . $this->expand( $names->item( 0 ), $flags );
1250                                                 if ( $attrs->length > 0 ) {
1251                                                         $s .= $this->expand( $attrs->item( 0 ), $flags );
1252                                                 }
1253                                                 if ( $inners->length > 0 ) {
1254                                                         $s .= '>' . $this->expand( $inners->item( 0 ), $flags );
1255                                                         if ( $closes->length > 0 ) {
1256                                                                 $s .= $this->expand( $closes->item( 0 ), $flags );
1257                                                         }
1258                                                 } else {
1259                                                         $s .= '/>';
1260                                                 }
1261                                                 $out .= $s;
1262                                         } else {
1263                                                 $params = array(
1264                                                         'name' => new PPNode_DOM( $names->item( 0 ) ),
1265                                                         'attr' => $attrs->length > 0 ? new PPNode_DOM( $attrs->item( 0 ) ) : null,
1266                                                         'inner' => $inners->length > 0 ? new PPNode_DOM( $inners->item( 0 ) ) : null,
1267                                                         'close' => $closes->length > 0 ? new PPNode_DOM( $closes->item( 0 ) ) : null,
1268                                                 );
1269                                                 $out .= $this->parser->extensionSubstitution( $params, $this );
1270                                         }
1271                                 } elseif ( $contextNode->nodeName == 'h' ) {
1272                                         # Heading
1273                                         $s = $this->expand( $contextNode->childNodes, $flags );
1274
1275                                         # Insert a heading marker only for <h> children of <root>
1276                                         # This is to stop extractSections from going over multiple tree levels
1277                                         if ( $contextNode->parentNode->nodeName == 'root' && $this->parser->ot['html'] ) {
1278                                                 # Insert heading index marker
1279                                                 $headingIndex = $contextNode->getAttribute( 'i' );
1280                                                 $titleText = $this->title->getPrefixedDBkey();
1281                                                 $this->parser->mHeadings[] = array( $titleText, $headingIndex );
1282                                                 $serial = count( $this->parser->mHeadings ) - 1;
1283                                                 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
1284                                                 $count = $contextNode->getAttribute( 'level' );
1285                                                 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
1286                                                 $this->parser->mStripState->addGeneral( $marker, '' );
1287                                         }
1288                                         $out .= $s;
1289                                 } else {
1290                                         # Generic recursive expansion
1291                                         $newIterator = $contextNode->childNodes;
1292                                 }
1293                         } else {
1294                                 wfProfileOut( __METHOD__ );
1295                                 throw new MWException( __METHOD__ . ': Invalid parameter type' );
1296                         }
1297
1298                         if ( $newIterator !== false ) {
1299                                 if ( $newIterator instanceof PPNode_DOM ) {
1300                                         $newIterator = $newIterator->node;
1301                                 }
1302                                 $outStack[] = '';
1303                                 $iteratorStack[] = $newIterator;
1304                                 $indexStack[] = 0;
1305                         } elseif ( $iteratorStack[$level] === false ) {
1306                                 // Return accumulated value to parent
1307                                 // With tail recursion
1308                                 while ( $iteratorStack[$level] === false && $level > 0 ) {
1309                                         $outStack[$level - 1] .= $out;
1310                                         array_pop( $outStack );
1311                                         array_pop( $iteratorStack );
1312                                         array_pop( $indexStack );
1313                                         $level--;
1314                                 }
1315                         }
1316                 }
1317                 --$expansionDepth;
1318                 wfProfileOut( __METHOD__ );
1319                 return $outStack[0];
1320         }
1321
1322         /**
1323          * @param string $sep
1324          * @param int $flags
1325          * @param string|PPNode_DOM|DOMDocument $args,...
1326          * @return string
1327          */
1328         public function implodeWithFlags( $sep, $flags /*, ... */ ) {
1329                 $args = array_slice( func_get_args(), 2 );
1330
1331                 $first = true;
1332                 $s = '';
1333                 foreach ( $args as $root ) {
1334                         if ( $root instanceof PPNode_DOM ) {
1335                                 $root = $root->node;
1336                         }
1337                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1338                                 $root = array( $root );
1339                         }
1340                         foreach ( $root as $node ) {
1341                                 if ( $first ) {
1342                                         $first = false;
1343                                 } else {
1344                                         $s .= $sep;
1345                                 }
1346                                 $s .= $this->expand( $node, $flags );
1347                         }
1348                 }
1349                 return $s;
1350         }
1351
1352         /**
1353          * Implode with no flags specified
1354          * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1355          *
1356          * @param string $sep
1357          * @param string|PPNode_DOM|DOMDocument $args,...
1358          * @return string
1359          */
1360         public function implode( $sep /*, ... */ ) {
1361                 $args = array_slice( func_get_args(), 1 );
1362
1363                 $first = true;
1364                 $s = '';
1365                 foreach ( $args as $root ) {
1366                         if ( $root instanceof PPNode_DOM ) {
1367                                 $root = $root->node;
1368                         }
1369                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1370                                 $root = array( $root );
1371                         }
1372                         foreach ( $root as $node ) {
1373                                 if ( $first ) {
1374                                         $first = false;
1375                                 } else {
1376                                         $s .= $sep;
1377                                 }
1378                                 $s .= $this->expand( $node );
1379                         }
1380                 }
1381                 return $s;
1382         }
1383
1384         /**
1385          * Makes an object that, when expand()ed, will be the same as one obtained
1386          * with implode()
1387          *
1388          * @param string $sep
1389          * @param string|PPNode_DOM|DOMDocument $args,...
1390          * @return array
1391          */
1392         public function virtualImplode( $sep /*, ... */ ) {
1393                 $args = array_slice( func_get_args(), 1 );
1394                 $out = array();
1395                 $first = true;
1396
1397                 foreach ( $args as $root ) {
1398                         if ( $root instanceof PPNode_DOM ) {
1399                                 $root = $root->node;
1400                         }
1401                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1402                                 $root = array( $root );
1403                         }
1404                         foreach ( $root as $node ) {
1405                                 if ( $first ) {
1406                                         $first = false;
1407                                 } else {
1408                                         $out[] = $sep;
1409                                 }
1410                                 $out[] = $node;
1411                         }
1412                 }
1413                 return $out;
1414         }
1415
1416         /**
1417          * Virtual implode with brackets
1418          * @param string $start
1419          * @param string $sep
1420          * @param string $end
1421          * @param string|PPNode_DOM|DOMDocument $args,...
1422          * @return array
1423          */
1424         public function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1425                 $args = array_slice( func_get_args(), 3 );
1426                 $out = array( $start );
1427                 $first = true;
1428
1429                 foreach ( $args as $root ) {
1430                         if ( $root instanceof PPNode_DOM ) {
1431                                 $root = $root->node;
1432                         }
1433                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
1434                                 $root = array( $root );
1435                         }
1436                         foreach ( $root as $node ) {
1437                                 if ( $first ) {
1438                                         $first = false;
1439                                 } else {
1440                                         $out[] = $sep;
1441                                 }
1442                                 $out[] = $node;
1443                         }
1444                 }
1445                 $out[] = $end;
1446                 return $out;
1447         }
1448
1449         public function __toString() {
1450                 return 'frame{}';
1451         }
1452
1453         public function getPDBK( $level = false ) {
1454                 if ( $level === false ) {
1455                         return $this->title->getPrefixedDBkey();
1456                 } else {
1457                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1458                 }
1459         }
1460
1461         /**
1462          * @return array
1463          */
1464         public function getArguments() {
1465                 return array();
1466         }
1467
1468         /**
1469          * @return array
1470          */
1471         public function getNumberedArguments() {
1472                 return array();
1473         }
1474
1475         /**
1476          * @return array
1477          */
1478         public function getNamedArguments() {
1479                 return array();
1480         }
1481
1482         /**
1483          * Returns true if there are no arguments in this frame
1484          *
1485          * @return bool
1486          */
1487         public function isEmpty() {
1488                 return true;
1489         }
1490
1491         public function getArgument( $name ) {
1492                 return false;
1493         }
1494
1495         /**
1496          * Returns true if the infinite loop check is OK, false if a loop is detected
1497          *
1498          * @param Title $title
1499          * @return bool
1500          */
1501         public function loopCheck( $title ) {
1502                 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1503         }
1504
1505         /**
1506          * Return true if the frame is a template frame
1507          *
1508          * @return bool
1509          */
1510         public function isTemplate() {
1511                 return false;
1512         }
1513
1514         /**
1515          * Get a title of frame
1516          *
1517          * @return Title
1518          */
1519         public function getTitle() {
1520                 return $this->title;
1521         }
1522
1523         /**
1524          * Set the volatile flag
1525          *
1526          * @param bool $flag
1527          */
1528         public function setVolatile( $flag = true ) {
1529                 $this->volatile = $flag;
1530         }
1531
1532         /**
1533          * Get the volatile flag
1534          *
1535          * @return bool
1536          */
1537         public function isVolatile() {
1538                 return $this->volatile;
1539         }
1540
1541         /**
1542          * Set the TTL
1543          *
1544          * @param int $ttl
1545          */
1546         public function setTTL( $ttl ) {
1547                 if ( $ttl !== null && ( $this->ttl === null || $ttl < $this->ttl ) ) {
1548                         $this->ttl = $ttl;
1549                 }
1550         }
1551
1552         /**
1553          * Get the TTL
1554          *
1555          * @return int|null
1556          */
1557         public function getTTL() {
1558                 return $this->ttl;
1559         }
1560 }
1561
1562 /**
1563  * Expansion frame with template arguments
1564  * @ingroup Parser
1565  * @codingStandardsIgnoreStart
1566  */
1567 class PPTemplateFrame_DOM extends PPFrame_DOM {
1568         // @codingStandardsIgnoreEnd
1569
1570         public $numberedArgs, $namedArgs;
1571
1572         /**
1573          * @var PPFrame_DOM
1574          */
1575         public $parent;
1576         public $numberedExpansionCache, $namedExpansionCache;
1577
1578         /**
1579          * @param Preprocessor $preprocessor
1580          * @param bool|PPFrame_DOM $parent
1581          * @param array $numberedArgs
1582          * @param array $namedArgs
1583          * @param bool|Title $title
1584          */
1585         public function __construct( $preprocessor, $parent = false, $numberedArgs = array(),
1586                 $namedArgs = array(), $title = false
1587         ) {
1588                 parent::__construct( $preprocessor );
1589
1590                 $this->parent = $parent;
1591                 $this->numberedArgs = $numberedArgs;
1592                 $this->namedArgs = $namedArgs;
1593                 $this->title = $title;
1594                 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1595                 $this->titleCache = $parent->titleCache;
1596                 $this->titleCache[] = $pdbk;
1597                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1598                 if ( $pdbk !== false ) {
1599                         $this->loopCheckHash[$pdbk] = true;
1600                 }
1601                 $this->depth = $parent->depth + 1;
1602                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1603         }
1604
1605         public function __toString() {
1606                 $s = 'tplframe{';
1607                 $first = true;
1608                 $args = $this->numberedArgs + $this->namedArgs;
1609                 foreach ( $args as $name => $value ) {
1610                         if ( $first ) {
1611                                 $first = false;
1612                         } else {
1613                                 $s .= ', ';
1614                         }
1615                         $s .= "\"$name\":\"" .
1616                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
1617                 }
1618                 $s .= '}';
1619                 return $s;
1620         }
1621
1622         /**
1623          * @throws MWException
1624          * @param string|int $key
1625          * @param string|PPNode_DOM|DOMDocument $root
1626          * @param int $flags
1627          * @return string
1628          */
1629         public function cachedExpand( $key, $root, $flags = 0 ) {
1630                 if ( isset( $this->parent->childExpansionCache[$key] ) ) {
1631                         return $this->parent->childExpansionCache[$key];
1632                 }
1633                 $retval = $this->expand( $root, $flags );
1634                 if ( !$this->isVolatile() ) {
1635                         $this->parent->childExpansionCache[$key] = $retval;
1636                 }
1637                 return $retval;
1638         }
1639
1640         /**
1641          * Returns true if there are no arguments in this frame
1642          *
1643          * @return bool
1644          */
1645         public function isEmpty() {
1646                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1647         }
1648
1649         public function getArguments() {
1650                 $arguments = array();
1651                 foreach ( array_merge(
1652                                 array_keys( $this->numberedArgs ),
1653                                 array_keys( $this->namedArgs ) ) as $key ) {
1654                         $arguments[$key] = $this->getArgument( $key );
1655                 }
1656                 return $arguments;
1657         }
1658
1659         public function getNumberedArguments() {
1660                 $arguments = array();
1661                 foreach ( array_keys( $this->numberedArgs ) as $key ) {
1662                         $arguments[$key] = $this->getArgument( $key );
1663                 }
1664                 return $arguments;
1665         }
1666
1667         public function getNamedArguments() {
1668                 $arguments = array();
1669                 foreach ( array_keys( $this->namedArgs ) as $key ) {
1670                         $arguments[$key] = $this->getArgument( $key );
1671                 }
1672                 return $arguments;
1673         }
1674
1675         public function getNumberedArgument( $index ) {
1676                 if ( !isset( $this->numberedArgs[$index] ) ) {
1677                         return false;
1678                 }
1679                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1680                         # No trimming for unnamed arguments
1681                         $this->numberedExpansionCache[$index] = $this->parent->expand(
1682                                 $this->numberedArgs[$index],
1683                                 PPFrame::STRIP_COMMENTS
1684                         );
1685                 }
1686                 return $this->numberedExpansionCache[$index];
1687         }
1688
1689         public function getNamedArgument( $name ) {
1690                 if ( !isset( $this->namedArgs[$name] ) ) {
1691                         return false;
1692                 }
1693                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1694                         # Trim named arguments post-expand, for backwards compatibility
1695                         $this->namedExpansionCache[$name] = trim(
1696                                 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1697                 }
1698                 return $this->namedExpansionCache[$name];
1699         }
1700
1701         public function getArgument( $name ) {
1702                 $text = $this->getNumberedArgument( $name );
1703                 if ( $text === false ) {
1704                         $text = $this->getNamedArgument( $name );
1705                 }
1706                 return $text;
1707         }
1708
1709         /**
1710          * Return true if the frame is a template frame
1711          *
1712          * @return bool
1713          */
1714         public function isTemplate() {
1715                 return true;
1716         }
1717
1718         public function setVolatile( $flag = true ) {
1719                 parent::setVolatile( $flag );
1720                 $this->parent->setVolatile( $flag );
1721         }
1722
1723         public function setTTL( $ttl ) {
1724                 parent::setTTL( $ttl );
1725                 $this->parent->setTTL( $ttl );
1726         }
1727 }
1728
1729 /**
1730  * Expansion frame with custom arguments
1731  * @ingroup Parser
1732  * @codingStandardsIgnoreStart
1733  */
1734 class PPCustomFrame_DOM extends PPFrame_DOM {
1735         // @codingStandardsIgnoreEnd
1736
1737         public $args;
1738
1739         public function __construct( $preprocessor, $args ) {
1740                 parent::__construct( $preprocessor );
1741                 $this->args = $args;
1742         }
1743
1744         public function __toString() {
1745                 $s = 'cstmframe{';
1746                 $first = true;
1747                 foreach ( $this->args as $name => $value ) {
1748                         if ( $first ) {
1749                                 $first = false;
1750                         } else {
1751                                 $s .= ', ';
1752                         }
1753                         $s .= "\"$name\":\"" .
1754                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1755                 }
1756                 $s .= '}';
1757                 return $s;
1758         }
1759
1760         /**
1761          * @return bool
1762          */
1763         public function isEmpty() {
1764                 return !count( $this->args );
1765         }
1766
1767         public function getArgument( $index ) {
1768                 if ( !isset( $this->args[$index] ) ) {
1769                         return false;
1770                 }
1771                 return $this->args[$index];
1772         }
1773
1774         public function getArguments() {
1775                 return $this->args;
1776         }
1777 }
1778
1779 /**
1780  * @ingroup Parser
1781  * @codingStandardsIgnoreStart
1782  */
1783 class PPNode_DOM implements PPNode {
1784         // @codingStandardsIgnoreEnd
1785
1786         /**
1787          * @var DOMElement
1788          */
1789         public $node;
1790         public $xpath;
1791
1792         public function __construct( $node, $xpath = false ) {
1793                 $this->node = $node;
1794         }
1795
1796         /**
1797          * @return DOMXPath
1798          */
1799         public function getXPath() {
1800                 if ( $this->xpath === null ) {
1801                         $this->xpath = new DOMXPath( $this->node->ownerDocument );
1802                 }
1803                 return $this->xpath;
1804         }
1805
1806         public function __toString() {
1807                 if ( $this->node instanceof DOMNodeList ) {
1808                         $s = '';
1809                         foreach ( $this->node as $node ) {
1810                                 $s .= $node->ownerDocument->saveXML( $node );
1811                         }
1812                 } else {
1813                         $s = $this->node->ownerDocument->saveXML( $this->node );
1814                 }
1815                 return $s;
1816         }
1817
1818         /**
1819          * @return bool|PPNode_DOM
1820          */
1821         public function getChildren() {
1822                 return $this->node->childNodes ? new self( $this->node->childNodes ) : false;
1823         }
1824
1825         /**
1826          * @return bool|PPNode_DOM
1827          */
1828         public function getFirstChild() {
1829                 return $this->node->firstChild ? new self( $this->node->firstChild ) : false;
1830         }
1831
1832         /**
1833          * @return bool|PPNode_DOM
1834          */
1835         public function getNextSibling() {
1836                 return $this->node->nextSibling ? new self( $this->node->nextSibling ) : false;
1837         }
1838
1839         /**
1840          * @param string $type
1841          *
1842          * @return bool|PPNode_DOM
1843          */
1844         public function getChildrenOfType( $type ) {
1845                 return new self( $this->getXPath()->query( $type, $this->node ) );
1846         }
1847
1848         /**
1849          * @return int
1850          */
1851         public function getLength() {
1852                 if ( $this->node instanceof DOMNodeList ) {
1853                         return $this->node->length;
1854                 } else {
1855                         return false;
1856                 }
1857         }
1858
1859         /**
1860          * @param int $i
1861          * @return bool|PPNode_DOM
1862          */
1863         public function item( $i ) {
1864                 $item = $this->node->item( $i );
1865                 return $item ? new self( $item ) : false;
1866         }
1867
1868         /**
1869          * @return string
1870          */
1871         public function getName() {
1872                 if ( $this->node instanceof DOMNodeList ) {
1873                         return '#nodelist';
1874                 } else {
1875                         return $this->node->nodeName;
1876                 }
1877         }
1878
1879         /**
1880          * Split a "<part>" node into an associative array containing:
1881          *  - name          PPNode name
1882          *  - index         String index
1883          *  - value         PPNode value
1884          *
1885          * @throws MWException
1886          * @return array
1887          */
1888         public function splitArg() {
1889                 $xpath = $this->getXPath();
1890                 $names = $xpath->query( 'name', $this->node );
1891                 $values = $xpath->query( 'value', $this->node );
1892                 if ( !$names->length || !$values->length ) {
1893                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1894                 }
1895                 $name = $names->item( 0 );
1896                 $index = $name->getAttribute( 'index' );
1897                 return array(
1898                         'name' => new self( $name ),
1899                         'index' => $index,
1900                         'value' => new self( $values->item( 0 ) ) );
1901         }
1902
1903         /**
1904          * Split an "<ext>" node into an associative array containing name, attr, inner and close
1905          * All values in the resulting array are PPNodes. Inner and close are optional.
1906          *
1907          * @throws MWException
1908          * @return array
1909          */
1910         public function splitExt() {
1911                 $xpath = $this->getXPath();
1912                 $names = $xpath->query( 'name', $this->node );
1913                 $attrs = $xpath->query( 'attr', $this->node );
1914                 $inners = $xpath->query( 'inner', $this->node );
1915                 $closes = $xpath->query( 'close', $this->node );
1916                 if ( !$names->length || !$attrs->length ) {
1917                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1918                 }
1919                 $parts = array(
1920                         'name' => new self( $names->item( 0 ) ),
1921                         'attr' => new self( $attrs->item( 0 ) ) );
1922                 if ( $inners->length ) {
1923                         $parts['inner'] = new self( $inners->item( 0 ) );
1924                 }
1925                 if ( $closes->length ) {
1926                         $parts['close'] = new self( $closes->item( 0 ) );
1927                 }
1928                 return $parts;
1929         }
1930
1931         /**
1932          * Split a "<h>" node
1933          * @throws MWException
1934          * @return array
1935          */
1936         public function splitHeading() {
1937                 if ( $this->getName() !== 'h' ) {
1938                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1939                 }
1940                 return array(
1941                         'i' => $this->node->getAttribute( 'i' ),
1942                         'level' => $this->node->getAttribute( 'level' ),
1943                         'contents' => $this->getChildren()
1944                 );
1945         }
1946 }