includes/parser/Preprocessor_Hash.php

   1 <?php
   2 /**
   3  * Preprocessor using PHP arrays
   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  * Differences from DOM schema:
  26  *   * attribute nodes are children
  27  *   * "<h>" nodes that aren't at the top are replaced with <possible-h>
  28  * @ingroup Parser
  29  */
  30 class Preprocessor_Hash implements Preprocessor {
  31         /**
  32          * @var Parser
  33          */
  34         var $parser;
  35
  36         const CACHE_VERSION = 1;
  37
  38         function __construct( $parser ) {
  39                 $this->parser = $parser;
  40         }
  41
  42         /**
  43          * @return PPFrame_Hash
  44          */
  45         function newFrame() {
  46                 return new PPFrame_Hash( $this );
  47         }
  48
  49         /**
  50          * @param array $args
  51          * @return PPCustomFrame_Hash
  52          */
  53         function newCustomFrame( $args ) {
  54                 return new PPCustomFrame_Hash( $this, $args );
  55         }
  56
  57         /**
  58          * @param array $values
  59          * @return PPNode_Hash_Array
  60          */
  61         function newPartNodeArray( $values ) {
  62                 $list = array();
  63
  64                 foreach ( $values as $k => $val ) {
  65                         $partNode = new PPNode_Hash_Tree( 'part' );
  66                         $nameNode = new PPNode_Hash_Tree( 'name' );
  67
  68                         if ( is_int( $k ) ) {
  69                                 $nameNode->addChild( new PPNode_Hash_Attr( 'index', $k ) );
  70                                 $partNode->addChild( $nameNode );
  71                         } else {
  72                                 $nameNode->addChild( new PPNode_Hash_Text( $k ) );
  73                                 $partNode->addChild( $nameNode );
  74                                 $partNode->addChild( new PPNode_Hash_Text( '=' ) );
  75                         }
  76
  77                         $valueNode = new PPNode_Hash_Tree( 'value' );
  78                         $valueNode->addChild( new PPNode_Hash_Text( $val ) );
  79                         $partNode->addChild( $valueNode );
  80
  81                         $list[] = $partNode;
  82                 }
  83
  84                 $node = new PPNode_Hash_Array( $list );
  85                 return $node;
  86         }
  87
  88         /**
  89          * Preprocess some wikitext and return the document tree.
  90          * This is the ghost of Parser::replace_variables().
  91          *
  92          * @param string $text The text to parse
  93          * @param int $flags Bitwise combination of:
  94          *    Parser::PTD_FOR_INCLUSION    Handle "<noinclude>" and "<includeonly>" as if the text is being
  95          *                                 included. Default is to assume a direct page view.
  96          *
  97          * The generated DOM tree must depend only on the input text and the flags.
  98          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
  99          *
 100          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
 101          * change in the DOM tree for a given text, must be passed through the section identifier
 102          * in the section edit link and thus back to extractSections().
 103          *
 104          * The output of this function is currently only cached in process memory, but a persistent
 105          * cache may be implemented at a later date which takes further advantage of these strict
 106          * dependency requirements.
 107          *
 108          * @throws MWException
 109          * @return PPNode_Hash_Tree
 110          */
 111         function preprocessToObj( $text, $flags = 0 ) {
 112                 wfProfileIn( __METHOD__ );
 113
 114                 // Check cache.
 115                 global $wgMemc, $wgPreprocessorCacheThreshold;
 116
 117                 $cacheable = $wgPreprocessorCacheThreshold !== false
 118                         && strlen( $text ) > $wgPreprocessorCacheThreshold;
 119
 120                 if ( $cacheable ) {
 121                         wfProfileIn( __METHOD__ . '-cacheable' );
 122
 123                         $cacheKey = wfMemcKey( 'preprocess-hash', md5( $text ), $flags );
 124                         $cacheValue = $wgMemc->get( $cacheKey );
 125                         if ( $cacheValue ) {
 126                                 $version = substr( $cacheValue, 0, 8 );
 127                                 if ( intval( $version ) == self::CACHE_VERSION ) {
 128                                         $hash = unserialize( substr( $cacheValue, 8 ) );
 129                                         // From the cache
 130                                         wfDebugLog( "Preprocessor",
 131                                                 "Loaded preprocessor hash from memcached (key $cacheKey)" );
 132                                         wfProfileOut( __METHOD__ . '-cacheable' );
 133                                         wfProfileOut( __METHOD__ );
 134                                         return $hash;
 135                                 }
 136                         }
 137                         wfProfileIn( __METHOD__ . '-cache-miss' );
 138                 }
 139
 140                 $rules = array(
 141                         '{' => array(
 142                                 'end' => '}',
 143                                 'names' => array(
 144                                         2 => 'template',
 145                                         3 => 'tplarg',
 146                                 ),
 147                                 'min' => 2,
 148                                 'max' => 3,
 149                         ),
 150                         '[' => array(
 151                                 'end' => ']',
 152                                 'names' => array( 2 => null ),
 153                                 'min' => 2,
 154                                 'max' => 2,
 155                         )
 156                 );
 157
 158                 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
 159
 160                 $xmlishElements = $this->parser->getStripList();
 161                 $enableOnlyinclude = false;
 162                 if ( $forInclusion ) {
 163                         $ignoredTags = array( 'includeonly', '/includeonly' );
 164                         $ignoredElements = array( 'noinclude' );
 165                         $xmlishElements[] = 'noinclude';
 166                         if ( strpos( $text, '<onlyinclude>' ) !== false
 167                                 && strpos( $text, '</onlyinclude>' ) !== false
 168                         ) {
 169                                 $enableOnlyinclude = true;
 170                         }
 171                 } else {
 172                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
 173                         $ignoredElements = array( 'includeonly' );
 174                         $xmlishElements[] = 'includeonly';
 175                 }
 176                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
 177
 178                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
 179                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
 180
 181                 $stack = new PPDStack_Hash;
 182
 183                 $searchBase = "[{<\n";
 184                 // For fast reverse searches
 185                 $revText = strrev( $text );
 186                 $lengthText = strlen( $text );
 187
 188                 // Input pointer, starts out pointing to a pseudo-newline before the start
 189                 $i = 0;
 190                 // Current accumulator
 191                 $accum =& $stack->getAccum();
 192                 // True to find equals signs in arguments
 193                 $findEquals = false;
 194                 // True to take notice of pipe characters
 195                 $findPipe = false;
 196                 $headingIndex = 1;
 197                 // True if $i is inside a possible heading
 198                 $inHeading = false;
 199                 // True if there are no more greater-than (>) signs right of $i
 200                 $noMoreGT = false;
 201                 // True to ignore all input up to the next <onlyinclude>
 202                 $findOnlyinclude = $enableOnlyinclude;
 203                 // Do a line-start run without outputting an LF character
 204                 $fakeLineStart = true;
 205
 206                 while ( true ) {
 207                         //$this->memCheck();
 208
 209                         if ( $findOnlyinclude ) {
 210                                 // Ignore all input up to the next <onlyinclude>
 211                                 $startPos = strpos( $text, '<onlyinclude>', $i );
 212                                 if ( $startPos === false ) {
 213                                         // Ignored section runs to the end
 214                                         $accum->addNodeWithText( 'ignore', substr( $text, $i ) );
 215                                         break;
 216                                 }
 217                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
 218                                 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i ) );
 219                                 $i = $tagEndPos;
 220                                 $findOnlyinclude = false;
 221                         }
 222
 223                         if ( $fakeLineStart ) {
 224                                 $found = 'line-start';
 225                                 $curChar = '';
 226                         } else {
 227                                 # Find next opening brace, closing brace or pipe
 228                                 $search = $searchBase;
 229                                 if ( $stack->top === false ) {
 230                                         $currentClosing = '';
 231                                 } else {
 232                                         $currentClosing = $stack->top->close;
 233                                         $search .= $currentClosing;
 234                                 }
 235                                 if ( $findPipe ) {
 236                                         $search .= '|';
 237                                 }
 238                                 if ( $findEquals ) {
 239                                         // First equals will be for the template
 240                                         $search .= '=';
 241                                 }
 242                                 $rule = null;
 243                                 # Output literal section, advance input counter
 244                                 $literalLength = strcspn( $text, $search, $i );
 245                                 if ( $literalLength > 0 ) {
 246                                         $accum->addLiteral( substr( $text, $i, $literalLength ) );
 247                                         $i += $literalLength;
 248                                 }
 249                                 if ( $i >= $lengthText ) {
 250                                         if ( $currentClosing == "\n" ) {
 251                                                 // Do a past-the-end run to finish off the heading
 252                                                 $curChar = '';
 253                                                 $found = 'line-end';
 254                                         } else {
 255                                                 # All done
 256                                                 break;
 257                                         }
 258                                 } else {
 259                                         $curChar = $text[$i];
 260                                         if ( $curChar == '|' ) {
 261                                                 $found = 'pipe';
 262                                         } elseif ( $curChar == '=' ) {
 263                                                 $found = 'equals';
 264                                         } elseif ( $curChar == '<' ) {
 265                                                 $found = 'angle';
 266                                         } elseif ( $curChar == "\n" ) {
 267                                                 if ( $inHeading ) {
 268                                                         $found = 'line-end';
 269                                                 } else {
 270                                                         $found = 'line-start';
 271                                                 }
 272                                         } elseif ( $curChar == $currentClosing ) {
 273                                                 $found = 'close';
 274                                         } elseif ( isset( $rules[$curChar] ) ) {
 275                                                 $found = 'open';
 276                                                 $rule = $rules[$curChar];
 277                                         } else {
 278                                                 # Some versions of PHP have a strcspn which stops on null characters
 279                                                 # Ignore and continue
 280                                                 ++$i;
 281                                                 continue;
 282                                         }
 283                                 }
 284                         }
 285
 286                         if ( $found == 'angle' ) {
 287                                 $matches = false;
 288                                 // Handle </onlyinclude>
 289                                 if ( $enableOnlyinclude
 290                                         && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>'
 291                                 ) {
 292                                         $findOnlyinclude = true;
 293                                         continue;
 294                                 }
 295
 296                                 // Determine element name
 297                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
 298                                         // Element name missing or not listed
 299                                         $accum->addLiteral( '<' );
 300                                         ++$i;
 301                                         continue;
 302                                 }
 303                                 // Handle comments
 304                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
 305
 306                                         // To avoid leaving blank lines, when a sequence of
 307                                         // space-separated comments is both preceded and followed by
 308                                         // a newline (ignoring spaces), then
 309                                         // trim leading and trailing spaces and the trailing newline.
 310
 311                                         // Find the end
 312                                         $endPos = strpos( $text, '-->', $i + 4 );
 313                                         if ( $endPos === false ) {
 314                                                 // Unclosed comment in input, runs to end
 315                                                 $inner = substr( $text, $i );
 316                                                 $accum->addNodeWithText( 'comment', $inner );
 317                                                 $i = $lengthText;
 318                                         } else {
 319                                                 // Search backwards for leading whitespace
 320                                                 $wsStart = $i ? ( $i - strspn( $revText, " \t", $lengthText - $i ) ) : 0;
 321
 322                                                 // Search forwards for trailing whitespace
 323                                                 // $wsEnd will be the position of the last space (or the '>' if there's none)
 324                                                 $wsEnd = $endPos + 2 + strspn( $text, " \t", $endPos + 3 );
 325
 326                                                 // Keep looking forward as long as we're finding more
 327                                                 // comments.
 328                                                 $comments = array( array( $wsStart, $wsEnd ) );
 329                                                 while ( substr( $text, $wsEnd + 1, 4 ) == '<!--' ) {
 330                                                         $c = strpos( $text, '-->', $wsEnd + 4 );
 331                                                         if ( $c === false ) {
 332                                                                 break;
 333                                                         }
 334                                                         $c = $c + 2 + strspn( $text, " \t", $c + 3 );
 335                                                         $comments[] = array( $wsEnd + 1, $c );
 336                                                         $wsEnd = $c;
 337                                                 }
 338
 339                                                 // Eat the line if possible
 340                                                 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
 341                                                 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
 342                                                 // it's a possible beneficial b/c break.
 343                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
 344                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n"
 345                                                 ) {
 346                                                         // Remove leading whitespace from the end of the accumulator
 347                                                         // Sanity check first though
 348                                                         $wsLength = $i - $wsStart;
 349                                                         if ( $wsLength > 0
 350                                                                 && $accum->lastNode instanceof PPNode_Hash_Text
 351                                                                 && strspn( $accum->lastNode->value, " \t", -$wsLength ) === $wsLength
 352                                                         ) {
 353                                                                 $accum->lastNode->value = substr( $accum->lastNode->value, 0, -$wsLength );
 354                                                         }
 355
 356                                                         // Dump all but the last comment to the accumulator
 357                                                         foreach ( $comments as $j => $com ) {
 358                                                                 $startPos = $com[0];
 359                                                                 $endPos = $com[1] + 1;
 360                                                                 if ( $j == ( count( $comments ) - 1 ) ) {
 361                                                                         break;
 362                                                                 }
 363                                                                 $inner = substr( $text, $startPos, $endPos - $startPos );
 364                                                                 $accum->addNodeWithText( 'comment', $inner );
 365                                                         }
 366
 367                                                         // Do a line-start run next time to look for headings after the comment
 368                                                         $fakeLineStart = true;
 369                                                 } else {
 370                                                         // No line to eat, just take the comment itself
 371                                                         $startPos = $i;
 372                                                         $endPos += 2;
 373                                                 }
 374
 375                                                 if ( $stack->top ) {
 376                                                         $part = $stack->top->getCurrentPart();
 377                                                         if ( !( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) ) {
 378                                                                 $part->visualEnd = $wsStart;
 379                                                         }
 380                                                         // Else comments abutting, no change in visual end
 381                                                         $part->commentEnd = $endPos;
 382                                                 }
 383                                                 $i = $endPos + 1;
 384                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
 385                                                 $accum->addNodeWithText( 'comment', $inner );
 386                                         }
 387                                         continue;
 388                                 }
 389                                 $name = $matches[1];
 390                                 $lowerName = strtolower( $name );
 391                                 $attrStart = $i + strlen( $name ) + 1;
 392
 393                                 // Find end of tag
 394                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
 395                                 if ( $tagEndPos === false ) {
 396                                         // Infinite backtrack
 397                                         // Disable tag search to prevent worst-case O(N^2) performance
 398                                         $noMoreGT = true;
 399                                         $accum->addLiteral( '<' );
 400                                         ++$i;
 401                                         continue;
 402                                 }
 403
 404                                 // Handle ignored tags
 405                                 if ( in_array( $lowerName, $ignoredTags ) ) {
 406                                         $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i + 1 ) );
 407                                         $i = $tagEndPos + 1;
 408                                         continue;
 409                                 }
 410
 411                                 $tagStartPos = $i;
 412                                 if ( $text[$tagEndPos - 1] == '/' ) {
 413                                         // Short end tag
 414                                         $attrEnd = $tagEndPos - 1;
 415                                         $inner = null;
 416                                         $i = $tagEndPos + 1;
 417                                         $close = null;
 418                                 } else {
 419                                         $attrEnd = $tagEndPos;
 420                                         // Find closing tag
 421                                         if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
 422                                                         $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 )
 423                                         ) {
 424                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
 425                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
 426                                                 $close = $matches[0][0];
 427                                         } else {
 428                                                 // No end tag -- let it run out to the end of the text.
 429                                                 $inner = substr( $text, $tagEndPos + 1 );
 430                                                 $i = $lengthText;
 431                                                 $close = null;
 432                                         }
 433                                 }
 434                                 // <includeonly> and <noinclude> just become <ignore> tags
 435                                 if ( in_array( $lowerName, $ignoredElements ) ) {
 436                                         $accum->addNodeWithText( 'ignore', substr( $text, $tagStartPos, $i - $tagStartPos ) );
 437                                         continue;
 438                                 }
 439
 440                                 if ( $attrEnd <= $attrStart ) {
 441                                         $attr = '';
 442                                 } else {
 443                                         // Note that the attr element contains the whitespace between name and attribute,
 444                                         // this is necessary for precise reconstruction during pre-save transform.
 445                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
 446                                 }
 447
 448                                 $extNode = new PPNode_Hash_Tree( 'ext' );
 449                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'name', $name ) );
 450                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'attr', $attr ) );
 451                                 if ( $inner !== null ) {
 452                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'inner', $inner ) );
 453                                 }
 454                                 if ( $close !== null ) {
 455                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'close', $close ) );
 456                                 }
 457                                 $accum->addNode( $extNode );
 458                         } elseif ( $found == 'line-start' ) {
 459                                 // Is this the start of a heading?
 460                                 // Line break belongs before the heading element in any case
 461                                 if ( $fakeLineStart ) {
 462                                         $fakeLineStart = false;
 463                                 } else {
 464                                         $accum->addLiteral( $curChar );
 465                                         $i++;
 466                                 }
 467
 468                                 $count = strspn( $text, '=', $i, 6 );
 469                                 if ( $count == 1 && $findEquals ) {
 470                                         // DWIM: This looks kind of like a name/value separator.
 471                                         // Let's let the equals handler have it and break the potential
 472                                         // heading. This is heuristic, but AFAICT the methods for
 473                                         // completely correct disambiguation are very complex.
 474                                 } elseif ( $count > 0 ) {
 475                                         $piece = array(
 476                                                 'open' => "\n",
 477                                                 'close' => "\n",
 478                                                 'parts' => array( new PPDPart_Hash( str_repeat( '=', $count ) ) ),
 479                                                 'startPos' => $i,
 480                                                 'count' => $count );
 481                                         $stack->push( $piece );
 482                                         $accum =& $stack->getAccum();
 483                                         extract( $stack->getFlags() );
 484                                         $i += $count;
 485                                 }
 486                         } elseif ( $found == 'line-end' ) {
 487                                 $piece = $stack->top;
 488                                 // A heading must be open, otherwise \n wouldn't have been in the search list
 489                                 assert( '$piece->open == "\n"' );
 490                                 $part = $piece->getCurrentPart();
 491                                 // Search back through the input to see if it has a proper close.
 492                                 // Do this using the reversed string since the other solutions
 493                                 // (end anchor, etc.) are inefficient.
 494                                 $wsLength = strspn( $revText, " \t", $lengthText - $i );
 495                                 $searchStart = $i - $wsLength;
 496                                 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
 497                                         // Comment found at line end
 498                                         // Search for equals signs before the comment
 499                                         $searchStart = $part->visualEnd;
 500                                         $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
 501                                 }
 502                                 $count = $piece->count;
 503                                 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
 504                                 if ( $equalsLength > 0 ) {
 505                                         if ( $searchStart - $equalsLength == $piece->startPos ) {
 506                                                 // This is just a single string of equals signs on its own line
 507                                                 // Replicate the doHeadings behavior /={count}(.+)={count}/
 508                                                 // First find out how many equals signs there really are (don't stop at 6)
 509                                                 $count = $equalsLength;
 510                                                 if ( $count < 3 ) {
 511                                                         $count = 0;
 512                                                 } else {
 513                                                         $count = min( 6, intval( ( $count - 1 ) / 2 ) );
 514                                                 }
 515                                         } else {
 516                                                 $count = min( $equalsLength, $count );
 517                                         }
 518                                         if ( $count > 0 ) {
 519                                                 // Normal match, output <h>
 520                                                 $element = new PPNode_Hash_Tree( 'possible-h' );
 521                                                 $element->addChild( new PPNode_Hash_Attr( 'level', $count ) );
 522                                                 $element->addChild( new PPNode_Hash_Attr( 'i', $headingIndex++ ) );
 523                                                 $element->lastChild->nextSibling = $accum->firstNode;
 524                                                 $element->lastChild = $accum->lastNode;
 525                                         } else {
 526                                                 // Single equals sign on its own line, count=0
 527                                                 $element = $accum;
 528                                         }
 529                                 } else {
 530                                         // No match, no <h>, just pass down the inner text
 531                                         $element = $accum;
 532                                 }
 533                                 // Unwind the stack
 534                                 $stack->pop();
 535                                 $accum =& $stack->getAccum();
 536                                 extract( $stack->getFlags() );
 537
 538                                 // Append the result to the enclosing accumulator
 539                                 if ( $element instanceof PPNode ) {
 540                                         $accum->addNode( $element );
 541                                 } else {
 542                                         $accum->addAccum( $element );
 543                                 }
 544                                 // Note that we do NOT increment the input pointer.
 545                                 // This is because the closing linebreak could be the opening linebreak of
 546                                 // another heading. Infinite loops are avoided because the next iteration MUST
 547                                 // hit the heading open case above, which unconditionally increments the
 548                                 // input pointer.
 549                         } elseif ( $found == 'open' ) {
 550                                 # count opening brace characters
 551                                 $count = strspn( $text, $curChar, $i );
 552
 553                                 # we need to add to stack only if opening brace count is enough for one of the rules
 554                                 if ( $count >= $rule['min'] ) {
 555                                         # Add it to the stack
 556                                         $piece = array(
 557                                                 'open' => $curChar,
 558                                                 'close' => $rule['end'],
 559                                                 'count' => $count,
 560                                                 'lineStart' => ( $i > 0 && $text[$i - 1] == "\n" ),
 561                                         );
 562
 563                                         $stack->push( $piece );
 564                                         $accum =& $stack->getAccum();
 565                                         extract( $stack->getFlags() );
 566                                 } else {
 567                                         # Add literal brace(s)
 568                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 569                                 }
 570                                 $i += $count;
 571                         } elseif ( $found == 'close' ) {
 572                                 $piece = $stack->top;
 573                                 # lets check if there are enough characters for closing brace
 574                                 $maxCount = $piece->count;
 575                                 $count = strspn( $text, $curChar, $i, $maxCount );
 576
 577                                 # check for maximum matching characters (if there are 5 closing
 578                                 # characters, we will probably need only 3 - depending on the rules)
 579                                 $rule = $rules[$piece->open];
 580                                 if ( $count > $rule['max'] ) {
 581                                         # The specified maximum exists in the callback array, unless the caller
 582                                         # has made an error
 583                                         $matchingCount = $rule['max'];
 584                                 } else {
 585                                         # Count is less than the maximum
 586                                         # Skip any gaps in the callback array to find the true largest match
 587                                         # Need to use array_key_exists not isset because the callback can be null
 588                                         $matchingCount = $count;
 589                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
 590                                                 --$matchingCount;
 591                                         }
 592                                 }
 593
 594                                 if ( $matchingCount <= 0 ) {
 595                                         # No matching element found in callback array
 596                                         # Output a literal closing brace and continue
 597                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 598                                         $i += $count;
 599                                         continue;
 600                                 }
 601                                 $name = $rule['names'][$matchingCount];
 602                                 if ( $name === null ) {
 603                                         // No element, just literal text
 604                                         $element = $piece->breakSyntax( $matchingCount );
 605                                         $element->addLiteral( str_repeat( $rule['end'], $matchingCount ) );
 606                                 } else {
 607                                         # Create XML element
 608                                         # Note: $parts is already XML, does not need to be encoded further
 609                                         $parts = $piece->parts;
 610                                         $titleAccum = $parts[0]->out;
 611                                         unset( $parts[0] );
 612
 613                                         $element = new PPNode_Hash_Tree( $name );
 614
 615                                         # The invocation is at the start of the line if lineStart is set in
 616                                         # the stack, and all opening brackets are used up.
 617                                         if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
 618                                                 $element->addChild( new PPNode_Hash_Attr( 'lineStart', 1 ) );
 619                                         }
 620                                         $titleNode = new PPNode_Hash_Tree( 'title' );
 621                                         $titleNode->firstChild = $titleAccum->firstNode;
 622                                         $titleNode->lastChild = $titleAccum->lastNode;
 623                                         $element->addChild( $titleNode );
 624                                         $argIndex = 1;
 625                                         foreach ( $parts as $part ) {
 626                                                 if ( isset( $part->eqpos ) ) {
 627                                                         // Find equals
 628                                                         $lastNode = false;
 629                                                         for ( $node = $part->out->firstNode; $node; $node = $node->nextSibling ) {
 630                                                                 if ( $node === $part->eqpos ) {
 631                                                                         break;
 632                                                                 }
 633                                                                 $lastNode = $node;
 634                                                         }
 635                                                         if ( !$node ) {
 636                                                                 if ( $cacheable ) {
 637                                                                         wfProfileOut( __METHOD__ . '-cache-miss' );
 638                                                                         wfProfileOut( __METHOD__ . '-cacheable' );
 639                                                                 }
 640                                                                 wfProfileOut( __METHOD__ );
 641                                                                 throw new MWException( __METHOD__ . ': eqpos not found' );
 642                                                         }
 643                                                         if ( $node->name !== 'equals' ) {
 644                                                                 if ( $cacheable ) {
 645                                                                         wfProfileOut( __METHOD__ . '-cache-miss' );
 646                                                                         wfProfileOut( __METHOD__ . '-cacheable' );
 647                                                                 }
 648                                                                 wfProfileOut( __METHOD__ );
 649                                                                 throw new MWException( __METHOD__ . ': eqpos is not equals' );
 650                                                         }
 651                                                         $equalsNode = $node;
 652
 653                                                         // Construct name node
 654                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 655                                                         if ( $lastNode !== false ) {
 656                                                                 $lastNode->nextSibling = false;
 657                                                                 $nameNode->firstChild = $part->out->firstNode;
 658                                                                 $nameNode->lastChild = $lastNode;
 659                                                         }
 660
 661                                                         // Construct value node
 662                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 663                                                         if ( $equalsNode->nextSibling !== false ) {
 664                                                                 $valueNode->firstChild = $equalsNode->nextSibling;
 665                                                                 $valueNode->lastChild = $part->out->lastNode;
 666                                                         }
 667                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 668                                                         $partNode->addChild( $nameNode );
 669                                                         $partNode->addChild( $equalsNode->firstChild );
 670                                                         $partNode->addChild( $valueNode );
 671                                                         $element->addChild( $partNode );
 672                                                 } else {
 673                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 674                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 675                                                         $nameNode->addChild( new PPNode_Hash_Attr( 'index', $argIndex++ ) );
 676                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 677                                                         $valueNode->firstChild = $part->out->firstNode;
 678                                                         $valueNode->lastChild = $part->out->lastNode;
 679                                                         $partNode->addChild( $nameNode );
 680                                                         $partNode->addChild( $valueNode );
 681                                                         $element->addChild( $partNode );
 682                                                 }
 683                                         }
 684                                 }
 685
 686                                 # Advance input pointer
 687                                 $i += $matchingCount;
 688
 689                                 # Unwind the stack
 690                                 $stack->pop();
 691                                 $accum =& $stack->getAccum();
 692
 693                                 # Re-add the old stack element if it still has unmatched opening characters remaining
 694                                 if ( $matchingCount < $piece->count ) {
 695                                         $piece->parts = array( new PPDPart_Hash );
 696                                         $piece->count -= $matchingCount;
 697                                         # do we still qualify for any callback with remaining count?
 698                                         $min = $rules[$piece->open]['min'];
 699                                         if ( $piece->count >= $min ) {
 700                                                 $stack->push( $piece );
 701                                                 $accum =& $stack->getAccum();
 702                                         } else {
 703                                                 $accum->addLiteral( str_repeat( $piece->open, $piece->count ) );
 704                                         }
 705                                 }
 706
 707                                 extract( $stack->getFlags() );
 708
 709                                 # Add XML element to the enclosing accumulator
 710                                 if ( $element instanceof PPNode ) {
 711                                         $accum->addNode( $element );
 712                                 } else {
 713                                         $accum->addAccum( $element );
 714                                 }
 715                         } elseif ( $found == 'pipe' ) {
 716                                 $findEquals = true; // shortcut for getFlags()
 717                                 $stack->addPart();
 718                                 $accum =& $stack->getAccum();
 719                                 ++$i;
 720                         } elseif ( $found == 'equals' ) {
 721                                 $findEquals = false; // shortcut for getFlags()
 722                                 $accum->addNodeWithText( 'equals', '=' );
 723                                 $stack->getCurrentPart()->eqpos = $accum->lastNode;
 724                                 ++$i;
 725                         }
 726                 }
 727
 728                 # Output any remaining unclosed brackets
 729                 foreach ( $stack->stack as $piece ) {
 730                         $stack->rootAccum->addAccum( $piece->breakSyntax() );
 731                 }
 732
 733                 # Enable top-level headings
 734                 for ( $node = $stack->rootAccum->firstNode; $node; $node = $node->nextSibling ) {
 735                         if ( isset( $node->name ) && $node->name === 'possible-h' ) {
 736                                 $node->name = 'h';
 737                         }
 738                 }
 739
 740                 $rootNode = new PPNode_Hash_Tree( 'root' );
 741                 $rootNode->firstChild = $stack->rootAccum->firstNode;
 742                 $rootNode->lastChild = $stack->rootAccum->lastNode;
 743
 744                 // Cache
 745                 if ( $cacheable ) {
 746                         $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . serialize( $rootNode );
 747                         $wgMemc->set( $cacheKey, $cacheValue, 86400 );
 748                         wfProfileOut( __METHOD__ . '-cache-miss' );
 749                         wfProfileOut( __METHOD__ . '-cacheable' );
 750                         wfDebugLog( "Preprocessor", "Saved preprocessor Hash to memcached (key $cacheKey)" );
 751                 }
 752
 753                 wfProfileOut( __METHOD__ );
 754                 return $rootNode;
 755         }
 756 }
 757
 758 /**
 759  * Stack class to help Preprocessor::preprocessToObj()
 760  * @ingroup Parser
 761  */
 762 class PPDStack_Hash extends PPDStack {
 763         function __construct() {
 764                 $this->elementClass = 'PPDStackElement_Hash';
 765                 parent::__construct();
 766                 $this->rootAccum = new PPDAccum_Hash;
 767         }
 768 }
 769
 770 /**
 771  * @ingroup Parser
 772  */
 773 class PPDStackElement_Hash extends PPDStackElement {
 774         function __construct( $data = array() ) {
 775                 $this->partClass = 'PPDPart_Hash';
 776                 parent::__construct( $data );
 777         }
 778
 779         /**
 780          * Get the accumulator that would result if the close is not found.
 781          *
 782          * @param int|bool $openingCount
 783          * @return PPDAccum_Hash
 784          */
 785         function breakSyntax( $openingCount = false ) {
 786                 if ( $this->open == "\n" ) {
 787                         $accum = $this->parts[0]->out;
 788                 } else {
 789                         if ( $openingCount === false ) {
 790                                 $openingCount = $this->count;
 791                         }
 792                         $accum = new PPDAccum_Hash;
 793                         $accum->addLiteral( str_repeat( $this->open, $openingCount ) );
 794                         $first = true;
 795                         foreach ( $this->parts as $part ) {
 796                                 if ( $first ) {
 797                                         $first = false;
 798                                 } else {
 799                                         $accum->addLiteral( '|' );
 800                                 }
 801                                 $accum->addAccum( $part->out );
 802                         }
 803                 }
 804                 return $accum;
 805         }
 806 }
 807
 808 /**
 809  * @ingroup Parser
 810  */
 811 class PPDPart_Hash extends PPDPart {
 812         function __construct( $out = '' ) {
 813                 $accum = new PPDAccum_Hash;
 814                 if ( $out !== '' ) {
 815                         $accum->addLiteral( $out );
 816                 }
 817                 parent::__construct( $accum );
 818         }
 819 }
 820
 821 /**
 822  * @ingroup Parser
 823  */
 824 class PPDAccum_Hash {
 825         var $firstNode, $lastNode;
 826
 827         function __construct() {
 828                 $this->firstNode = $this->lastNode = false;
 829         }
 830
 831         /**
 832          * Append a string literal
 833          * @param string $s
 834          */
 835         function addLiteral( $s ) {
 836                 if ( $this->lastNode === false ) {
 837                         $this->firstNode = $this->lastNode = new PPNode_Hash_Text( $s );
 838                 } elseif ( $this->lastNode instanceof PPNode_Hash_Text ) {
 839                         $this->lastNode->value .= $s;
 840                 } else {
 841                         $this->lastNode->nextSibling = new PPNode_Hash_Text( $s );
 842                         $this->lastNode = $this->lastNode->nextSibling;
 843                 }
 844         }
 845
 846         /**
 847          * Append a PPNode
 848          * @param PPNode $node
 849          */
 850         function addNode( PPNode $node ) {
 851                 if ( $this->lastNode === false ) {
 852                         $this->firstNode = $this->lastNode = $node;
 853                 } else {
 854                         $this->lastNode->nextSibling = $node;
 855                         $this->lastNode = $node;
 856                 }
 857         }
 858
 859         /**
 860          * Append a tree node with text contents
 861          * @param string $name
 862          * @param string $value
 863          */
 864         function addNodeWithText( $name, $value ) {
 865                 $node = PPNode_Hash_Tree::newWithText( $name, $value );
 866                 $this->addNode( $node );
 867         }
 868
 869         /**
 870          * Append a PPDAccum_Hash
 871          * Takes over ownership of the nodes in the source argument. These nodes may
 872          * subsequently be modified, especially nextSibling.
 873          * @param PPDAccum_Hash $accum
 874          */
 875         function addAccum( $accum ) {
 876                 if ( $accum->lastNode === false ) {
 877                         // nothing to add
 878                 } elseif ( $this->lastNode === false ) {
 879                         $this->firstNode = $accum->firstNode;
 880                         $this->lastNode = $accum->lastNode;
 881                 } else {
 882                         $this->lastNode->nextSibling = $accum->firstNode;
 883                         $this->lastNode = $accum->lastNode;
 884                 }
 885         }
 886 }
 887
 888 /**
 889  * An expansion frame, used as a context to expand the result of preprocessToObj()
 890  * @ingroup Parser
 891  */
 892 class PPFrame_Hash implements PPFrame {
 893
 894         /**
 895          * @var Parser
 896          */
 897         var $parser;
 898
 899         /**
 900          * @var Preprocessor
 901          */
 902         var $preprocessor;
 903
 904         /**
 905          * @var Title
 906          */
 907         var $title;
 908         var $titleCache;
 909
 910         /**
 911          * Hashtable listing templates which are disallowed for expansion in this frame,
 912          * having been encountered previously in parent frames.
 913          */
 914         var $loopCheckHash;
 915
 916         /**
 917          * Recursion depth of this frame, top = 0
 918          * Note that this is NOT the same as expansion depth in expand()
 919          */
 920         var $depth;
 921
 922         private $volatile = false;
 923
 924         /**
 925          * @var array
 926          */
 927         protected $childExpansionCache;
 928
 929         /**
 930          * Construct a new preprocessor frame.
 931          * @param Preprocessor $preprocessor The parent preprocessor
 932          */
 933         function __construct( $preprocessor ) {
 934                 $this->preprocessor = $preprocessor;
 935                 $this->parser = $preprocessor->parser;
 936                 $this->title = $this->parser->mTitle;
 937                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
 938                 $this->loopCheckHash = array();
 939                 $this->depth = 0;
 940                 $this->childExpansionCache = array();
 941         }
 942
 943         /**
 944          * Create a new child frame
 945          * $args is optionally a multi-root PPNode or array containing the template arguments
 946          *
 947          * @param array|bool|PPNode_Hash_Array $args
 948          * @param Title|bool $title
 949          * @param int $indexOffset
 950          * @throws MWException
 951          * @return PPTemplateFrame_Hash
 952          */
 953         function newChild( $args = false, $title = false, $indexOffset = 0 ) {
 954                 $namedArgs = array();
 955                 $numberedArgs = array();
 956                 if ( $title === false ) {
 957                         $title = $this->title;
 958                 }
 959                 if ( $args !== false ) {
 960                         if ( $args instanceof PPNode_Hash_Array ) {
 961                                 $args = $args->value;
 962                         } elseif ( !is_array( $args ) ) {
 963                                 throw new MWException( __METHOD__ . ': $args must be array or PPNode_Hash_Array' );
 964                         }
 965                         foreach ( $args as $arg ) {
 966                                 $bits = $arg->splitArg();
 967                                 if ( $bits['index'] !== '' ) {
 968                                         // Numbered parameter
 969                                         $index = $bits['index'] - $indexOffset;
 970                                         $numberedArgs[$index] = $bits['value'];
 971                                         unset( $namedArgs[$index] );
 972                                 } else {
 973                                         // Named parameter
 974                                         $name = trim( $this->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
 975                                         $namedArgs[$name] = $bits['value'];
 976                                         unset( $numberedArgs[$name] );
 977                                 }
 978                         }
 979                 }
 980                 return new PPTemplateFrame_Hash( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
 981         }
 982
 983         /**
 984          * @throws MWException
 985          * @param string|int $key
 986          * @param string|PPNode_Hash|DOMDocument $root
 987          * @param int $flags
 988          * @return string
 989          */
 990         function cachedExpand( $key, $root, $flags = 0 ) {
 991                 // we don't have a parent, so we don't have a cache
 992                 return $this->expand( $root, $flags );
 993         }
 994
 995         /**
 996          * @throws MWException
 997          * @param string|PPNode$root
 998          * @param int $flags
 999          * @return string
1000          */
1001         function expand( $root, $flags = 0 ) {
1002                 static $expansionDepth = 0;
1003                 if ( is_string( $root ) ) {
1004                         return $root;
1005                 }
1006
1007                 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) {
1008                         $this->parser->limitationWarn( 'node-count-exceeded',
1009                                         $this->parser->mPPNodeCount,
1010                                         $this->parser->mOptions->getMaxPPNodeCount()
1011                         );
1012                         return '<span class="error">Node-count limit exceeded</span>';
1013                 }
1014                 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
1015                         $this->parser->limitationWarn( 'expansion-depth-exceeded',
1016                                         $expansionDepth,
1017                                         $this->parser->mOptions->getMaxPPExpandDepth()
1018                         );
1019                         return '<span class="error">Expansion depth limit exceeded</span>';
1020                 }
1021                 ++$expansionDepth;
1022                 if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
1023                         $this->parser->mHighestExpansionDepth = $expansionDepth;
1024                 }
1025
1026                 $outStack = array( '', '' );
1027                 $iteratorStack = array( false, $root );
1028                 $indexStack = array( 0, 0 );
1029
1030                 while ( count( $iteratorStack ) > 1 ) {
1031                         $level = count( $outStack ) - 1;
1032                         $iteratorNode =& $iteratorStack[$level];
1033                         $out =& $outStack[$level];
1034                         $index =& $indexStack[$level];
1035
1036                         if ( is_array( $iteratorNode ) ) {
1037                                 if ( $index >= count( $iteratorNode ) ) {
1038                                         // All done with this iterator
1039                                         $iteratorStack[$level] = false;
1040                                         $contextNode = false;
1041                                 } else {
1042                                         $contextNode = $iteratorNode[$index];
1043                                         $index++;
1044                                 }
1045                         } elseif ( $iteratorNode instanceof PPNode_Hash_Array ) {
1046                                 if ( $index >= $iteratorNode->getLength() ) {
1047                                         // All done with this iterator
1048                                         $iteratorStack[$level] = false;
1049                                         $contextNode = false;
1050                                 } else {
1051                                         $contextNode = $iteratorNode->item( $index );
1052                                         $index++;
1053                                 }
1054                         } else {
1055                                 // Copy to $contextNode and then delete from iterator stack,
1056                                 // because this is not an iterator but we do have to execute it once
1057                                 $contextNode = $iteratorStack[$level];
1058                                 $iteratorStack[$level] = false;
1059                         }
1060
1061                         $newIterator = false;
1062
1063                         if ( $contextNode === false ) {
1064                                 // nothing to do
1065                         } elseif ( is_string( $contextNode ) ) {
1066                                 $out .= $contextNode;
1067                         } elseif ( is_array( $contextNode ) || $contextNode instanceof PPNode_Hash_Array ) {
1068                                 $newIterator = $contextNode;
1069                         } elseif ( $contextNode instanceof PPNode_Hash_Attr ) {
1070                                 // No output
1071                         } elseif ( $contextNode instanceof PPNode_Hash_Text ) {
1072                                 $out .= $contextNode->value;
1073                         } elseif ( $contextNode instanceof PPNode_Hash_Tree ) {
1074                                 if ( $contextNode->name == 'template' ) {
1075                                         # Double-brace expansion
1076                                         $bits = $contextNode->splitTemplate();
1077                                         if ( $flags & PPFrame::NO_TEMPLATES ) {
1078                                                 $newIterator = $this->virtualBracketedImplode(
1079                                                         '{{', '|', '}}',
1080                                                         $bits['title'],
1081                                                         $bits['parts']
1082                                                 );
1083                                         } else {
1084                                                 $ret = $this->parser->braceSubstitution( $bits, $this );
1085                                                 if ( isset( $ret['object'] ) ) {
1086                                                         $newIterator = $ret['object'];
1087                                                 } else {
1088                                                         $out .= $ret['text'];
1089                                                 }
1090                                         }
1091                                 } elseif ( $contextNode->name == 'tplarg' ) {
1092                                         # Triple-brace expansion
1093                                         $bits = $contextNode->splitTemplate();
1094                                         if ( $flags & PPFrame::NO_ARGS ) {
1095                                                 $newIterator = $this->virtualBracketedImplode(
1096                                                         '{{{', '|', '}}}',
1097                                                         $bits['title'],
1098                                                         $bits['parts']
1099                                                 );
1100                                         } else {
1101                                                 $ret = $this->parser->argSubstitution( $bits, $this );
1102                                                 if ( isset( $ret['object'] ) ) {
1103                                                         $newIterator = $ret['object'];
1104                                                 } else {
1105                                                         $out .= $ret['text'];
1106                                                 }
1107                                         }
1108                                 } elseif ( $contextNode->name == 'comment' ) {
1109                                         # HTML-style comment
1110                                         # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1111                                         if ( $this->parser->ot['html']
1112                                                 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
1113                                                 || ( $flags & PPFrame::STRIP_COMMENTS )
1114                                         ) {
1115                                                 $out .= '';
1116                                         } elseif ( $this->parser->ot['wiki'] && !( $flags & PPFrame::RECOVER_COMMENTS ) ) {
1117                                                 # Add a strip marker in PST mode so that pstPass2() can
1118                                                 # run some old-fashioned regexes on the result.
1119                                                 # Not in RECOVER_COMMENTS mode (extractSections) though.
1120                                                 $out .= $this->parser->insertStripItem( $contextNode->firstChild->value );
1121                                         } else {
1122                                                 # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1123                                                 $out .= $contextNode->firstChild->value;
1124                                         }
1125                                 } elseif ( $contextNode->name == 'ignore' ) {
1126                                         # Output suppression used by <includeonly> etc.
1127                                         # OT_WIKI will only respect <ignore> in substed templates.
1128                                         # The other output types respect it unless NO_IGNORE is set.
1129                                         # extractSections() sets NO_IGNORE and so never respects it.
1130                                         if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] )
1131                                                 || ( $flags & PPFrame::NO_IGNORE )
1132                                         ) {
1133                                                 $out .= $contextNode->firstChild->value;
1134                                         } else {
1135                                                 //$out .= '';
1136                                         }
1137                                 } elseif ( $contextNode->name == 'ext' ) {
1138                                         # Extension tag
1139                                         $bits = $contextNode->splitExt() + array( 'attr' => null, 'inner' => null, 'close' => null );
1140                                         $out .= $this->parser->extensionSubstitution( $bits, $this );
1141                                 } elseif ( $contextNode->name == 'h' ) {
1142                                         # Heading
1143                                         if ( $this->parser->ot['html'] ) {
1144                                                 # Expand immediately and insert heading index marker
1145                                                 $s = '';
1146                                                 for ( $node = $contextNode->firstChild; $node; $node = $node->nextSibling ) {
1147                                                         $s .= $this->expand( $node, $flags );
1148                                                 }
1149
1150                                                 $bits = $contextNode->splitHeading();
1151                                                 $titleText = $this->title->getPrefixedDBkey();
1152                                                 $this->parser->mHeadings[] = array( $titleText, $bits['i'] );
1153                                                 $serial = count( $this->parser->mHeadings ) - 1;
1154                                                 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
1155                                                 $s = substr( $s, 0, $bits['level'] ) . $marker . substr( $s, $bits['level'] );
1156                                                 $this->parser->mStripState->addGeneral( $marker, '' );
1157                                                 $out .= $s;
1158                                         } else {
1159                                                 # Expand in virtual stack
1160                                                 $newIterator = $contextNode->getChildren();
1161                                         }
1162                                 } else {
1163                                         # Generic recursive expansion
1164                                         $newIterator = $contextNode->getChildren();
1165                                 }
1166                         } else {
1167                                 throw new MWException( __METHOD__ . ': Invalid parameter type' );
1168                         }
1169
1170                         if ( $newIterator !== false ) {
1171                                 $outStack[] = '';
1172                                 $iteratorStack[] = $newIterator;
1173                                 $indexStack[] = 0;
1174                         } elseif ( $iteratorStack[$level] === false ) {
1175                                 // Return accumulated value to parent
1176                                 // With tail recursion
1177                                 while ( $iteratorStack[$level] === false && $level > 0 ) {
1178                                         $outStack[$level - 1] .= $out;
1179                                         array_pop( $outStack );
1180                                         array_pop( $iteratorStack );
1181                                         array_pop( $indexStack );
1182                                         $level--;
1183                                 }
1184                         }
1185                 }
1186                 --$expansionDepth;
1187                 return $outStack[0];
1188         }
1189
1190         /**
1191          * @param string $sep
1192          * @param int $flags
1193          * @return string
1194          */
1195         function implodeWithFlags( $sep, $flags /*, ... */ ) {
1196                 $args = array_slice( func_get_args(), 2 );
1197
1198                 $first = true;
1199                 $s = '';
1200                 foreach ( $args as $root ) {
1201                         if ( $root instanceof PPNode_Hash_Array ) {
1202                                 $root = $root->value;
1203                         }
1204                         if ( !is_array( $root ) ) {
1205                                 $root = array( $root );
1206                         }
1207                         foreach ( $root as $node ) {
1208                                 if ( $first ) {
1209                                         $first = false;
1210                                 } else {
1211                                         $s .= $sep;
1212                                 }
1213                                 $s .= $this->expand( $node, $flags );
1214                         }
1215                 }
1216                 return $s;
1217         }
1218
1219         /**
1220          * Implode with no flags specified
1221          * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1222          * @param string $sep
1223          * @return string
1224          */
1225         function implode( $sep /*, ... */ ) {
1226                 $args = array_slice( func_get_args(), 1 );
1227
1228                 $first = true;
1229                 $s = '';
1230                 foreach ( $args as $root ) {
1231                         if ( $root instanceof PPNode_Hash_Array ) {
1232                                 $root = $root->value;
1233                         }
1234                         if ( !is_array( $root ) ) {
1235                                 $root = array( $root );
1236                         }
1237                         foreach ( $root as $node ) {
1238                                 if ( $first ) {
1239                                         $first = false;
1240                                 } else {
1241                                         $s .= $sep;
1242                                 }
1243                                 $s .= $this->expand( $node );
1244                         }
1245                 }
1246                 return $s;
1247         }
1248
1249         /**
1250          * Makes an object that, when expand()ed, will be the same as one obtained
1251          * with implode()
1252          *
1253          * @param string $sep
1254          * @return PPNode_Hash_Array
1255          */
1256         function virtualImplode( $sep /*, ... */ ) {
1257                 $args = array_slice( func_get_args(), 1 );
1258                 $out = array();
1259                 $first = true;
1260
1261                 foreach ( $args as $root ) {
1262                         if ( $root instanceof PPNode_Hash_Array ) {
1263                                 $root = $root->value;
1264                         }
1265                         if ( !is_array( $root ) ) {
1266                                 $root = array( $root );
1267                         }
1268                         foreach ( $root as $node ) {
1269                                 if ( $first ) {
1270                                         $first = false;
1271                                 } else {
1272                                         $out[] = $sep;
1273                                 }
1274                                 $out[] = $node;
1275                         }
1276                 }
1277                 return new PPNode_Hash_Array( $out );
1278         }
1279
1280         /**
1281          * Virtual implode with brackets
1282          *
1283          * @param string $start
1284          * @param string $sep
1285          * @param string $end
1286          * @return PPNode_Hash_Array
1287          */
1288         function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1289                 $args = array_slice( func_get_args(), 3 );
1290                 $out = array( $start );
1291                 $first = true;
1292
1293                 foreach ( $args as $root ) {
1294                         if ( $root instanceof PPNode_Hash_Array ) {
1295                                 $root = $root->value;
1296                         }
1297                         if ( !is_array( $root ) ) {
1298                                 $root = array( $root );
1299                         }
1300                         foreach ( $root as $node ) {
1301                                 if ( $first ) {
1302                                         $first = false;
1303                                 } else {
1304                                         $out[] = $sep;
1305                                 }
1306                                 $out[] = $node;
1307                         }
1308                 }
1309                 $out[] = $end;
1310                 return new PPNode_Hash_Array( $out );
1311         }
1312
1313         function __toString() {
1314                 return 'frame{}';
1315         }
1316
1317         /**
1318          * @param bool $level
1319          * @return array|bool|string
1320          */
1321         function getPDBK( $level = false ) {
1322                 if ( $level === false ) {
1323                         return $this->title->getPrefixedDBkey();
1324                 } else {
1325                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1326                 }
1327         }
1328
1329         /**
1330          * @return array
1331          */
1332         function getArguments() {
1333                 return array();
1334         }
1335
1336         /**
1337          * @return array
1338          */
1339         function getNumberedArguments() {
1340                 return array();
1341         }
1342
1343         /**
1344          * @return array
1345          */
1346         function getNamedArguments() {
1347                 return array();
1348         }
1349
1350         /**
1351          * Returns true if there are no arguments in this frame
1352          *
1353          * @return bool
1354          */
1355         function isEmpty() {
1356                 return true;
1357         }
1358
1359         /**
1360          * @param string $name
1361          * @return bool
1362          */
1363         function getArgument( $name ) {
1364                 return false;
1365         }
1366
1367         /**
1368          * Returns true if the infinite loop check is OK, false if a loop is detected
1369          *
1370          * @param Title $title
1371          *
1372          * @return bool
1373          */
1374         function loopCheck( $title ) {
1375                 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1376         }
1377
1378         /**
1379          * Return true if the frame is a template frame
1380          *
1381          * @return bool
1382          */
1383         function isTemplate() {
1384                 return false;
1385         }
1386
1387         /**
1388          * Get a title of frame
1389          *
1390          * @return Title
1391          */
1392         function getTitle() {
1393                 return $this->title;
1394         }
1395
1396         /**
1397          * Set the volatile flag
1398          *
1399          * @param bool $flag
1400          */
1401         function setVolatile( $flag = true ) {
1402                 $this->volatile = $flag;
1403         }
1404
1405         /**
1406          * Get the volatile flag
1407          *
1408          * @return bool
1409          */
1410         function isVolatile() {
1411                 return $this->volatile;
1412         }
1413 }
1414
1415 /**
1416  * Expansion frame with template arguments
1417  * @ingroup Parser
1418  */
1419 class PPTemplateFrame_Hash extends PPFrame_Hash {
1420         var $numberedArgs, $namedArgs, $parent;
1421         var $numberedExpansionCache, $namedExpansionCache;
1422
1423         /**
1424          * @param Preprocessor $preprocessor
1425          * @param bool|PPFrame $parent
1426          * @param array $numberedArgs
1427          * @param array $namedArgs
1428          * @param bool|Title $title
1429          */
1430         function __construct( $preprocessor, $parent = false, $numberedArgs = array(),
1431                 $namedArgs = array(), $title = false
1432         ) {
1433                 parent::__construct( $preprocessor );
1434
1435                 $this->parent = $parent;
1436                 $this->numberedArgs = $numberedArgs;
1437                 $this->namedArgs = $namedArgs;
1438                 $this->title = $title;
1439                 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1440                 $this->titleCache = $parent->titleCache;
1441                 $this->titleCache[] = $pdbk;
1442                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1443                 if ( $pdbk !== false ) {
1444                         $this->loopCheckHash[$pdbk] = true;
1445                 }
1446                 $this->depth = $parent->depth + 1;
1447                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1448         }
1449
1450         function __toString() {
1451                 $s = 'tplframe{';
1452                 $first = true;
1453                 $args = $this->numberedArgs + $this->namedArgs;
1454                 foreach ( $args as $name => $value ) {
1455                         if ( $first ) {
1456                                 $first = false;
1457                         } else {
1458                                 $s .= ', ';
1459                         }
1460                         $s .= "\"$name\":\"" .
1461                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1462                 }
1463                 $s .= '}';
1464                 return $s;
1465         }
1466
1467         /**
1468          * @throws MWException
1469          * @param string|int $key
1470          * @param string|PPNode_Hash|DOMDocument $root
1471          * @param int $flags
1472          * @return string
1473          */
1474         function cachedExpand( $key, $root, $flags = 0 ) {
1475                 if ( isset( $this->parent->childExpansionCache[$key] ) ) {
1476                         return $this->parent->childExpansionCache[$key];
1477                 }
1478                 $retval = $this->expand( $root, $flags );
1479                 if ( !$this->isVolatile() ) {
1480                         $this->parent->childExpansionCache[$key] = $retval;
1481                 }
1482                 return $retval;
1483         }
1484
1485         /**
1486          * Returns true if there are no arguments in this frame
1487          *
1488          * @return bool
1489          */
1490         function isEmpty() {
1491                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1492         }
1493
1494         /**
1495          * @return array
1496          */
1497         function getArguments() {
1498                 $arguments = array();
1499                 foreach ( array_merge(
1500                                 array_keys( $this->numberedArgs ),
1501                                 array_keys( $this->namedArgs ) ) as $key ) {
1502                         $arguments[$key] = $this->getArgument( $key );
1503                 }
1504                 return $arguments;
1505         }
1506
1507         /**
1508          * @return array
1509          */
1510         function getNumberedArguments() {
1511                 $arguments = array();
1512                 foreach ( array_keys( $this->numberedArgs ) as $key ) {
1513                         $arguments[$key] = $this->getArgument( $key );
1514                 }
1515                 return $arguments;
1516         }
1517
1518         /**
1519          * @return array
1520          */
1521         function getNamedArguments() {
1522                 $arguments = array();
1523                 foreach ( array_keys( $this->namedArgs ) as $key ) {
1524                         $arguments[$key] = $this->getArgument( $key );
1525                 }
1526                 return $arguments;
1527         }
1528
1529         /**
1530          * @param int $index
1531          * @return array|bool
1532          */
1533         function getNumberedArgument( $index ) {
1534                 if ( !isset( $this->numberedArgs[$index] ) ) {
1535                         return false;
1536                 }
1537                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1538                         # No trimming for unnamed arguments
1539                         $this->numberedExpansionCache[$index] = $this->parent->expand(
1540                                 $this->numberedArgs[$index],
1541                                 PPFrame::STRIP_COMMENTS
1542                         );
1543                 }
1544                 return $this->numberedExpansionCache[$index];
1545         }
1546
1547         /**
1548          * @param string $name
1549          * @return bool
1550          */
1551         function getNamedArgument( $name ) {
1552                 if ( !isset( $this->namedArgs[$name] ) ) {
1553                         return false;
1554                 }
1555                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1556                         # Trim named arguments post-expand, for backwards compatibility
1557                         $this->namedExpansionCache[$name] = trim(
1558                                 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1559                 }
1560                 return $this->namedExpansionCache[$name];
1561         }
1562
1563         /**
1564          * @param string $name
1565          * @return array|bool
1566          */
1567         function getArgument( $name ) {
1568                 $text = $this->getNumberedArgument( $name );
1569                 if ( $text === false ) {
1570                         $text = $this->getNamedArgument( $name );
1571                 }
1572                 return $text;
1573         }
1574
1575         /**
1576          * Return true if the frame is a template frame
1577          *
1578          * @return bool
1579          */
1580         function isTemplate() {
1581                 return true;
1582         }
1583
1584         function setVolatile( $flag = true ) {
1585                 parent::setVolatile( $flag );
1586                 $this->parent->setVolatile( $flag );
1587         }
1588 }
1589
1590 /**
1591  * Expansion frame with custom arguments
1592  * @ingroup Parser
1593  */
1594 class PPCustomFrame_Hash extends PPFrame_Hash {
1595         var $args;
1596
1597         function __construct( $preprocessor, $args ) {
1598                 parent::__construct( $preprocessor );
1599                 $this->args = $args;
1600         }
1601
1602         function __toString() {
1603                 $s = 'cstmframe{';
1604                 $first = true;
1605                 foreach ( $this->args as $name => $value ) {
1606                         if ( $first ) {
1607                                 $first = false;
1608                         } else {
1609                                 $s .= ', ';
1610                         }
1611                         $s .= "\"$name\":\"" .
1612                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1613                 }
1614                 $s .= '}';
1615                 return $s;
1616         }
1617
1618         /**
1619          * @return bool
1620          */
1621         function isEmpty() {
1622                 return !count( $this->args );
1623         }
1624
1625         /**
1626          * @param int $index
1627          * @return bool
1628          */
1629         function getArgument( $index ) {
1630                 if ( !isset( $this->args[$index] ) ) {
1631                         return false;
1632                 }
1633                 return $this->args[$index];
1634         }
1635
1636         function getArguments() {
1637                 return $this->args;
1638         }
1639 }
1640
1641 /**
1642  * @ingroup Parser
1643  */
1644 class PPNode_Hash_Tree implements PPNode {
1645         var $name, $firstChild, $lastChild, $nextSibling;
1646
1647         function __construct( $name ) {
1648                 $this->name = $name;
1649                 $this->firstChild = $this->lastChild = $this->nextSibling = false;
1650         }
1651
1652         function __toString() {
1653                 $inner = '';
1654                 $attribs = '';
1655                 for ( $node = $this->firstChild; $node; $node = $node->nextSibling ) {
1656                         if ( $node instanceof PPNode_Hash_Attr ) {
1657                                 $attribs .= ' ' . $node->name . '="' . htmlspecialchars( $node->value ) . '"';
1658                         } else {
1659                                 $inner .= $node->__toString();
1660                         }
1661                 }
1662                 if ( $inner === '' ) {
1663                         return "<{$this->name}$attribs/>";
1664                 } else {
1665                         return "<{$this->name}$attribs>$inner</{$this->name}>";
1666                 }
1667         }
1668
1669         /**
1670          * @param string $name
1671          * @param string $text
1672          * @return PPNode_Hash_Tree
1673          */
1674         static function newWithText( $name, $text ) {
1675                 $obj = new self( $name );
1676                 $obj->addChild( new PPNode_Hash_Text( $text ) );
1677                 return $obj;
1678         }
1679
1680         function addChild( $node ) {
1681                 if ( $this->lastChild === false ) {
1682                         $this->firstChild = $this->lastChild = $node;
1683                 } else {
1684                         $this->lastChild->nextSibling = $node;
1685                         $this->lastChild = $node;
1686                 }
1687         }
1688
1689         /**
1690          * @return PPNode_Hash_Array
1691          */
1692         function getChildren() {
1693                 $children = array();
1694                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1695                         $children[] = $child;
1696                 }
1697                 return new PPNode_Hash_Array( $children );
1698         }
1699
1700         function getFirstChild() {
1701                 return $this->firstChild;
1702         }
1703
1704         function getNextSibling() {
1705                 return $this->nextSibling;
1706         }
1707
1708         function getChildrenOfType( $name ) {
1709                 $children = array();
1710                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1711                         if ( isset( $child->name ) && $child->name === $name ) {
1712                                 $children[] = $child;
1713                         }
1714                 }
1715                 return $children;
1716         }
1717
1718         /**
1719          * @return bool
1720          */
1721         function getLength() {
1722                 return false;
1723         }
1724
1725         /**
1726          * @param int $i
1727          * @return bool
1728          */
1729         function item( $i ) {
1730                 return false;
1731         }
1732
1733         /**
1734          * @return string
1735          */
1736         function getName() {
1737                 return $this->name;
1738         }
1739
1740         /**
1741          * Split a "<part>" node into an associative array containing:
1742          *  - name          PPNode name
1743          *  - index         String index
1744          *  - value         PPNode value
1745          *
1746          * @throws MWException
1747          * @return array
1748          */
1749         function splitArg() {
1750                 $bits = array();
1751                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1752                         if ( !isset( $child->name ) ) {
1753                                 continue;
1754                         }
1755                         if ( $child->name === 'name' ) {
1756                                 $bits['name'] = $child;
1757                                 if ( $child->firstChild instanceof PPNode_Hash_Attr
1758                                         && $child->firstChild->name === 'index'
1759                                 ) {
1760                                         $bits['index'] = $child->firstChild->value;
1761                                 }
1762                         } elseif ( $child->name === 'value' ) {
1763                                 $bits['value'] = $child;
1764                         }
1765                 }
1766
1767                 if ( !isset( $bits['name'] ) ) {
1768                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1769                 }
1770                 if ( !isset( $bits['index'] ) ) {
1771                         $bits['index'] = '';
1772                 }
1773                 return $bits;
1774         }
1775
1776         /**
1777          * Split an "<ext>" node into an associative array containing name, attr, inner and close
1778          * All values in the resulting array are PPNodes. Inner and close are optional.
1779          *
1780          * @throws MWException
1781          * @return array
1782          */
1783         function splitExt() {
1784                 $bits = array();
1785                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1786                         if ( !isset( $child->name ) ) {
1787                                 continue;
1788                         }
1789                         if ( $child->name == 'name' ) {
1790                                 $bits['name'] = $child;
1791                         } elseif ( $child->name == 'attr' ) {
1792                                 $bits['attr'] = $child;
1793                         } elseif ( $child->name == 'inner' ) {
1794                                 $bits['inner'] = $child;
1795                         } elseif ( $child->name == 'close' ) {
1796                                 $bits['close'] = $child;
1797                         }
1798                 }
1799                 if ( !isset( $bits['name'] ) ) {
1800                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1801                 }
1802                 return $bits;
1803         }
1804
1805         /**
1806          * Split an "<h>" node
1807          *
1808          * @throws MWException
1809          * @return array
1810          */
1811         function splitHeading() {
1812                 if ( $this->name !== 'h' ) {
1813                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1814                 }
1815                 $bits = array();
1816                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1817                         if ( !isset( $child->name ) ) {
1818                                 continue;
1819                         }
1820                         if ( $child->name == 'i' ) {
1821                                 $bits['i'] = $child->value;
1822                         } elseif ( $child->name == 'level' ) {
1823                                 $bits['level'] = $child->value;
1824                         }
1825                 }
1826                 if ( !isset( $bits['i'] ) ) {
1827                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1828                 }
1829                 return $bits;
1830         }
1831
1832         /**
1833          * Split a "<template>" or "<tplarg>" node
1834          *
1835          * @throws MWException
1836          * @return array
1837          */
1838         function splitTemplate() {
1839                 $parts = array();
1840                 $bits = array( 'lineStart' => '' );
1841                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1842                         if ( !isset( $child->name ) ) {
1843                                 continue;
1844                         }
1845                         if ( $child->name == 'title' ) {
1846                                 $bits['title'] = $child;
1847                         }
1848                         if ( $child->name == 'part' ) {
1849                                 $parts[] = $child;
1850                         }
1851                         if ( $child->name == 'lineStart' ) {
1852                                 $bits['lineStart'] = '1';
1853                         }
1854                 }
1855                 if ( !isset( $bits['title'] ) ) {
1856                         throw new MWException( 'Invalid node passed to ' . __METHOD__ );
1857                 }
1858                 $bits['parts'] = new PPNode_Hash_Array( $parts );
1859                 return $bits;
1860         }
1861 }
1862
1863 /**
1864  * @ingroup Parser
1865  */
1866 class PPNode_Hash_Text implements PPNode {
1867         var $value, $nextSibling;
1868
1869         function __construct( $value ) {
1870                 if ( is_object( $value ) ) {
1871                         throw new MWException( __CLASS__ . ' given object instead of string' );
1872                 }
1873                 $this->value = $value;
1874         }
1875
1876         function __toString() {
1877                 return htmlspecialchars( $this->value );
1878         }
1879
1880         function getNextSibling() {
1881                 return $this->nextSibling;
1882         }
1883
1884         function getChildren() {
1885                 return false;
1886         }
1887
1888         function getFirstChild() {
1889                 return false;
1890         }
1891
1892         function getChildrenOfType( $name ) {
1893                 return false;
1894         }
1895
1896         function getLength() {
1897                 return false;
1898         }
1899
1900         function item( $i ) {
1901                 return false;
1902         }
1903
1904         function getName() {
1905                 return '#text';
1906         }
1907
1908         function splitArg() {
1909                 throw new MWException( __METHOD__ . ': not supported' );
1910         }
1911
1912         function splitExt() {
1913                 throw new MWException( __METHOD__ . ': not supported' );
1914         }
1915
1916         function splitHeading() {
1917                 throw new MWException( __METHOD__ . ': not supported' );
1918         }
1919 }
1920
1921 /**
1922  * @ingroup Parser
1923  */
1924 class PPNode_Hash_Array implements PPNode {
1925         var $value, $nextSibling;
1926
1927         function __construct( $value ) {
1928                 $this->value = $value;
1929         }
1930
1931         function __toString() {
1932                 return var_export( $this, true );
1933         }
1934
1935         function getLength() {
1936                 return count( $this->value );
1937         }
1938
1939         function item( $i ) {
1940                 return $this->value[$i];
1941         }
1942
1943         function getName() {
1944                 return '#nodelist';
1945         }
1946
1947         function getNextSibling() {
1948                 return $this->nextSibling;
1949         }
1950
1951         function getChildren() {
1952                 return false;
1953         }
1954
1955         function getFirstChild() {
1956                 return false;
1957         }
1958
1959         function getChildrenOfType( $name ) {
1960                 return false;
1961         }
1962
1963         function splitArg() {
1964                 throw new MWException( __METHOD__ . ': not supported' );
1965         }
1966
1967         function splitExt() {
1968                 throw new MWException( __METHOD__ . ': not supported' );
1969         }
1970
1971         function splitHeading() {
1972                 throw new MWException( __METHOD__ . ': not supported' );
1973         }
1974 }
1975
1976 /**
1977  * @ingroup Parser
1978  */
1979 class PPNode_Hash_Attr implements PPNode {
1980         var $name, $value, $nextSibling;
1981
1982         function __construct( $name, $value ) {
1983                 $this->name = $name;
1984                 $this->value = $value;
1985         }
1986
1987         function __toString() {
1988                 return "<@{$this->name}>" . htmlspecialchars( $this->value ) . "</@{$this->name}>";
1989         }
1990
1991         function getName() {
1992                 return $this->name;
1993         }
1994
1995         function getNextSibling() {
1996                 return $this->nextSibling;
1997         }
1998
1999         function getChildren() {
2000                 return false;
2001         }
2002
2003         function getFirstChild() {
2004                 return false;
2005         }
2006
2007         function getChildrenOfType( $name ) {
2008                 return false;
2009         }
2010
2011         function getLength() {
2012                 return false;
2013         }
2014
2015         function item( $i ) {
2016                 return false;
2017         }
2018
2019         function splitArg() {
2020                 throw new MWException( __METHOD__ . ': not supported' );
2021         }
2022
2023         function splitExt() {
2024                 throw new MWException( __METHOD__ . ': not supported' );
2025         }
2026
2027         function splitHeading() {
2028                 throw new MWException( __METHOD__ . ': not supported' );
2029         }
2030 }