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