includes/parser/Preprocessor_Hash.php

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