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