includes/parser/Preprocessor_Hash.php

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