parser/Preprocessor_Hash.php

   1 <?php
   2
   3 /**
   4  * Differences from DOM schema:
   5  *   * attribute nodes are children
   6  *   * <h> nodes that aren't at the top are replaced with <possible-h>
   7  * @ingroup Parser
   8  */
   9 class Preprocessor_Hash implements Preprocessor {
  10         var $parser;
  11
  12         function __construct( $parser ) {
  13                 $this->parser = $parser;
  14         }
  15
  16         function newFrame() {
  17                 return new PPFrame_Hash( $this );
  18         }
  19
  20         function newCustomFrame( $args ) {
  21                 return new PPCustomFrame_Hash( $this, $args );
  22         }
  23
  24         /**
  25          * Preprocess some wikitext and return the document tree.
  26          * This is the ghost of Parser::replace_variables().
  27          *
  28          * @param string $text The text to parse
  29          * @param integer flags Bitwise combination of:
  30          *          Parser::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
  31          *                                     included. Default is to assume a direct page view.
  32          *
  33          * The generated DOM tree must depend only on the input text and the flags.
  34          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
  35          *
  36          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
  37          * change in the DOM tree for a given text, must be passed through the section identifier
  38          * in the section edit link and thus back to extractSections().
  39          *
  40          * The output of this function is currently only cached in process memory, but a persistent
  41          * cache may be implemented at a later date which takes further advantage of these strict
  42          * dependency requirements.
  43          *
  44          * @private
  45          */
  46         function preprocessToObj( $text, $flags = 0 ) {
  47                 wfProfileIn( __METHOD__ );
  48
  49                 global $wgMemc;
  50                 $cacheKey = wfMemckey( 'preprocessor-hash', md5( $text ), $flags );
  51
  52                 if ( $obj = $wgMemc->get( $cacheKey ) ) {
  53                         wfDebugLog( "Preprocessor", "Got preprocessor_hash output from cache" );
  54                         wfProfileOut( __METHOD__ );
  55                         return $obj;
  56                 }
  57
  58                 $rules = array(
  59                         '{' => array(
  60                                 'end' => '}',
  61                                 'names' => array(
  62                                         2 => 'template',
  63                                         3 => 'tplarg',
  64                                 ),
  65                                 'min' => 2,
  66                                 'max' => 3,
  67                         ),
  68                         '[' => array(
  69                                 'end' => ']',
  70                                 'names' => array( 2 => null ),
  71                                 'min' => 2,
  72                                 'max' => 2,
  73                         )
  74                 );
  75
  76                 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
  77
  78                 $xmlishElements = $this->parser->getStripList();
  79                 $enableOnlyinclude = false;
  80                 if ( $forInclusion ) {
  81                         $ignoredTags = array( 'includeonly', '/includeonly' );
  82                         $ignoredElements = array( 'noinclude' );
  83                         $xmlishElements[] = 'noinclude';
  84                         if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
  85                                 $enableOnlyinclude = true;
  86                         }
  87                 } else {
  88                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
  89                         $ignoredElements = array( 'includeonly' );
  90                         $xmlishElements[] = 'includeonly';
  91                 }
  92                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
  93
  94                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
  95                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
  96
  97                 $stack = new PPDStack_Hash;
  98
  99                 $searchBase = "[{<\n";
 100                 $revText = strrev( $text ); // For fast reverse searches
 101
 102                 $i = 0;                     # Input pointer, starts out pointing to a pseudo-newline before the start
 103                 $accum =& $stack->getAccum();   # Current accumulator
 104                 $findEquals = false;            # True to find equals signs in arguments
 105                 $findPipe = false;              # True to take notice of pipe characters
 106                 $headingIndex = 1;
 107                 $inHeading = false;        # True if $i is inside a possible heading
 108                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
 109                 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
 110                 $fakeLineStart = true;     # Do a line-start run without outputting an LF character
 111
 112                 while ( true ) {
 113                         //$this->memCheck();
 114
 115                         if ( $findOnlyinclude ) {
 116                                 // Ignore all input up to the next <onlyinclude>
 117                                 $startPos = strpos( $text, '<onlyinclude>', $i );
 118                                 if ( $startPos === false ) {
 119                                         // Ignored section runs to the end
 120                                         $accum->addNodeWithText( 'ignore', substr( $text, $i ) );
 121                                         break;
 122                                 }
 123                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
 124                                 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i ) );
 125                                 $i = $tagEndPos;
 126                                 $findOnlyinclude = false;
 127                         }
 128
 129                         if ( $fakeLineStart ) {
 130                                 $found = 'line-start';
 131                                 $curChar = '';
 132                         } else {
 133                                 # Find next opening brace, closing brace or pipe
 134                                 $search = $searchBase;
 135                                 if ( $stack->top === false ) {
 136                                         $currentClosing = '';
 137                                 } else {
 138                                         $currentClosing = $stack->top->close;
 139                                         $search .= $currentClosing;
 140                                 }
 141                                 if ( $findPipe ) {
 142                                         $search .= '|';
 143                                 }
 144                                 if ( $findEquals ) {
 145                                         // First equals will be for the template
 146                                         $search .= '=';
 147                                 }
 148                                 $rule = null;
 149                                 # Output literal section, advance input counter
 150                                 $literalLength = strcspn( $text, $search, $i );
 151                                 if ( $literalLength > 0 ) {
 152                                         $accum->addLiteral( substr( $text, $i, $literalLength ) );
 153                                         $i += $literalLength;
 154                                 }
 155                                 if ( $i >= strlen( $text ) ) {
 156                                         if ( $currentClosing == "\n" ) {
 157                                                 // Do a past-the-end run to finish off the heading
 158                                                 $curChar = '';
 159                                                 $found = 'line-end';
 160                                         } else {
 161                                                 # All done
 162                                                 break;
 163                                         }
 164                                 } else {
 165                                         $curChar = $text[$i];
 166                                         if ( $curChar == '|' ) {
 167                                                 $found = 'pipe';
 168                                         } elseif ( $curChar == '=' ) {
 169                                                 $found = 'equals';
 170                                         } elseif ( $curChar == '<' ) {
 171                                                 $found = 'angle';
 172                                         } elseif ( $curChar == "\n" ) {
 173                                                 if ( $inHeading ) {
 174                                                         $found = 'line-end';
 175                                                 } else {
 176                                                         $found = 'line-start';
 177                                                 }
 178                                         } elseif ( $curChar == $currentClosing ) {
 179                                                 $found = 'close';
 180                                         } elseif ( isset( $rules[$curChar] ) ) {
 181                                                 $found = 'open';
 182                                                 $rule = $rules[$curChar];
 183                                         } else {
 184                                                 # Some versions of PHP have a strcspn which stops on null characters
 185                                                 # Ignore and continue
 186                                                 ++$i;
 187                                                 continue;
 188                                         }
 189                                 }
 190                         }
 191
 192                         if ( $found == 'angle' ) {
 193                                 $matches = false;
 194                                 // Handle </onlyinclude>
 195                                 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
 196                                         $findOnlyinclude = true;
 197                                         continue;
 198                                 }
 199
 200                                 // Determine element name
 201                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
 202                                         // Element name missing or not listed
 203                                         $accum->addLiteral( '<' );
 204                                         ++$i;
 205                                         continue;
 206                                 }
 207                                 // Handle comments
 208                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
 209                                         // To avoid leaving blank lines, when a comment is both preceded
 210                                         // and followed by a newline (ignoring spaces), trim leading and
 211                                         // trailing spaces and one of the newlines.
 212
 213                                         // Find the end
 214                                         $endPos = strpos( $text, '-->', $i + 4 );
 215                                         if ( $endPos === false ) {
 216                                                 // Unclosed comment in input, runs to end
 217                                                 $inner = substr( $text, $i );
 218                                                 $accum->addNodeWithText( 'comment', $inner );
 219                                                 $i = strlen( $text );
 220                                         } else {
 221                                                 // Search backwards for leading whitespace
 222                                                 $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
 223                                                 // Search forwards for trailing whitespace
 224                                                 // $wsEnd will be the position of the last space
 225                                                 $wsEnd = $endPos + 2 + strspn( $text, ' ', $endPos + 3 );
 226                                                 // Eat the line if possible
 227                                                 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
 228                                                 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
 229                                                 // it's a possible beneficial b/c break.
 230                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
 231                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n" )
 232                                                 {
 233                                                         $startPos = $wsStart;
 234                                                         $endPos = $wsEnd + 1;
 235                                                         // Remove leading whitespace from the end of the accumulator
 236                                                         // Sanity check first though
 237                                                         $wsLength = $i - $wsStart;
 238                                                         if ( $wsLength > 0
 239                                                                 && $accum->lastNode instanceof PPNode_Hash_Text
 240                                                                 && substr( $accum->lastNode->value, -$wsLength ) === str_repeat( ' ', $wsLength ) )
 241                                                         {
 242                                                                 $accum->lastNode->value = substr( $accum->lastNode->value, 0, -$wsLength );
 243                                                         }
 244                                                         // Do a line-start run next time to look for headings after the comment
 245                                                         $fakeLineStart = true;
 246                                                 } else {
 247                                                         // No line to eat, just take the comment itself
 248                                                         $startPos = $i;
 249                                                         $endPos += 2;
 250                                                 }
 251
 252                                                 if ( $stack->top ) {
 253                                                         $part = $stack->top->getCurrentPart();
 254                                                         if ( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) {
 255                                                                 // Comments abutting, no change in visual end
 256                                                                 $part->commentEnd = $wsEnd;
 257                                                         } else {
 258                                                                 $part->visualEnd = $wsStart;
 259                                                                 $part->commentEnd = $endPos;
 260                                                         }
 261                                                 }
 262                                                 $i = $endPos + 1;
 263                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
 264                                                 $accum->addNodeWithText( 'comment', $inner );
 265                                         }
 266                                         continue;
 267                                 }
 268                                 $name = $matches[1];
 269                                 $lowerName = strtolower( $name );
 270                                 $attrStart = $i + strlen( $name ) + 1;
 271
 272                                 // Find end of tag
 273                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
 274                                 if ( $tagEndPos === false ) {
 275                                         // Infinite backtrack
 276                                         // Disable tag search to prevent worst-case O(N^2) performance
 277                                         $noMoreGT = true;
 278                                         $accum->addLiteral( '<' );
 279                                         ++$i;
 280                                         continue;
 281                                 }
 282
 283                                 // Handle ignored tags
 284                                 if ( in_array( $lowerName, $ignoredTags ) ) {
 285                                         $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i + 1 ) );
 286                                         $i = $tagEndPos + 1;
 287                                         continue;
 288                                 }
 289
 290                                 $tagStartPos = $i;
 291                                 if ( $text[$tagEndPos-1] == '/' ) {
 292                                         // Short end tag
 293                                         $attrEnd = $tagEndPos - 1;
 294                                         $inner = null;
 295                                         $i = $tagEndPos + 1;
 296                                         $close = null;
 297                                 } else {
 298                                         $attrEnd = $tagEndPos;
 299                                         // Find closing tag
 300                                         if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
 301                                                         $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) )
 302                                         {
 303                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
 304                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
 305                                                 $close = $matches[0][0];
 306                                         } else {
 307                                                 // No end tag -- let it run out to the end of the text.
 308                                                 $inner = substr( $text, $tagEndPos + 1 );
 309                                                 $i = strlen( $text );
 310                                                 $close = null;
 311                                         }
 312                                 }
 313                                 // <includeonly> and <noinclude> just become <ignore> tags
 314                                 if ( in_array( $lowerName, $ignoredElements ) ) {
 315                                         $accum->addNodeWithText(  'ignore', substr( $text, $tagStartPos, $i - $tagStartPos ) );
 316                                         continue;
 317                                 }
 318
 319                                 if ( $attrEnd <= $attrStart ) {
 320                                         $attr = '';
 321                                 } else {
 322                                         // Note that the attr element contains the whitespace between name and attribute,
 323                                         // this is necessary for precise reconstruction during pre-save transform.
 324                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
 325                                 }
 326
 327                                 $extNode = new PPNode_Hash_Tree( 'ext' );
 328                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'name', $name ) );
 329                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'attr', $attr ) );
 330                                 if ( $inner !== null ) {
 331                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'inner', $inner ) );
 332                                 }
 333                                 if ( $close !== null ) {
 334                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'close', $close ) );
 335                                 }
 336                                 $accum->addNode( $extNode );
 337                         }
 338
 339                         elseif ( $found == 'line-start' ) {
 340                                 // Is this the start of a heading?
 341                                 // Line break belongs before the heading element in any case
 342                                 if ( $fakeLineStart ) {
 343                                         $fakeLineStart = false;
 344                                 } else {
 345                                         $accum->addLiteral( $curChar );
 346                                         $i++;
 347                                 }
 348
 349                                 $count = strspn( $text, '=', $i, 6 );
 350                                 if ( $count == 1 && $findEquals ) {
 351                                         // DWIM: This looks kind of like a name/value separator
 352                                         // Let's let the equals handler have it and break the potential heading
 353                                         // This is heuristic, but AFAICT the methods for completely correct disambiguation are very complex.
 354                                 } elseif ( $count > 0 ) {
 355                                         $piece = array(
 356                                                 'open' => "\n",
 357                                                 'close' => "\n",
 358                                                 'parts' => array( new PPDPart_Hash( str_repeat( '=', $count ) ) ),
 359                                                 'startPos' => $i,
 360                                                 'count' => $count );
 361                                         $stack->push( $piece );
 362                                         $accum =& $stack->getAccum();
 363                                         extract( $stack->getFlags() );
 364                                         $i += $count;
 365                                 }
 366                         }
 367
 368                         elseif ( $found == 'line-end' ) {
 369                                 $piece = $stack->top;
 370                                 // A heading must be open, otherwise \n wouldn't have been in the search list
 371                                 assert( $piece->open == "\n" );
 372                                 $part = $piece->getCurrentPart();
 373                                 // Search back through the input to see if it has a proper close
 374                                 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
 375                                 $wsLength = strspn( $revText, " \t", strlen( $text ) - $i );
 376                                 $searchStart = $i - $wsLength;
 377                                 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
 378                                         // Comment found at line end
 379                                         // Search for equals signs before the comment
 380                                         $searchStart = $part->visualEnd;
 381                                         $searchStart -= strspn( $revText, " \t", strlen( $text ) - $searchStart );
 382                                 }
 383                                 $count = $piece->count;
 384                                 $equalsLength = strspn( $revText, '=', strlen( $text ) - $searchStart );
 385                                 if ( $equalsLength > 0 ) {
 386                                         if ( $i - $equalsLength == $piece->startPos ) {
 387                                                 // This is just a single string of equals signs on its own line
 388                                                 // Replicate the doHeadings behaviour /={count}(.+)={count}/
 389                                                 // First find out how many equals signs there really are (don't stop at 6)
 390                                                 $count = $equalsLength;
 391                                                 if ( $count < 3 ) {
 392                                                         $count = 0;
 393                                                 } else {
 394                                                         $count = min( 6, intval( ( $count - 1 ) / 2 ) );
 395                                                 }
 396                                         } else {
 397                                                 $count = min( $equalsLength, $count );
 398                                         }
 399                                         if ( $count > 0 ) {
 400                                                 // Normal match, output <h>
 401                                                 $element = new PPNode_Hash_Tree( 'possible-h' );
 402                                                 $element->addChild( new PPNode_Hash_Attr( 'level', $count ) );
 403                                                 $element->addChild( new PPNode_Hash_Attr( 'i', $headingIndex++ ) );
 404                                                 $element->lastChild->nextSibling = $accum->firstNode;
 405                                                 $element->lastChild = $accum->lastNode;
 406                                         } else {
 407                                                 // Single equals sign on its own line, count=0
 408                                                 $element = $accum;
 409                                         }
 410                                 } else {
 411                                         // No match, no <h>, just pass down the inner text
 412                                         $element = $accum;
 413                                 }
 414                                 // Unwind the stack
 415                                 $stack->pop();
 416                                 $accum =& $stack->getAccum();
 417                                 extract( $stack->getFlags() );
 418
 419                                 // Append the result to the enclosing accumulator
 420                                 if ( $element instanceof PPNode ) {
 421                                         $accum->addNode( $element );
 422                                 } else {
 423                                         $accum->addAccum( $element );
 424                                 }
 425                                 // Note that we do NOT increment the input pointer.
 426                                 // This is because the closing linebreak could be the opening linebreak of
 427                                 // another heading. Infinite loops are avoided because the next iteration MUST
 428                                 // hit the heading open case above, which unconditionally increments the
 429                                 // input pointer.
 430                         }
 431
 432                         elseif ( $found == 'open' ) {
 433                                 # count opening brace characters
 434                                 $count = strspn( $text, $curChar, $i );
 435
 436                                 # we need to add to stack only if opening brace count is enough for one of the rules
 437                                 if ( $count >= $rule['min'] ) {
 438                                         # Add it to the stack
 439                                         $piece = array(
 440                                                 'open' => $curChar,
 441                                                 'close' => $rule['end'],
 442                                                 'count' => $count,
 443                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
 444                                         );
 445
 446                                         $stack->push( $piece );
 447                                         $accum =& $stack->getAccum();
 448                                         extract( $stack->getFlags() );
 449                                 } else {
 450                                         # Add literal brace(s)
 451                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 452                                 }
 453                                 $i += $count;
 454                         }
 455
 456                         elseif ( $found == 'close' ) {
 457                                 $piece = $stack->top;
 458                                 # lets check if there are enough characters for closing brace
 459                                 $maxCount = $piece->count;
 460                                 $count = strspn( $text, $curChar, $i, $maxCount );
 461
 462                                 # check for maximum matching characters (if there are 5 closing
 463                                 # characters, we will probably need only 3 - depending on the rules)
 464                                 $matchingCount = 0;
 465                                 $rule = $rules[$piece->open];
 466                                 if ( $count > $rule['max'] ) {
 467                                         # The specified maximum exists in the callback array, unless the caller
 468                                         # has made an error
 469                                         $matchingCount = $rule['max'];
 470                                 } else {
 471                                         # Count is less than the maximum
 472                                         # Skip any gaps in the callback array to find the true largest match
 473                                         # Need to use array_key_exists not isset because the callback can be null
 474                                         $matchingCount = $count;
 475                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
 476                                                 --$matchingCount;
 477                                         }
 478                                 }
 479
 480                                 if ($matchingCount <= 0) {
 481                                         # No matching element found in callback array
 482                                         # Output a literal closing brace and continue
 483                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 484                                         $i += $count;
 485                                         continue;
 486                                 }
 487                                 $name = $rule['names'][$matchingCount];
 488                                 if ( $name === null ) {
 489                                         // No element, just literal text
 490                                         $element = $piece->breakSyntax( $matchingCount );
 491                                         $element->addLiteral( str_repeat( $rule['end'], $matchingCount ) );
 492                                 } else {
 493                                         # Create XML element
 494                                         # Note: $parts is already XML, does not need to be encoded further
 495                                         $parts = $piece->parts;
 496                                         $titleAccum = $parts[0]->out;
 497                                         unset( $parts[0] );
 498
 499                                         $element = new PPNode_Hash_Tree( $name );
 500
 501                                         # The invocation is at the start of the line if lineStart is set in
 502                                         # the stack, and all opening brackets are used up.
 503                                         if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
 504                                                 $element->addChild( new PPNode_Hash_Attr( 'lineStart', 1 ) );
 505                                         }
 506                                         $titleNode = new PPNode_Hash_Tree( 'title' );
 507                                         $titleNode->firstChild = $titleAccum->firstNode;
 508                                         $titleNode->lastChild = $titleAccum->lastNode;
 509                                         $element->addChild( $titleNode );
 510                                         $argIndex = 1;
 511                                         foreach ( $parts as $partIndex => $part ) {
 512                                                 if ( isset( $part->eqpos ) ) {
 513                                                         // Find equals
 514                                                         $lastNode = false;
 515                                                         for ( $node = $part->out->firstNode; $node; $node = $node->nextSibling ) {
 516                                                                 if ( $node === $part->eqpos ) {
 517                                                                         break;
 518                                                                 }
 519                                                                 $lastNode = $node;
 520                                                         }
 521                                                         if ( !$node ) {
 522                                                                 throw new MWException( __METHOD__. ': eqpos not found' );
 523                                                         }
 524                                                         if ( $node->name !== 'equals' ) {
 525                                                                 throw new MWException( __METHOD__ .': eqpos is not equals' );
 526                                                         }
 527                                                         $equalsNode = $node;
 528
 529                                                         // Construct name node
 530                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 531                                                         if ( $lastNode !== false ) {
 532                                                                 $lastNode->nextSibling = false;
 533                                                                 $nameNode->firstChild = $part->out->firstNode;
 534                                                                 $nameNode->lastChild = $lastNode;
 535                                                         }
 536
 537                                                         // Construct value node
 538                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 539                                                         if ( $equalsNode->nextSibling !== false ) {
 540                                                                 $valueNode->firstChild = $equalsNode->nextSibling;
 541                                                                 $valueNode->lastChild = $part->out->lastNode;
 542                                                         }
 543                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 544                                                         $partNode->addChild( $nameNode );
 545                                                         $partNode->addChild( $equalsNode->firstChild );
 546                                                         $partNode->addChild( $valueNode );
 547                                                         $element->addChild( $partNode );
 548                                                 } else {
 549                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 550                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 551                                                         $nameNode->addChild( new PPNode_Hash_Attr( 'index', $argIndex++ ) );
 552                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 553                                                         $valueNode->firstChild = $part->out->firstNode;
 554                                                         $valueNode->lastChild = $part->out->lastNode;
 555                                                         $partNode->addChild( $nameNode );
 556                                                         $partNode->addChild( $valueNode );
 557                                                         $element->addChild( $partNode );
 558                                                 }
 559                                         }
 560                                 }
 561
 562                                 # Advance input pointer
 563                                 $i += $matchingCount;
 564
 565                                 # Unwind the stack
 566                                 $stack->pop();
 567                                 $accum =& $stack->getAccum();
 568
 569                                 # Re-add the old stack element if it still has unmatched opening characters remaining
 570                                 if ($matchingCount < $piece->count) {
 571                                         $piece->parts = array( new PPDPart_Hash );
 572                                         $piece->count -= $matchingCount;
 573                                         # do we still qualify for any callback with remaining count?
 574                                         $names = $rules[$piece->open]['names'];
 575                                         $skippedBraces = 0;
 576                                         $enclosingAccum =& $accum;
 577                                         while ( $piece->count ) {
 578                                                 if ( array_key_exists( $piece->count, $names ) ) {
 579                                                         $stack->push( $piece );
 580                                                         $accum =& $stack->getAccum();
 581                                                         break;
 582                                                 }
 583                                                 --$piece->count;
 584                                                 $skippedBraces ++;
 585                                         }
 586                                         $enclosingAccum->addLiteral( str_repeat( $piece->open, $skippedBraces ) );
 587                                 }
 588
 589                                 extract( $stack->getFlags() );
 590
 591                                 # Add XML element to the enclosing accumulator
 592                                 if ( $element instanceof PPNode ) {
 593                                         $accum->addNode( $element );
 594                                 } else {
 595                                         $accum->addAccum( $element );
 596                                 }
 597                         }
 598
 599                         elseif ( $found == 'pipe' ) {
 600                                 $findEquals = true; // shortcut for getFlags()
 601                                 $stack->addPart();
 602                                 $accum =& $stack->getAccum();
 603                                 ++$i;
 604                         }
 605
 606                         elseif ( $found == 'equals' ) {
 607                                 $findEquals = false; // shortcut for getFlags()
 608                                 $accum->addNodeWithText( 'equals', '=' );
 609                                 $stack->getCurrentPart()->eqpos = $accum->lastNode;
 610                                 ++$i;
 611                         }
 612                 }
 613
 614                 # Output any remaining unclosed brackets
 615                 foreach ( $stack->stack as $piece ) {
 616                         $stack->rootAccum->addAccum( $piece->breakSyntax() );
 617                 }
 618
 619                 # Enable top-level headings
 620                 for ( $node = $stack->rootAccum->firstNode; $node; $node = $node->nextSibling ) {
 621                         if ( isset( $node->name ) && $node->name === 'possible-h' ) {
 622                                 $node->name = 'h';
 623                         }
 624                 }
 625
 626                 $rootNode = new PPNode_Hash_Tree( 'root' );
 627                 $rootNode->firstChild = $stack->rootAccum->firstNode;
 628                 $rootNode->lastChild = $stack->rootAccum->lastNode;
 629                 wfProfileOut( __METHOD__ );
 630
 631                 $wgMemc->set( $cacheKey, $rootNode, 86400 );
 632
 633                 return $rootNode;
 634         }
 635 }
 636
 637 /**
 638  * Stack class to help Preprocessor::preprocessToObj()
 639  * @ingroup Parser
 640  */
 641 class PPDStack_Hash extends PPDStack {
 642         function __construct() {
 643                 $this->elementClass = 'PPDStackElement_Hash';
 644                 parent::__construct();
 645                 $this->rootAccum = new PPDAccum_Hash;
 646         }
 647 }
 648
 649 /**
 650  * @ingroup Parser
 651  */
 652 class PPDStackElement_Hash extends PPDStackElement {
 653         function __construct( $data = array() ) {
 654                 $this->partClass = 'PPDPart_Hash';
 655                 parent::__construct( $data );
 656         }
 657
 658         /**
 659          * Get the accumulator that would result if the close is not found.
 660          */
 661         function breakSyntax( $openingCount = false ) {
 662                 if ( $this->open == "\n" ) {
 663                         $accum = $this->parts[0]->out;
 664                 } else {
 665                         if ( $openingCount === false ) {
 666                                 $openingCount = $this->count;
 667                         }
 668                         $accum = new PPDAccum_Hash;
 669                         $accum->addLiteral( str_repeat( $this->open, $openingCount ) );
 670                         $first = true;
 671                         foreach ( $this->parts as $part ) {
 672                                 if ( $first ) {
 673                                         $first = false;
 674                                 } else {
 675                                         $accum->addLiteral( '|' );
 676                                 }
 677                                 $accum->addAccum( $part->out );
 678                         }
 679                 }
 680                 return $accum;
 681         }
 682 }
 683
 684 /**
 685  * @ingroup Parser
 686  */
 687 class PPDPart_Hash extends PPDPart {
 688         function __construct( $out = '' ) {
 689                 $accum = new PPDAccum_Hash;
 690                 if ( $out !== '' ) {
 691                         $accum->addLiteral( $out );
 692                 }
 693                 parent::__construct( $accum );
 694         }
 695 }
 696
 697 /**
 698  * @ingroup Parser
 699  */
 700 class PPDAccum_Hash {
 701         var $firstNode, $lastNode;
 702
 703         function __construct() {
 704                 $this->firstNode = $this->lastNode = false;
 705         }
 706
 707         /**
 708          * Append a string literal
 709          */
 710         function addLiteral( $s ) {
 711                 if ( $this->lastNode === false ) {
 712                         $this->firstNode = $this->lastNode = new PPNode_Hash_Text( $s );
 713                 } elseif ( $this->lastNode instanceof PPNode_Hash_Text ) {
 714                         $this->lastNode->value .= $s;
 715                 } else {
 716                         $this->lastNode->nextSibling = new PPNode_Hash_Text( $s );
 717                         $this->lastNode = $this->lastNode->nextSibling;
 718                 }
 719         }
 720
 721         /**
 722          * Append a PPNode
 723          */
 724         function addNode( PPNode $node ) {
 725                 if ( $this->lastNode === false ) {
 726                         $this->firstNode = $this->lastNode = $node;
 727                 } else {
 728                         $this->lastNode->nextSibling = $node;
 729                         $this->lastNode = $node;
 730                 }
 731         }
 732
 733         /**
 734          * Append a tree node with text contents
 735          */
 736         function addNodeWithText( $name, $value ) {
 737                 $node = PPNode_Hash_Tree::newWithText( $name, $value );
 738                 $this->addNode( $node );
 739         }
 740
 741         /**
 742          * Append a PPAccum_Hash
 743          * Takes over ownership of the nodes in the source argument. These nodes may
 744          * subsequently be modified, especially nextSibling.
 745          */
 746         function addAccum( $accum ) {
 747                 if ( $accum->lastNode === false ) {
 748                         // nothing to add
 749                 } elseif ( $this->lastNode === false ) {
 750                         $this->firstNode = $accum->firstNode;
 751                         $this->lastNode = $accum->lastNode;
 752                 } else {
 753                         $this->lastNode->nextSibling = $accum->firstNode;
 754                         $this->lastNode = $accum->lastNode;
 755                 }
 756         }
 757 }
 758
 759 /**
 760  * An expansion frame, used as a context to expand the result of preprocessToObj()
 761  * @ingroup Parser
 762  */
 763 class PPFrame_Hash implements PPFrame {
 764         var $preprocessor, $parser, $title;
 765         var $titleCache;
 766
 767         /**
 768          * Hashtable listing templates which are disallowed for expansion in this frame,
 769          * having been encountered previously in parent frames.
 770          */
 771         var $loopCheckHash;
 772
 773         /**
 774          * Recursion depth of this frame, top = 0
 775          * Note that this is NOT the same as expansion depth in expand()
 776          */
 777         var $depth;
 778
 779
 780         /**
 781          * Construct a new preprocessor frame.
 782          * @param Preprocessor $preprocessor The parent preprocessor
 783          */
 784         function __construct( $preprocessor ) {
 785                 $this->preprocessor = $preprocessor;
 786                 $this->parser = $preprocessor->parser;
 787                 $this->title = $this->parser->mTitle;
 788                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
 789                 $this->loopCheckHash = array();
 790                 $this->depth = 0;
 791         }
 792
 793         /**
 794          * Create a new child frame
 795          * $args is optionally a multi-root PPNode or array containing the template arguments
 796          */
 797         function newChild( $args = false, $title = false ) {
 798                 $namedArgs = array();
 799                 $numberedArgs = array();
 800                 if ( $title === false ) {
 801                         $title = $this->title;
 802                 }
 803                 if ( $args !== false ) {
 804                         $xpath = false;
 805                         if ( $args instanceof PPNode_Hash_Array ) {
 806                                 $args = $args->value;
 807                         } elseif ( !is_array( $args ) ) {
 808                                 throw new MWException( __METHOD__ . ': $args must be array or PPNode_Hash_Array' );
 809                         }
 810                         foreach ( $args as $arg ) {
 811                                 $bits = $arg->splitArg();
 812                                 if ( $bits['index'] !== '' ) {
 813                                         // Numbered parameter
 814                                         $numberedArgs[$bits['index']] = $bits['value'];
 815                                         unset( $namedArgs[$bits['index']] );
 816                                 } else {
 817                                         // Named parameter
 818                                         $name = trim( $this->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
 819                                         $namedArgs[$name] = $bits['value'];
 820                                         unset( $numberedArgs[$name] );
 821                                 }
 822                         }
 823                 }
 824                 return new PPTemplateFrame_Hash( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
 825         }
 826
 827         function expand( $root, $flags = 0 ) {
 828                 static $expansionDepth = 0;
 829                 if ( is_string( $root ) ) {
 830                         return $root;
 831                 }
 832
 833                 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
 834                 {
 835                         return '<span class="error">Node-count limit exceeded</span>';
 836                 }
 837                 if ( $expansionDepth > $this->parser->mOptions->mMaxPPExpandDepth ) {
 838                         return '<span class="error">Expansion depth limit exceeded</span>';
 839                 }
 840                 ++$expansionDepth;
 841
 842                 $outStack = array( '', '' );
 843                 $iteratorStack = array( false, $root );
 844                 $indexStack = array( 0, 0 );
 845
 846                 while ( count( $iteratorStack ) > 1 ) {
 847                         $level = count( $outStack ) - 1;
 848                         $iteratorNode =& $iteratorStack[ $level ];
 849                         $out =& $outStack[$level];
 850                         $index =& $indexStack[$level];
 851
 852                         if ( is_array( $iteratorNode ) ) {
 853                                 if ( $index >= count( $iteratorNode ) ) {
 854                                         // All done with this iterator
 855                                         $iteratorStack[$level] = false;
 856                                         $contextNode = false;
 857                                 } else {
 858                                         $contextNode = $iteratorNode[$index];
 859                                         $index++;
 860                                 }
 861                         } elseif ( $iteratorNode instanceof PPNode_Hash_Array ) {
 862                                 if ( $index >= $iteratorNode->getLength() ) {
 863                                         // All done with this iterator
 864                                         $iteratorStack[$level] = false;
 865                                         $contextNode = false;
 866                                 } else {
 867                                         $contextNode = $iteratorNode->item( $index );
 868                                         $index++;
 869                                 }
 870                         } else {
 871                                 // Copy to $contextNode and then delete from iterator stack,
 872                                 // because this is not an iterator but we do have to execute it once
 873                                 $contextNode = $iteratorStack[$level];
 874                                 $iteratorStack[$level] = false;
 875                         }
 876
 877                         $newIterator = false;
 878
 879                         if ( $contextNode === false ) {
 880                                 // nothing to do
 881                         } elseif ( is_string( $contextNode ) ) {
 882                                 $out .= $contextNode;
 883                         } elseif ( is_array( $contextNode ) || $contextNode instanceof PPNode_Hash_Array ) {
 884                                 $newIterator = $contextNode;
 885                         } elseif ( $contextNode instanceof PPNode_Hash_Attr ) {
 886                                 // No output
 887                         } elseif ( $contextNode instanceof PPNode_Hash_Text ) {
 888                                 $out .= $contextNode->value;
 889                         } elseif ( $contextNode instanceof PPNode_Hash_Tree ) {
 890                                 if ( $contextNode->name == 'template' ) {
 891                                         # Double-brace expansion
 892                                         $bits = $contextNode->splitTemplate();
 893                                         if ( $flags & self::NO_TEMPLATES ) {
 894                                                 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $bits['title'], $bits['parts'] );
 895                                         } else {
 896                                                 $ret = $this->parser->braceSubstitution( $bits, $this );
 897                                                 if ( isset( $ret['object'] ) ) {
 898                                                         $newIterator = $ret['object'];
 899                                                 } else {
 900                                                         $out .= $ret['text'];
 901                                                 }
 902                                         }
 903                                 } elseif ( $contextNode->name == 'tplarg' ) {
 904                                         # Triple-brace expansion
 905                                         $bits = $contextNode->splitTemplate();
 906                                         if ( $flags & self::NO_ARGS ) {
 907                                                 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $bits['title'], $bits['parts'] );
 908                                         } else {
 909                                                 $ret = $this->parser->argSubstitution( $bits, $this );
 910                                                 if ( isset( $ret['object'] ) ) {
 911                                                         $newIterator = $ret['object'];
 912                                                 } else {
 913                                                         $out .= $ret['text'];
 914                                                 }
 915                                         }
 916                                 } elseif ( $contextNode->name == 'comment' ) {
 917                                         # HTML-style comment
 918                                         # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
 919                                         if ( $this->parser->ot['html']
 920                                                 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
 921                                                 || ( $flags & self::STRIP_COMMENTS ) )
 922                                         {
 923                                                 $out .= '';
 924                                         }
 925                                         # Add a strip marker in PST mode so that pstPass2() can run some old-fashioned regexes on the result
 926                                         # Not in RECOVER_COMMENTS mode (extractSections) though
 927                                         elseif ( $this->parser->ot['wiki'] && ! ( $flags & self::RECOVER_COMMENTS ) ) {
 928                                                 $out .= $this->parser->insertStripItem( $contextNode->firstChild->value );
 929                                         }
 930                                         # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
 931                                         else {
 932                                                 $out .= $contextNode->firstChild->value;
 933                                         }
 934                                 } elseif ( $contextNode->name == 'ignore' ) {
 935                                         # Output suppression used by <includeonly> etc.
 936                                         # OT_WIKI will only respect <ignore> in substed templates.
 937                                         # The other output types respect it unless NO_IGNORE is set.
 938                                         # extractSections() sets NO_IGNORE and so never respects it.
 939                                         if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & self::NO_IGNORE ) ) {
 940                                                 $out .= $contextNode->firstChild->value;
 941                                         } else {
 942                                                 //$out .= '';
 943                                         }
 944                                 } elseif ( $contextNode->name == 'ext' ) {
 945                                         # Extension tag
 946                                         $bits = $contextNode->splitExt() + array( 'attr' => null, 'inner' => null, 'close' => null );
 947                                         $out .= $this->parser->extensionSubstitution( $bits, $this );
 948                                 } elseif ( $contextNode->name == 'h' ) {
 949                                         # Heading
 950                                         if ( $this->parser->ot['html'] ) {
 951                                                 # Expand immediately and insert heading index marker
 952                                                 $s = '';
 953                                                 for ( $node = $contextNode->firstChild; $node; $node = $node->nextSibling ) {
 954                                                         $s .= $this->expand( $node, $flags );
 955                                                 }
 956
 957                                                 $bits = $contextNode->splitHeading();
 958                                                 $titleText = $this->title->getPrefixedDBkey();
 959                                                 $this->parser->mHeadings[] = array( $titleText, $bits['i'] );
 960                                                 $serial = count( $this->parser->mHeadings ) - 1;
 961                                                 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
 962                                                 $s = substr( $s, 0, $bits['level'] ) . $marker . substr( $s, $bits['level'] );
 963                                                 $this->parser->mStripState->general->setPair( $marker, '' );
 964                                                 $out .= $s;
 965                                         } else {
 966                                                 # Expand in virtual stack
 967                                                 $newIterator = $contextNode->getChildren();
 968                                         }
 969                                 } else {
 970                                         # Generic recursive expansion
 971                                         $newIterator = $contextNode->getChildren();
 972                                 }
 973                         } else {
 974                                 throw new MWException( __METHOD__.': Invalid parameter type' );
 975                         }
 976
 977                         if ( $newIterator !== false ) {
 978                                 $outStack[] = '';
 979                                 $iteratorStack[] = $newIterator;
 980                                 $indexStack[] = 0;
 981                         } elseif ( $iteratorStack[$level] === false ) {
 982                                 // Return accumulated value to parent
 983                                 // With tail recursion
 984                                 while ( $iteratorStack[$level] === false && $level > 0 ) {
 985                                         $outStack[$level - 1] .= $out;
 986                                         array_pop( $outStack );
 987                                         array_pop( $iteratorStack );
 988                                         array_pop( $indexStack );
 989                                         $level--;
 990                                 }
 991                         }
 992                 }
 993                 --$expansionDepth;
 994                 return $outStack[0];
 995         }
 996
 997         function implodeWithFlags( $sep, $flags /*, ... */ ) {
 998                 $args = array_slice( func_get_args(), 2 );
 999
1000                 $first = true;
1001                 $s = '';
1002                 foreach ( $args as $root ) {
1003                         if ( $root instanceof PPNode_Hash_Array ) {
1004                                 $root = $root->value;
1005                         }
1006                         if ( !is_array( $root ) ) {
1007                                 $root = array( $root );
1008                         }
1009                         foreach ( $root as $node ) {
1010                                 if ( $first ) {
1011                                         $first = false;
1012                                 } else {
1013                                         $s .= $sep;
1014                                 }
1015                                 $s .= $this->expand( $node, $flags );
1016                         }
1017                 }
1018                 return $s;
1019         }
1020
1021         /**
1022          * Implode with no flags specified
1023          * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1024          */
1025         function implode( $sep /*, ... */ ) {
1026                 $args = array_slice( func_get_args(), 1 );
1027
1028                 $first = true;
1029                 $s = '';
1030                 foreach ( $args as $root ) {
1031                         if ( $root instanceof PPNode_Hash_Array ) {
1032                                 $root = $root->value;
1033                         }
1034                         if ( !is_array( $root ) ) {
1035                                 $root = array( $root );
1036                         }
1037                         foreach ( $root as $node ) {
1038                                 if ( $first ) {
1039                                         $first = false;
1040                                 } else {
1041                                         $s .= $sep;
1042                                 }
1043                                 $s .= $this->expand( $node );
1044                         }
1045                 }
1046                 return $s;
1047         }
1048
1049         /**
1050          * Makes an object that, when expand()ed, will be the same as one obtained
1051          * with implode()
1052          */
1053         function virtualImplode( $sep /*, ... */ ) {
1054                 $args = array_slice( func_get_args(), 1 );
1055                 $out = array();
1056                 $first = true;
1057
1058                 foreach ( $args as $root ) {
1059                         if ( $root instanceof PPNode_Hash_Array ) {
1060                                 $root = $root->value;
1061                         }
1062                         if ( !is_array( $root ) ) {
1063                                 $root = array( $root );
1064                         }
1065                         foreach ( $root as $node ) {
1066                                 if ( $first ) {
1067                                         $first = false;
1068                                 } else {
1069                                         $out[] = $sep;
1070                                 }
1071                                 $out[] = $node;
1072                         }
1073                 }
1074                 return new PPNode_Hash_Array( $out );
1075         }
1076
1077         /**
1078          * Virtual implode with brackets
1079          */
1080         function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1081                 $args = array_slice( func_get_args(), 3 );
1082                 $out = array( $start );
1083                 $first = true;
1084
1085                 foreach ( $args as $root ) {
1086                         if ( $root instanceof PPNode_Hash_Array ) {
1087                                 $root = $root->value;
1088                         }
1089                         if ( !is_array( $root ) ) {
1090                                 $root = array( $root );
1091                         }
1092                         foreach ( $root as $node ) {
1093                                 if ( $first ) {
1094                                         $first = false;
1095                                 } else {
1096                                         $out[] = $sep;
1097                                 }
1098                                 $out[] = $node;
1099                         }
1100                 }
1101                 $out[] = $end;
1102                 return new PPNode_Hash_Array( $out );
1103         }
1104
1105         function __toString() {
1106                 return 'frame{}';
1107         }
1108
1109         function getPDBK( $level = false ) {
1110                 if ( $level === false ) {
1111                         return $this->title->getPrefixedDBkey();
1112                 } else {
1113                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1114                 }
1115         }
1116
1117         /**
1118          * Returns true if there are no arguments in this frame
1119          */
1120         function isEmpty() {
1121                 return true;
1122         }
1123
1124         function getArgument( $name ) {
1125                 return false;
1126         }
1127
1128         /**
1129          * Returns true if the infinite loop check is OK, false if a loop is detected
1130          */
1131         function loopCheck( $title ) {
1132                 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1133         }
1134
1135         /**
1136          * Return true if the frame is a template frame
1137          */
1138         function isTemplate() {
1139                 return false;
1140         }
1141 }
1142
1143 /**
1144  * Expansion frame with template arguments
1145  * @ingroup Parser
1146  */
1147 class PPTemplateFrame_Hash extends PPFrame_Hash {
1148         var $numberedArgs, $namedArgs, $parent;
1149         var $numberedExpansionCache, $namedExpansionCache;
1150
1151         function __construct( $preprocessor, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
1152                 $this->preprocessor = $preprocessor;
1153                 $this->parser = $preprocessor->parser;
1154                 $this->parent = $parent;
1155                 $this->numberedArgs = $numberedArgs;
1156                 $this->namedArgs = $namedArgs;
1157                 $this->title = $title;
1158                 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1159                 $this->titleCache = $parent->titleCache;
1160                 $this->titleCache[] = $pdbk;
1161                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1162                 if ( $pdbk !== false ) {
1163                         $this->loopCheckHash[$pdbk] = true;
1164                 }
1165                 $this->depth = $parent->depth + 1;
1166                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1167         }
1168
1169         function __toString() {
1170                 $s = 'tplframe{';
1171                 $first = true;
1172                 $args = $this->numberedArgs + $this->namedArgs;
1173                 foreach ( $args as $name => $value ) {
1174                         if ( $first ) {
1175                                 $first = false;
1176                         } else {
1177                                 $s .= ', ';
1178                         }
1179                         $s .= "\"$name\":\"" .
1180                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1181                 }
1182                 $s .= '}';
1183                 return $s;
1184         }
1185         /**
1186          * Returns true if there are no arguments in this frame
1187          */
1188         function isEmpty() {
1189                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1190         }
1191
1192         function getArguments() {
1193                 $arguments = array();
1194                 foreach ( array_merge(
1195                                 array_keys($this->numberedArgs),
1196                                 array_keys($this->namedArgs)) as $key ) {
1197                         $arguments[$key] = $this->getArgument($key);
1198                 }
1199                 return $arguments;
1200         }
1201
1202         function getNumberedArguments() {
1203                 $arguments = array();
1204                 foreach ( array_keys($this->numberedArgs) as $key ) {
1205                         $arguments[$key] = $this->getArgument($key);
1206                 }
1207                 return $arguments;
1208         }
1209
1210         function getNamedArguments() {
1211                 $arguments = array();
1212                 foreach ( array_keys($this->namedArgs) as $key ) {
1213                         $arguments[$key] = $this->getArgument($key);
1214                 }
1215                 return $arguments;
1216         }
1217
1218         function getNumberedArgument( $index ) {
1219                 if ( !isset( $this->numberedArgs[$index] ) ) {
1220                         return false;
1221                 }
1222                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1223                         # No trimming for unnamed arguments
1224                         $this->numberedExpansionCache[$index] = $this->parent->expand( $this->numberedArgs[$index], self::STRIP_COMMENTS );
1225                 }
1226                 return $this->numberedExpansionCache[$index];
1227         }
1228
1229         function getNamedArgument( $name ) {
1230                 if ( !isset( $this->namedArgs[$name] ) ) {
1231                         return false;
1232                 }
1233                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1234                         # Trim named arguments post-expand, for backwards compatibility
1235                         $this->namedExpansionCache[$name] = trim(
1236                                 $this->parent->expand( $this->namedArgs[$name], self::STRIP_COMMENTS ) );
1237                 }
1238                 return $this->namedExpansionCache[$name];
1239         }
1240
1241         function getArgument( $name ) {
1242                 $text = $this->getNumberedArgument( $name );
1243                 if ( $text === false ) {
1244                         $text = $this->getNamedArgument( $name );
1245                 }
1246                 return $text;
1247         }
1248
1249         /**
1250          * Return true if the frame is a template frame
1251          */
1252         function isTemplate() {
1253                 return true;
1254         }
1255 }
1256
1257 /**
1258  * Expansion frame with custom arguments
1259  * @ingroup Parser
1260  */
1261 class PPCustomFrame_Hash extends PPFrame_Hash {
1262         var $args;
1263
1264         function __construct( $preprocessor, $args ) {
1265                 $this->preprocessor = $preprocessor;
1266                 $this->parser = $preprocessor->parser;
1267                 $this->args = $args;
1268         }
1269
1270         function __toString() {
1271                 $s = 'cstmframe{';
1272                 $first = true;
1273                 foreach ( $this->args as $name => $value ) {
1274                         if ( $first ) {
1275                                 $first = false;
1276                         } else {
1277                                 $s .= ', ';
1278                         }
1279                         $s .= "\"$name\":\"" .
1280                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1281                 }
1282                 $s .= '}';
1283                 return $s;
1284         }
1285
1286         function isEmpty() {
1287                 return !count( $this->args );
1288         }
1289
1290         function getArgument( $index ) {
1291                 if ( !isset( $this->args[$index] ) ) {
1292                         return false;
1293                 }
1294                 return $this->args[$index];
1295         }
1296 }
1297
1298 /**
1299  * @ingroup Parser
1300  */
1301 class PPNode_Hash_Tree implements PPNode {
1302         var $name, $firstChild, $lastChild, $nextSibling;
1303
1304         function __construct( $name ) {
1305                 $this->name = $name;
1306                 $this->firstChild = $this->lastChild = $this->nextSibling = false;
1307         }
1308
1309         function __toString() {
1310                 $inner = '';
1311                 $attribs = '';
1312                 for ( $node = $this->firstChild; $node; $node = $node->nextSibling ) {
1313                         if ( $node instanceof PPNode_Hash_Attr ) {
1314                                 $attribs .= ' ' . $node->name . '="' . htmlspecialchars( $node->value ) . '"';
1315                         } else {
1316                                 $inner .= $node->__toString();
1317                         }
1318                 }
1319                 if ( $inner === '' ) {
1320                         return "<{$this->name}$attribs/>";
1321                 } else {
1322                         return "<{$this->name}$attribs>$inner</{$this->name}>";
1323                 }
1324         }
1325
1326         static function newWithText( $name, $text ) {
1327                 $obj = new self( $name );
1328                 $obj->addChild( new PPNode_Hash_Text( $text ) );
1329                 return $obj;
1330         }
1331
1332         function addChild( $node ) {
1333                 if ( $this->lastChild === false ) {
1334                         $this->firstChild = $this->lastChild = $node;
1335                 } else {
1336                         $this->lastChild->nextSibling = $node;
1337                         $this->lastChild = $node;
1338                 }
1339         }
1340
1341         function getChildren() {
1342                 $children = array();
1343                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1344                         $children[] = $child;
1345                 }
1346                 return new PPNode_Hash_Array( $children );
1347         }
1348
1349         function getFirstChild() {
1350                 return $this->firstChild;
1351         }
1352
1353         function getNextSibling() {
1354                 return $this->nextSibling;
1355         }
1356
1357         function getChildrenOfType( $name ) {
1358                 $children = array();
1359                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1360                         if ( isset( $child->name ) && $child->name === $name ) {
1361                                 $children[] = $name;
1362                         }
1363                 }
1364                 return $children;
1365         }
1366
1367         function getLength() { return false; }
1368         function item( $i ) { return false; }
1369
1370         function getName() {
1371                 return $this->name;
1372         }
1373
1374         /**
1375          * Split a <part> node into an associative array containing:
1376          *    name          PPNode name
1377          *    index         String index
1378          *    value         PPNode value
1379          */
1380         function splitArg() {
1381                 $bits = array();
1382                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1383                         if ( !isset( $child->name ) ) {
1384                                 continue;
1385                         }
1386                         if ( $child->name === 'name' ) {
1387                                 $bits['name'] = $child;
1388                                 if ( $child->firstChild instanceof PPNode_Hash_Attr
1389                                         && $child->firstChild->name === 'index' )
1390                                 {
1391                                         $bits['index'] = $child->firstChild->value;
1392                                 }
1393                         } elseif ( $child->name === 'value' ) {
1394                                 $bits['value'] = $child;
1395                         }
1396                 }
1397
1398                 if ( !isset( $bits['name'] ) ) {
1399                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1400                 }
1401                 if ( !isset( $bits['index'] ) ) {
1402                         $bits['index'] = '';
1403                 }
1404                 return $bits;
1405         }
1406
1407         /**
1408          * Split an <ext> node into an associative array containing name, attr, inner and close
1409          * All values in the resulting array are PPNodes. Inner and close are optional.
1410          */
1411         function splitExt() {
1412                 $bits = array();
1413                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1414                         if ( !isset( $child->name ) ) {
1415                                 continue;
1416                         }
1417                         if ( $child->name == 'name' ) {
1418                                 $bits['name'] = $child;
1419                         } elseif ( $child->name == 'attr' ) {
1420                                 $bits['attr'] = $child;
1421                         } elseif ( $child->name == 'inner' ) {
1422                                 $bits['inner'] = $child;
1423                         } elseif ( $child->name == 'close' ) {
1424                                 $bits['close'] = $child;
1425                         }
1426                 }
1427                 if ( !isset( $bits['name'] ) ) {
1428                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1429                 }
1430                 return $bits;
1431         }
1432
1433         /**
1434          * Split an <h> node
1435          */
1436         function splitHeading() {
1437                 if ( $this->name !== 'h' ) {
1438                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1439                 }
1440                 $bits = array();
1441                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1442                         if ( !isset( $child->name ) ) {
1443                                 continue;
1444                         }
1445                         if ( $child->name == 'i' ) {
1446                                 $bits['i'] = $child->value;
1447                         } elseif ( $child->name == 'level' ) {
1448                                 $bits['level'] = $child->value;
1449                         }
1450                 }
1451                 if ( !isset( $bits['i'] ) ) {
1452                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1453                 }
1454                 return $bits;
1455         }
1456
1457         /**
1458          * Split a <template> or <tplarg> node
1459          */
1460         function splitTemplate() {
1461                 $parts = array();
1462                 $bits = array( 'lineStart' => '' );
1463                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1464                         if ( !isset( $child->name ) ) {
1465                                 continue;
1466                         }
1467                         if ( $child->name == 'title' ) {
1468                                 $bits['title'] = $child;
1469                         }
1470                         if ( $child->name == 'part' ) {
1471                                 $parts[] = $child;
1472                         }
1473                         if ( $child->name == 'lineStart' ) {
1474                                 $bits['lineStart'] = '1';
1475                         }
1476                 }
1477                 if ( !isset( $bits['title'] ) ) {
1478                         throw new MWException( 'Invalid node passed to ' . __METHOD__ );
1479                 }
1480                 $bits['parts'] = new PPNode_Hash_Array( $parts );
1481                 return $bits;
1482         }
1483 }
1484
1485 /**
1486  * @ingroup Parser
1487  */
1488 class PPNode_Hash_Text implements PPNode {
1489         var $value, $nextSibling;
1490
1491         function __construct( $value ) {
1492                 if ( is_object( $value ) ) {
1493                         throw new MWException( __CLASS__ . ' given object instead of string' );
1494                 }
1495                 $this->value = $value;
1496         }
1497
1498         function __toString() {
1499                 return htmlspecialchars( $this->value );
1500         }
1501
1502         function getNextSibling() {
1503                 return $this->nextSibling;
1504         }
1505
1506         function getChildren() { return false; }
1507         function getFirstChild() { return false; }
1508         function getChildrenOfType( $name ) { return false; }
1509         function getLength() { return false; }
1510         function item( $i ) { return false; }
1511         function getName() { return '#text'; }
1512         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1513         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1514         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1515 }
1516
1517 /**
1518  * @ingroup Parser
1519  */
1520 class PPNode_Hash_Array implements PPNode {
1521         var $value, $nextSibling;
1522
1523         function __construct( $value ) {
1524                 $this->value = $value;
1525         }
1526
1527         function __toString() {
1528                 return var_export( $this, true );
1529         }
1530
1531         function getLength() {
1532                 return count( $this->value );
1533         }
1534
1535         function item( $i ) {
1536                 return $this->value[$i];
1537         }
1538
1539         function getName() { return '#nodelist'; }
1540
1541         function getNextSibling() {
1542                 return $this->nextSibling;
1543         }
1544
1545         function getChildren() { return false; }
1546         function getFirstChild() { return false; }
1547         function getChildrenOfType( $name ) { return false; }
1548         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1549         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1550         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1551 }
1552
1553 /**
1554  * @ingroup Parser
1555  */
1556 class PPNode_Hash_Attr implements PPNode {
1557         var $name, $value, $nextSibling;
1558
1559         function __construct( $name, $value ) {
1560                 $this->name = $name;
1561                 $this->value = $value;
1562         }
1563
1564         function __toString() {
1565                 return "<@{$this->name}>" . htmlspecialchars( $this->value ) . "</@{$this->name}>";
1566         }
1567
1568         function getName() {
1569                 return $this->name;
1570         }
1571
1572         function getNextSibling() {
1573                 return $this->nextSibling;
1574         }
1575
1576         function getChildren() { return false; }
1577         function getFirstChild() { return false; }
1578         function getChildrenOfType( $name ) { return false; }
1579         function getLength() { return false; }
1580         function item( $i ) { return false; }
1581         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1582         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1583         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1584 }