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