includes/parser/Preprocessor_Hash.php

   1 <?php
   2 /**
   3  * Preprocessor using PHP arrays
   4  *
   5  * @file
   6  * @ingroup Parser
   7  */
   8
   9 /**
  10  * Differences from DOM schema:
  11  *   * attribute nodes are children
  12  *   * <h> nodes that aren't at the top are replaced with <possible-h>
  13  * @ingroup Parser
  14  */
  15 class Preprocessor_Hash implements Preprocessor {
  16         /**
  17          * @var Parser
  18          */
  19         var $parser;
  20
  21         const CACHE_VERSION = 1;
  22
  23         function __construct( $parser ) {
  24                 $this->parser = $parser;
  25         }
  26
  27         function newFrame() {
  28                 return new PPFrame_Hash( $this );
  29         }
  30
  31         function newCustomFrame( $args ) {
  32                 return new PPCustomFrame_Hash( $this, $args );
  33         }
  34
  35         function newPartNodeArray( $values ) {
  36                 $list = array();
  37
  38                 foreach ( $values as $k => $val ) {
  39                         $partNode = new PPNode_Hash_Tree( 'part' );
  40                         $nameNode = new PPNode_Hash_Tree( 'name' );
  41
  42                         if ( is_int( $k ) ) {
  43                                 $nameNode->addChild( new PPNode_Hash_Attr( 'index', $k ) );
  44                                 $partNode->addChild( $nameNode );
  45                         } else {
  46                                 $nameNode->addChild( new PPNode_Hash_Text( $k ) );
  47                                 $partNode->addChild( $nameNode );
  48                                 $partNode->addChild( new PPNode_Hash_Text( '=' ) );
  49                         }
  50
  51                         $valueNode = new PPNode_Hash_Tree( 'value' );
  52                         $valueNode->addChild( new PPNode_Hash_Text( $val ) );
  53                         $partNode->addChild( $valueNode );
  54
  55                         $list[] = $partNode;
  56                 }
  57
  58                 $node = new PPNode_Hash_Array( $list );
  59                 return $node;
  60         }
  61
  62         /**
  63          * Preprocess some wikitext and return the document tree.
  64          * This is the ghost of Parser::replace_variables().
  65          *
  66          * @param $text String: the text to parse
  67          * @param $flags Integer: bitwise combination of:
  68          *          Parser::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
  69          *                                     included. Default is to assume a direct page view.
  70          *
  71          * The generated DOM tree must depend only on the input text and the flags.
  72          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
  73          *
  74          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
  75          * change in the DOM tree for a given text, must be passed through the section identifier
  76          * in the section edit link and thus back to extractSections().
  77          *
  78          * The output of this function is currently only cached in process memory, but a persistent
  79          * cache may be implemented at a later date which takes further advantage of these strict
  80          * dependency requirements.
  81          *
  82          * @private
  83          */
  84         function preprocessToObj( $text, $flags = 0 ) {
  85                 wfProfileIn( __METHOD__ );
  86
  87                 // Check cache.
  88                 global $wgMemc, $wgPreprocessorCacheThreshold;
  89
  90                 $cacheable = $wgPreprocessorCacheThreshold !== false && strlen( $text ) > $wgPreprocessorCacheThreshold;
  91                 if ( $cacheable ) {
  92                         wfProfileIn( __METHOD__.'-cacheable' );
  93
  94                         $cacheKey = wfMemcKey( 'preprocess-hash', md5($text), $flags );
  95                         $cacheValue = $wgMemc->get( $cacheKey );
  96                         if ( $cacheValue ) {
  97                                 $version = substr( $cacheValue, 0, 8 );
  98                                 if ( intval( $version ) == self::CACHE_VERSION ) {
  99                                         $hash = unserialize( substr( $cacheValue, 8 ) );
 100                                         // From the cache
 101                                         wfDebugLog( "Preprocessor",
 102                                                 "Loaded preprocessor hash from memcached (key $cacheKey)" );
 103                                         wfProfileOut( __METHOD__.'-cacheable' );
 104                                         wfProfileOut( __METHOD__ );
 105                                         return $hash;
 106                                 }
 107                         }
 108                         wfProfileIn( __METHOD__.'-cache-miss' );
 109                 }
 110
 111                 $rules = array(
 112                         '{' => array(
 113                                 'end' => '}',
 114                                 'names' => array(
 115                                         2 => 'template',
 116                                         3 => 'tplarg',
 117                                 ),
 118                                 'min' => 2,
 119                                 'max' => 3,
 120                         ),
 121                         '[' => array(
 122                                 'end' => ']',
 123                                 'names' => array( 2 => null ),
 124                                 'min' => 2,
 125                                 'max' => 2,
 126                         )
 127                 );
 128
 129                 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
 130
 131                 $xmlishElements = $this->parser->getStripList();
 132                 $enableOnlyinclude = false;
 133                 if ( $forInclusion ) {
 134                         $ignoredTags = array( 'includeonly', '/includeonly' );
 135                         $ignoredElements = array( 'noinclude' );
 136                         $xmlishElements[] = 'noinclude';
 137                         if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
 138                                 $enableOnlyinclude = true;
 139                         }
 140                 } else {
 141                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
 142                         $ignoredElements = array( 'includeonly' );
 143                         $xmlishElements[] = 'includeonly';
 144                 }
 145                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
 146
 147                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
 148                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
 149
 150                 $stack = new PPDStack_Hash;
 151
 152                 $searchBase = "[{<\n";
 153                 $revText = strrev( $text ); // For fast reverse searches
 154
 155                 $i = 0;                     # Input pointer, starts out pointing to a pseudo-newline before the start
 156                 $accum =& $stack->getAccum();   # Current accumulator
 157                 $findEquals = false;            # True to find equals signs in arguments
 158                 $findPipe = false;              # True to take notice of pipe characters
 159                 $headingIndex = 1;
 160                 $inHeading = false;        # True if $i is inside a possible heading
 161                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
 162                 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
 163                 $fakeLineStart = true;     # Do a line-start run without outputting an LF character
 164
 165                 while ( true ) {
 166                         //$this->memCheck();
 167
 168                         if ( $findOnlyinclude ) {
 169                                 // Ignore all input up to the next <onlyinclude>
 170                                 $startPos = strpos( $text, '<onlyinclude>', $i );
 171                                 if ( $startPos === false ) {
 172                                         // Ignored section runs to the end
 173                                         $accum->addNodeWithText( 'ignore', substr( $text, $i ) );
 174                                         break;
 175                                 }
 176                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
 177                                 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i ) );
 178                                 $i = $tagEndPos;
 179                                 $findOnlyinclude = false;
 180                         }
 181
 182                         if ( $fakeLineStart ) {
 183                                 $found = 'line-start';
 184                                 $curChar = '';
 185                         } else {
 186                                 # Find next opening brace, closing brace or pipe
 187                                 $search = $searchBase;
 188                                 if ( $stack->top === false ) {
 189                                         $currentClosing = '';
 190                                 } else {
 191                                         $currentClosing = $stack->top->close;
 192                                         $search .= $currentClosing;
 193                                 }
 194                                 if ( $findPipe ) {
 195                                         $search .= '|';
 196                                 }
 197                                 if ( $findEquals ) {
 198                                         // First equals will be for the template
 199                                         $search .= '=';
 200                                 }
 201                                 $rule = null;
 202                                 # Output literal section, advance input counter
 203                                 $literalLength = strcspn( $text, $search, $i );
 204                                 if ( $literalLength > 0 ) {
 205                                         $accum->addLiteral( substr( $text, $i, $literalLength ) );
 206                                         $i += $literalLength;
 207                                 }
 208                                 if ( $i >= strlen( $text ) ) {
 209                                         if ( $currentClosing == "\n" ) {
 210                                                 // Do a past-the-end run to finish off the heading
 211                                                 $curChar = '';
 212                                                 $found = 'line-end';
 213                                         } else {
 214                                                 # All done
 215                                                 break;
 216                                         }
 217                                 } else {
 218                                         $curChar = $text[$i];
 219                                         if ( $curChar == '|' ) {
 220                                                 $found = 'pipe';
 221                                         } elseif ( $curChar == '=' ) {
 222                                                 $found = 'equals';
 223                                         } elseif ( $curChar == '<' ) {
 224                                                 $found = 'angle';
 225                                         } elseif ( $curChar == "\n" ) {
 226                                                 if ( $inHeading ) {
 227                                                         $found = 'line-end';
 228                                                 } else {
 229                                                         $found = 'line-start';
 230                                                 }
 231                                         } elseif ( $curChar == $currentClosing ) {
 232                                                 $found = 'close';
 233                                         } elseif ( isset( $rules[$curChar] ) ) {
 234                                                 $found = 'open';
 235                                                 $rule = $rules[$curChar];
 236                                         } else {
 237                                                 # Some versions of PHP have a strcspn which stops on null characters
 238                                                 # Ignore and continue
 239                                                 ++$i;
 240                                                 continue;
 241                                         }
 242                                 }
 243                         }
 244
 245                         if ( $found == 'angle' ) {
 246                                 $matches = false;
 247                                 // Handle </onlyinclude>
 248                                 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
 249                                         $findOnlyinclude = true;
 250                                         continue;
 251                                 }
 252
 253                                 // Determine element name
 254                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
 255                                         // Element name missing or not listed
 256                                         $accum->addLiteral( '<' );
 257                                         ++$i;
 258                                         continue;
 259                                 }
 260                                 // Handle comments
 261                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
 262                                         // To avoid leaving blank lines, when a comment is both preceded
 263                                         // and followed by a newline (ignoring spaces), trim leading and
 264                                         // trailing spaces and one of the newlines.
 265
 266                                         // Find the end
 267                                         $endPos = strpos( $text, '-->', $i + 4 );
 268                                         if ( $endPos === false ) {
 269                                                 // Unclosed comment in input, runs to end
 270                                                 $inner = substr( $text, $i );
 271                                                 $accum->addNodeWithText( 'comment', $inner );
 272                                                 $i = strlen( $text );
 273                                         } else {
 274                                                 // Search backwards for leading whitespace
 275                                                 $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
 276                                                 // Search forwards for trailing whitespace
 277                                                 // $wsEnd will be the position of the last space (or the '>' if there's none)
 278                                                 $wsEnd = $endPos + 2 + strspn( $text, ' ', $endPos + 3 );
 279                                                 // Eat the line if possible
 280                                                 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
 281                                                 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
 282                                                 // it's a possible beneficial b/c break.
 283                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
 284                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n" )
 285                                                 {
 286                                                         $startPos = $wsStart;
 287                                                         $endPos = $wsEnd + 1;
 288                                                         // Remove leading whitespace from the end of the accumulator
 289                                                         // Sanity check first though
 290                                                         $wsLength = $i - $wsStart;
 291                                                         if ( $wsLength > 0
 292                                                                 && $accum->lastNode instanceof PPNode_Hash_Text
 293                                                                 && substr( $accum->lastNode->value, -$wsLength ) === str_repeat( ' ', $wsLength ) )
 294                                                         {
 295                                                                 $accum->lastNode->value = substr( $accum->lastNode->value, 0, -$wsLength );
 296                                                         }
 297                                                         // Do a line-start run next time to look for headings after the comment
 298                                                         $fakeLineStart = true;
 299                                                 } else {
 300                                                         // No line to eat, just take the comment itself
 301                                                         $startPos = $i;
 302                                                         $endPos += 2;
 303                                                 }
 304
 305                                                 if ( $stack->top ) {
 306                                                         $part = $stack->top->getCurrentPart();
 307                                                         if ( isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 ) {
 308                                                                 // Comments abutting, no change in visual end
 309                                                                 $part->commentEnd = $wsEnd;
 310                                                         } else {
 311                                                                 $part->visualEnd = $wsStart;
 312                                                                 $part->commentEnd = $endPos;
 313                                                         }
 314                                                 }
 315                                                 $i = $endPos + 1;
 316                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
 317                                                 $accum->addNodeWithText( 'comment', $inner );
 318                                         }
 319                                         continue;
 320                                 }
 321                                 $name = $matches[1];
 322                                 $lowerName = strtolower( $name );
 323                                 $attrStart = $i + strlen( $name ) + 1;
 324
 325                                 // Find end of tag
 326                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
 327                                 if ( $tagEndPos === false ) {
 328                                         // Infinite backtrack
 329                                         // Disable tag search to prevent worst-case O(N^2) performance
 330                                         $noMoreGT = true;
 331                                         $accum->addLiteral( '<' );
 332                                         ++$i;
 333                                         continue;
 334                                 }
 335
 336                                 // Handle ignored tags
 337                                 if ( in_array( $lowerName, $ignoredTags ) ) {
 338                                         $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i + 1 ) );
 339                                         $i = $tagEndPos + 1;
 340                                         continue;
 341                                 }
 342
 343                                 $tagStartPos = $i;
 344                                 if ( $text[$tagEndPos-1] == '/' ) {
 345                                         // Short end tag
 346                                         $attrEnd = $tagEndPos - 1;
 347                                         $inner = null;
 348                                         $i = $tagEndPos + 1;
 349                                         $close = null;
 350                                 } else {
 351                                         $attrEnd = $tagEndPos;
 352                                         // Find closing tag
 353                                         if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
 354                                                         $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) )
 355                                         {
 356                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
 357                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
 358                                                 $close = $matches[0][0];
 359                                         } else {
 360                                                 // No end tag -- let it run out to the end of the text.
 361                                                 $inner = substr( $text, $tagEndPos + 1 );
 362                                                 $i = strlen( $text );
 363                                                 $close = null;
 364                                         }
 365                                 }
 366                                 // <includeonly> and <noinclude> just become <ignore> tags
 367                                 if ( in_array( $lowerName, $ignoredElements ) ) {
 368                                         $accum->addNodeWithText(  'ignore', substr( $text, $tagStartPos, $i - $tagStartPos ) );
 369                                         continue;
 370                                 }
 371
 372                                 if ( $attrEnd <= $attrStart ) {
 373                                         $attr = '';
 374                                 } else {
 375                                         // Note that the attr element contains the whitespace between name and attribute,
 376                                         // this is necessary for precise reconstruction during pre-save transform.
 377                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
 378                                 }
 379
 380                                 $extNode = new PPNode_Hash_Tree( 'ext' );
 381                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'name', $name ) );
 382                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'attr', $attr ) );
 383                                 if ( $inner !== null ) {
 384                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'inner', $inner ) );
 385                                 }
 386                                 if ( $close !== null ) {
 387                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'close', $close ) );
 388                                 }
 389                                 $accum->addNode( $extNode );
 390                         }
 391
 392                         elseif ( $found == 'line-start' ) {
 393                                 // Is this the start of a heading?
 394                                 // Line break belongs before the heading element in any case
 395                                 if ( $fakeLineStart ) {
 396                                         $fakeLineStart = false;
 397                                 } else {
 398                                         $accum->addLiteral( $curChar );
 399                                         $i++;
 400                                 }
 401
 402                                 $count = strspn( $text, '=', $i, 6 );
 403                                 if ( $count == 1 && $findEquals ) {
 404                                         // DWIM: This looks kind of like a name/value separator
 405                                         // Let's let the equals handler have it and break the potential heading
 406                                         // This is heuristic, but AFAICT the methods for completely correct disambiguation are very complex.
 407                                 } elseif ( $count > 0 ) {
 408                                         $piece = array(
 409                                                 'open' => "\n",
 410                                                 'close' => "\n",
 411                                                 'parts' => array( new PPDPart_Hash( str_repeat( '=', $count ) ) ),
 412                                                 'startPos' => $i,
 413                                                 'count' => $count );
 414                                         $stack->push( $piece );
 415                                         $accum =& $stack->getAccum();
 416                                         extract( $stack->getFlags() );
 417                                         $i += $count;
 418                                 }
 419                         }
 420
 421                         elseif ( $found == 'line-end' ) {
 422                                 $piece = $stack->top;
 423                                 // A heading must be open, otherwise \n wouldn't have been in the search list
 424                                 assert( $piece->open == "\n" );
 425                                 $part = $piece->getCurrentPart();
 426                                 // Search back through the input to see if it has a proper close
 427                                 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
 428                                 $wsLength = strspn( $revText, " \t", strlen( $text ) - $i );
 429                                 $searchStart = $i - $wsLength;
 430                                 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
 431                                         // Comment found at line end
 432                                         // Search for equals signs before the comment
 433                                         $searchStart = $part->visualEnd;
 434                                         $searchStart -= strspn( $revText, " \t", strlen( $text ) - $searchStart );
 435                                 }
 436                                 $count = $piece->count;
 437                                 $equalsLength = strspn( $revText, '=', strlen( $text ) - $searchStart );
 438                                 if ( $equalsLength > 0 ) {
 439                                         if ( $searchStart - $equalsLength == $piece->startPos ) {
 440                                                 // This is just a single string of equals signs on its own line
 441                                                 // Replicate the doHeadings behaviour /={count}(.+)={count}/
 442                                                 // First find out how many equals signs there really are (don't stop at 6)
 443                                                 $count = $equalsLength;
 444                                                 if ( $count < 3 ) {
 445                                                         $count = 0;
 446                                                 } else {
 447                                                         $count = min( 6, intval( ( $count - 1 ) / 2 ) );
 448                                                 }
 449                                         } else {
 450                                                 $count = min( $equalsLength, $count );
 451                                         }
 452                                         if ( $count > 0 ) {
 453                                                 // Normal match, output <h>
 454                                                 $element = new PPNode_Hash_Tree( 'possible-h' );
 455                                                 $element->addChild( new PPNode_Hash_Attr( 'level', $count ) );
 456                                                 $element->addChild( new PPNode_Hash_Attr( 'i', $headingIndex++ ) );
 457                                                 $element->lastChild->nextSibling = $accum->firstNode;
 458                                                 $element->lastChild = $accum->lastNode;
 459                                         } else {
 460                                                 // Single equals sign on its own line, count=0
 461                                                 $element = $accum;
 462                                         }
 463                                 } else {
 464                                         // No match, no <h>, just pass down the inner text
 465                                         $element = $accum;
 466                                 }
 467                                 // Unwind the stack
 468                                 $stack->pop();
 469                                 $accum =& $stack->getAccum();
 470                                 extract( $stack->getFlags() );
 471
 472                                 // Append the result to the enclosing accumulator
 473                                 if ( $element instanceof PPNode ) {
 474                                         $accum->addNode( $element );
 475                                 } else {
 476                                         $accum->addAccum( $element );
 477                                 }
 478                                 // Note that we do NOT increment the input pointer.
 479                                 // This is because the closing linebreak could be the opening linebreak of
 480                                 // another heading. Infinite loops are avoided because the next iteration MUST
 481                                 // hit the heading open case above, which unconditionally increments the
 482                                 // input pointer.
 483                         }
 484
 485                         elseif ( $found == 'open' ) {
 486                                 # count opening brace characters
 487                                 $count = strspn( $text, $curChar, $i );
 488
 489                                 # we need to add to stack only if opening brace count is enough for one of the rules
 490                                 if ( $count >= $rule['min'] ) {
 491                                         # Add it to the stack
 492                                         $piece = array(
 493                                                 'open' => $curChar,
 494                                                 'close' => $rule['end'],
 495                                                 'count' => $count,
 496                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
 497                                         );
 498
 499                                         $stack->push( $piece );
 500                                         $accum =& $stack->getAccum();
 501                                         extract( $stack->getFlags() );
 502                                 } else {
 503                                         # Add literal brace(s)
 504                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 505                                 }
 506                                 $i += $count;
 507                         }
 508
 509                         elseif ( $found == 'close' ) {
 510                                 $piece = $stack->top;
 511                                 # lets check if there are enough characters for closing brace
 512                                 $maxCount = $piece->count;
 513                                 $count = strspn( $text, $curChar, $i, $maxCount );
 514
 515                                 # check for maximum matching characters (if there are 5 closing
 516                                 # characters, we will probably need only 3 - depending on the rules)
 517                                 $rule = $rules[$piece->open];
 518                                 if ( $count > $rule['max'] ) {
 519                                         # The specified maximum exists in the callback array, unless the caller
 520                                         # has made an error
 521                                         $matchingCount = $rule['max'];
 522                                 } else {
 523                                         # Count is less than the maximum
 524                                         # Skip any gaps in the callback array to find the true largest match
 525                                         # Need to use array_key_exists not isset because the callback can be null
 526                                         $matchingCount = $count;
 527                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
 528                                                 --$matchingCount;
 529                                         }
 530                                 }
 531
 532                                 if ($matchingCount <= 0) {
 533                                         # No matching element found in callback array
 534                                         # Output a literal closing brace and continue
 535                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 536                                         $i += $count;
 537                                         continue;
 538                                 }
 539                                 $name = $rule['names'][$matchingCount];
 540                                 if ( $name === null ) {
 541                                         // No element, just literal text
 542                                         $element = $piece->breakSyntax( $matchingCount );
 543                                         $element->addLiteral( str_repeat( $rule['end'], $matchingCount ) );
 544                                 } else {
 545                                         # Create XML element
 546                                         # Note: $parts is already XML, does not need to be encoded further
 547                                         $parts = $piece->parts;
 548                                         $titleAccum = $parts[0]->out;
 549                                         unset( $parts[0] );
 550
 551                                         $element = new PPNode_Hash_Tree( $name );
 552
 553                                         # The invocation is at the start of the line if lineStart is set in
 554                                         # the stack, and all opening brackets are used up.
 555                                         if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
 556                                                 $element->addChild( new PPNode_Hash_Attr( 'lineStart', 1 ) );
 557                                         }
 558                                         $titleNode = new PPNode_Hash_Tree( 'title' );
 559                                         $titleNode->firstChild = $titleAccum->firstNode;
 560                                         $titleNode->lastChild = $titleAccum->lastNode;
 561                                         $element->addChild( $titleNode );
 562                                         $argIndex = 1;
 563                                         foreach ( $parts as $part ) {
 564                                                 if ( isset( $part->eqpos ) ) {
 565                                                         // Find equals
 566                                                         $lastNode = false;
 567                                                         for ( $node = $part->out->firstNode; $node; $node = $node->nextSibling ) {
 568                                                                 if ( $node === $part->eqpos ) {
 569                                                                         break;
 570                                                                 }
 571                                                                 $lastNode = $node;
 572                                                         }
 573                                                         if ( !$node ) {
 574                                                                 throw new MWException( __METHOD__. ': eqpos not found' );
 575                                                         }
 576                                                         if ( $node->name !== 'equals' ) {
 577                                                                 throw new MWException( __METHOD__ .': eqpos is not equals' );
 578                                                         }
 579                                                         $equalsNode = $node;
 580
 581                                                         // Construct name node
 582                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 583                                                         if ( $lastNode !== false ) {
 584                                                                 $lastNode->nextSibling = false;
 585                                                                 $nameNode->firstChild = $part->out->firstNode;
 586                                                                 $nameNode->lastChild = $lastNode;
 587                                                         }
 588
 589                                                         // Construct value node
 590                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 591                                                         if ( $equalsNode->nextSibling !== false ) {
 592                                                                 $valueNode->firstChild = $equalsNode->nextSibling;
 593                                                                 $valueNode->lastChild = $part->out->lastNode;
 594                                                         }
 595                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 596                                                         $partNode->addChild( $nameNode );
 597                                                         $partNode->addChild( $equalsNode->firstChild );
 598                                                         $partNode->addChild( $valueNode );
 599                                                         $element->addChild( $partNode );
 600                                                 } else {
 601                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 602                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 603                                                         $nameNode->addChild( new PPNode_Hash_Attr( 'index', $argIndex++ ) );
 604                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 605                                                         $valueNode->firstChild = $part->out->firstNode;
 606                                                         $valueNode->lastChild = $part->out->lastNode;
 607                                                         $partNode->addChild( $nameNode );
 608                                                         $partNode->addChild( $valueNode );
 609                                                         $element->addChild( $partNode );
 610                                                 }
 611                                         }
 612                                 }
 613
 614                                 # Advance input pointer
 615                                 $i += $matchingCount;
 616
 617                                 # Unwind the stack
 618                                 $stack->pop();
 619                                 $accum =& $stack->getAccum();
 620
 621                                 # Re-add the old stack element if it still has unmatched opening characters remaining
 622                                 if ($matchingCount < $piece->count) {
 623                                         $piece->parts = array( new PPDPart_Hash );
 624                                         $piece->count -= $matchingCount;
 625                                         # do we still qualify for any callback with remaining count?
 626                                         $names = $rules[$piece->open]['names'];
 627                                         $skippedBraces = 0;
 628                                         $enclosingAccum =& $accum;
 629                                         while ( $piece->count ) {
 630                                                 if ( array_key_exists( $piece->count, $names ) ) {
 631                                                         $stack->push( $piece );
 632                                                         $accum =& $stack->getAccum();
 633                                                         break;
 634                                                 }
 635                                                 --$piece->count;
 636                                                 $skippedBraces ++;
 637                                         }
 638                                         $enclosingAccum->addLiteral( str_repeat( $piece->open, $skippedBraces ) );
 639                                 }
 640
 641                                 extract( $stack->getFlags() );
 642
 643                                 # Add XML element to the enclosing accumulator
 644                                 if ( $element instanceof PPNode ) {
 645                                         $accum->addNode( $element );
 646                                 } else {
 647                                         $accum->addAccum( $element );
 648                                 }
 649                         }
 650
 651                         elseif ( $found == 'pipe' ) {
 652                                 $findEquals = true; // shortcut for getFlags()
 653                                 $stack->addPart();
 654                                 $accum =& $stack->getAccum();
 655                                 ++$i;
 656                         }
 657
 658                         elseif ( $found == 'equals' ) {
 659                                 $findEquals = false; // shortcut for getFlags()
 660                                 $accum->addNodeWithText( 'equals', '=' );
 661                                 $stack->getCurrentPart()->eqpos = $accum->lastNode;
 662                                 ++$i;
 663                         }
 664                 }
 665
 666                 # Output any remaining unclosed brackets
 667                 foreach ( $stack->stack as $piece ) {
 668                         $stack->rootAccum->addAccum( $piece->breakSyntax() );
 669                 }
 670
 671                 # Enable top-level headings
 672                 for ( $node = $stack->rootAccum->firstNode; $node; $node = $node->nextSibling ) {
 673                         if ( isset( $node->name ) && $node->name === 'possible-h' ) {
 674                                 $node->name = 'h';
 675                         }
 676                 }
 677
 678                 $rootNode = new PPNode_Hash_Tree( 'root' );
 679                 $rootNode->firstChild = $stack->rootAccum->firstNode;
 680                 $rootNode->lastChild = $stack->rootAccum->lastNode;
 681
 682                 // Cache
 683                 if ($cacheable) {
 684                         $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . serialize( $rootNode );
 685                         $wgMemc->set( $cacheKey, $cacheValue, 86400 );
 686                         wfProfileOut( __METHOD__.'-cache-miss' );
 687                         wfProfileOut( __METHOD__.'-cacheable' );
 688                         wfDebugLog( "Preprocessor", "Saved preprocessor Hash to memcached (key $cacheKey)" );
 689                 }
 690
 691                 wfProfileOut( __METHOD__ );
 692                 return $rootNode;
 693         }
 694 }
 695
 696 /**
 697  * Stack class to help Preprocessor::preprocessToObj()
 698  * @ingroup Parser
 699  */
 700 class PPDStack_Hash extends PPDStack {
 701         function __construct() {
 702                 $this->elementClass = 'PPDStackElement_Hash';
 703                 parent::__construct();
 704                 $this->rootAccum = new PPDAccum_Hash;
 705         }
 706 }
 707
 708 /**
 709  * @ingroup Parser
 710  */
 711 class PPDStackElement_Hash extends PPDStackElement {
 712         function __construct( $data = array() ) {
 713                 $this->partClass = 'PPDPart_Hash';
 714                 parent::__construct( $data );
 715         }
 716
 717         /**
 718          * Get the accumulator that would result if the close is not found.
 719          */
 720         function breakSyntax( $openingCount = false ) {
 721                 if ( $this->open == "\n" ) {
 722                         $accum = $this->parts[0]->out;
 723                 } else {
 724                         if ( $openingCount === false ) {
 725                                 $openingCount = $this->count;
 726                         }
 727                         $accum = new PPDAccum_Hash;
 728                         $accum->addLiteral( str_repeat( $this->open, $openingCount ) );
 729                         $first = true;
 730                         foreach ( $this->parts as $part ) {
 731                                 if ( $first ) {
 732                                         $first = false;
 733                                 } else {
 734                                         $accum->addLiteral( '|' );
 735                                 }
 736                                 $accum->addAccum( $part->out );
 737                         }
 738                 }
 739                 return $accum;
 740         }
 741 }
 742
 743 /**
 744  * @ingroup Parser
 745  */
 746 class PPDPart_Hash extends PPDPart {
 747         function __construct( $out = '' ) {
 748                 $accum = new PPDAccum_Hash;
 749                 if ( $out !== '' ) {
 750                         $accum->addLiteral( $out );
 751                 }
 752                 parent::__construct( $accum );
 753         }
 754 }
 755
 756 /**
 757  * @ingroup Parser
 758  */
 759 class PPDAccum_Hash {
 760         var $firstNode, $lastNode;
 761
 762         function __construct() {
 763                 $this->firstNode = $this->lastNode = false;
 764         }
 765
 766         /**
 767          * Append a string literal
 768          */
 769         function addLiteral( $s ) {
 770                 if ( $this->lastNode === false ) {
 771                         $this->firstNode = $this->lastNode = new PPNode_Hash_Text( $s );
 772                 } elseif ( $this->lastNode instanceof PPNode_Hash_Text ) {
 773                         $this->lastNode->value .= $s;
 774                 } else {
 775                         $this->lastNode->nextSibling = new PPNode_Hash_Text( $s );
 776                         $this->lastNode = $this->lastNode->nextSibling;
 777                 }
 778         }
 779
 780         /**
 781          * Append a PPNode
 782          */
 783         function addNode( PPNode $node ) {
 784                 if ( $this->lastNode === false ) {
 785                         $this->firstNode = $this->lastNode = $node;
 786                 } else {
 787                         $this->lastNode->nextSibling = $node;
 788                         $this->lastNode = $node;
 789                 }
 790         }
 791
 792         /**
 793          * Append a tree node with text contents
 794          */
 795         function addNodeWithText( $name, $value ) {
 796                 $node = PPNode_Hash_Tree::newWithText( $name, $value );
 797                 $this->addNode( $node );
 798         }
 799
 800         /**
 801          * Append a PPAccum_Hash
 802          * Takes over ownership of the nodes in the source argument. These nodes may
 803          * subsequently be modified, especially nextSibling.
 804          */
 805         function addAccum( $accum ) {
 806                 if ( $accum->lastNode === false ) {
 807                         // nothing to add
 808                 } elseif ( $this->lastNode === false ) {
 809                         $this->firstNode = $accum->firstNode;
 810                         $this->lastNode = $accum->lastNode;
 811                 } else {
 812                         $this->lastNode->nextSibling = $accum->firstNode;
 813                         $this->lastNode = $accum->lastNode;
 814                 }
 815         }
 816 }
 817
 818 /**
 819  * An expansion frame, used as a context to expand the result of preprocessToObj()
 820  * @ingroup Parser
 821  */
 822 class PPFrame_Hash implements PPFrame {
 823
 824         /**
 825          * @var Parser
 826          */
 827         var $parser;
 828
 829         var $preprocessor, $title;
 830         var $titleCache;
 831
 832         /**
 833          * Hashtable listing templates which are disallowed for expansion in this frame,
 834          * having been encountered previously in parent frames.
 835          */
 836         var $loopCheckHash;
 837
 838         /**
 839          * Recursion depth of this frame, top = 0
 840          * Note that this is NOT the same as expansion depth in expand()
 841          */
 842         var $depth;
 843
 844
 845         /**
 846          * Construct a new preprocessor frame.
 847          * @param $preprocessor Preprocessor: the parent preprocessor
 848          */
 849         function __construct( $preprocessor ) {
 850                 $this->preprocessor = $preprocessor;
 851                 $this->parser = $preprocessor->parser;
 852                 $this->title = $this->parser->mTitle;
 853                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
 854                 $this->loopCheckHash = array();
 855                 $this->depth = 0;
 856         }
 857
 858         /**
 859          * Create a new child frame
 860          * $args is optionally a multi-root PPNode or array containing the template arguments
 861          */
 862         function newChild( $args = false, $title = false ) {
 863                 $namedArgs = array();
 864                 $numberedArgs = array();
 865                 if ( $title === false ) {
 866                         $title = $this->title;
 867                 }
 868                 if ( $args !== false ) {
 869                         if ( $args instanceof PPNode_Hash_Array ) {
 870                                 $args = $args->value;
 871                         } elseif ( !is_array( $args ) ) {
 872                                 throw new MWException( __METHOD__ . ': $args must be array or PPNode_Hash_Array' );
 873                         }
 874                         foreach ( $args as $arg ) {
 875                                 $bits = $arg->splitArg();
 876                                 if ( $bits['index'] !== '' ) {
 877                                         // Numbered parameter
 878                                         $numberedArgs[$bits['index']] = $bits['value'];
 879                                         unset( $namedArgs[$bits['index']] );
 880                                 } else {
 881                                         // Named parameter
 882                                         $name = trim( $this->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
 883                                         $namedArgs[$name] = $bits['value'];
 884                                         unset( $numberedArgs[$name] );
 885                                 }
 886                         }
 887                 }
 888                 return new PPTemplateFrame_Hash( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
 889         }
 890
 891         function expand( $root, $flags = 0 ) {
 892                 static $expansionDepth = 0;
 893                 if ( is_string( $root ) ) {
 894                         return $root;
 895                 }
 896
 897                 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() )
 898                 {
 899                         return '<span class="error">Node-count limit exceeded</span>';
 900                 }
 901                 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
 902                         return '<span class="error">Expansion depth limit exceeded</span>';
 903                 }
 904                 ++$expansionDepth;
 905
 906                 $outStack = array( '', '' );
 907                 $iteratorStack = array( false, $root );
 908                 $indexStack = array( 0, 0 );
 909
 910                 while ( count( $iteratorStack ) > 1 ) {
 911                         $level = count( $outStack ) - 1;
 912                         $iteratorNode =& $iteratorStack[ $level ];
 913                         $out =& $outStack[$level];
 914                         $index =& $indexStack[$level];
 915
 916                         if ( is_array( $iteratorNode ) ) {
 917                                 if ( $index >= count( $iteratorNode ) ) {
 918                                         // All done with this iterator
 919                                         $iteratorStack[$level] = false;
 920                                         $contextNode = false;
 921                                 } else {
 922                                         $contextNode = $iteratorNode[$index];
 923                                         $index++;
 924                                 }
 925                         } elseif ( $iteratorNode instanceof PPNode_Hash_Array ) {
 926                                 if ( $index >= $iteratorNode->getLength() ) {
 927                                         // All done with this iterator
 928                                         $iteratorStack[$level] = false;
 929                                         $contextNode = false;
 930                                 } else {
 931                                         $contextNode = $iteratorNode->item( $index );
 932                                         $index++;
 933                                 }
 934                         } else {
 935                                 // Copy to $contextNode and then delete from iterator stack,
 936                                 // because this is not an iterator but we do have to execute it once
 937                                 $contextNode = $iteratorStack[$level];
 938                                 $iteratorStack[$level] = false;
 939                         }
 940
 941                         $newIterator = false;
 942
 943                         if ( $contextNode === false ) {
 944                                 // nothing to do
 945                         } elseif ( is_string( $contextNode ) ) {
 946                                 $out .= $contextNode;
 947                         } elseif ( is_array( $contextNode ) || $contextNode instanceof PPNode_Hash_Array ) {
 948                                 $newIterator = $contextNode;
 949                         } elseif ( $contextNode instanceof PPNode_Hash_Attr ) {
 950                                 // No output
 951                         } elseif ( $contextNode instanceof PPNode_Hash_Text ) {
 952                                 $out .= $contextNode->value;
 953                         } elseif ( $contextNode instanceof PPNode_Hash_Tree ) {
 954                                 if ( $contextNode->name == 'template' ) {
 955                                         # Double-brace expansion
 956                                         $bits = $contextNode->splitTemplate();
 957                                         if ( $flags & PPFrame::NO_TEMPLATES ) {
 958                                                 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $bits['title'], $bits['parts'] );
 959                                         } else {
 960                                                 $ret = $this->parser->braceSubstitution( $bits, $this );
 961                                                 if ( isset( $ret['object'] ) ) {
 962                                                         $newIterator = $ret['object'];
 963                                                 } else {
 964                                                         $out .= $ret['text'];
 965                                                 }
 966                                         }
 967                                 } elseif ( $contextNode->name == 'tplarg' ) {
 968                                         # Triple-brace expansion
 969                                         $bits = $contextNode->splitTemplate();
 970                                         if ( $flags & PPFrame::NO_ARGS ) {
 971                                                 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $bits['title'], $bits['parts'] );
 972                                         } else {
 973                                                 $ret = $this->parser->argSubstitution( $bits, $this );
 974                                                 if ( isset( $ret['object'] ) ) {
 975                                                         $newIterator = $ret['object'];
 976                                                 } else {
 977                                                         $out .= $ret['text'];
 978                                                 }
 979                                         }
 980                                 } elseif ( $contextNode->name == 'comment' ) {
 981                                         # HTML-style comment
 982                                         # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
 983                                         if ( $this->parser->ot['html']
 984                                                 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
 985                                                 || ( $flags & PPFrame::STRIP_COMMENTS ) )
 986                                         {
 987                                                 $out .= '';
 988                                         }
 989                                         # Add a strip marker in PST mode so that pstPass2() can run some old-fashioned regexes on the result
 990                                         # Not in RECOVER_COMMENTS mode (extractSections) though
 991                                         elseif ( $this->parser->ot['wiki'] && ! ( $flags & PPFrame::RECOVER_COMMENTS ) ) {
 992                                                 $out .= $this->parser->insertStripItem( $contextNode->firstChild->value );
 993                                         }
 994                                         # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
 995                                         else {
 996                                                 $out .= $contextNode->firstChild->value;
 997                                         }
 998                                 } elseif ( $contextNode->name == 'ignore' ) {
 999                                         # Output suppression used by <includeonly> etc.
1000                                         # OT_WIKI will only respect <ignore> in substed templates.
1001                                         # The other output types respect it unless NO_IGNORE is set.
1002                                         # extractSections() sets NO_IGNORE and so never respects it.
1003                                         if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & PPFrame::NO_IGNORE ) ) {
1004                                                 $out .= $contextNode->firstChild->value;
1005                                         } else {
1006                                                 //$out .= '';
1007                                         }
1008                                 } elseif ( $contextNode->name == 'ext' ) {
1009                                         # Extension tag
1010                                         $bits = $contextNode->splitExt() + array( 'attr' => null, 'inner' => null, 'close' => null );
1011                                         $out .= $this->parser->extensionSubstitution( $bits, $this );
1012                                 } elseif ( $contextNode->name == 'h' ) {
1013                                         # Heading
1014                                         if ( $this->parser->ot['html'] ) {
1015                                                 # Expand immediately and insert heading index marker
1016                                                 $s = '';
1017                                                 for ( $node = $contextNode->firstChild; $node; $node = $node->nextSibling ) {
1018                                                         $s .= $this->expand( $node, $flags );
1019                                                 }
1020
1021                                                 $bits = $contextNode->splitHeading();
1022                                                 $titleText = $this->title->getPrefixedDBkey();
1023                                                 $this->parser->mHeadings[] = array( $titleText, $bits['i'] );
1024                                                 $serial = count( $this->parser->mHeadings ) - 1;
1025                                                 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
1026                                                 $s = substr( $s, 0, $bits['level'] ) . $marker . substr( $s, $bits['level'] );
1027                                                 $this->parser->mStripState->setGeneral( $marker, '' );
1028                                                 $out .= $s;
1029                                         } else {
1030                                                 # Expand in virtual stack
1031                                                 $newIterator = $contextNode->getChildren();
1032                                         }
1033                                 } else {
1034                                         # Generic recursive expansion
1035                                         $newIterator = $contextNode->getChildren();
1036                                 }
1037                         } else {
1038                                 throw new MWException( __METHOD__.': Invalid parameter type' );
1039                         }
1040
1041                         if ( $newIterator !== false ) {
1042                                 $outStack[] = '';
1043                                 $iteratorStack[] = $newIterator;
1044                                 $indexStack[] = 0;
1045                         } elseif ( $iteratorStack[$level] === false ) {
1046                                 // Return accumulated value to parent
1047                                 // With tail recursion
1048                                 while ( $iteratorStack[$level] === false && $level > 0 ) {
1049                                         $outStack[$level - 1] .= $out;
1050                                         array_pop( $outStack );
1051                                         array_pop( $iteratorStack );
1052                                         array_pop( $indexStack );
1053                                         $level--;
1054                                 }
1055                         }
1056                 }
1057                 --$expansionDepth;
1058                 return $outStack[0];
1059         }
1060
1061         function implodeWithFlags( $sep, $flags /*, ... */ ) {
1062                 $args = array_slice( func_get_args(), 2 );
1063
1064                 $first = true;
1065                 $s = '';
1066                 foreach ( $args as $root ) {
1067                         if ( $root instanceof PPNode_Hash_Array ) {
1068                                 $root = $root->value;
1069                         }
1070                         if ( !is_array( $root ) ) {
1071                                 $root = array( $root );
1072                         }
1073                         foreach ( $root as $node ) {
1074                                 if ( $first ) {
1075                                         $first = false;
1076                                 } else {
1077                                         $s .= $sep;
1078                                 }
1079                                 $s .= $this->expand( $node, $flags );
1080                         }
1081                 }
1082                 return $s;
1083         }
1084
1085         /**
1086          * Implode with no flags specified
1087          * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1088          */
1089         function implode( $sep /*, ... */ ) {
1090                 $args = array_slice( func_get_args(), 1 );
1091
1092                 $first = true;
1093                 $s = '';
1094                 foreach ( $args as $root ) {
1095                         if ( $root instanceof PPNode_Hash_Array ) {
1096                                 $root = $root->value;
1097                         }
1098                         if ( !is_array( $root ) ) {
1099                                 $root = array( $root );
1100                         }
1101                         foreach ( $root as $node ) {
1102                                 if ( $first ) {
1103                                         $first = false;
1104                                 } else {
1105                                         $s .= $sep;
1106                                 }
1107                                 $s .= $this->expand( $node );
1108                         }
1109                 }
1110                 return $s;
1111         }
1112
1113         /**
1114          * Makes an object that, when expand()ed, will be the same as one obtained
1115          * with implode()
1116          */
1117         function virtualImplode( $sep /*, ... */ ) {
1118                 $args = array_slice( func_get_args(), 1 );
1119                 $out = array();
1120                 $first = true;
1121
1122                 foreach ( $args as $root ) {
1123                         if ( $root instanceof PPNode_Hash_Array ) {
1124                                 $root = $root->value;
1125                         }
1126                         if ( !is_array( $root ) ) {
1127                                 $root = array( $root );
1128                         }
1129                         foreach ( $root as $node ) {
1130                                 if ( $first ) {
1131                                         $first = false;
1132                                 } else {
1133                                         $out[] = $sep;
1134                                 }
1135                                 $out[] = $node;
1136                         }
1137                 }
1138                 return new PPNode_Hash_Array( $out );
1139         }
1140
1141         /**
1142          * Virtual implode with brackets
1143          */
1144         function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1145                 $args = array_slice( func_get_args(), 3 );
1146                 $out = array( $start );
1147                 $first = true;
1148
1149                 foreach ( $args as $root ) {
1150                         if ( $root instanceof PPNode_Hash_Array ) {
1151                                 $root = $root->value;
1152                         }
1153                         if ( !is_array( $root ) ) {
1154                                 $root = array( $root );
1155                         }
1156                         foreach ( $root as $node ) {
1157                                 if ( $first ) {
1158                                         $first = false;
1159                                 } else {
1160                                         $out[] = $sep;
1161                                 }
1162                                 $out[] = $node;
1163                         }
1164                 }
1165                 $out[] = $end;
1166                 return new PPNode_Hash_Array( $out );
1167         }
1168
1169         function __toString() {
1170                 return 'frame{}';
1171         }
1172
1173         function getPDBK( $level = false ) {
1174                 if ( $level === false ) {
1175                         return $this->title->getPrefixedDBkey();
1176                 } else {
1177                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1178                 }
1179         }
1180
1181         function getArguments() {
1182                 return array();
1183         }
1184
1185         function getNumberedArguments() {
1186                 return array();
1187         }
1188
1189         function getNamedArguments() {
1190                 return array();
1191         }
1192
1193         /**
1194          * Returns true if there are no arguments in this frame
1195          */
1196         function isEmpty() {
1197                 return true;
1198         }
1199
1200         function getArgument( $name ) {
1201                 return false;
1202         }
1203
1204         /**
1205          * Returns true if the infinite loop check is OK, false if a loop is detected
1206          */
1207         function loopCheck( $title ) {
1208                 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1209         }
1210
1211         /**
1212          * Return true if the frame is a template frame
1213          */
1214         function isTemplate() {
1215                 return false;
1216         }
1217 }
1218
1219 /**
1220  * Expansion frame with template arguments
1221  * @ingroup Parser
1222  */
1223 class PPTemplateFrame_Hash extends PPFrame_Hash {
1224         var $numberedArgs, $namedArgs, $parent;
1225         var $numberedExpansionCache, $namedExpansionCache;
1226
1227         function __construct( $preprocessor, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
1228                 parent::__construct( $preprocessor );
1229
1230                 $this->parent = $parent;
1231                 $this->numberedArgs = $numberedArgs;
1232                 $this->namedArgs = $namedArgs;
1233                 $this->title = $title;
1234                 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1235                 $this->titleCache = $parent->titleCache;
1236                 $this->titleCache[] = $pdbk;
1237                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1238                 if ( $pdbk !== false ) {
1239                         $this->loopCheckHash[$pdbk] = true;
1240                 }
1241                 $this->depth = $parent->depth + 1;
1242                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1243         }
1244
1245         function __toString() {
1246                 $s = 'tplframe{';
1247                 $first = true;
1248                 $args = $this->numberedArgs + $this->namedArgs;
1249                 foreach ( $args as $name => $value ) {
1250                         if ( $first ) {
1251                                 $first = false;
1252                         } else {
1253                                 $s .= ', ';
1254                         }
1255                         $s .= "\"$name\":\"" .
1256                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1257                 }
1258                 $s .= '}';
1259                 return $s;
1260         }
1261         /**
1262          * Returns true if there are no arguments in this frame
1263          */
1264         function isEmpty() {
1265                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1266         }
1267
1268         function getArguments() {
1269                 $arguments = array();
1270                 foreach ( array_merge(
1271                                 array_keys($this->numberedArgs),
1272                                 array_keys($this->namedArgs)) as $key ) {
1273                         $arguments[$key] = $this->getArgument($key);
1274                 }
1275                 return $arguments;
1276         }
1277
1278         function getNumberedArguments() {
1279                 $arguments = array();
1280                 foreach ( array_keys($this->numberedArgs) as $key ) {
1281                         $arguments[$key] = $this->getArgument($key);
1282                 }
1283                 return $arguments;
1284         }
1285
1286         function getNamedArguments() {
1287                 $arguments = array();
1288                 foreach ( array_keys($this->namedArgs) as $key ) {
1289                         $arguments[$key] = $this->getArgument($key);
1290                 }
1291                 return $arguments;
1292         }
1293
1294         function getNumberedArgument( $index ) {
1295                 if ( !isset( $this->numberedArgs[$index] ) ) {
1296                         return false;
1297                 }
1298                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1299                         # No trimming for unnamed arguments
1300                         $this->numberedExpansionCache[$index] = $this->parent->expand( $this->numberedArgs[$index], PPFrame::STRIP_COMMENTS );
1301                 }
1302                 return $this->numberedExpansionCache[$index];
1303         }
1304
1305         function getNamedArgument( $name ) {
1306                 if ( !isset( $this->namedArgs[$name] ) ) {
1307                         return false;
1308                 }
1309                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1310                         # Trim named arguments post-expand, for backwards compatibility
1311                         $this->namedExpansionCache[$name] = trim(
1312                                 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1313                 }
1314                 return $this->namedExpansionCache[$name];
1315         }
1316
1317         function getArgument( $name ) {
1318                 $text = $this->getNumberedArgument( $name );
1319                 if ( $text === false ) {
1320                         $text = $this->getNamedArgument( $name );
1321                 }
1322                 return $text;
1323         }
1324
1325         /**
1326          * Return true if the frame is a template frame
1327          */
1328         function isTemplate() {
1329                 return true;
1330         }
1331 }
1332
1333 /**
1334  * Expansion frame with custom arguments
1335  * @ingroup Parser
1336  */
1337 class PPCustomFrame_Hash extends PPFrame_Hash {
1338         var $args;
1339
1340         function __construct( $preprocessor, $args ) {
1341                 parent::__construct( $preprocessor );
1342                 $this->args = $args;
1343         }
1344
1345         function __toString() {
1346                 $s = 'cstmframe{';
1347                 $first = true;
1348                 foreach ( $this->args as $name => $value ) {
1349                         if ( $first ) {
1350                                 $first = false;
1351                         } else {
1352                                 $s .= ', ';
1353                         }
1354                         $s .= "\"$name\":\"" .
1355                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1356                 }
1357                 $s .= '}';
1358                 return $s;
1359         }
1360
1361         function isEmpty() {
1362                 return !count( $this->args );
1363         }
1364
1365         function getArgument( $index ) {
1366                 if ( !isset( $this->args[$index] ) ) {
1367                         return false;
1368                 }
1369                 return $this->args[$index];
1370         }
1371 }
1372
1373 /**
1374  * @ingroup Parser
1375  */
1376 class PPNode_Hash_Tree implements PPNode {
1377         var $name, $firstChild, $lastChild, $nextSibling;
1378
1379         function __construct( $name ) {
1380                 $this->name = $name;
1381                 $this->firstChild = $this->lastChild = $this->nextSibling = false;
1382         }
1383
1384         function __toString() {
1385                 $inner = '';
1386                 $attribs = '';
1387                 for ( $node = $this->firstChild; $node; $node = $node->nextSibling ) {
1388                         if ( $node instanceof PPNode_Hash_Attr ) {
1389                                 $attribs .= ' ' . $node->name . '="' . htmlspecialchars( $node->value ) . '"';
1390                         } else {
1391                                 $inner .= $node->__toString();
1392                         }
1393                 }
1394                 if ( $inner === '' ) {
1395                         return "<{$this->name}$attribs/>";
1396                 } else {
1397                         return "<{$this->name}$attribs>$inner</{$this->name}>";
1398                 }
1399         }
1400
1401         static function newWithText( $name, $text ) {
1402                 $obj = new self( $name );
1403                 $obj->addChild( new PPNode_Hash_Text( $text ) );
1404                 return $obj;
1405         }
1406
1407         function addChild( $node ) {
1408                 if ( $this->lastChild === false ) {
1409                         $this->firstChild = $this->lastChild = $node;
1410                 } else {
1411                         $this->lastChild->nextSibling = $node;
1412                         $this->lastChild = $node;
1413                 }
1414         }
1415
1416         function getChildren() {
1417                 $children = array();
1418                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1419                         $children[] = $child;
1420                 }
1421                 return new PPNode_Hash_Array( $children );
1422         }
1423
1424         function getFirstChild() {
1425                 return $this->firstChild;
1426         }
1427
1428         function getNextSibling() {
1429                 return $this->nextSibling;
1430         }
1431
1432         function getChildrenOfType( $name ) {
1433                 $children = array();
1434                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1435                         if ( isset( $child->name ) && $child->name === $name ) {
1436                                 $children[] = $name;
1437                         }
1438                 }
1439                 return $children;
1440         }
1441
1442         function getLength() { return false; }
1443         function item( $i ) { return false; }
1444
1445         function getName() {
1446                 return $this->name;
1447         }
1448
1449         /**
1450          * Split a <part> node into an associative array containing:
1451          *    name          PPNode name
1452          *    index         String index
1453          *    value         PPNode value
1454          */
1455         function splitArg() {
1456                 $bits = array();
1457                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1458                         if ( !isset( $child->name ) ) {
1459                                 continue;
1460                         }
1461                         if ( $child->name === 'name' ) {
1462                                 $bits['name'] = $child;
1463                                 if ( $child->firstChild instanceof PPNode_Hash_Attr
1464                                         && $child->firstChild->name === 'index' )
1465                                 {
1466                                         $bits['index'] = $child->firstChild->value;
1467                                 }
1468                         } elseif ( $child->name === 'value' ) {
1469                                 $bits['value'] = $child;
1470                         }
1471                 }
1472
1473                 if ( !isset( $bits['name'] ) ) {
1474                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1475                 }
1476                 if ( !isset( $bits['index'] ) ) {
1477                         $bits['index'] = '';
1478                 }
1479                 return $bits;
1480         }
1481
1482         /**
1483          * Split an <ext> node into an associative array containing name, attr, inner and close
1484          * All values in the resulting array are PPNodes. Inner and close are optional.
1485          */
1486         function splitExt() {
1487                 $bits = array();
1488                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1489                         if ( !isset( $child->name ) ) {
1490                                 continue;
1491                         }
1492                         if ( $child->name == 'name' ) {
1493                                 $bits['name'] = $child;
1494                         } elseif ( $child->name == 'attr' ) {
1495                                 $bits['attr'] = $child;
1496                         } elseif ( $child->name == 'inner' ) {
1497                                 $bits['inner'] = $child;
1498                         } elseif ( $child->name == 'close' ) {
1499                                 $bits['close'] = $child;
1500                         }
1501                 }
1502                 if ( !isset( $bits['name'] ) ) {
1503                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1504                 }
1505                 return $bits;
1506         }
1507
1508         /**
1509          * Split an <h> node
1510          */
1511         function splitHeading() {
1512                 if ( $this->name !== 'h' ) {
1513                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1514                 }
1515                 $bits = array();
1516                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1517                         if ( !isset( $child->name ) ) {
1518                                 continue;
1519                         }
1520                         if ( $child->name == 'i' ) {
1521                                 $bits['i'] = $child->value;
1522                         } elseif ( $child->name == 'level' ) {
1523                                 $bits['level'] = $child->value;
1524                         }
1525                 }
1526                 if ( !isset( $bits['i'] ) ) {
1527                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1528                 }
1529                 return $bits;
1530         }
1531
1532         /**
1533          * Split a <template> or <tplarg> node
1534          */
1535         function splitTemplate() {
1536                 $parts = array();
1537                 $bits = array( 'lineStart' => '' );
1538                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1539                         if ( !isset( $child->name ) ) {
1540                                 continue;
1541                         }
1542                         if ( $child->name == 'title' ) {
1543                                 $bits['title'] = $child;
1544                         }
1545                         if ( $child->name == 'part' ) {
1546                                 $parts[] = $child;
1547                         }
1548                         if ( $child->name == 'lineStart' ) {
1549                                 $bits['lineStart'] = '1';
1550                         }
1551                 }
1552                 if ( !isset( $bits['title'] ) ) {
1553                         throw new MWException( 'Invalid node passed to ' . __METHOD__ );
1554                 }
1555                 $bits['parts'] = new PPNode_Hash_Array( $parts );
1556                 return $bits;
1557         }
1558 }
1559
1560 /**
1561  * @ingroup Parser
1562  */
1563 class PPNode_Hash_Text implements PPNode {
1564         var $value, $nextSibling;
1565
1566         function __construct( $value ) {
1567                 if ( is_object( $value ) ) {
1568                         throw new MWException( __CLASS__ . ' given object instead of string' );
1569                 }
1570                 $this->value = $value;
1571         }
1572
1573         function __toString() {
1574                 return htmlspecialchars( $this->value );
1575         }
1576
1577         function getNextSibling() {
1578                 return $this->nextSibling;
1579         }
1580
1581         function getChildren() { return false; }
1582         function getFirstChild() { return false; }
1583         function getChildrenOfType( $name ) { return false; }
1584         function getLength() { return false; }
1585         function item( $i ) { return false; }
1586         function getName() { return '#text'; }
1587         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1588         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1589         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1590 }
1591
1592 /**
1593  * @ingroup Parser
1594  */
1595 class PPNode_Hash_Array implements PPNode {
1596         var $value, $nextSibling;
1597
1598         function __construct( $value ) {
1599                 $this->value = $value;
1600         }
1601
1602         function __toString() {
1603                 return var_export( $this, true );
1604         }
1605
1606         function getLength() {
1607                 return count( $this->value );
1608         }
1609
1610         function item( $i ) {
1611                 return $this->value[$i];
1612         }
1613
1614         function getName() { return '#nodelist'; }
1615
1616         function getNextSibling() {
1617                 return $this->nextSibling;
1618         }
1619
1620         function getChildren() { return false; }
1621         function getFirstChild() { return false; }
1622         function getChildrenOfType( $name ) { return false; }
1623         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1624         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1625         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1626 }
1627
1628 /**
1629  * @ingroup Parser
1630  */
1631 class PPNode_Hash_Attr implements PPNode {
1632         var $name, $value, $nextSibling;
1633
1634         function __construct( $name, $value ) {
1635                 $this->name = $name;
1636                 $this->value = $value;
1637         }
1638
1639         function __toString() {
1640                 return "<@{$this->name}>" . htmlspecialchars( $this->value ) . "</@{$this->name}>";
1641         }
1642
1643         function getName() {
1644                 return $this->name;
1645         }
1646
1647         function getNextSibling() {
1648                 return $this->nextSibling;
1649         }
1650
1651         function getChildren() { return false; }
1652         function getFirstChild() { return false; }
1653         function getChildrenOfType( $name ) { return false; }
1654         function getLength() { return false; }
1655         function item( $i ) { return false; }
1656         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1657         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1658         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1659 }