includes/parser/Preprocessor_Hash.php

   1 <?php
   2 /**
   3  * Preprocessor using PHP arrays
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Parser
  22  */
  23
  24 /**
  25  * Differences from DOM schema:
  26  *   * attribute nodes are children
  27  *   * "<h>" nodes that aren't at the top are replaced with <possible-h>
  28  * @ingroup Parser
  29  */
  30 class Preprocessor_Hash implements Preprocessor {
  31         /**
  32          * @var Parser
  33          */
  34         var $parser;
  35
  36         const CACHE_VERSION = 1;
  37
  38         function __construct( $parser ) {
  39                 $this->parser = $parser;
  40         }
  41
  42         /**
  43          * @return PPFrame_Hash
  44          */
  45         function newFrame() {
  46                 return new PPFrame_Hash( $this );
  47         }
  48
  49         /**
  50          * @param $args array
  51          * @return PPCustomFrame_Hash
  52          */
  53         function newCustomFrame( $args ) {
  54                 return new PPCustomFrame_Hash( $this, $args );
  55         }
  56
  57         /**
  58          * @param $values array
  59          * @return PPNode_Hash_Array
  60          */
  61         function newPartNodeArray( $values ) {
  62                 $list = array();
  63
  64                 foreach ( $values as $k => $val ) {
  65                         $partNode = new PPNode_Hash_Tree( 'part' );
  66                         $nameNode = new PPNode_Hash_Tree( 'name' );
  67
  68                         if ( is_int( $k ) ) {
  69                                 $nameNode->addChild( new PPNode_Hash_Attr( 'index', $k ) );
  70                                 $partNode->addChild( $nameNode );
  71                         } else {
  72                                 $nameNode->addChild( new PPNode_Hash_Text( $k ) );
  73                                 $partNode->addChild( $nameNode );
  74                                 $partNode->addChild( new PPNode_Hash_Text( '=' ) );
  75                         }
  76
  77                         $valueNode = new PPNode_Hash_Tree( 'value' );
  78                         $valueNode->addChild( new PPNode_Hash_Text( $val ) );
  79                         $partNode->addChild( $valueNode );
  80
  81                         $list[] = $partNode;
  82                 }
  83
  84                 $node = new PPNode_Hash_Array( $list );
  85                 return $node;
  86         }
  87
  88         /**
  89          * Preprocess some wikitext and return the document tree.
  90          * This is the ghost of Parser::replace_variables().
  91          *
  92          * @param $text String: the text to parse
  93          * @param $flags Integer: bitwise combination of:
  94          *          Parser::PTD_FOR_INCLUSION    Handle "<noinclude>" and "<includeonly>" as if the text is being
  95          *                                     included. Default is to assume a direct page view.
  96          *
  97          * The generated DOM tree must depend only on the input text and the flags.
  98          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
  99          *
 100          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
 101          * change in the DOM tree for a given text, must be passed through the section identifier
 102          * in the section edit link and thus back to extractSections().
 103          *
 104          * The output of this function is currently only cached in process memory, but a persistent
 105          * cache may be implemented at a later date which takes further advantage of these strict
 106          * dependency requirements.
 107          *
 108          * @throws MWException
 109          * @return PPNode_Hash_Tree
 110          */
 111         function preprocessToObj( $text, $flags = 0 ) {
 112                 wfProfileIn( __METHOD__ );
 113
 114                 // Check cache.
 115                 global $wgMemc, $wgPreprocessorCacheThreshold;
 116
 117                 $cacheable = $wgPreprocessorCacheThreshold !== false && strlen( $text ) > $wgPreprocessorCacheThreshold;
 118                 if ( $cacheable ) {
 119                         wfProfileIn( __METHOD__ . '-cacheable' );
 120
 121                         $cacheKey = wfMemcKey( 'preprocess-hash', md5( $text ), $flags );
 122                         $cacheValue = $wgMemc->get( $cacheKey );
 123                         if ( $cacheValue ) {
 124                                 $version = substr( $cacheValue, 0, 8 );
 125                                 if ( intval( $version ) == self::CACHE_VERSION ) {
 126                                         $hash = unserialize( substr( $cacheValue, 8 ) );
 127                                         // From the cache
 128                                         wfDebugLog( "Preprocessor",
 129                                                 "Loaded preprocessor hash from memcached (key $cacheKey)" );
 130                                         wfProfileOut( __METHOD__ . '-cacheable' );
 131                                         wfProfileOut( __METHOD__ );
 132                                         return $hash;
 133                                 }
 134                         }
 135                         wfProfileIn( __METHOD__ . '-cache-miss' );
 136                 }
 137
 138                 $rules = array(
 139                         '{' => array(
 140                                 'end' => '}',
 141                                 'names' => array(
 142                                         2 => 'template',
 143                                         3 => 'tplarg',
 144                                 ),
 145                                 'min' => 2,
 146                                 'max' => 3,
 147                         ),
 148                         '[' => array(
 149                                 'end' => ']',
 150                                 'names' => array( 2 => null ),
 151                                 'min' => 2,
 152                                 'max' => 2,
 153                         )
 154                 );
 155
 156                 $forInclusion = $flags & Parser::PTD_FOR_INCLUSION;
 157
 158                 $xmlishElements = $this->parser->getStripList();
 159                 $enableOnlyinclude = false;
 160                 if ( $forInclusion ) {
 161                         $ignoredTags = array( 'includeonly', '/includeonly' );
 162                         $ignoredElements = array( 'noinclude' );
 163                         $xmlishElements[] = 'noinclude';
 164                         if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
 165                                 $enableOnlyinclude = true;
 166                         }
 167                 } else {
 168                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
 169                         $ignoredElements = array( 'includeonly' );
 170                         $xmlishElements[] = 'includeonly';
 171                 }
 172                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
 173
 174                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
 175                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
 176
 177                 $stack = new PPDStack_Hash;
 178
 179                 $searchBase = "[{<\n";
 180                 $revText = strrev( $text ); // For fast reverse searches
 181                 $lengthText = strlen( $text );
 182
 183                 $i = 0;                     # Input pointer, starts out pointing to a pseudo-newline before the start
 184                 $accum =& $stack->getAccum();   # Current accumulator
 185                 $findEquals = false;            # True to find equals signs in arguments
 186                 $findPipe = false;              # True to take notice of pipe characters
 187                 $headingIndex = 1;
 188                 $inHeading = false;        # True if $i is inside a possible heading
 189                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
 190                 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
 191                 $fakeLineStart = true;     # Do a line-start run without outputting an LF character
 192
 193                 while ( true ) {
 194                         //$this->memCheck();
 195
 196                         if ( $findOnlyinclude ) {
 197                                 // Ignore all input up to the next <onlyinclude>
 198                                 $startPos = strpos( $text, '<onlyinclude>', $i );
 199                                 if ( $startPos === false ) {
 200                                         // Ignored section runs to the end
 201                                         $accum->addNodeWithText( 'ignore', substr( $text, $i ) );
 202                                         break;
 203                                 }
 204                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
 205                                 $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i ) );
 206                                 $i = $tagEndPos;
 207                                 $findOnlyinclude = false;
 208                         }
 209
 210                         if ( $fakeLineStart ) {
 211                                 $found = 'line-start';
 212                                 $curChar = '';
 213                         } else {
 214                                 # Find next opening brace, closing brace or pipe
 215                                 $search = $searchBase;
 216                                 if ( $stack->top === false ) {
 217                                         $currentClosing = '';
 218                                 } else {
 219                                         $currentClosing = $stack->top->close;
 220                                         $search .= $currentClosing;
 221                                 }
 222                                 if ( $findPipe ) {
 223                                         $search .= '|';
 224                                 }
 225                                 if ( $findEquals ) {
 226                                         // First equals will be for the template
 227                                         $search .= '=';
 228                                 }
 229                                 $rule = null;
 230                                 # Output literal section, advance input counter
 231                                 $literalLength = strcspn( $text, $search, $i );
 232                                 if ( $literalLength > 0 ) {
 233                                         $accum->addLiteral( substr( $text, $i, $literalLength ) );
 234                                         $i += $literalLength;
 235                                 }
 236                                 if ( $i >= $lengthText ) {
 237                                         if ( $currentClosing == "\n" ) {
 238                                                 // Do a past-the-end run to finish off the heading
 239                                                 $curChar = '';
 240                                                 $found = 'line-end';
 241                                         } else {
 242                                                 # All done
 243                                                 break;
 244                                         }
 245                                 } else {
 246                                         $curChar = $text[$i];
 247                                         if ( $curChar == '|' ) {
 248                                                 $found = 'pipe';
 249                                         } elseif ( $curChar == '=' ) {
 250                                                 $found = 'equals';
 251                                         } elseif ( $curChar == '<' ) {
 252                                                 $found = 'angle';
 253                                         } elseif ( $curChar == "\n" ) {
 254                                                 if ( $inHeading ) {
 255                                                         $found = 'line-end';
 256                                                 } else {
 257                                                         $found = 'line-start';
 258                                                 }
 259                                         } elseif ( $curChar == $currentClosing ) {
 260                                                 $found = 'close';
 261                                         } elseif ( isset( $rules[$curChar] ) ) {
 262                                                 $found = 'open';
 263                                                 $rule = $rules[$curChar];
 264                                         } else {
 265                                                 # Some versions of PHP have a strcspn which stops on null characters
 266                                                 # Ignore and continue
 267                                                 ++$i;
 268                                                 continue;
 269                                         }
 270                                 }
 271                         }
 272
 273                         if ( $found == 'angle' ) {
 274                                 $matches = false;
 275                                 // Handle </onlyinclude>
 276                                 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
 277                                         $findOnlyinclude = true;
 278                                         continue;
 279                                 }
 280
 281                                 // Determine element name
 282                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
 283                                         // Element name missing or not listed
 284                                         $accum->addLiteral( '<' );
 285                                         ++$i;
 286                                         continue;
 287                                 }
 288                                 // Handle comments
 289                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
 290                                         // To avoid leaving blank lines, when a comment is both preceded
 291                                         // and followed by a newline (ignoring spaces), trim leading and
 292                                         // trailing spaces and one of the newlines.
 293
 294                                         // Find the end
 295                                         $endPos = strpos( $text, '-->', $i + 4 );
 296                                         if ( $endPos === false ) {
 297                                                 // Unclosed comment in input, runs to end
 298                                                 $inner = substr( $text, $i );
 299                                                 $accum->addNodeWithText( 'comment', $inner );
 300                                                 $i = $lengthText;
 301                                         } else {
 302                                                 // Search backwards for leading whitespace
 303                                                 $wsStart = $i ? ( $i - strspn( $revText, ' ', $lengthText - $i ) ) : 0;
 304                                                 // Search forwards for trailing whitespace
 305                                                 // $wsEnd will be the position of the last space (or the '>' if there's none)
 306                                                 $wsEnd = $endPos + 2 + strspn( $text, ' ', $endPos + 3 );
 307                                                 // Eat the line if possible
 308                                                 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
 309                                                 // the overall start. That's not how Sanitizer::removeHTMLcomments() did it, but
 310                                                 // it's a possible beneficial b/c break.
 311                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
 312                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n" )
 313                                                 {
 314                                                         $startPos = $wsStart;
 315                                                         $endPos = $wsEnd + 1;
 316                                                         // Remove leading whitespace from the end of the accumulator
 317                                                         // Sanity check first though
 318                                                         $wsLength = $i - $wsStart;
 319                                                         if ( $wsLength > 0
 320                                                                 && $accum->lastNode instanceof PPNode_Hash_Text
 321                                                                 && substr( $accum->lastNode->value, -$wsLength ) === str_repeat( ' ', $wsLength ) )
 322                                                         {
 323                                                                 $accum->lastNode->value = substr( $accum->lastNode->value, 0, -$wsLength );
 324                                                         }
 325                                                         // Do a line-start run next time to look for headings after the comment
 326                                                         $fakeLineStart = true;
 327                                                 } else {
 328                                                         // No line to eat, just take the comment itself
 329                                                         $startPos = $i;
 330                                                         $endPos += 2;
 331                                                 }
 332
 333                                                 if ( $stack->top ) {
 334                                                         $part = $stack->top->getCurrentPart();
 335                                                         if ( !(isset( $part->commentEnd ) && $part->commentEnd == $wsStart - 1 )) {
 336                                                                 $part->visualEnd = $wsStart;
 337                                                         }
 338                                                         // Else comments abutting, no change in visual end
 339                                                         $part->commentEnd = $endPos;
 340                                                 }
 341                                                 $i = $endPos + 1;
 342                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
 343                                                 $accum->addNodeWithText( 'comment', $inner );
 344                                         }
 345                                         continue;
 346                                 }
 347                                 $name = $matches[1];
 348                                 $lowerName = strtolower( $name );
 349                                 $attrStart = $i + strlen( $name ) + 1;
 350
 351                                 // Find end of tag
 352                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
 353                                 if ( $tagEndPos === false ) {
 354                                         // Infinite backtrack
 355                                         // Disable tag search to prevent worst-case O(N^2) performance
 356                                         $noMoreGT = true;
 357                                         $accum->addLiteral( '<' );
 358                                         ++$i;
 359                                         continue;
 360                                 }
 361
 362                                 // Handle ignored tags
 363                                 if ( in_array( $lowerName, $ignoredTags ) ) {
 364                                         $accum->addNodeWithText( 'ignore', substr( $text, $i, $tagEndPos - $i + 1 ) );
 365                                         $i = $tagEndPos + 1;
 366                                         continue;
 367                                 }
 368
 369                                 $tagStartPos = $i;
 370                                 if ( $text[$tagEndPos-1] == '/' ) {
 371                                         // Short end tag
 372                                         $attrEnd = $tagEndPos - 1;
 373                                         $inner = null;
 374                                         $i = $tagEndPos + 1;
 375                                         $close = null;
 376                                 } else {
 377                                         $attrEnd = $tagEndPos;
 378                                         // Find closing tag
 379                                         if ( preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
 380                                                         $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) )
 381                                         {
 382                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
 383                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
 384                                                 $close = $matches[0][0];
 385                                         } else {
 386                                                 // No end tag -- let it run out to the end of the text.
 387                                                 $inner = substr( $text, $tagEndPos + 1 );
 388                                                 $i = $lengthText;
 389                                                 $close = null;
 390                                         }
 391                                 }
 392                                 // <includeonly> and <noinclude> just become <ignore> tags
 393                                 if ( in_array( $lowerName, $ignoredElements ) ) {
 394                                         $accum->addNodeWithText( 'ignore', substr( $text, $tagStartPos, $i - $tagStartPos ) );
 395                                         continue;
 396                                 }
 397
 398                                 if ( $attrEnd <= $attrStart ) {
 399                                         $attr = '';
 400                                 } else {
 401                                         // Note that the attr element contains the whitespace between name and attribute,
 402                                         // this is necessary for precise reconstruction during pre-save transform.
 403                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
 404                                 }
 405
 406                                 $extNode = new PPNode_Hash_Tree( 'ext' );
 407                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'name', $name ) );
 408                                 $extNode->addChild( PPNode_Hash_Tree::newWithText( 'attr', $attr ) );
 409                                 if ( $inner !== null ) {
 410                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'inner', $inner ) );
 411                                 }
 412                                 if ( $close !== null ) {
 413                                         $extNode->addChild( PPNode_Hash_Tree::newWithText( 'close', $close ) );
 414                                 }
 415                                 $accum->addNode( $extNode );
 416                         }
 417
 418                         elseif ( $found == 'line-start' ) {
 419                                 // Is this the start of a heading?
 420                                 // Line break belongs before the heading element in any case
 421                                 if ( $fakeLineStart ) {
 422                                         $fakeLineStart = false;
 423                                 } else {
 424                                         $accum->addLiteral( $curChar );
 425                                         $i++;
 426                                 }
 427
 428                                 $count = strspn( $text, '=', $i, 6 );
 429                                 if ( $count == 1 && $findEquals ) {
 430                                         // DWIM: This looks kind of like a name/value separator
 431                                         // Let's let the equals handler have it and break the potential heading
 432                                         // This is heuristic, but AFAICT the methods for completely correct disambiguation are very complex.
 433                                 } elseif ( $count > 0 ) {
 434                                         $piece = array(
 435                                                 'open' => "\n",
 436                                                 'close' => "\n",
 437                                                 'parts' => array( new PPDPart_Hash( str_repeat( '=', $count ) ) ),
 438                                                 'startPos' => $i,
 439                                                 'count' => $count );
 440                                         $stack->push( $piece );
 441                                         $accum =& $stack->getAccum();
 442                                         extract( $stack->getFlags() );
 443                                         $i += $count;
 444                                 }
 445                         } elseif ( $found == 'line-end' ) {
 446                                 $piece = $stack->top;
 447                                 // A heading must be open, otherwise \n wouldn't have been in the search list
 448                                 assert( '$piece->open == "\n"' );
 449                                 $part = $piece->getCurrentPart();
 450                                 // Search back through the input to see if it has a proper close
 451                                 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
 452                                 $wsLength = strspn( $revText, " \t", $lengthText - $i );
 453                                 $searchStart = $i - $wsLength;
 454                                 if ( isset( $part->commentEnd ) && $searchStart - 1 == $part->commentEnd ) {
 455                                         // Comment found at line end
 456                                         // Search for equals signs before the comment
 457                                         $searchStart = $part->visualEnd;
 458                                         $searchStart -= strspn( $revText, " \t", $lengthText - $searchStart );
 459                                 }
 460                                 $count = $piece->count;
 461                                 $equalsLength = strspn( $revText, '=', $lengthText - $searchStart );
 462                                 if ( $equalsLength > 0 ) {
 463                                         if ( $searchStart - $equalsLength == $piece->startPos ) {
 464                                                 // This is just a single string of equals signs on its own line
 465                                                 // Replicate the doHeadings behaviour /={count}(.+)={count}/
 466                                                 // First find out how many equals signs there really are (don't stop at 6)
 467                                                 $count = $equalsLength;
 468                                                 if ( $count < 3 ) {
 469                                                         $count = 0;
 470                                                 } else {
 471                                                         $count = min( 6, intval( ( $count - 1 ) / 2 ) );
 472                                                 }
 473                                         } else {
 474                                                 $count = min( $equalsLength, $count );
 475                                         }
 476                                         if ( $count > 0 ) {
 477                                                 // Normal match, output <h>
 478                                                 $element = new PPNode_Hash_Tree( 'possible-h' );
 479                                                 $element->addChild( new PPNode_Hash_Attr( 'level', $count ) );
 480                                                 $element->addChild( new PPNode_Hash_Attr( 'i', $headingIndex++ ) );
 481                                                 $element->lastChild->nextSibling = $accum->firstNode;
 482                                                 $element->lastChild = $accum->lastNode;
 483                                         } else {
 484                                                 // Single equals sign on its own line, count=0
 485                                                 $element = $accum;
 486                                         }
 487                                 } else {
 488                                         // No match, no <h>, just pass down the inner text
 489                                         $element = $accum;
 490                                 }
 491                                 // Unwind the stack
 492                                 $stack->pop();
 493                                 $accum =& $stack->getAccum();
 494                                 extract( $stack->getFlags() );
 495
 496                                 // Append the result to the enclosing accumulator
 497                                 if ( $element instanceof PPNode ) {
 498                                         $accum->addNode( $element );
 499                                 } else {
 500                                         $accum->addAccum( $element );
 501                                 }
 502                                 // Note that we do NOT increment the input pointer.
 503                                 // This is because the closing linebreak could be the opening linebreak of
 504                                 // another heading. Infinite loops are avoided because the next iteration MUST
 505                                 // hit the heading open case above, which unconditionally increments the
 506                                 // input pointer.
 507                         } elseif ( $found == 'open' ) {
 508                                 # count opening brace characters
 509                                 $count = strspn( $text, $curChar, $i );
 510
 511                                 # we need to add to stack only if opening brace count is enough for one of the rules
 512                                 if ( $count >= $rule['min'] ) {
 513                                         # Add it to the stack
 514                                         $piece = array(
 515                                                 'open' => $curChar,
 516                                                 'close' => $rule['end'],
 517                                                 'count' => $count,
 518                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
 519                                         );
 520
 521                                         $stack->push( $piece );
 522                                         $accum =& $stack->getAccum();
 523                                         extract( $stack->getFlags() );
 524                                 } else {
 525                                         # Add literal brace(s)
 526                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 527                                 }
 528                                 $i += $count;
 529                         } elseif ( $found == 'close' ) {
 530                                 $piece = $stack->top;
 531                                 # lets check if there are enough characters for closing brace
 532                                 $maxCount = $piece->count;
 533                                 $count = strspn( $text, $curChar, $i, $maxCount );
 534
 535                                 # check for maximum matching characters (if there are 5 closing
 536                                 # characters, we will probably need only 3 - depending on the rules)
 537                                 $rule = $rules[$piece->open];
 538                                 if ( $count > $rule['max'] ) {
 539                                         # The specified maximum exists in the callback array, unless the caller
 540                                         # has made an error
 541                                         $matchingCount = $rule['max'];
 542                                 } else {
 543                                         # Count is less than the maximum
 544                                         # Skip any gaps in the callback array to find the true largest match
 545                                         # Need to use array_key_exists not isset because the callback can be null
 546                                         $matchingCount = $count;
 547                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
 548                                                 --$matchingCount;
 549                                         }
 550                                 }
 551
 552                                 if ( $matchingCount <= 0 ) {
 553                                         # No matching element found in callback array
 554                                         # Output a literal closing brace and continue
 555                                         $accum->addLiteral( str_repeat( $curChar, $count ) );
 556                                         $i += $count;
 557                                         continue;
 558                                 }
 559                                 $name = $rule['names'][$matchingCount];
 560                                 if ( $name === null ) {
 561                                         // No element, just literal text
 562                                         $element = $piece->breakSyntax( $matchingCount );
 563                                         $element->addLiteral( str_repeat( $rule['end'], $matchingCount ) );
 564                                 } else {
 565                                         # Create XML element
 566                                         # Note: $parts is already XML, does not need to be encoded further
 567                                         $parts = $piece->parts;
 568                                         $titleAccum = $parts[0]->out;
 569                                         unset( $parts[0] );
 570
 571                                         $element = new PPNode_Hash_Tree( $name );
 572
 573                                         # The invocation is at the start of the line if lineStart is set in
 574                                         # the stack, and all opening brackets are used up.
 575                                         if ( $maxCount == $matchingCount && !empty( $piece->lineStart ) ) {
 576                                                 $element->addChild( new PPNode_Hash_Attr( 'lineStart', 1 ) );
 577                                         }
 578                                         $titleNode = new PPNode_Hash_Tree( 'title' );
 579                                         $titleNode->firstChild = $titleAccum->firstNode;
 580                                         $titleNode->lastChild = $titleAccum->lastNode;
 581                                         $element->addChild( $titleNode );
 582                                         $argIndex = 1;
 583                                         foreach ( $parts as $part ) {
 584                                                 if ( isset( $part->eqpos ) ) {
 585                                                         // Find equals
 586                                                         $lastNode = false;
 587                                                         for ( $node = $part->out->firstNode; $node; $node = $node->nextSibling ) {
 588                                                                 if ( $node === $part->eqpos ) {
 589                                                                         break;
 590                                                                 }
 591                                                                 $lastNode = $node;
 592                                                         }
 593                                                         if ( !$node ) {
 594                                                                 throw new MWException( __METHOD__ . ': eqpos not found' );
 595                                                         }
 596                                                         if ( $node->name !== 'equals' ) {
 597                                                                 throw new MWException( __METHOD__ . ': eqpos is not equals' );
 598                                                         }
 599                                                         $equalsNode = $node;
 600
 601                                                         // Construct name node
 602                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 603                                                         if ( $lastNode !== false ) {
 604                                                                 $lastNode->nextSibling = false;
 605                                                                 $nameNode->firstChild = $part->out->firstNode;
 606                                                                 $nameNode->lastChild = $lastNode;
 607                                                         }
 608
 609                                                         // Construct value node
 610                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 611                                                         if ( $equalsNode->nextSibling !== false ) {
 612                                                                 $valueNode->firstChild = $equalsNode->nextSibling;
 613                                                                 $valueNode->lastChild = $part->out->lastNode;
 614                                                         }
 615                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 616                                                         $partNode->addChild( $nameNode );
 617                                                         $partNode->addChild( $equalsNode->firstChild );
 618                                                         $partNode->addChild( $valueNode );
 619                                                         $element->addChild( $partNode );
 620                                                 } else {
 621                                                         $partNode = new PPNode_Hash_Tree( 'part' );
 622                                                         $nameNode = new PPNode_Hash_Tree( 'name' );
 623                                                         $nameNode->addChild( new PPNode_Hash_Attr( 'index', $argIndex++ ) );
 624                                                         $valueNode = new PPNode_Hash_Tree( 'value' );
 625                                                         $valueNode->firstChild = $part->out->firstNode;
 626                                                         $valueNode->lastChild = $part->out->lastNode;
 627                                                         $partNode->addChild( $nameNode );
 628                                                         $partNode->addChild( $valueNode );
 629                                                         $element->addChild( $partNode );
 630                                                 }
 631                                         }
 632                                 }
 633
 634                                 # Advance input pointer
 635                                 $i += $matchingCount;
 636
 637                                 # Unwind the stack
 638                                 $stack->pop();
 639                                 $accum =& $stack->getAccum();
 640
 641                                 # Re-add the old stack element if it still has unmatched opening characters remaining
 642                                 if ( $matchingCount < $piece->count ) {
 643                                         $piece->parts = array( new PPDPart_Hash );
 644                                         $piece->count -= $matchingCount;
 645                                         # do we still qualify for any callback with remaining count?
 646                                         $names = $rules[$piece->open]['names'];
 647                                         $skippedBraces = 0;
 648                                         $enclosingAccum =& $accum;
 649                                         while ( $piece->count ) {
 650                                                 if ( array_key_exists( $piece->count, $names ) ) {
 651                                                         $stack->push( $piece );
 652                                                         $accum =& $stack->getAccum();
 653                                                         break;
 654                                                 }
 655                                                 --$piece->count;
 656                                                 $skippedBraces ++;
 657                                         }
 658                                         $enclosingAccum->addLiteral( str_repeat( $piece->open, $skippedBraces ) );
 659                                 }
 660
 661                                 extract( $stack->getFlags() );
 662
 663                                 # Add XML element to the enclosing accumulator
 664                                 if ( $element instanceof PPNode ) {
 665                                         $accum->addNode( $element );
 666                                 } else {
 667                                         $accum->addAccum( $element );
 668                                 }
 669                         } elseif ( $found == 'pipe' ) {
 670                                 $findEquals = true; // shortcut for getFlags()
 671                                 $stack->addPart();
 672                                 $accum =& $stack->getAccum();
 673                                 ++$i;
 674                         } elseif ( $found == 'equals' ) {
 675                                 $findEquals = false; // shortcut for getFlags()
 676                                 $accum->addNodeWithText( 'equals', '=' );
 677                                 $stack->getCurrentPart()->eqpos = $accum->lastNode;
 678                                 ++$i;
 679                         }
 680                 }
 681
 682                 # Output any remaining unclosed brackets
 683                 foreach ( $stack->stack as $piece ) {
 684                         $stack->rootAccum->addAccum( $piece->breakSyntax() );
 685                 }
 686
 687                 # Enable top-level headings
 688                 for ( $node = $stack->rootAccum->firstNode; $node; $node = $node->nextSibling ) {
 689                         if ( isset( $node->name ) && $node->name === 'possible-h' ) {
 690                                 $node->name = 'h';
 691                         }
 692                 }
 693
 694                 $rootNode = new PPNode_Hash_Tree( 'root' );
 695                 $rootNode->firstChild = $stack->rootAccum->firstNode;
 696                 $rootNode->lastChild = $stack->rootAccum->lastNode;
 697
 698                 // Cache
 699                 if ( $cacheable ) {
 700                         $cacheValue = sprintf( "%08d", self::CACHE_VERSION ) . serialize( $rootNode );
 701                         $wgMemc->set( $cacheKey, $cacheValue, 86400 );
 702                         wfProfileOut( __METHOD__ . '-cache-miss' );
 703                         wfProfileOut( __METHOD__ . '-cacheable' );
 704                         wfDebugLog( "Preprocessor", "Saved preprocessor Hash to memcached (key $cacheKey)" );
 705                 }
 706
 707                 wfProfileOut( __METHOD__ );
 708                 return $rootNode;
 709         }
 710 }
 711
 712 /**
 713  * Stack class to help Preprocessor::preprocessToObj()
 714  * @ingroup Parser
 715  */
 716 class PPDStack_Hash extends PPDStack {
 717         function __construct() {
 718                 $this->elementClass = 'PPDStackElement_Hash';
 719                 parent::__construct();
 720                 $this->rootAccum = new PPDAccum_Hash;
 721         }
 722 }
 723
 724 /**
 725  * @ingroup Parser
 726  */
 727 class PPDStackElement_Hash extends PPDStackElement {
 728         function __construct( $data = array() ) {
 729                 $this->partClass = 'PPDPart_Hash';
 730                 parent::__construct( $data );
 731         }
 732
 733         /**
 734          * Get the accumulator that would result if the close is not found.
 735          *
 736          * @return PPDAccum_Hash
 737          */
 738         function breakSyntax( $openingCount = false ) {
 739                 if ( $this->open == "\n" ) {
 740                         $accum = $this->parts[0]->out;
 741                 } else {
 742                         if ( $openingCount === false ) {
 743                                 $openingCount = $this->count;
 744                         }
 745                         $accum = new PPDAccum_Hash;
 746                         $accum->addLiteral( str_repeat( $this->open, $openingCount ) );
 747                         $first = true;
 748                         foreach ( $this->parts as $part ) {
 749                                 if ( $first ) {
 750                                         $first = false;
 751                                 } else {
 752                                         $accum->addLiteral( '|' );
 753                                 }
 754                                 $accum->addAccum( $part->out );
 755                         }
 756                 }
 757                 return $accum;
 758         }
 759 }
 760
 761 /**
 762  * @ingroup Parser
 763  */
 764 class PPDPart_Hash extends PPDPart {
 765         function __construct( $out = '' ) {
 766                 $accum = new PPDAccum_Hash;
 767                 if ( $out !== '' ) {
 768                         $accum->addLiteral( $out );
 769                 }
 770                 parent::__construct( $accum );
 771         }
 772 }
 773
 774 /**
 775  * @ingroup Parser
 776  */
 777 class PPDAccum_Hash {
 778         var $firstNode, $lastNode;
 779
 780         function __construct() {
 781                 $this->firstNode = $this->lastNode = false;
 782         }
 783
 784         /**
 785          * Append a string literal
 786          */
 787         function addLiteral( $s ) {
 788                 if ( $this->lastNode === false ) {
 789                         $this->firstNode = $this->lastNode = new PPNode_Hash_Text( $s );
 790                 } elseif ( $this->lastNode instanceof PPNode_Hash_Text ) {
 791                         $this->lastNode->value .= $s;
 792                 } else {
 793                         $this->lastNode->nextSibling = new PPNode_Hash_Text( $s );
 794                         $this->lastNode = $this->lastNode->nextSibling;
 795                 }
 796         }
 797
 798         /**
 799          * Append a PPNode
 800          */
 801         function addNode( PPNode $node ) {
 802                 if ( $this->lastNode === false ) {
 803                         $this->firstNode = $this->lastNode = $node;
 804                 } else {
 805                         $this->lastNode->nextSibling = $node;
 806                         $this->lastNode = $node;
 807                 }
 808         }
 809
 810         /**
 811          * Append a tree node with text contents
 812          */
 813         function addNodeWithText( $name, $value ) {
 814                 $node = PPNode_Hash_Tree::newWithText( $name, $value );
 815                 $this->addNode( $node );
 816         }
 817
 818         /**
 819          * Append a PPAccum_Hash
 820          * Takes over ownership of the nodes in the source argument. These nodes may
 821          * subsequently be modified, especially nextSibling.
 822          */
 823         function addAccum( $accum ) {
 824                 if ( $accum->lastNode === false ) {
 825                         // nothing to add
 826                 } elseif ( $this->lastNode === false ) {
 827                         $this->firstNode = $accum->firstNode;
 828                         $this->lastNode = $accum->lastNode;
 829                 } else {
 830                         $this->lastNode->nextSibling = $accum->firstNode;
 831                         $this->lastNode = $accum->lastNode;
 832                 }
 833         }
 834 }
 835
 836 /**
 837  * An expansion frame, used as a context to expand the result of preprocessToObj()
 838  * @ingroup Parser
 839  */
 840 class PPFrame_Hash implements PPFrame {
 841
 842         /**
 843          * @var Parser
 844          */
 845         var $parser;
 846
 847         /**
 848          * @var Preprocessor
 849          */
 850         var $preprocessor;
 851
 852         /**
 853          * @var Title
 854          */
 855         var $title;
 856         var $titleCache;
 857
 858         /**
 859          * Hashtable listing templates which are disallowed for expansion in this frame,
 860          * having been encountered previously in parent frames.
 861          */
 862         var $loopCheckHash;
 863
 864         /**
 865          * Recursion depth of this frame, top = 0
 866          * Note that this is NOT the same as expansion depth in expand()
 867          */
 868         var $depth;
 869
 870
 871         /**
 872          * Construct a new preprocessor frame.
 873          * @param $preprocessor Preprocessor: the parent preprocessor
 874          */
 875         function __construct( $preprocessor ) {
 876                 $this->preprocessor = $preprocessor;
 877                 $this->parser = $preprocessor->parser;
 878                 $this->title = $this->parser->mTitle;
 879                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
 880                 $this->loopCheckHash = array();
 881                 $this->depth = 0;
 882         }
 883
 884         /**
 885          * Create a new child frame
 886          * $args is optionally a multi-root PPNode or array containing the template arguments
 887          *
 888          * @param array|bool|\PPNode_Hash_Array $args PPNode_Hash_Array|array
 889          * @param $title Title|bool
 890          *
 891          * @param int $indexOffset
 892          * @throws MWException
 893          * @return PPTemplateFrame_Hash
 894          */
 895         function newChild( $args = false, $title = false, $indexOffset = 0 ) {
 896                 $namedArgs = array();
 897                 $numberedArgs = array();
 898                 if ( $title === false ) {
 899                         $title = $this->title;
 900                 }
 901                 if ( $args !== false ) {
 902                         if ( $args instanceof PPNode_Hash_Array ) {
 903                                 $args = $args->value;
 904                         } elseif ( !is_array( $args ) ) {
 905                                 throw new MWException( __METHOD__ . ': $args must be array or PPNode_Hash_Array' );
 906                         }
 907                         foreach ( $args as $arg ) {
 908                                 $bits = $arg->splitArg();
 909                                 if ( $bits['index'] !== '' ) {
 910                                         // Numbered parameter
 911                                         $index = $bits['index'] - $indexOffset;
 912                                         $numberedArgs[$index] = $bits['value'];
 913                                         unset( $namedArgs[$index] );
 914                                 } else {
 915                                         // Named parameter
 916                                         $name = trim( $this->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
 917                                         $namedArgs[$name] = $bits['value'];
 918                                         unset( $numberedArgs[$name] );
 919                                 }
 920                         }
 921                 }
 922                 return new PPTemplateFrame_Hash( $this->preprocessor, $this, $numberedArgs, $namedArgs, $title );
 923         }
 924
 925         /**
 926          * @throws MWException
 927          * @param $root
 928          * @param $flags int
 929          * @return string
 930          */
 931         function expand( $root, $flags = 0 ) {
 932                 static $expansionDepth = 0;
 933                 if ( is_string( $root ) ) {
 934                         return $root;
 935                 }
 936
 937                 if ( ++$this->parser->mPPNodeCount > $this->parser->mOptions->getMaxPPNodeCount() ) {
 938                         $this->parser->limitationWarn( 'node-count-exceeded',
 939                                         $this->parser->mPPNodeCount,
 940                                         $this->parser->mOptions->getMaxPPNodeCount()
 941                         );
 942                         return '<span class="error">Node-count limit exceeded</span>';
 943                 }
 944                 if ( $expansionDepth > $this->parser->mOptions->getMaxPPExpandDepth() ) {
 945                         $this->parser->limitationWarn( 'expansion-depth-exceeded',
 946                                         $expansionDepth,
 947                                         $this->parser->mOptions->getMaxPPExpandDepth()
 948                         );
 949                         return '<span class="error">Expansion depth limit exceeded</span>';
 950                 }
 951                 ++$expansionDepth;
 952                 if ( $expansionDepth > $this->parser->mHighestExpansionDepth ) {
 953                         $this->parser->mHighestExpansionDepth = $expansionDepth;
 954                 }
 955
 956                 $outStack = array( '', '' );
 957                 $iteratorStack = array( false, $root );
 958                 $indexStack = array( 0, 0 );
 959
 960                 while ( count( $iteratorStack ) > 1 ) {
 961                         $level = count( $outStack ) - 1;
 962                         $iteratorNode =& $iteratorStack[ $level ];
 963                         $out =& $outStack[$level];
 964                         $index =& $indexStack[$level];
 965
 966                         if ( is_array( $iteratorNode ) ) {
 967                                 if ( $index >= count( $iteratorNode ) ) {
 968                                         // All done with this iterator
 969                                         $iteratorStack[$level] = false;
 970                                         $contextNode = false;
 971                                 } else {
 972                                         $contextNode = $iteratorNode[$index];
 973                                         $index++;
 974                                 }
 975                         } elseif ( $iteratorNode instanceof PPNode_Hash_Array ) {
 976                                 if ( $index >= $iteratorNode->getLength() ) {
 977                                         // All done with this iterator
 978                                         $iteratorStack[$level] = false;
 979                                         $contextNode = false;
 980                                 } else {
 981                                         $contextNode = $iteratorNode->item( $index );
 982                                         $index++;
 983                                 }
 984                         } else {
 985                                 // Copy to $contextNode and then delete from iterator stack,
 986                                 // because this is not an iterator but we do have to execute it once
 987                                 $contextNode = $iteratorStack[$level];
 988                                 $iteratorStack[$level] = false;
 989                         }
 990
 991                         $newIterator = false;
 992
 993                         if ( $contextNode === false ) {
 994                                 // nothing to do
 995                         } elseif ( is_string( $contextNode ) ) {
 996                                 $out .= $contextNode;
 997                         } elseif ( is_array( $contextNode ) || $contextNode instanceof PPNode_Hash_Array ) {
 998                                 $newIterator = $contextNode;
 999                         } elseif ( $contextNode instanceof PPNode_Hash_Attr ) {
1000                                 // No output
1001                         } elseif ( $contextNode instanceof PPNode_Hash_Text ) {
1002                                 $out .= $contextNode->value;
1003                         } elseif ( $contextNode instanceof PPNode_Hash_Tree ) {
1004                                 if ( $contextNode->name == 'template' ) {
1005                                         # Double-brace expansion
1006                                         $bits = $contextNode->splitTemplate();
1007                                         if ( $flags & PPFrame::NO_TEMPLATES ) {
1008                                                 $newIterator = $this->virtualBracketedImplode( '{{', '|', '}}', $bits['title'], $bits['parts'] );
1009                                         } else {
1010                                                 $ret = $this->parser->braceSubstitution( $bits, $this );
1011                                                 if ( isset( $ret['object'] ) ) {
1012                                                         $newIterator = $ret['object'];
1013                                                 } else {
1014                                                         $out .= $ret['text'];
1015                                                 }
1016                                         }
1017                                 } elseif ( $contextNode->name == 'tplarg' ) {
1018                                         # Triple-brace expansion
1019                                         $bits = $contextNode->splitTemplate();
1020                                         if ( $flags & PPFrame::NO_ARGS ) {
1021                                                 $newIterator = $this->virtualBracketedImplode( '{{{', '|', '}}}', $bits['title'], $bits['parts'] );
1022                                         } else {
1023                                                 $ret = $this->parser->argSubstitution( $bits, $this );
1024                                                 if ( isset( $ret['object'] ) ) {
1025                                                         $newIterator = $ret['object'];
1026                                                 } else {
1027                                                         $out .= $ret['text'];
1028                                                 }
1029                                         }
1030                                 } elseif ( $contextNode->name == 'comment' ) {
1031                                         # HTML-style comment
1032                                         # Remove it in HTML, pre+remove and STRIP_COMMENTS modes
1033                                         if ( $this->parser->ot['html']
1034                                                 || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
1035                                                 || ( $flags & PPFrame::STRIP_COMMENTS ) )
1036                                         {
1037                                                 $out .= '';
1038                                         }
1039                                         # Add a strip marker in PST mode so that pstPass2() can run some old-fashioned regexes on the result
1040                                         # Not in RECOVER_COMMENTS mode (extractSections) though
1041                                         elseif ( $this->parser->ot['wiki'] && !( $flags & PPFrame::RECOVER_COMMENTS ) ) {
1042                                                 $out .= $this->parser->insertStripItem( $contextNode->firstChild->value );
1043                                         }
1044                                         # Recover the literal comment in RECOVER_COMMENTS and pre+no-remove
1045                                         else {
1046                                                 $out .= $contextNode->firstChild->value;
1047                                         }
1048                                 } elseif ( $contextNode->name == 'ignore' ) {
1049                                         # Output suppression used by <includeonly> etc.
1050                                         # OT_WIKI will only respect <ignore> in substed templates.
1051                                         # The other output types respect it unless NO_IGNORE is set.
1052                                         # extractSections() sets NO_IGNORE and so never respects it.
1053                                         if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & PPFrame::NO_IGNORE ) ) {
1054                                                 $out .= $contextNode->firstChild->value;
1055                                         } else {
1056                                                 //$out .= '';
1057                                         }
1058                                 } elseif ( $contextNode->name == 'ext' ) {
1059                                         # Extension tag
1060                                         $bits = $contextNode->splitExt() + array( 'attr' => null, 'inner' => null, 'close' => null );
1061                                         $out .= $this->parser->extensionSubstitution( $bits, $this );
1062                                 } elseif ( $contextNode->name == 'h' ) {
1063                                         # Heading
1064                                         if ( $this->parser->ot['html'] ) {
1065                                                 # Expand immediately and insert heading index marker
1066                                                 $s = '';
1067                                                 for ( $node = $contextNode->firstChild; $node; $node = $node->nextSibling ) {
1068                                                         $s .= $this->expand( $node, $flags );
1069                                                 }
1070
1071                                                 $bits = $contextNode->splitHeading();
1072                                                 $titleText = $this->title->getPrefixedDBkey();
1073                                                 $this->parser->mHeadings[] = array( $titleText, $bits['i'] );
1074                                                 $serial = count( $this->parser->mHeadings ) - 1;
1075                                                 $marker = "{$this->parser->mUniqPrefix}-h-$serial-" . Parser::MARKER_SUFFIX;
1076                                                 $s = substr( $s, 0, $bits['level'] ) . $marker . substr( $s, $bits['level'] );
1077                                                 $this->parser->mStripState->addGeneral( $marker, '' );
1078                                                 $out .= $s;
1079                                         } else {
1080                                                 # Expand in virtual stack
1081                                                 $newIterator = $contextNode->getChildren();
1082                                         }
1083                                 } else {
1084                                         # Generic recursive expansion
1085                                         $newIterator = $contextNode->getChildren();
1086                                 }
1087                         } else {
1088                                 throw new MWException( __METHOD__ . ': Invalid parameter type' );
1089                         }
1090
1091                         if ( $newIterator !== false ) {
1092                                 $outStack[] = '';
1093                                 $iteratorStack[] = $newIterator;
1094                                 $indexStack[] = 0;
1095                         } elseif ( $iteratorStack[$level] === false ) {
1096                                 // Return accumulated value to parent
1097                                 // With tail recursion
1098                                 while ( $iteratorStack[$level] === false && $level > 0 ) {
1099                                         $outStack[$level - 1] .= $out;
1100                                         array_pop( $outStack );
1101                                         array_pop( $iteratorStack );
1102                                         array_pop( $indexStack );
1103                                         $level--;
1104                                 }
1105                         }
1106                 }
1107                 --$expansionDepth;
1108                 return $outStack[0];
1109         }
1110
1111         /**
1112          * @param $sep
1113          * @param $flags
1114          * @return string
1115          */
1116         function implodeWithFlags( $sep, $flags /*, ... */ ) {
1117                 $args = array_slice( func_get_args(), 2 );
1118
1119                 $first = true;
1120                 $s = '';
1121                 foreach ( $args as $root ) {
1122                         if ( $root instanceof PPNode_Hash_Array ) {
1123                                 $root = $root->value;
1124                         }
1125                         if ( !is_array( $root ) ) {
1126                                 $root = array( $root );
1127                         }
1128                         foreach ( $root as $node ) {
1129                                 if ( $first ) {
1130                                         $first = false;
1131                                 } else {
1132                                         $s .= $sep;
1133                                 }
1134                                 $s .= $this->expand( $node, $flags );
1135                         }
1136                 }
1137                 return $s;
1138         }
1139
1140         /**
1141          * Implode with no flags specified
1142          * This previously called implodeWithFlags but has now been inlined to reduce stack depth
1143          * @return string
1144          */
1145         function implode( $sep /*, ... */ ) {
1146                 $args = array_slice( func_get_args(), 1 );
1147
1148                 $first = true;
1149                 $s = '';
1150                 foreach ( $args as $root ) {
1151                         if ( $root instanceof PPNode_Hash_Array ) {
1152                                 $root = $root->value;
1153                         }
1154                         if ( !is_array( $root ) ) {
1155                                 $root = array( $root );
1156                         }
1157                         foreach ( $root as $node ) {
1158                                 if ( $first ) {
1159                                         $first = false;
1160                                 } else {
1161                                         $s .= $sep;
1162                                 }
1163                                 $s .= $this->expand( $node );
1164                         }
1165                 }
1166                 return $s;
1167         }
1168
1169         /**
1170          * Makes an object that, when expand()ed, will be the same as one obtained
1171          * with implode()
1172          *
1173          * @return PPNode_Hash_Array
1174          */
1175         function virtualImplode( $sep /*, ... */ ) {
1176                 $args = array_slice( func_get_args(), 1 );
1177                 $out = array();
1178                 $first = true;
1179
1180                 foreach ( $args as $root ) {
1181                         if ( $root instanceof PPNode_Hash_Array ) {
1182                                 $root = $root->value;
1183                         }
1184                         if ( !is_array( $root ) ) {
1185                                 $root = array( $root );
1186                         }
1187                         foreach ( $root as $node ) {
1188                                 if ( $first ) {
1189                                         $first = false;
1190                                 } else {
1191                                         $out[] = $sep;
1192                                 }
1193                                 $out[] = $node;
1194                         }
1195                 }
1196                 return new PPNode_Hash_Array( $out );
1197         }
1198
1199         /**
1200          * Virtual implode with brackets
1201          *
1202          * @return PPNode_Hash_Array
1203          */
1204         function virtualBracketedImplode( $start, $sep, $end /*, ... */ ) {
1205                 $args = array_slice( func_get_args(), 3 );
1206                 $out = array( $start );
1207                 $first = true;
1208
1209                 foreach ( $args as $root ) {
1210                         if ( $root instanceof PPNode_Hash_Array ) {
1211                                 $root = $root->value;
1212                         }
1213                         if ( !is_array( $root ) ) {
1214                                 $root = array( $root );
1215                         }
1216                         foreach ( $root as $node ) {
1217                                 if ( $first ) {
1218                                         $first = false;
1219                                 } else {
1220                                         $out[] = $sep;
1221                                 }
1222                                 $out[] = $node;
1223                         }
1224                 }
1225                 $out[] = $end;
1226                 return new PPNode_Hash_Array( $out );
1227         }
1228
1229         function __toString() {
1230                 return 'frame{}';
1231         }
1232
1233         /**
1234          * @param $level bool
1235          * @return array|bool|String
1236          */
1237         function getPDBK( $level = false ) {
1238                 if ( $level === false ) {
1239                         return $this->title->getPrefixedDBkey();
1240                 } else {
1241                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
1242                 }
1243         }
1244
1245         /**
1246          * @return array
1247          */
1248         function getArguments() {
1249                 return array();
1250         }
1251
1252         /**
1253          * @return array
1254          */
1255         function getNumberedArguments() {
1256                 return array();
1257         }
1258
1259         /**
1260          * @return array
1261          */
1262         function getNamedArguments() {
1263                 return array();
1264         }
1265
1266         /**
1267          * Returns true if there are no arguments in this frame
1268          *
1269          * @return bool
1270          */
1271         function isEmpty() {
1272                 return true;
1273         }
1274
1275         /**
1276          * @param $name
1277          * @return bool
1278          */
1279         function getArgument( $name ) {
1280                 return false;
1281         }
1282
1283         /**
1284          * Returns true if the infinite loop check is OK, false if a loop is detected
1285          *
1286          * @param $title Title
1287          *
1288          * @return bool
1289          */
1290         function loopCheck( $title ) {
1291                 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
1292         }
1293
1294         /**
1295          * Return true if the frame is a template frame
1296          *
1297          * @return bool
1298          */
1299         function isTemplate() {
1300                 return false;
1301         }
1302
1303         /**
1304          * Get a title of frame
1305          *
1306          * @return Title
1307          */
1308         function getTitle() {
1309                 return $this->title;
1310         }
1311 }
1312
1313 /**
1314  * Expansion frame with template arguments
1315  * @ingroup Parser
1316  */
1317 class PPTemplateFrame_Hash extends PPFrame_Hash {
1318         var $numberedArgs, $namedArgs, $parent;
1319         var $numberedExpansionCache, $namedExpansionCache;
1320
1321         /**
1322          * @param $preprocessor
1323          * @param $parent
1324          * @param $numberedArgs array
1325          * @param $namedArgs array
1326          * @param $title Title
1327          */
1328         function __construct( $preprocessor, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
1329                 parent::__construct( $preprocessor );
1330
1331                 $this->parent = $parent;
1332                 $this->numberedArgs = $numberedArgs;
1333                 $this->namedArgs = $namedArgs;
1334                 $this->title = $title;
1335                 $pdbk = $title ? $title->getPrefixedDBkey() : false;
1336                 $this->titleCache = $parent->titleCache;
1337                 $this->titleCache[] = $pdbk;
1338                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
1339                 if ( $pdbk !== false ) {
1340                         $this->loopCheckHash[$pdbk] = true;
1341                 }
1342                 $this->depth = $parent->depth + 1;
1343                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
1344         }
1345
1346         function __toString() {
1347                 $s = 'tplframe{';
1348                 $first = true;
1349                 $args = $this->numberedArgs + $this->namedArgs;
1350                 foreach ( $args as $name => $value ) {
1351                         if ( $first ) {
1352                                 $first = false;
1353                         } else {
1354                                 $s .= ', ';
1355                         }
1356                         $s .= "\"$name\":\"" .
1357                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1358                 }
1359                 $s .= '}';
1360                 return $s;
1361         }
1362         /**
1363          * Returns true if there are no arguments in this frame
1364          *
1365          * @return bool
1366          */
1367         function isEmpty() {
1368                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
1369         }
1370
1371         /**
1372          * @return array
1373          */
1374         function getArguments() {
1375                 $arguments = array();
1376                 foreach ( array_merge(
1377                                 array_keys( $this->numberedArgs ),
1378                                 array_keys( $this->namedArgs ) ) as $key ) {
1379                         $arguments[$key] = $this->getArgument( $key );
1380                 }
1381                 return $arguments;
1382         }
1383
1384         /**
1385          * @return array
1386          */
1387         function getNumberedArguments() {
1388                 $arguments = array();
1389                 foreach ( array_keys( $this->numberedArgs ) as $key ) {
1390                         $arguments[$key] = $this->getArgument( $key );
1391                 }
1392                 return $arguments;
1393         }
1394
1395         /**
1396          * @return array
1397          */
1398         function getNamedArguments() {
1399                 $arguments = array();
1400                 foreach ( array_keys( $this->namedArgs ) as $key ) {
1401                         $arguments[$key] = $this->getArgument( $key );
1402                 }
1403                 return $arguments;
1404         }
1405
1406         /**
1407          * @param $index
1408          * @return array|bool
1409          */
1410         function getNumberedArgument( $index ) {
1411                 if ( !isset( $this->numberedArgs[$index] ) ) {
1412                         return false;
1413                 }
1414                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
1415                         # No trimming for unnamed arguments
1416                         $this->numberedExpansionCache[$index] = $this->parent->expand( $this->numberedArgs[$index], PPFrame::STRIP_COMMENTS );
1417                 }
1418                 return $this->numberedExpansionCache[$index];
1419         }
1420
1421         /**
1422          * @param $name
1423          * @return bool
1424          */
1425         function getNamedArgument( $name ) {
1426                 if ( !isset( $this->namedArgs[$name] ) ) {
1427                         return false;
1428                 }
1429                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
1430                         # Trim named arguments post-expand, for backwards compatibility
1431                         $this->namedExpansionCache[$name] = trim(
1432                                 $this->parent->expand( $this->namedArgs[$name], PPFrame::STRIP_COMMENTS ) );
1433                 }
1434                 return $this->namedExpansionCache[$name];
1435         }
1436
1437         /**
1438          * @param $name
1439          * @return array|bool
1440          */
1441         function getArgument( $name ) {
1442                 $text = $this->getNumberedArgument( $name );
1443                 if ( $text === false ) {
1444                         $text = $this->getNamedArgument( $name );
1445                 }
1446                 return $text;
1447         }
1448
1449         /**
1450          * Return true if the frame is a template frame
1451          *
1452          * @return bool
1453          */
1454         function isTemplate() {
1455                 return true;
1456         }
1457 }
1458
1459 /**
1460  * Expansion frame with custom arguments
1461  * @ingroup Parser
1462  */
1463 class PPCustomFrame_Hash extends PPFrame_Hash {
1464         var $args;
1465
1466         function __construct( $preprocessor, $args ) {
1467                 parent::__construct( $preprocessor );
1468                 $this->args = $args;
1469         }
1470
1471         function __toString() {
1472                 $s = 'cstmframe{';
1473                 $first = true;
1474                 foreach ( $this->args as $name => $value ) {
1475                         if ( $first ) {
1476                                 $first = false;
1477                         } else {
1478                                 $s .= ', ';
1479                         }
1480                         $s .= "\"$name\":\"" .
1481                                 str_replace( '"', '\\"', $value->__toString() ) . '"';
1482                 }
1483                 $s .= '}';
1484                 return $s;
1485         }
1486
1487         /**
1488          * @return bool
1489          */
1490         function isEmpty() {
1491                 return !count( $this->args );
1492         }
1493
1494         /**
1495          * @param $index
1496          * @return bool
1497          */
1498         function getArgument( $index ) {
1499                 if ( !isset( $this->args[$index] ) ) {
1500                         return false;
1501                 }
1502                 return $this->args[$index];
1503         }
1504
1505         function getArguments() {
1506                 return $this->args;
1507         }
1508 }
1509
1510 /**
1511  * @ingroup Parser
1512  */
1513 class PPNode_Hash_Tree implements PPNode {
1514         var $name, $firstChild, $lastChild, $nextSibling;
1515
1516         function __construct( $name ) {
1517                 $this->name = $name;
1518                 $this->firstChild = $this->lastChild = $this->nextSibling = false;
1519         }
1520
1521         function __toString() {
1522                 $inner = '';
1523                 $attribs = '';
1524                 for ( $node = $this->firstChild; $node; $node = $node->nextSibling ) {
1525                         if ( $node instanceof PPNode_Hash_Attr ) {
1526                                 $attribs .= ' ' . $node->name . '="' . htmlspecialchars( $node->value ) . '"';
1527                         } else {
1528                                 $inner .= $node->__toString();
1529                         }
1530                 }
1531                 if ( $inner === '' ) {
1532                         return "<{$this->name}$attribs/>";
1533                 } else {
1534                         return "<{$this->name}$attribs>$inner</{$this->name}>";
1535                 }
1536         }
1537
1538         /**
1539          * @param $name
1540          * @param $text
1541          * @return PPNode_Hash_Tree
1542          */
1543         static function newWithText( $name, $text ) {
1544                 $obj = new self( $name );
1545                 $obj->addChild( new PPNode_Hash_Text( $text ) );
1546                 return $obj;
1547         }
1548
1549         function addChild( $node ) {
1550                 if ( $this->lastChild === false ) {
1551                         $this->firstChild = $this->lastChild = $node;
1552                 } else {
1553                         $this->lastChild->nextSibling = $node;
1554                         $this->lastChild = $node;
1555                 }
1556         }
1557
1558         /**
1559          * @return PPNode_Hash_Array
1560          */
1561         function getChildren() {
1562                 $children = array();
1563                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1564                         $children[] = $child;
1565                 }
1566                 return new PPNode_Hash_Array( $children );
1567         }
1568
1569         function getFirstChild() {
1570                 return $this->firstChild;
1571         }
1572
1573         function getNextSibling() {
1574                 return $this->nextSibling;
1575         }
1576
1577         function getChildrenOfType( $name ) {
1578                 $children = array();
1579                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1580                         if ( isset( $child->name ) && $child->name === $name ) {
1581                                 $children[] = $child;
1582                         }
1583                 }
1584                 return $children;
1585         }
1586
1587         /**
1588          * @return bool
1589          */
1590         function getLength() {
1591                 return false;
1592         }
1593
1594         /**
1595          * @param  $i
1596          * @return bool
1597          */
1598         function item( $i ) {
1599                 return false;
1600         }
1601
1602         /**
1603          * @return string
1604          */
1605         function getName() {
1606                 return $this->name;
1607         }
1608
1609         /**
1610          * Split a "<part>" node into an associative array containing:
1611          *  - name          PPNode name
1612          *  - index         String index
1613          *  - value         PPNode value
1614          *
1615          * @throws MWException
1616          * @return array
1617          */
1618         function splitArg() {
1619                 $bits = array();
1620                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1621                         if ( !isset( $child->name ) ) {
1622                                 continue;
1623                         }
1624                         if ( $child->name === 'name' ) {
1625                                 $bits['name'] = $child;
1626                                 if ( $child->firstChild instanceof PPNode_Hash_Attr
1627                                         && $child->firstChild->name === 'index' )
1628                                 {
1629                                         $bits['index'] = $child->firstChild->value;
1630                                 }
1631                         } elseif ( $child->name === 'value' ) {
1632                                 $bits['value'] = $child;
1633                         }
1634                 }
1635
1636                 if ( !isset( $bits['name'] ) ) {
1637                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
1638                 }
1639                 if ( !isset( $bits['index'] ) ) {
1640                         $bits['index'] = '';
1641                 }
1642                 return $bits;
1643         }
1644
1645         /**
1646          * Split an "<ext>" node into an associative array containing name, attr, inner and close
1647          * All values in the resulting array are PPNodes. Inner and close are optional.
1648          *
1649          * @throws MWException
1650          * @return array
1651          */
1652         function splitExt() {
1653                 $bits = array();
1654                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1655                         if ( !isset( $child->name ) ) {
1656                                 continue;
1657                         }
1658                         if ( $child->name == 'name' ) {
1659                                 $bits['name'] = $child;
1660                         } elseif ( $child->name == 'attr' ) {
1661                                 $bits['attr'] = $child;
1662                         } elseif ( $child->name == 'inner' ) {
1663                                 $bits['inner'] = $child;
1664                         } elseif ( $child->name == 'close' ) {
1665                                 $bits['close'] = $child;
1666                         }
1667                 }
1668                 if ( !isset( $bits['name'] ) ) {
1669                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
1670                 }
1671                 return $bits;
1672         }
1673
1674         /**
1675          * Split an "<h>" node
1676          *
1677          * @throws MWException
1678          * @return array
1679          */
1680         function splitHeading() {
1681                 if ( $this->name !== 'h' ) {
1682                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1683                 }
1684                 $bits = array();
1685                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1686                         if ( !isset( $child->name ) ) {
1687                                 continue;
1688                         }
1689                         if ( $child->name == 'i' ) {
1690                                 $bits['i'] = $child->value;
1691                         } elseif ( $child->name == 'level' ) {
1692                                 $bits['level'] = $child->value;
1693                         }
1694                 }
1695                 if ( !isset( $bits['i'] ) ) {
1696                         throw new MWException( 'Invalid h node passed to ' . __METHOD__ );
1697                 }
1698                 return $bits;
1699         }
1700
1701         /**
1702          * Split a "<template>" or "<tplarg>" node
1703          *
1704          * @throws MWException
1705          * @return array
1706          */
1707         function splitTemplate() {
1708                 $parts = array();
1709                 $bits = array( 'lineStart' => '' );
1710                 for ( $child = $this->firstChild; $child; $child = $child->nextSibling ) {
1711                         if ( !isset( $child->name ) ) {
1712                                 continue;
1713                         }
1714                         if ( $child->name == 'title' ) {
1715                                 $bits['title'] = $child;
1716                         }
1717                         if ( $child->name == 'part' ) {
1718                                 $parts[] = $child;
1719                         }
1720                         if ( $child->name == 'lineStart' ) {
1721                                 $bits['lineStart'] = '1';
1722                         }
1723                 }
1724                 if ( !isset( $bits['title'] ) ) {
1725                         throw new MWException( 'Invalid node passed to ' . __METHOD__ );
1726                 }
1727                 $bits['parts'] = new PPNode_Hash_Array( $parts );
1728                 return $bits;
1729         }
1730 }
1731
1732 /**
1733  * @ingroup Parser
1734  */
1735 class PPNode_Hash_Text implements PPNode {
1736         var $value, $nextSibling;
1737
1738         function __construct( $value ) {
1739                 if ( is_object( $value ) ) {
1740                         throw new MWException( __CLASS__ . ' given object instead of string' );
1741                 }
1742                 $this->value = $value;
1743         }
1744
1745         function __toString() {
1746                 return htmlspecialchars( $this->value );
1747         }
1748
1749         function getNextSibling() {
1750                 return $this->nextSibling;
1751         }
1752
1753         function getChildren() { return false; }
1754         function getFirstChild() { return false; }
1755         function getChildrenOfType( $name ) { return false; }
1756         function getLength() { return false; }
1757         function item( $i ) { return false; }
1758         function getName() { return '#text'; }
1759         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1760         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1761         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1762 }
1763
1764 /**
1765  * @ingroup Parser
1766  */
1767 class PPNode_Hash_Array implements PPNode {
1768         var $value, $nextSibling;
1769
1770         function __construct( $value ) {
1771                 $this->value = $value;
1772         }
1773
1774         function __toString() {
1775                 return var_export( $this, true );
1776         }
1777
1778         function getLength() {
1779                 return count( $this->value );
1780         }
1781
1782         function item( $i ) {
1783                 return $this->value[$i];
1784         }
1785
1786         function getName() { return '#nodelist'; }
1787
1788         function getNextSibling() {
1789                 return $this->nextSibling;
1790         }
1791
1792         function getChildren() { return false; }
1793         function getFirstChild() { return false; }
1794         function getChildrenOfType( $name ) { return false; }
1795         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1796         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1797         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1798 }
1799
1800 /**
1801  * @ingroup Parser
1802  */
1803 class PPNode_Hash_Attr implements PPNode {
1804         var $name, $value, $nextSibling;
1805
1806         function __construct( $name, $value ) {
1807                 $this->name = $name;
1808                 $this->value = $value;
1809         }
1810
1811         function __toString() {
1812                 return "<@{$this->name}>" . htmlspecialchars( $this->value ) . "</@{$this->name}>";
1813         }
1814
1815         function getName() {
1816                 return $this->name;
1817         }
1818
1819         function getNextSibling() {
1820                 return $this->nextSibling;
1821         }
1822
1823         function getChildren() { return false; }
1824         function getFirstChild() { return false; }
1825         function getChildrenOfType( $name ) { return false; }
1826         function getLength() { return false; }
1827         function item( $i ) { return false; }
1828         function splitArg() { throw new MWException( __METHOD__ . ': not supported' ); }
1829         function splitExt() { throw new MWException( __METHOD__ . ': not supported' ); }
1830         function splitHeading() { throw new MWException( __METHOD__ . ': not supported' ); }
1831 }