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