maintenance/parserTests.inc

   1 <?php
   2 # Copyright (C) 2004 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   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 # 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 /**
  21  * @todo Make this more independent of the configuration (and if possible the database)
  22  * @todo document
  23  * @package MediaWiki
  24  * @subpackage Maintenance
  25  */
  26
  27 /** */
  28 $options = array( 'quick', 'color', 'quiet', 'help' );
  29 $optionsWithArgs = array( 'regex' );
  30
  31 require_once( 'commandLine.inc' );
  32 require_once( "$IP/includes/ObjectCache.php" );
  33 require_once( "$IP/includes/BagOStuff.php" );
  34 require_once( "$IP/languages/LanguageUtf8.php" );
  35 require_once( "$IP/includes/Hooks.php" );
  36 require_once( "$IP/maintenance/parserTestsParserHook.php" );
  37 require_once( "$IP/maintenance/parserTestsParserTime.php" );
  38
  39 /**
  40  * @package MediaWiki
  41  * @subpackage Maintenance
  42  */
  43 class ParserTest {
  44         /**
  45          * boolean $color whereas output should be colorized
  46          * @access private
  47          */
  48         var $color;
  49
  50         /**
  51          * boolean $lightcolor whereas output should use light colors
  52          * @access private
  53          */
  54         var $lightcolor;
  55
  56         /**
  57          * Sets terminal colorization and diff/quick modes depending on OS and
  58          * command-line options (--color and --quick).
  59          *
  60          * @access public
  61          */
  62         function ParserTest() {
  63                 global $options;
  64
  65                 # Only colorize output if stdout is a terminal.
  66                 $this->lightcolor = false;
  67                 $this->color = !wfIsWindows() && posix_isatty(1);
  68
  69                 if( isset( $options['color'] ) ) {
  70                         switch( $options['color'] ) {
  71                         case 'no':
  72                                 $this->color = false;
  73                                 break;
  74                         case 'light':
  75                                 $this->lightcolor = true;
  76                                 # Fall through
  77                         case 'yes':
  78                         default:
  79                                 $this->color = true;
  80                                 break;
  81                         }
  82                 }
  83
  84                 $this->showDiffs = !isset( $options['quick'] );
  85
  86                 $this->quiet = isset( $options['quiet'] );
  87
  88                 if (isset($options['regex'])) {
  89                         $this->regex = $options['regex'];
  90                 } else {
  91                         # Matches anything
  92                         $this->regex = '';
  93                 }
  94         }
  95
  96         /**
  97          * Remove last character if it is a newline
  98          * @access private
  99          */
 100         function chomp($s) {
 101                 if (substr($s, -1) === "\n") {
 102                         return substr($s, 0, -1);
 103                 }
 104                 else {
 105                         return $s;
 106                 }
 107         }
 108
 109         /**
 110          * Run a series of tests listed in the given text file.
 111          * Each test consists of a brief description, wikitext input,
 112          * and the expected HTML output.
 113          *
 114          * Prints status updates on stdout and counts up the total
 115          * number and percentage of passed tests.
 116          *
 117          * @param string $filename
 118          * @return bool True if passed all tests, false if any tests failed.
 119          * @access public
 120          */
 121         function runTestsFromFile( $filename ) {
 122                 $infile = fopen( $filename, 'rt' );
 123                 if( !$infile ) {
 124                         die( "Couldn't open parserTests.txt\n" );
 125                 }
 126
 127                 $data = array();
 128                 $section = null;
 129                 $success = 0;
 130                 $total = 0;
 131                 $n = 0;
 132                 while( false !== ($line = fgets( $infile ) ) ) {
 133                         $n++;
 134                         if( preg_match( '/^!!\s*(\w+)/', $line, $matches ) ) {
 135                                 $section = strtolower( $matches[1] );
 136                                 if( $section == 'endarticle') {
 137                                         if( !isset( $data['text'] ) ) {
 138                                                 die( "'endarticle' without 'text' at line $n\n" );
 139                                         }
 140                                         if( !isset( $data['article'] ) ) {
 141                                                 die( "'endarticle' without 'article' at line $n\n" );
 142                                         }
 143                                         $this->addArticle($this->chomp($data['article']), $this->chomp($data['text']), $n);
 144                                         $data = array();
 145                                         $section = null;
 146                                         continue;
 147                                 }
 148                                 if( $section == 'end' ) {
 149                                         if( !isset( $data['test'] ) ) {
 150                                                 die( "'end' without 'test' at line $n\n" );
 151                                         }
 152                                         if( !isset( $data['input'] ) ) {
 153                                                 die( "'end' without 'input' at line $n\n" );
 154                                         }
 155                                         if( !isset( $data['result'] ) ) {
 156                                                 die( "'end' without 'result' at line $n\n" );
 157                                         }
 158                                         if( !isset( $data['options'] ) ) {
 159                                                 $data['options'] = '';
 160                                         }
 161                                         else {
 162                                                 $data['options'] = $this->chomp( $data['options'] );
 163                                         }
 164                                         if (preg_match('/\\bdisabled\\b/i', $data['options'])
 165                                                 || !preg_match("/{$this->regex}/i", $data['test'])) {
 166                                                 # disabled test
 167                                                 $data = array();
 168                                                 $section = null;
 169                                                 continue;
 170                                         }
 171                                         if( $this->runTest(
 172                                                 $this->chomp( $data['test'] ),
 173                                                 $this->chomp( $data['input'] ),
 174                                                 $this->chomp( $data['result'] ),
 175                                                 $this->chomp( $data['options'] ) ) ) {
 176                                                 $success++;
 177                                         }
 178                                         $total++;
 179                                         $data = array();
 180                                         $section = null;
 181                                         continue;
 182                                 }
 183                                 if ( isset ($data[$section] ) ) {
 184                                         die ( "duplicate section '$section' at line $n\n" );
 185                                 }
 186                                 $data[$section] = '';
 187                                 continue;
 188                         }
 189                         if( $section ) {
 190                                 $data[$section] .= $line;
 191                         }
 192                 }
 193                 if( $total > 0 ) {
 194                         $ratio = wfPercent( 100 * $success / $total );
 195                         print $this->termColor( 1 ) . "\nPassed $success of $total tests ($ratio) ";
 196                         if( $success == $total ) {
 197                                 print $this->termColor( 32 ) . "PASSED!";
 198                         } else {
 199                                 print $this->termColor( 31 ) . "FAILED!";
 200                         }
 201                         print $this->termReset() . "\n";
 202                         return ($success == $total);
 203                 } else {
 204                         die( "No tests found.\n" );
 205                 }
 206         }
 207
 208         /**
 209          * Run a given wikitext input through a freshly-constructed wiki parser,
 210          * and compare the output against the expected results.
 211          * Prints status and explanatory messages to stdout.
 212          *
 213          * @param string $input Wikitext to try rendering
 214          * @param string $result Result to output
 215          * @return bool
 216          */
 217         function runTest( $desc, $input, $result, $opts ) {
 218                 if( !$this->quiet ) {
 219                         $this->showTesting( $desc );
 220                 }
 221
 222                 $this->setupGlobals($opts);
 223
 224                 $user =& new User();
 225                 $options = ParserOptions::newFromUser( $user );
 226
 227                 if (preg_match('/\\bmath\\b/i', $opts)) {
 228                         # XXX this should probably be done by the ParserOptions
 229                         require_once('Math.php');
 230
 231                         $options->setUseTex(true);
 232                 }
 233
 234                 if (preg_match('/title=\[\[(.*)\]\]/', $opts, $m)) {
 235                         $titleText = $m[1];
 236                 }
 237                 else {
 238                         $titleText = 'Parser test';
 239                 }
 240
 241                 $noxml = (bool)preg_match( '~\\b noxml \\b~x', $opts );
 242
 243                 $parser =& new Parser();
 244                 wfRunHooks( 'ParserTestParser', array( &$parser ) );
 245
 246                 $title =& Title::makeTitle( NS_MAIN, $titleText );
 247
 248                 if (preg_match('/\\bpst\\b/i', $opts)) {
 249                         $out = $parser->preSaveTransform( $input, $title, $user, $options );
 250                 } elseif (preg_match('/\\bmsg\\b/i', $opts)) {
 251                         $out = $parser->transformMsg( $input, $options );
 252                 } else {
 253                         $output = $parser->parse( $input, $title, $options, true, true, 1337 );
 254                         $out = $output->getText();
 255
 256                         if (preg_match('/\\bill\\b/i', $opts)) {
 257                                 $out = $this->tidy( implode( ' ', $output->getLanguageLinks() ) );
 258                         } else if (preg_match('/\\bcat\\b/i', $opts)) {
 259                                 $out = $this->tidy ( implode( ' ', $output->getCategoryLinks() ) );
 260                         }
 261
 262                         $result = $this->tidy($result);
 263                 }
 264
 265                 $this->teardownGlobals();
 266
 267                 if( $result === $out && ( $noxml === true || $this->wellFormed( $out ) ) ) {
 268                         return $this->showSuccess( $desc );
 269                 } else {
 270                         return $this->showFailure( $desc, $result, $out );
 271                 }
 272         }
 273
 274         /**
 275          * Set up the global variables for a consistent environment for each test.
 276          * Ideally this should replace the global configuration entirely.
 277          *
 278          * @access private
 279          */
 280         function setupGlobals($opts = '') {
 281                 # Save the prefixed / quoted table names for later use when we make the temporaries.
 282                 $db =& wfGetDB( DB_READ );
 283                 $this->oldTableNames = array();
 284                 foreach( $this->listTables() as $table ) {
 285                         $this->oldTableNames[$table] = $db->tableName( $table );
 286                 }
 287                 if( !isset( $this->uploadDir ) ) {
 288                         $this->uploadDir = $this->setupUploadDir();
 289                 }
 290
 291                 $settings = array(
 292                         'wgServer' => 'http://localhost',
 293                         'wgScript' => '/index.php',
 294                         'wgScriptPath' => '/',
 295                         'wgArticlePath' => '/wiki/$1',
 296                         'wgUploadPath' => 'http://example.com/images',
 297                         'wgUploadDirectory' => $this->uploadDir,
 298                         'wgStyleSheetPath' => '/skins',
 299                         'wgSitename' => 'MediaWiki',
 300                         'wgServerName' => 'Britney Spears',
 301                         'wgLanguageCode' => 'en',
 302                         'wgContLanguageCode' => 'en',
 303                         'wgDBprefix' => 'parsertest',
 304                         'wgDefaultUserOptions' => array(),
 305
 306                         'wgLang' => new LanguageUtf8(),
 307                         'wgContLang' => new LanguageUtf8(),
 308                         'wgNamespacesWithSubpages' => array( 0 => preg_match('/\\bsubpage\\b/i', $opts)),
 309                         'wgMaxTocLevel' => 999,
 310                         'wgCapitalLinks' => true,
 311                         'wgDefaultUserOptions' => array(),
 312                         'wgNoFollowLinks' => true,
 313                         'wgThumbnailScriptPath' => false,
 314                         'wgUseTeX' => false,
 315                         );
 316                 $this->savedGlobals = array();
 317                 foreach( $settings as $var => $val ) {
 318                         $this->savedGlobals[$var] = $GLOBALS[$var];
 319                         $GLOBALS[$var] = $val;
 320                 }
 321                 $GLOBALS['wgLoadBalancer']->loadMasterPos();
 322                 $GLOBALS['wgMessageCache']->initialise( new BagOStuff(), false, 0, $GLOBALS['wgDBname'] );
 323                 $this->setupDatabase();
 324
 325                 global $wgUser;
 326                 $wgUser = new User();
 327         }
 328
 329         # List of temporary tables to create, without prefix
 330         # Some of these probably aren't necessary
 331         function listTables() {
 332                 return array('user', 'page', 'revision', 'text',
 333                         'pagelinks', 'imagelinks', 'categorylinks', 'templatelinks',
 334                         'site_stats', 'hitcounter',
 335                         'ipblocks', 'image', 'oldimage',
 336                         'recentchanges',
 337                         'watchlist', 'math', 'searchindex',
 338                         'interwiki', 'querycache',
 339                         'objectcache'
 340                 );
 341         }
 342
 343         /**
 344          * Set up a temporary set of wiki tables to work with for the tests.
 345          * Currently this will only be done once per run, and any changes to
 346          * the db will be visible to later tests in the run.
 347          *
 348          * @access private
 349          */
 350         function setupDatabase() {
 351                 static $setupDB = false;
 352                 global $wgDBprefix;
 353
 354                 # Make sure we don't mess with the live DB
 355                 if (!$setupDB && $wgDBprefix === 'parsertest') {
 356                         # oh teh horror
 357                         $GLOBALS['wgLoadBalancer'] = LoadBalancer::newFromParams( $GLOBALS['wgDBservers'] );
 358                         $db =& wfGetDB( DB_MASTER );
 359
 360                         $tables = $this->listTables();
 361
 362                         if (!(strcmp($db->getServerVersion(), '4.1') < 0 and stristr($db->getSoftwareLink(), 'MySQL'))) {
 363                                 # Database that supports CREATE TABLE ... LIKE
 364                                 global $wgDBtype;
 365                                 if( $wgDBtype == 'PostgreSQL' ) {
 366                                         $def = 'INCLUDING DEFAULTS';
 367                                 } else {
 368                                         $def = '';
 369                                 }
 370                                 foreach ($tables as $tbl) {
 371                                         $newTableName = $db->tableName( $tbl );
 372                                         $tableName = $this->oldTableNames[$tbl];
 373                                         $db->query("CREATE TEMPORARY TABLE $newTableName (LIKE $tableName $def)");
 374                                 }
 375                         } else {
 376                                 # Hack for MySQL versions < 4.1, which don't support
 377                                 # "CREATE TABLE ... LIKE". Note that
 378                                 # "CREATE TEMPORARY TABLE ... SELECT * FROM ... LIMIT 0"
 379                                 # would not create the indexes we need....
 380                                 foreach ($tables as $tbl) {
 381                                         $res = $db->query("SHOW CREATE TABLE {$this->oldTableNames[$tbl]}");
 382                                         $row = $db->fetchRow($res);
 383                                         $create = $row[1];
 384                                         $create_tmp = preg_replace('/CREATE TABLE `(.*?)`/', 'CREATE TEMPORARY TABLE `'
 385                                                 . $wgDBprefix . $tbl .'`', $create);
 386                                         if ($create === $create_tmp) {
 387                                                 # Couldn't do replacement
 388                                                 die("could not create temporary table $tbl");
 389                                         }
 390                                         $db->query($create_tmp);
 391                                 }
 392
 393                         }
 394
 395                         # Hack: insert a few Wikipedia in-project interwiki prefixes,
 396                         # for testing inter-language links
 397                         $db->insert( 'interwiki', array(
 398                                 array( 'iw_prefix' => 'Wikipedia',
 399                                        'iw_url'    => 'http://en.wikipedia.org/wiki/$1',
 400                                        'iw_local'  => 0 ),
 401                                 array( 'iw_prefix' => 'MeatBall',
 402                                        'iw_url'    => 'http://www.usemod.com/cgi-bin/mb.pl?$1',
 403                                        'iw_local'  => 0 ),
 404                                 array( 'iw_prefix' => 'zh',
 405                                        'iw_url'    => 'http://zh.wikipedia.org/wiki/$1',
 406                                        'iw_local'  => 1 ),
 407                                 array( 'iw_prefix' => 'es',
 408                                        'iw_url'    => 'http://es.wikipedia.org/wiki/$1',
 409                                        'iw_local'  => 1 ),
 410                                 array( 'iw_prefix' => 'fr',
 411                                        'iw_url'    => 'http://fr.wikipedia.org/wiki/$1',
 412                                        'iw_local'  => 1 ),
 413                                 array( 'iw_prefix' => 'ru',
 414                                        'iw_url'    => 'http://ru.wikipedia.org/wiki/$1',
 415                                        'iw_local'  => 1 ),
 416                                 ) );
 417
 418                         # Hack: Insert an image to work with
 419                         $db->insert( 'image', array(
 420                                 'img_name'        => 'Foobar.jpg',
 421                                 'img_size'        => 12345,
 422                                 'img_description' => 'Some lame file',
 423                                 'img_user'        => 1,
 424                                 'img_user_text'   => 'WikiSysop',
 425                                 'img_timestamp'   => $db->timestamp( '20010115123500' ),
 426                                 'img_width'       => 1941,
 427                                 'img_height'      => 220,
 428                                 'img_bits'        => 24,
 429                                 'img_media_type'  => MEDIATYPE_BITMAP,
 430                                 'img_major_mime'  => "image",
 431                                 'img_minor_mime'  => "jpeg",
 432                                 ) );
 433
 434                         $setupDB = true;
 435                 }
 436         }
 437
 438         /**
 439          * Create a dummy uploads directory which will contain a couple
 440          * of files in order to pass existence tests.
 441          * @return string The directory
 442          * @access private
 443          */
 444         function setupUploadDir() {
 445                 global $IP;
 446
 447                 $dir = wfTempDir() . "/mwParser-" . mt_rand() . "-images";
 448                 mkdir( $dir );
 449                 mkdir( $dir . '/3' );
 450                 mkdir( $dir . '/3/3a' );
 451
 452                 $img = "$IP/skins/monobook/headbg.jpg";
 453                 $h = fopen($img, 'r');
 454                 $c = fread($h, filesize($img));
 455                 fclose($h);
 456
 457                 $f = fopen( $dir . '/3/3a/Foobar.jpg', 'wb' );
 458                 fwrite( $f, $c );
 459                 fclose( $f );
 460                 return $dir;
 461         }
 462
 463         /**
 464          * Restore default values and perform any necessary clean-up
 465          * after each test runs.
 466          *
 467          * @access private
 468          */
 469         function teardownGlobals() {
 470                 foreach( $this->savedGlobals as $var => $val ) {
 471                         $GLOBALS[$var] = $val;
 472                 }
 473                 if( isset( $this->uploadDir ) ) {
 474                         $this->teardownUploadDir( $this->uploadDir );
 475                         unset( $this->uploadDir );
 476                 }
 477         }
 478
 479         /**
 480          * Remove the dummy uploads directory
 481          * @access private
 482          */
 483         function teardownUploadDir( $dir ) {
 484                 unlink( "$dir/3/3a/Foobar.jpg" );
 485                 rmdir( "$dir/3/3a" );
 486                 rmdir( "$dir/3" );
 487
 488                 @unlink( "$dir/thumb/3/3a/Foobar.jpg/180px-Foobar.jpg" );
 489                 @rmdir( "$dir/thumb/3/3a/Foobar.jpg" );
 490                 @rmdir( "$dir/thumb/3/3a" );
 491                 @rmdir( "$dir/thumb/3/39" ); # wtf?
 492                 @rmdir( "$dir/thumb/3" );
 493                 @rmdir( "$dir/thumb" );
 494                 rmdir( "$dir" );
 495         }
 496
 497         /**
 498          * "Running test $desc..."
 499          * @access private
 500          */
 501         function showTesting( $desc ) {
 502                 print "Running test $desc... ";
 503         }
 504
 505         /**
 506          * Print a happy success message.
 507          *
 508          * @param string $desc The test name
 509          * @return bool
 510          * @access private
 511          */
 512         function showSuccess( $desc ) {
 513                 if( !$this->quiet ) {
 514                         print $this->termColor( '1;32' ) . 'PASSED' . $this->termReset() . "\n";
 515                 }
 516                 return true;
 517         }
 518
 519         /**
 520          * Print a failure message and provide some explanatory output
 521          * about what went wrong if so configured.
 522          *
 523          * @param string $desc The test name
 524          * @param string $result Expected HTML output
 525          * @param string $html Actual HTML output
 526          * @return bool
 527          * @access private
 528          */
 529         function showFailure( $desc, $result, $html ) {
 530                 if( $this->quiet ) {
 531                         # In quiet mode we didn't show the 'Testing' message before the
 532                         # test, in case it succeeded. Show it now:
 533                         $this->showTesting( $desc );
 534                 }
 535                 print $this->termColor( '1;31' ) . 'FAILED!' . $this->termReset() . "\n";
 536                 if( $this->showDiffs ) {
 537                         print $this->quickDiff( $result, $html );
 538                 }
 539                 if( !$this->wellFormed( $html ) ) {
 540                         print "XML error: $this->mXmlError\n";
 541                 }
 542                 return false;
 543         }
 544
 545         /**
 546          * Run given strings through a diff and return the (colorized) output.
 547          * Requires writable /tmp directory and a 'diff' command in the PATH.
 548          *
 549          * @param string $input
 550          * @param string $output
 551          * @param string $inFileTail Tailing for the input file name
 552          * @param string $outFileTail Tailing for the output file name
 553          * @return string
 554          * @access private
 555          */
 556         function quickDiff( $input, $output, $inFileTail='expected', $outFileTail='actual' ) {
 557                 $prefix = wfTempDir() . "/mwParser-" . mt_rand();
 558
 559                 $infile = "$prefix-$inFileTail";
 560                 $this->dumpToFile( $input, $infile );
 561
 562                 $outfile = "$prefix-$outFileTail";
 563                 $this->dumpToFile( $output, $outfile );
 564
 565                 $diff = `diff -au $infile $outfile`;
 566                 unlink( $infile );
 567                 unlink( $outfile );
 568
 569                 return $this->colorDiff( $diff );
 570         }
 571
 572         /**
 573          * Write the given string to a file, adding a final newline.
 574          *
 575          * @param string $data
 576          * @param string $filename
 577          * @access private
 578          */
 579         function dumpToFile( $data, $filename ) {
 580                 $file = fopen( $filename, "wt" );
 581                 fwrite( $file, $data . "\n" );
 582                 fclose( $file );
 583         }
 584
 585         /**
 586          * Return ANSI terminal escape code for changing text attribs/color,
 587          * or empty string if color output is disabled.
 588          *
 589          * @param string $color Semicolon-separated list of attribute/color codes
 590          * @return string
 591          * @access private
 592          */
 593         function termColor( $color ) {
 594                 if($this->lightcolor) {
 595                         return $this->color ? "\x1b[1;{$color}m" : '';
 596                 } else {
 597                         return $this->color ? "\x1b[{$color}m" : '';
 598                 }
 599         }
 600
 601         /**
 602          * Return ANSI terminal escape code for restoring default text attributes,
 603          * or empty string if color output is disabled.
 604          *
 605          * @return string
 606          * @access private
 607          */
 608         function termReset() {
 609                 return $this->color ? "\x1b[0m" : '';
 610         }
 611
 612         /**
 613          * Colorize unified diff output if set for ANSI color output.
 614          * Subtractions are colored blue, additions red.
 615          *
 616          * @param string $text
 617          * @return string
 618          * @access private
 619          */
 620         function colorDiff( $text ) {
 621                 return preg_replace(
 622                         array( '/^(-.*)$/m', '/^(\+.*)$/m' ),
 623                         array( $this->termColor( 34 ) . '$1' . $this->termReset(),
 624                                $this->termColor( 31 ) . '$1' . $this->termReset() ),
 625                         $text );
 626         }
 627
 628         /**
 629          * Insert a temporary test article
 630          * @param string $name the title, including any prefix
 631          * @param string $text the article text
 632          * @param int $line the input line number, for reporting errors
 633          * @static
 634          * @access private
 635          */
 636         function addArticle($name, $text, $line) {
 637                 $this->setupGlobals();
 638                 $title = Title::newFromText( $name );
 639                 if ( is_null($title) ) {
 640                         die( "invalid title at line $line\n" );
 641                 }
 642
 643                 $aid = $title->getArticleID( GAID_FOR_UPDATE );
 644                 if ($aid != 0) {
 645                         die( "duplicate article at line $line\n" );
 646                 }
 647
 648                 $art = new Article($title);
 649                 $art->insertNewArticle($text, '', false, false );
 650                 $this->teardownGlobals();
 651         }
 652
 653         /*
 654          * Run the "tidy" command on text if the $wgUseTidy
 655          * global is true
 656          *
 657          * @param string $text the text to tidy
 658          * @return string
 659          * @static
 660          * @access private
 661          */
 662         function tidy( $text ) {
 663                 global $wgUseTidy;
 664                 if ($wgUseTidy) {
 665                         $text = Parser::tidy($text);
 666                 }
 667                 return $text;
 668         }
 669
 670         /**
 671          * Hack up a private DOCTYPE with HTML's standard entity declarations.
 672          * PHP 4 seemed to know these if you gave it an HTML doctype, but
 673          * PHP 5.1 doesn't.
 674          * @return string
 675          * @access private
 676          */
 677         function hackDocType() {
 678                 global $wgHtmlEntities;
 679                 $out = "<!DOCTYPE html [\n";
 680                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
 681                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
 682                 }
 683                 $out .= "]>\n";
 684                 return $out;
 685         }
 686
 687         function wellFormed( $text ) {
 688                 $html =
 689                         $this->hackDocType() .
 690                         '<html>' .
 691                         $text .
 692                         '</html>';
 693
 694                 $parser = xml_parser_create( "UTF-8" );
 695
 696                 # case folding violates XML standard, turn it off
 697                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 698
 699                 if( !xml_parse( $parser, $html, true ) ) {
 700                         $err = xml_error_string( xml_get_error_code( $parser ) );
 701                         $position = xml_get_current_byte_index( $parser );
 702                         $fragment = $this->extractFragment( $html, $position );
 703                         $this->mXmlError = "$err at byte $position:\n$fragment";
 704                         xml_parser_free( $parser );
 705                         return false;
 706                 }
 707                 xml_parser_free( $parser );
 708                 return true;
 709         }
 710
 711         function extractFragment( $text, $position ) {
 712                 $start = max( 0, $position - 10 );
 713                 $before = $position - $start;
 714                 $fragment = '...' .
 715                         $this->termColor( 34 ) .
 716                         substr( $text, $start, $before ) .
 717                         $this->termColor( 0 ) .
 718                         $this->termColor( 31 ) .
 719                         $this->termColor( 1 ) .
 720                         substr( $text, $position, 1 ) .
 721                         $this->termColor( 0 ) .
 722                         $this->termColor( 34 ) .
 723                         substr( $text, $position + 1, 9 ) .
 724                         $this->termColor( 0 ) .
 725                         '...';
 726                 $display = str_replace( "\n", ' ', $fragment );
 727                 $caret = '   ' .
 728                         str_repeat( ' ', $before ) .
 729                         $this->termColor( 31 ) .
 730                         '^' .
 731                         $this->termColor( 0 );
 732                 return "$display\n$caret";
 733         }
 734
 735 }
 736
 737 ?>