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                         wfDie( "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                                                 wfDie( "'endarticle' without 'text' at line $n\n" );
 139                                         }
 140                                         if( !isset( $data['article'] ) ) {
 141                                                 wfDie( "'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                                                 wfDie( "'end' without 'test' at line $n\n" );
 151                                         }
 152                                         if( !isset( $data['input'] ) ) {
 153                                                 wfDie( "'end' without 'input' at line $n\n" );
 154                                         }
 155                                         if( !isset( $data['result'] ) ) {
 156                                                 wfDie( "'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[$section] ) ) {
 184                                         wfDie( "duplicate section '$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                         wfDie( "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                         'wgLocaltimezone' => 'UTC',
 316                         );
 317                 $this->savedGlobals = array();
 318                 foreach( $settings as $var => $val ) {
 319                         $this->savedGlobals[$var] = $GLOBALS[$var];
 320                         $GLOBALS[$var] = $val;
 321                 }
 322                 $GLOBALS['wgLoadBalancer']->loadMasterPos();
 323                 $GLOBALS['wgMessageCache']->initialise( new BagOStuff(), false, 0, $GLOBALS['wgDBname'] );
 324                 $this->setupDatabase();
 325
 326                 global $wgUser;
 327                 $wgUser = new User();
 328         }
 329
 330         # List of temporary tables to create, without prefix
 331         # Some of these probably aren't necessary
 332         function listTables() {
 333                 $tables = array('user', 'page', 'revision', 'text',
 334                         'pagelinks', 'imagelinks', 'categorylinks', 'templatelinks',
 335                         'site_stats', 'hitcounter',
 336                         'ipblocks', 'image', 'oldimage',
 337                         'recentchanges',
 338                         'watchlist', 'math', 'searchindex',
 339                         'interwiki', 'querycache',
 340                         'objectcache'
 341                 );
 342
 343                 // FIXME manually adding additional table for the tasks extension
 344                 // we probably need a better software wide system to register new
 345                 // tables.
 346                 global $wgExtensionFunctions;
 347                 if( in_array('wfTasksExtension' , $wgExtensionFunctions ) ) {
 348                         $tables[] = 'tasks';
 349                 }
 350
 351                 return $tables;
 352         }
 353
 354         /**
 355          * Set up a temporary set of wiki tables to work with for the tests.
 356          * Currently this will only be done once per run, and any changes to
 357          * the db will be visible to later tests in the run.
 358          *
 359          * @access private
 360          */
 361         function setupDatabase() {
 362                 static $setupDB = false;
 363                 global $wgDBprefix;
 364
 365                 # Make sure we don't mess with the live DB
 366                 if (!$setupDB && $wgDBprefix === 'parsertest_') {
 367                         # oh teh horror
 368                         $GLOBALS['wgLoadBalancer'] = LoadBalancer::newFromParams( $GLOBALS['wgDBservers'] );
 369                         $db =& wfGetDB( DB_MASTER );
 370
 371                         $tables = $this->listTables();
 372
 373                         if (!(strcmp($db->getServerVersion(), '4.1') < 0 and stristr($db->getSoftwareLink(), 'MySQL'))) {
 374                                 # Database that supports CREATE TABLE ... LIKE
 375                                 global $wgDBtype;
 376                                 if( $wgDBtype == 'PostgreSQL' ) {
 377                                         $def = 'INCLUDING DEFAULTS';
 378                                 } else {
 379                                         $def = '';
 380                                 }
 381                                 foreach ($tables as $tbl) {
 382                                         $newTableName = $db->tableName( $tbl );
 383                                         $tableName = $this->oldTableNames[$tbl];
 384                                         $db->query("CREATE TEMPORARY TABLE $newTableName (LIKE $tableName $def)");
 385                                 }
 386                         } else {
 387                                 # Hack for MySQL versions < 4.1, which don't support
 388                                 # "CREATE TABLE ... LIKE". Note that
 389                                 # "CREATE TEMPORARY TABLE ... SELECT * FROM ... LIMIT 0"
 390                                 # would not create the indexes we need....
 391                                 foreach ($tables as $tbl) {
 392                                         $res = $db->query("SHOW CREATE TABLE {$this->oldTableNames[$tbl]}");
 393                                         $row = $db->fetchRow($res);
 394                                         $create = $row[1];
 395                                         $create_tmp = preg_replace('/CREATE TABLE `(.*?)`/', 'CREATE TEMPORARY TABLE `'
 396                                                 . $wgDBprefix . $tbl .'`', $create);
 397                                         if ($create === $create_tmp) {
 398                                                 # Couldn't do replacement
 399                                                 wfDie("could not create temporary table $tbl");
 400                                         }
 401                                         $db->query($create_tmp);
 402                                 }
 403
 404                         }
 405
 406                         # Hack: insert a few Wikipedia in-project interwiki prefixes,
 407                         # for testing inter-language links
 408                         $db->insert( 'interwiki', array(
 409                                 array( 'iw_prefix' => 'Wikipedia',
 410                                        'iw_url'    => 'http://en.wikipedia.org/wiki/$1',
 411                                        'iw_local'  => 0 ),
 412                                 array( 'iw_prefix' => 'MeatBall',
 413                                        'iw_url'    => 'http://www.usemod.com/cgi-bin/mb.pl?$1',
 414                                        'iw_local'  => 0 ),
 415                                 array( 'iw_prefix' => 'zh',
 416                                        'iw_url'    => 'http://zh.wikipedia.org/wiki/$1',
 417                                        'iw_local'  => 1 ),
 418                                 array( 'iw_prefix' => 'es',
 419                                        'iw_url'    => 'http://es.wikipedia.org/wiki/$1',
 420                                        'iw_local'  => 1 ),
 421                                 array( 'iw_prefix' => 'fr',
 422                                        'iw_url'    => 'http://fr.wikipedia.org/wiki/$1',
 423                                        'iw_local'  => 1 ),
 424                                 array( 'iw_prefix' => 'ru',
 425                                        'iw_url'    => 'http://ru.wikipedia.org/wiki/$1',
 426                                        'iw_local'  => 1 ),
 427                                 ) );
 428
 429                         # Hack: Insert an image to work with
 430                         $db->insert( 'image', array(
 431                                 'img_name'        => 'Foobar.jpg',
 432                                 'img_size'        => 12345,
 433                                 'img_description' => 'Some lame file',
 434                                 'img_user'        => 1,
 435                                 'img_user_text'   => 'WikiSysop',
 436                                 'img_timestamp'   => $db->timestamp( '20010115123500' ),
 437                                 'img_width'       => 1941,
 438                                 'img_height'      => 220,
 439                                 'img_bits'        => 24,
 440                                 'img_media_type'  => MEDIATYPE_BITMAP,
 441                                 'img_major_mime'  => "image",
 442                                 'img_minor_mime'  => "jpeg",
 443                                 ) );
 444
 445                         $setupDB = true;
 446                 }
 447         }
 448
 449         /**
 450          * Create a dummy uploads directory which will contain a couple
 451          * of files in order to pass existence tests.
 452          * @return string The directory
 453          * @access private
 454          */
 455         function setupUploadDir() {
 456                 global $IP;
 457
 458                 $dir = wfTempDir() . "/mwParser-" . mt_rand() . "-images";
 459                 mkdir( $dir );
 460                 mkdir( $dir . '/3' );
 461                 mkdir( $dir . '/3/3a' );
 462
 463                 $img = "$IP/skins/monobook/headbg.jpg";
 464                 $h = fopen($img, 'r');
 465                 $c = fread($h, filesize($img));
 466                 fclose($h);
 467
 468                 $f = fopen( $dir . '/3/3a/Foobar.jpg', 'wb' );
 469                 fwrite( $f, $c );
 470                 fclose( $f );
 471                 return $dir;
 472         }
 473
 474         /**
 475          * Restore default values and perform any necessary clean-up
 476          * after each test runs.
 477          *
 478          * @access private
 479          */
 480         function teardownGlobals() {
 481                 foreach( $this->savedGlobals as $var => $val ) {
 482                         $GLOBALS[$var] = $val;
 483                 }
 484                 if( isset( $this->uploadDir ) ) {
 485                         $this->teardownUploadDir( $this->uploadDir );
 486                         unset( $this->uploadDir );
 487                 }
 488         }
 489
 490         /**
 491          * Remove the dummy uploads directory
 492          * @access private
 493          */
 494         function teardownUploadDir( $dir ) {
 495                 unlink( "$dir/3/3a/Foobar.jpg" );
 496                 rmdir( "$dir/3/3a" );
 497                 rmdir( "$dir/3" );
 498
 499                 @unlink( "$dir/thumb/3/3a/Foobar.jpg/180px-Foobar.jpg" );
 500                 @rmdir( "$dir/thumb/3/3a/Foobar.jpg" );
 501                 @rmdir( "$dir/thumb/3/3a" );
 502                 @rmdir( "$dir/thumb/3/39" ); # wtf?
 503                 @rmdir( "$dir/thumb/3" );
 504                 @rmdir( "$dir/thumb" );
 505                 rmdir( "$dir" );
 506         }
 507
 508         /**
 509          * "Running test $desc..."
 510          * @access private
 511          */
 512         function showTesting( $desc ) {
 513                 print "Running test $desc... ";
 514         }
 515
 516         /**
 517          * Print a happy success message.
 518          *
 519          * @param string $desc The test name
 520          * @return bool
 521          * @access private
 522          */
 523         function showSuccess( $desc ) {
 524                 if( !$this->quiet ) {
 525                         print $this->termColor( '1;32' ) . 'PASSED' . $this->termReset() . "\n";
 526                 }
 527                 return true;
 528         }
 529
 530         /**
 531          * Print a failure message and provide some explanatory output
 532          * about what went wrong if so configured.
 533          *
 534          * @param string $desc The test name
 535          * @param string $result Expected HTML output
 536          * @param string $html Actual HTML output
 537          * @return bool
 538          * @access private
 539          */
 540         function showFailure( $desc, $result, $html ) {
 541                 if( $this->quiet ) {
 542                         # In quiet mode we didn't show the 'Testing' message before the
 543                         # test, in case it succeeded. Show it now:
 544                         $this->showTesting( $desc );
 545                 }
 546                 print $this->termColor( '1;31' ) . 'FAILED!' . $this->termReset() . "\n";
 547                 if( $this->showDiffs ) {
 548                         print $this->quickDiff( $result, $html );
 549                 }
 550                 if( !$this->wellFormed( $html ) ) {
 551                         print "XML error: $this->mXmlError\n";
 552                 }
 553                 return false;
 554         }
 555
 556         /**
 557          * Run given strings through a diff and return the (colorized) output.
 558          * Requires writable /tmp directory and a 'diff' command in the PATH.
 559          *
 560          * @param string $input
 561          * @param string $output
 562          * @param string $inFileTail Tailing for the input file name
 563          * @param string $outFileTail Tailing for the output file name
 564          * @return string
 565          * @access private
 566          */
 567         function quickDiff( $input, $output, $inFileTail='expected', $outFileTail='actual' ) {
 568                 $prefix = wfTempDir() . "/mwParser-" . mt_rand();
 569
 570                 $infile = "$prefix-$inFileTail";
 571                 $this->dumpToFile( $input, $infile );
 572
 573                 $outfile = "$prefix-$outFileTail";
 574                 $this->dumpToFile( $output, $outfile );
 575
 576                 $diff = `diff -au $infile $outfile`;
 577                 unlink( $infile );
 578                 unlink( $outfile );
 579
 580                 return $this->colorDiff( $diff );
 581         }
 582
 583         /**
 584          * Write the given string to a file, adding a final newline.
 585          *
 586          * @param string $data
 587          * @param string $filename
 588          * @access private
 589          */
 590         function dumpToFile( $data, $filename ) {
 591                 $file = fopen( $filename, "wt" );
 592                 fwrite( $file, $data . "\n" );
 593                 fclose( $file );
 594         }
 595
 596         /**
 597          * Return ANSI terminal escape code for changing text attribs/color,
 598          * or empty string if color output is disabled.
 599          *
 600          * @param string $color Semicolon-separated list of attribute/color codes
 601          * @return string
 602          * @access private
 603          */
 604         function termColor( $color ) {
 605                 if($this->lightcolor) {
 606                         return $this->color ? "\x1b[1;{$color}m" : '';
 607                 } else {
 608                         return $this->color ? "\x1b[{$color}m" : '';
 609                 }
 610         }
 611
 612         /**
 613          * Return ANSI terminal escape code for restoring default text attributes,
 614          * or empty string if color output is disabled.
 615          *
 616          * @return string
 617          * @access private
 618          */
 619         function termReset() {
 620                 return $this->color ? "\x1b[0m" : '';
 621         }
 622
 623         /**
 624          * Colorize unified diff output if set for ANSI color output.
 625          * Subtractions are colored blue, additions red.
 626          *
 627          * @param string $text
 628          * @return string
 629          * @access private
 630          */
 631         function colorDiff( $text ) {
 632                 return preg_replace(
 633                         array( '/^(-.*)$/m', '/^(\+.*)$/m' ),
 634                         array( $this->termColor( 34 ) . '$1' . $this->termReset(),
 635                                $this->termColor( 31 ) . '$1' . $this->termReset() ),
 636                         $text );
 637         }
 638
 639         /**
 640          * Insert a temporary test article
 641          * @param string $name the title, including any prefix
 642          * @param string $text the article text
 643          * @param int $line the input line number, for reporting errors
 644          * @static
 645          * @access private
 646          */
 647         function addArticle($name, $text, $line) {
 648                 $this->setupGlobals();
 649                 $title = Title::newFromText( $name );
 650                 if ( is_null($title) ) {
 651                         wfDie( "invalid title at line $line\n" );
 652                 }
 653
 654                 $aid = $title->getArticleID( GAID_FOR_UPDATE );
 655                 if ($aid != 0) {
 656                         wfDie( "duplicate article at line $line\n" );
 657                 }
 658
 659                 $art = new Article($title);
 660                 $art->insertNewArticle($text, '', false, false );
 661                 $this->teardownGlobals();
 662         }
 663
 664         /*
 665          * Run the "tidy" command on text if the $wgUseTidy
 666          * global is true
 667          *
 668          * @param string $text the text to tidy
 669          * @return string
 670          * @static
 671          * @access private
 672          */
 673         function tidy( $text ) {
 674                 global $wgUseTidy;
 675                 if ($wgUseTidy) {
 676                         $text = Parser::tidy($text);
 677                 }
 678                 return $text;
 679         }
 680
 681         function wellFormed( $text ) {
 682                 $html =
 683                         Sanitizer::hackDocType() .
 684                         '<html>' .
 685                         $text .
 686                         '</html>';
 687
 688                 $parser = xml_parser_create( "UTF-8" );
 689
 690                 # case folding violates XML standard, turn it off
 691                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 692
 693                 if( !xml_parse( $parser, $html, true ) ) {
 694                         $err = xml_error_string( xml_get_error_code( $parser ) );
 695                         $position = xml_get_current_byte_index( $parser );
 696                         $fragment = $this->extractFragment( $html, $position );
 697                         $this->mXmlError = "$err at byte $position:\n$fragment";
 698                         xml_parser_free( $parser );
 699                         return false;
 700                 }
 701                 xml_parser_free( $parser );
 702                 return true;
 703         }
 704
 705         function extractFragment( $text, $position ) {
 706                 $start = max( 0, $position - 10 );
 707                 $before = $position - $start;
 708                 $fragment = '...' .
 709                         $this->termColor( 34 ) .
 710                         substr( $text, $start, $before ) .
 711                         $this->termColor( 0 ) .
 712                         $this->termColor( 31 ) .
 713                         $this->termColor( 1 ) .
 714                         substr( $text, $position, 1 ) .
 715                         $this->termColor( 0 ) .
 716                         $this->termColor( 34 ) .
 717                         substr( $text, $position + 1, 9 ) .
 718                         $this->termColor( 0 ) .
 719                         '...';
 720                 $display = str_replace( "\n", ' ', $fragment );
 721                 $caret = '   ' .
 722                         str_repeat( ' ', $before ) .
 723                         $this->termColor( 31 ) .
 724                         '^' .
 725                         $this->termColor( 0 );
 726                 return "$display\n$caret";
 727         }
 728
 729 }
 730
 731 ?>