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