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
  36 /**
  37  * @package MediaWiki
  38  * @subpackage Maintenance
  39  */
  40 class ParserTest {
  41         /**
  42          * boolean $color whereas output should be colorized
  43          * @access private
  44          */
  45         var $color;
  46
  47         /**
  48          * boolean $lightcolor whereas output should use light colors
  49          * @access private
  50          */
  51         var $lightcolor;
  52
  53         /**
  54          * Sets terminal colorization and diff/quick modes depending on OS and
  55          * command-line options (--color and --quick).
  56          *
  57          * @access public
  58          */
  59         function ParserTest() {
  60                 global $options;
  61
  62                 # Only colorize output if stdout is a terminal.
  63                 $this->lightcolor = false;
  64                 $this->color = !wfIsWindows() && posix_isatty(1);
  65
  66                 if( isset( $options['color'] ) ) {
  67                         switch( $options['color'] ) {
  68                         case 'no':
  69                                 $this->color = false;
  70                                 break;
  71                         case 'light':
  72                                 $this->lightcolor = true;
  73                                 # Fall through
  74                         case 'yes':
  75                         default:
  76                                 $this->color = true;
  77                                 break;
  78                         }
  79                 }
  80
  81                 $this->showDiffs = !isset( $options['quick'] );
  82
  83                 $this->quiet = isset( $options['quiet'] );
  84
  85                 if (isset($options['regex'])) {
  86                         $this->regex = $options['regex'];
  87                 } else {
  88                         # Matches anything
  89                         $this->regex = '';
  90                 }
  91         }
  92
  93         /**
  94          * Remove last character if it is a newline
  95          * @access private
  96          */
  97         function chomp($s) {
  98                 if (substr($s, -1) === "\n") {
  99                         return substr($s, 0, -1);
 100                 }
 101                 else {
 102                         return $s;
 103                 }
 104         }
 105
 106         /**
 107          * Run a series of tests listed in the given text file.
 108          * Each test consists of a brief description, wikitext input,
 109          * and the expected HTML output.
 110          *
 111          * Prints status updates on stdout and counts up the total
 112          * number and percentage of passed tests.
 113          *
 114          * @param string $filename
 115          * @return bool True if passed all tests, false if any tests failed.
 116          * @access public
 117          */
 118         function runTestsFromFile( $filename ) {
 119                 $infile = fopen( $filename, 'rt' );
 120                 if( !$infile ) {
 121                         die( "Couldn't open parserTests.txt\n" );
 122                 }
 123
 124                 $data = array();
 125                 $section = null;
 126                 $success = 0;
 127                 $total = 0;
 128                 $n = 0;
 129                 while( false !== ($line = fgets( $infile ) ) ) {
 130                         $n++;
 131                         if( preg_match( '/^!!\s*(\w+)/', $line, $matches ) ) {
 132                                 $section = strtolower( $matches[1] );
 133                                 if( $section == 'endarticle') {
 134                                         if( !isset( $data['text'] ) ) {
 135                                                 die( "'endarticle' without 'text' at line $n\n" );
 136                                         }
 137                                         if( !isset( $data['article'] ) ) {
 138                                                 die( "'endarticle' without 'article' at line $n\n" );
 139                                         }
 140                                         $this->addArticle($this->chomp($data['article']), $this->chomp($data['text']), $n);
 141                                         $data = array();
 142                                         $section = null;
 143                                         continue;
 144                                 }
 145                                 if( $section == 'end' ) {
 146                                         if( !isset( $data['test'] ) ) {
 147                                                 die( "'end' without 'test' at line $n\n" );
 148                                         }
 149                                         if( !isset( $data['input'] ) ) {
 150                                                 die( "'end' without 'input' at line $n\n" );
 151                                         }
 152                                         if( !isset( $data['result'] ) ) {
 153                                                 die( "'end' without 'result' at line $n\n" );
 154                                         }
 155                                         if( !isset( $data['options'] ) ) {
 156                                                 $data['options'] = '';
 157                                         }
 158                                         else {
 159                                                 $data['options'] = $this->chomp( $data['options'] );
 160                                         }
 161                                         if (preg_match('/\\bdisabled\\b/i', $data['options'])
 162                                                 || !preg_match("/{$this->regex}/i", $data['test'])) {
 163                                                 # disabled test
 164                                                 $data = array();
 165                                                 $section = null;
 166                                                 continue;
 167                                         }
 168                                         if( $this->runTest(
 169                                                 $this->chomp( $data['test'] ),
 170                                                 $this->chomp( $data['input'] ),
 171                                                 $this->chomp( $data['result'] ),
 172                                                 $this->chomp( $data['options'] ) ) ) {
 173                                                 $success++;
 174                                         }
 175                                         $total++;
 176                                         $data = array();
 177                                         $section = null;
 178                                         continue;
 179                                 }
 180                                 if ( isset ($data[$section] ) ) {
 181                                         die ( "duplicate section '$section' at line $n\n" );
 182                                 }
 183                                 $data[$section] = '';
 184                                 continue;
 185                         }
 186                         if( $section ) {
 187                                 $data[$section] .= $line;
 188                         }
 189                 }
 190                 if( $total > 0 ) {
 191                         $ratio = wfPercent( 100 * $success / $total );
 192                         print $this->termColor( 1 ) . "\nPassed $success of $total tests ($ratio) ";
 193                         if( $success == $total ) {
 194                                 print $this->termColor( 32 ) . "PASSED!";
 195                         } else {
 196                                 print $this->termColor( 31 ) . "FAILED!";
 197                         }
 198                         print $this->termReset() . "\n";
 199                         return ($success == $total);
 200                 } else {
 201                         die( "No tests found.\n" );
 202                 }
 203         }
 204
 205         /**
 206          * Run a given wikitext input through a freshly-constructed wiki parser,
 207          * and compare the output against the expected results.
 208          * Prints status and explanatory messages to stdout.
 209          *
 210          * @param string $input Wikitext to try rendering
 211          * @param string $result Result to output
 212          * @return bool
 213          */
 214         function runTest( $desc, $input, $result, $opts ) {
 215                 if( !$this->quiet ) {
 216                         $this->showTesting( $desc );
 217                 }
 218
 219                 $this->setupGlobals($opts);
 220
 221                 $user =& new User();
 222                 $options = ParserOptions::newFromUser( $user );
 223
 224                 if (preg_match('/\\bmath\\b/i', $opts)) {
 225                         # XXX this should probably be done by the ParserOptions
 226                         require_once('Math.php');
 227
 228                         $options->setUseTex(true);
 229                 }
 230
 231                 if (preg_match('/title=\[\[(.*)\]\]/', $opts, $m)) {
 232                         $titleText = $m[1];
 233                 }
 234                 else {
 235                         $titleText = 'Parser test';
 236                 }
 237
 238                 $parser =& new Parser();
 239                 $title =& Title::makeTitle( NS_MAIN, $titleText );
 240
 241                 if (preg_match('/\\bpst\\b/i', $opts)) {
 242                         $out = $parser->preSaveTransform( $input, $title, $user, $options );
 243                 } elseif (preg_match('/\\bmsg\\b/i', $opts)) {
 244                         $out = $parser->transformMsg( $input, $options );
 245                 } else {
 246                         $output = $parser->parse( $input, $title, $options );
 247                         $out = $output->getText();
 248
 249                         if (preg_match('/\\bill\\b/i', $opts)) {
 250                                 $out = $this->tidy( implode( ' ', $output->getLanguageLinks() ) );
 251                         } else if (preg_match('/\\bcat\\b/i', $opts)) {
 252                                 $out = $this->tidy ( implode( ' ', $output->getCategoryLinks() ) );
 253                         }
 254
 255                         $result = $this->tidy($result);
 256                 }
 257
 258                 $this->teardownGlobals();
 259
 260                 if( $result === $out && $this->wellFormed( $out ) ) {
 261                         return $this->showSuccess( $desc );
 262                 } else {
 263                         return $this->showFailure( $desc, $result, $out );
 264                 }
 265         }
 266
 267         /**
 268          * Set up the global variables for a consistent environment for each test.
 269          * Ideally this should replace the global configuration entirely.
 270          *
 271          * @access private
 272          */
 273         function setupGlobals($opts = '') {
 274                 # Save the prefixed / quoted table names for later use when we make the temporaries.
 275                 $db =& wfGetDB( DB_READ );
 276                 $this->oldTableNames = array();
 277                 foreach( $this->listTables() as $table ) {
 278                         $this->oldTableNames[$table] = $db->tableName( $table );
 279                 }
 280                 if( !isset( $this->uploadDir ) ) {
 281                         $this->uploadDir = $this->setupUploadDir();
 282                 }
 283
 284                 $settings = array(
 285                         'wgServer' => 'http://localhost',
 286                         'wgScript' => '/index.php',
 287                         'wgScriptPath' => '/',
 288                         'wgArticlePath' => '/wiki/$1',
 289                         'wgUploadPath' => 'http://example.com/images',
 290                         'wgUploadDirectory' => $this->uploadDir,
 291                         'wgStyleSheetPath' => '/skins',
 292                         'wgSitename' => 'MediaWiki',
 293                         'wgLanguageCode' => 'en',
 294                         'wgContLanguageCode' => 'en',
 295                         'wgDBprefix' => 'parsertest',
 296                         'wgDefaultUserOptions' => array(),
 297
 298                         'wgLang' => new LanguageUtf8(),
 299                         'wgContLang' => new LanguageUtf8(),
 300                         'wgNamespacesWithSubpages' => array( 0 => preg_match('/\\bsubpage\\b/i', $opts)),
 301                         'wgMaxTocLevel' => 999,
 302                         'wgCapitalLinks' => true,
 303                         'wgDefaultUserOptions' => array(),
 304                         'wgNoFollowLinks' => true,
 305                         'wgThumbnailScriptPath' => false,
 306                         );
 307                 $this->savedGlobals = array();
 308                 foreach( $settings as $var => $val ) {
 309                         $this->savedGlobals[$var] = $GLOBALS[$var];
 310                         $GLOBALS[$var] = $val;
 311                 }
 312                 $GLOBALS['wgLoadBalancer']->loadMasterPos();
 313                 $GLOBALS['wgMessageCache']->initialise( new BagOStuff(), false, 0, $GLOBALS['wgDBname'] );
 314                 $this->setupDatabase();
 315
 316                 global $wgUser;
 317                 $wgUser = new User();
 318         }
 319
 320         # List of temporary tables to create, without prefix
 321         # Some of these probably aren't necessary
 322         function listTables() {
 323                 return array('user', 'page', 'revision', 'text',
 324                         'pagelinks', 'imagelinks', 'categorylinks',
 325                         'site_stats', 'hitcounter',
 326                         'ipblocks', 'image', 'oldimage',
 327                         'recentchanges',
 328                         'watchlist', 'math', 'searchindex',
 329                         'interwiki', 'querycache',
 330                         'objectcache'
 331                 );
 332         }
 333
 334         /**
 335          * Set up a temporary set of wiki tables to work with for the tests.
 336          * Currently this will only be done once per run, and any changes to
 337          * the db will be visible to later tests in the run.
 338          *
 339          * @access private
 340          */
 341         function setupDatabase() {
 342                 static $setupDB = false;
 343                 global $wgDBprefix;
 344
 345                 # Make sure we don't mess with the live DB
 346                 if (!$setupDB && $wgDBprefix === 'parsertest') {
 347                         # oh teh horror
 348                         $GLOBALS['wgLoadBalancer'] = LoadBalancer::newFromParams( $GLOBALS['wgDBservers'] );
 349                         $db =& wfGetDB( DB_MASTER );
 350
 351                         $tables = $this->listTables();
 352
 353                         if (!(strcmp($db->getServerVersion(), '4.1') < 0 and stristr($db->getSoftwareLink(), 'MySQL'))) {
 354                                 # Database that supports CREATE TABLE ... LIKE
 355                                 global $wgDBtype;
 356                                 if( $wgDBtype == 'PostgreSQL' ) {
 357                                         $def = 'INCLUDING DEFAULTS';
 358                                 } else {
 359                                         $def = '';
 360                                 }
 361                                 foreach ($tables as $tbl) {
 362                                         $newTableName = $db->tableName( $tbl );
 363                                         $tableName = $this->oldTableNames[$tbl];
 364                                         $db->query("CREATE TEMPORARY TABLE $newTableName (LIKE $tableName $def)");
 365                                 }
 366                         } else {
 367                                 # Hack for MySQL versions < 4.1, which don't support
 368                                 # "CREATE TABLE ... LIKE". Note that
 369                                 # "CREATE TEMPORARY TABLE ... SELECT * FROM ... LIMIT 0"
 370                                 # would not create the indexes we need....
 371                                 foreach ($tables as $tbl) {
 372                                         $res = $db->query("SHOW CREATE TABLE {$this->oldTableNames[$tbl]}");
 373                                         $row = $db->fetchRow($res);
 374                                         $create = $row[1];
 375                                         $create_tmp = preg_replace('/CREATE TABLE `(.*?)`/', 'CREATE TEMPORARY TABLE `'
 376                                                 . $wgDBprefix . $tbl .'`', $create);
 377                                         if ($create === $create_tmp) {
 378                                                 # Couldn't do replacement
 379                                                 die("could not create temporary table $tbl");
 380                                         }
 381                                         $db->query($create_tmp);
 382                                 }
 383
 384                         }
 385
 386                         # Hack: insert a few Wikipedia in-project interwiki prefixes,
 387                         # for testing inter-language links
 388                         $db->insert( 'interwiki', array(
 389                                 array( 'iw_prefix' => 'Wikipedia',
 390                                        'iw_url'    => 'http://en.wikipedia.org/wiki/$1',
 391                                        'iw_local'  => 0 ),
 392                                 array( 'iw_prefix' => 'MeatBall',
 393                                        'iw_url'    => 'http://www.usemod.com/cgi-bin/mb.pl?$1',
 394                                        'iw_local'  => 0 ),
 395                                 array( 'iw_prefix' => 'zh',
 396                                        'iw_url'    => 'http://zh.wikipedia.org/wiki/$1',
 397                                        'iw_local'  => 1 ),
 398                                 array( 'iw_prefix' => 'es',
 399                                        'iw_url'    => 'http://es.wikipedia.org/wiki/$1',
 400                                        'iw_local'  => 1 ),
 401                                 array( 'iw_prefix' => 'fr',
 402                                        'iw_url'    => 'http://fr.wikipedia.org/wiki/$1',
 403                                        'iw_local'  => 1 ),
 404                                 array( 'iw_prefix' => 'ru',
 405                                        'iw_url'    => 'http://ru.wikipedia.org/wiki/$1',
 406                                        'iw_local'  => 1 ),
 407                                 ) );
 408
 409                         # Hack: Insert an image to work with
 410                         $db->insert( 'image', array(
 411                                 'img_name'        => 'Foobar.jpg',
 412                                 'img_size'        => 12345,
 413                                 'img_description' => 'Some lame file',
 414                                 'img_user'        => 1,
 415                                 'img_user_text'   => 'WikiSysop',
 416                                 'img_timestamp'   => $db->timestamp( '20010115123500' ),
 417                                 'img_width'       => 1941,
 418                                 'img_height'      => 220,
 419                                 'img_bits'        => 24,
 420                                 'img_media_type'  => MEDIATYPE_BITMAP,
 421                                 'img_major_mime'  => "image",
 422                                 'img_minor_mime'  => "jpeg",
 423                                 ) );
 424
 425                         $setupDB = true;
 426                 }
 427         }
 428
 429         /**
 430          * Create a dummy uploads directory which will contain a couple
 431          * of files in order to pass existence tests.
 432          * @return string The directory
 433          * @access private
 434          */
 435         function setupUploadDir() {
 436                 global $IP;
 437
 438                 $dir = wfTempDir() . "/mwParser-" . mt_rand() . "-images";
 439                 mkdir( $dir );
 440                 mkdir( $dir . '/3' );
 441                 mkdir( $dir . '/3/3a' );
 442
 443                 $img = "$IP/skins/monobook/headbg.jpg";
 444                 $h = fopen($img, 'r');
 445                 $c = fread($h, filesize($img));
 446                 fclose($h);
 447
 448                 $f = fopen( $dir . '/3/3a/Foobar.jpg', 'wb' );
 449                 fwrite( $f, $c );
 450                 fclose( $f );
 451                 return $dir;
 452         }
 453
 454         /**
 455          * Restore default values and perform any necessary clean-up
 456          * after each test runs.
 457          *
 458          * @access private
 459          */
 460         function teardownGlobals() {
 461                 foreach( $this->savedGlobals as $var => $val ) {
 462                         $GLOBALS[$var] = $val;
 463                 }
 464                 if( isset( $this->uploadDir ) ) {
 465                         $this->teardownUploadDir( $this->uploadDir );
 466                         unset( $this->uploadDir );
 467                 }
 468         }
 469
 470         /**
 471          * Remove the dummy uploads directory
 472          * @access private
 473          */
 474         function teardownUploadDir( $dir ) {
 475                 unlink( "$dir/3/3a/Foobar.jpg" );
 476                 rmdir( "$dir/3/3a" );
 477                 rmdir( "$dir/3" );
 478
 479                 @unlink( "$dir/thumb/3/3a/Foobar.jpg/180px-Foobar.jpg" );
 480                 @rmdir( "$dir/thumb/3/3a/Foobar.jpg" );
 481                 @rmdir( "$dir/thumb/3/3a" );
 482                 @rmdir( "$dir/thumb/3/39" ); # wtf?
 483                 @rmdir( "$dir/thumb/3" );
 484                 @rmdir( "$dir/thumb" );
 485                 rmdir( "$dir" );
 486         }
 487
 488         /**
 489          * "Running test $desc..."
 490          * @access private
 491          */
 492         function showTesting( $desc ) {
 493                 print "Running test $desc... ";
 494         }
 495
 496         /**
 497          * Print a happy success message.
 498          *
 499          * @param string $desc The test name
 500          * @return bool
 501          * @access private
 502          */
 503         function showSuccess( $desc ) {
 504                 if( !$this->quiet ) {
 505                         print $this->termColor( '1;32' ) . 'PASSED' . $this->termReset() . "\n";
 506                 }
 507                 return true;
 508         }
 509
 510         /**
 511          * Print a failure message and provide some explanatory output
 512          * about what went wrong if so configured.
 513          *
 514          * @param string $desc The test name
 515          * @param string $result Expected HTML output
 516          * @param string $html Actual HTML output
 517          * @return bool
 518          * @access private
 519          */
 520         function showFailure( $desc, $result, $html ) {
 521                 if( $this->quiet ) {
 522                         # In quiet mode we didn't show the 'Testing' message before the
 523                         # test, in case it succeeded. Show it now:
 524                         $this->showTesting( $desc );
 525                 }
 526                 print $this->termColor( '1;31' ) . 'FAILED!' . $this->termReset() . "\n";
 527                 if( $this->showDiffs ) {
 528                         print $this->quickDiff( $result, $html );
 529                 }
 530                 if( !$this->wellFormed( $html ) ) {
 531                         print "XML error: $this->mXmlError\n";
 532                 }
 533                 return false;
 534         }
 535
 536         /**
 537          * Run given strings through a diff and return the (colorized) output.
 538          * Requires writable /tmp directory and a 'diff' command in the PATH.
 539          *
 540          * @param string $input
 541          * @param string $output
 542          * @param string $inFileTail Tailing for the input file name
 543          * @param string $outFileTail Tailing for the output file name
 544          * @return string
 545          * @access private
 546          */
 547         function quickDiff( $input, $output, $inFileTail='expected', $outFileTail='actual' ) {
 548                 $prefix = wfTempDir() . "/mwParser-" . mt_rand();
 549
 550                 $infile = "$prefix-$inFileTail";
 551                 $this->dumpToFile( $input, $infile );
 552
 553                 $outfile = "$prefix-$outFileTail";
 554                 $this->dumpToFile( $output, $outfile );
 555
 556                 $diff = `diff -au $infile $outfile`;
 557                 unlink( $infile );
 558                 unlink( $outfile );
 559
 560                 return $this->colorDiff( $diff );
 561         }
 562
 563         /**
 564          * Write the given string to a file, adding a final newline.
 565          *
 566          * @param string $data
 567          * @param string $filename
 568          * @access private
 569          */
 570         function dumpToFile( $data, $filename ) {
 571                 $file = fopen( $filename, "wt" );
 572                 fwrite( $file, $data . "\n" );
 573                 fclose( $file );
 574         }
 575
 576         /**
 577          * Return ANSI terminal escape code for changing text attribs/color,
 578          * or empty string if color output is disabled.
 579          *
 580          * @param string $color Semicolon-separated list of attribute/color codes
 581          * @return string
 582          * @access private
 583          */
 584         function termColor( $color ) {
 585                 if($this->lightcolor) {
 586                         return $this->color ? "\x1b[1;{$color}m" : '';
 587                 } else {
 588                         return $this->color ? "\x1b[{$color}m" : '';
 589                 }
 590         }
 591
 592         /**
 593          * Return ANSI terminal escape code for restoring default text attributes,
 594          * or empty string if color output is disabled.
 595          *
 596          * @return string
 597          * @access private
 598          */
 599         function termReset() {
 600                 return $this->color ? "\x1b[0m" : '';
 601         }
 602
 603         /**
 604          * Colorize unified diff output if set for ANSI color output.
 605          * Subtractions are colored blue, additions red.
 606          *
 607          * @param string $text
 608          * @return string
 609          * @access private
 610          */
 611         function colorDiff( $text ) {
 612                 return preg_replace(
 613                         array( '/^(-.*)$/m', '/^(\+.*)$/m' ),
 614                         array( $this->termColor( 34 ) . '$1' . $this->termReset(),
 615                                $this->termColor( 31 ) . '$1' . $this->termReset() ),
 616                         $text );
 617         }
 618
 619         /**
 620          * Insert a temporary test article
 621          * @param string $name the title, including any prefix
 622          * @param string $text the article text
 623          * @param int $line the input line number, for reporting errors
 624          * @static
 625          * @access private
 626          */
 627         function addArticle($name, $text, $line) {
 628                 $this->setupGlobals();
 629                 $title = Title::newFromText( $name );
 630                 if ( is_null($title) ) {
 631                         die( "invalid title at line $line\n" );
 632                 }
 633
 634                 $aid = $title->getArticleID( GAID_FOR_UPDATE );
 635                 if ($aid != 0) {
 636                         die( "duplicate article at line $line\n" );
 637                 }
 638
 639                 $art = new Article($title);
 640                 $art->insertNewArticle($text, '', false, false );
 641                 $this->teardownGlobals();
 642         }
 643
 644         /*
 645          * Run the "tidy" command on text if the $wgUseTidy
 646          * global is true
 647          *
 648          * @param string $text the text to tidy
 649          * @return string
 650          * @static
 651          * @access private
 652          */
 653         function tidy( $text ) {
 654                 global $wgUseTidy;
 655                 if ($wgUseTidy) {
 656                         $text = Parser::tidy($text);
 657                 }
 658                 return $text;
 659         }
 660
 661         function wellFormed( $text ) {
 662                 $html =
 663                         '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" ' .
 664                         '"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">' .
 665                         '<html>' .
 666                         $text .
 667                         '</html>';
 668
 669                 $parser = xml_parser_create( "UTF-8" );
 670
 671                 # case folding violates XML standard, turn it off
 672                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 673
 674                 if( !xml_parse( $parser, $html, true ) ) {
 675                         $err = xml_error_string( xml_get_error_code( $parser ) );
 676                         $position = xml_get_current_byte_index( $parser );
 677                         $fragment = $this->extractFragment( $html, $position );
 678                         $this->mXmlError = "$err at byte $position:\n$fragment";
 679                         xml_parser_free( $parser );
 680                         return false;
 681                 }
 682                 xml_parser_free( $parser );
 683                 return true;
 684         }
 685
 686         function extractFragment( $text, $position ) {
 687                 $start = max( 0, $position - 10 );
 688                 $before = $position - $start;
 689                 $fragment = '...' .
 690                         $this->termColor( 34 ) .
 691                         substr( $text, $start, $before ) .
 692                         $this->termColor( 0 ) .
 693                         $this->termColor( 31 ) .
 694                         $this->termColor( 1 ) .
 695                         substr( $text, $position, 1 ) .
 696                         $this->termColor( 0 ) .
 697                         $this->termColor( 34 ) .
 698                         substr( $text, $position + 1, 9 ) .
 699                         $this->termColor( 0 ) .
 700                         '...';
 701                 $display = str_replace( "\n", ' ', $fragment );
 702                 $caret = '   ' .
 703                         str_repeat( ' ', $before ) .
 704                         $this->termColor( 31 ) .
 705                         '^' .
 706                         $this->termColor( 0 );
 707                 return "$display\n$caret";
 708         }
 709
 710 }
 711
 712 ?>