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