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                 $noxml = (bool)preg_match( '~\\b noxml \\b~x', $opts );
 241
 242                 $parser =& new Parser();
 243                 wfRunHooks( 'ParserTestParser', array( &$parser ) );
 244                 $title =& Title::makeTitle( NS_MAIN, $titleText );
 245
 246                 if (preg_match('/\\bpst\\b/i', $opts)) {
 247                         $out = $parser->preSaveTransform( $input, $title, $user, $options );
 248                 } elseif (preg_match('/\\bmsg\\b/i', $opts)) {
 249                         $out = $parser->transformMsg( $input, $options );
 250                 } else {
 251                         $output = $parser->parse( $input, $title, $options );
 252                         $out = $output->getText();
 253
 254                         if (preg_match('/\\bill\\b/i', $opts)) {
 255                                 $out = $this->tidy( implode( ' ', $output->getLanguageLinks() ) );
 256                         } else if (preg_match('/\\bcat\\b/i', $opts)) {
 257                                 $out = $this->tidy ( implode( ' ', $output->getCategoryLinks() ) );
 258                         }
 259
 260                         $result = $this->tidy($result);
 261                 }
 262
 263                 $this->teardownGlobals();
 264
 265                 if( $result === $out && ( $noxml === true || $this->wellFormed( $out ) ) ) {
 266                         return $this->showSuccess( $desc );
 267                 } else {
 268                         return $this->showFailure( $desc, $result, $out );
 269                 }
 270         }
 271
 272         /**
 273          * Set up the global variables for a consistent environment for each test.
 274          * Ideally this should replace the global configuration entirely.
 275          *
 276          * @access private
 277          */
 278         function setupGlobals($opts = '') {
 279                 # Save the prefixed / quoted table names for later use when we make the temporaries.
 280                 $db =& wfGetDB( DB_READ );
 281                 $this->oldTableNames = array();
 282                 foreach( $this->listTables() as $table ) {
 283                         $this->oldTableNames[$table] = $db->tableName( $table );
 284                 }
 285                 if( !isset( $this->uploadDir ) ) {
 286                         $this->uploadDir = $this->setupUploadDir();
 287                 }
 288
 289                 $settings = array(
 290                         'wgServer' => 'http://localhost',
 291                         'wgScript' => '/index.php',
 292                         'wgScriptPath' => '/',
 293                         'wgArticlePath' => '/wiki/$1',
 294                         'wgUploadPath' => 'http://example.com/images',
 295                         'wgUploadDirectory' => $this->uploadDir,
 296                         'wgStyleSheetPath' => '/skins',
 297                         'wgSitename' => 'MediaWiki',
 298                         'wgLanguageCode' => 'en',
 299                         'wgContLanguageCode' => 'en',
 300                         'wgDBprefix' => 'parsertest',
 301                         'wgDefaultUserOptions' => array(),
 302
 303                         'wgLang' => new LanguageUtf8(),
 304                         'wgContLang' => new LanguageUtf8(),
 305                         'wgNamespacesWithSubpages' => array( 0 => preg_match('/\\bsubpage\\b/i', $opts)),
 306                         'wgMaxTocLevel' => 999,
 307                         'wgCapitalLinks' => true,
 308                         'wgDefaultUserOptions' => array(),
 309                         'wgNoFollowLinks' => true,
 310                         'wgThumbnailScriptPath' => false,
 311                         'wgUseTeX' => false,
 312                         );
 313                 $this->savedGlobals = array();
 314                 foreach( $settings as $var => $val ) {
 315                         $this->savedGlobals[$var] = $GLOBALS[$var];
 316                         $GLOBALS[$var] = $val;
 317                 }
 318                 $GLOBALS['wgLoadBalancer']->loadMasterPos();
 319                 $GLOBALS['wgMessageCache']->initialise( new BagOStuff(), false, 0, $GLOBALS['wgDBname'] );
 320                 $this->setupDatabase();
 321
 322                 global $wgUser;
 323                 $wgUser = new User();
 324         }
 325
 326         # List of temporary tables to create, without prefix
 327         # Some of these probably aren't necessary
 328         function listTables() {
 329                 return array('user', 'page', 'revision', 'text',
 330                         'pagelinks', 'imagelinks', 'categorylinks', 'templatelinks',
 331                         'site_stats', 'hitcounter',
 332                         'ipblocks', 'image', 'oldimage',
 333                         'recentchanges',
 334                         'watchlist', 'math', 'searchindex',
 335                         'interwiki', 'querycache',
 336                         'objectcache'
 337                 );
 338         }
 339
 340         /**
 341          * Set up a temporary set of wiki tables to work with for the tests.
 342          * Currently this will only be done once per run, and any changes to
 343          * the db will be visible to later tests in the run.
 344          *
 345          * @access private
 346          */
 347         function setupDatabase() {
 348                 static $setupDB = false;
 349                 global $wgDBprefix;
 350
 351                 # Make sure we don't mess with the live DB
 352                 if (!$setupDB && $wgDBprefix === 'parsertest') {
 353                         # oh teh horror
 354                         $GLOBALS['wgLoadBalancer'] = LoadBalancer::newFromParams( $GLOBALS['wgDBservers'] );
 355                         $db =& wfGetDB( DB_MASTER );
 356
 357                         $tables = $this->listTables();
 358
 359                         if (!(strcmp($db->getServerVersion(), '4.1') < 0 and stristr($db->getSoftwareLink(), 'MySQL'))) {
 360                                 # Database that supports CREATE TABLE ... LIKE
 361                                 global $wgDBtype;
 362                                 if( $wgDBtype == 'PostgreSQL' ) {
 363                                         $def = 'INCLUDING DEFAULTS';
 364                                 } else {
 365                                         $def = '';
 366                                 }
 367                                 foreach ($tables as $tbl) {
 368                                         $newTableName = $db->tableName( $tbl );
 369                                         $tableName = $this->oldTableNames[$tbl];
 370                                         $db->query("CREATE TEMPORARY TABLE $newTableName (LIKE $tableName $def)");
 371                                 }
 372                         } else {
 373                                 # Hack for MySQL versions < 4.1, which don't support
 374                                 # "CREATE TABLE ... LIKE". Note that
 375                                 # "CREATE TEMPORARY TABLE ... SELECT * FROM ... LIMIT 0"
 376                                 # would not create the indexes we need....
 377                                 foreach ($tables as $tbl) {
 378                                         $res = $db->query("SHOW CREATE TABLE {$this->oldTableNames[$tbl]}");
 379                                         $row = $db->fetchRow($res);
 380                                         $create = $row[1];
 381                                         $create_tmp = preg_replace('/CREATE TABLE `(.*?)`/', 'CREATE TEMPORARY TABLE `'
 382                                                 . $wgDBprefix . $tbl .'`', $create);
 383                                         if ($create === $create_tmp) {
 384                                                 # Couldn't do replacement
 385                                                 die("could not create temporary table $tbl");
 386                                         }
 387                                         $db->query($create_tmp);
 388                                 }
 389
 390                         }
 391
 392                         # Hack: insert a few Wikipedia in-project interwiki prefixes,
 393                         # for testing inter-language links
 394                         $db->insert( 'interwiki', array(
 395                                 array( 'iw_prefix' => 'Wikipedia',
 396                                        'iw_url'    => 'http://en.wikipedia.org/wiki/$1',
 397                                        'iw_local'  => 0 ),
 398                                 array( 'iw_prefix' => 'MeatBall',
 399                                        'iw_url'    => 'http://www.usemod.com/cgi-bin/mb.pl?$1',
 400                                        'iw_local'  => 0 ),
 401                                 array( 'iw_prefix' => 'zh',
 402                                        'iw_url'    => 'http://zh.wikipedia.org/wiki/$1',
 403                                        'iw_local'  => 1 ),
 404                                 array( 'iw_prefix' => 'es',
 405                                        'iw_url'    => 'http://es.wikipedia.org/wiki/$1',
 406                                        'iw_local'  => 1 ),
 407                                 array( 'iw_prefix' => 'fr',
 408                                        'iw_url'    => 'http://fr.wikipedia.org/wiki/$1',
 409                                        'iw_local'  => 1 ),
 410                                 array( 'iw_prefix' => 'ru',
 411                                        'iw_url'    => 'http://ru.wikipedia.org/wiki/$1',
 412                                        'iw_local'  => 1 ),
 413                                 ) );
 414
 415                         # Hack: Insert an image to work with
 416                         $db->insert( 'image', array(
 417                                 'img_name'        => 'Foobar.jpg',
 418                                 'img_size'        => 12345,
 419                                 'img_description' => 'Some lame file',
 420                                 'img_user'        => 1,
 421                                 'img_user_text'   => 'WikiSysop',
 422                                 'img_timestamp'   => $db->timestamp( '20010115123500' ),
 423                                 'img_width'       => 1941,
 424                                 'img_height'      => 220,
 425                                 'img_bits'        => 24,
 426                                 'img_media_type'  => MEDIATYPE_BITMAP,
 427                                 'img_major_mime'  => "image",
 428                                 'img_minor_mime'  => "jpeg",
 429                                 ) );
 430
 431                         $setupDB = true;
 432                 }
 433         }
 434
 435         /**
 436          * Create a dummy uploads directory which will contain a couple
 437          * of files in order to pass existence tests.
 438          * @return string The directory
 439          * @access private
 440          */
 441         function setupUploadDir() {
 442                 global $IP;
 443
 444                 $dir = wfTempDir() . "/mwParser-" . mt_rand() . "-images";
 445                 mkdir( $dir );
 446                 mkdir( $dir . '/3' );
 447                 mkdir( $dir . '/3/3a' );
 448
 449                 $img = "$IP/skins/monobook/headbg.jpg";
 450                 $h = fopen($img, 'r');
 451                 $c = fread($h, filesize($img));
 452                 fclose($h);
 453
 454                 $f = fopen( $dir . '/3/3a/Foobar.jpg', 'wb' );
 455                 fwrite( $f, $c );
 456                 fclose( $f );
 457                 return $dir;
 458         }
 459
 460         /**
 461          * Restore default values and perform any necessary clean-up
 462          * after each test runs.
 463          *
 464          * @access private
 465          */
 466         function teardownGlobals() {
 467                 foreach( $this->savedGlobals as $var => $val ) {
 468                         $GLOBALS[$var] = $val;
 469                 }
 470                 if( isset( $this->uploadDir ) ) {
 471                         $this->teardownUploadDir( $this->uploadDir );
 472                         unset( $this->uploadDir );
 473                 }
 474         }
 475
 476         /**
 477          * Remove the dummy uploads directory
 478          * @access private
 479          */
 480         function teardownUploadDir( $dir ) {
 481                 unlink( "$dir/3/3a/Foobar.jpg" );
 482                 rmdir( "$dir/3/3a" );
 483                 rmdir( "$dir/3" );
 484
 485                 @unlink( "$dir/thumb/3/3a/Foobar.jpg/180px-Foobar.jpg" );
 486                 @rmdir( "$dir/thumb/3/3a/Foobar.jpg" );
 487                 @rmdir( "$dir/thumb/3/3a" );
 488                 @rmdir( "$dir/thumb/3/39" ); # wtf?
 489                 @rmdir( "$dir/thumb/3" );
 490                 @rmdir( "$dir/thumb" );
 491                 rmdir( "$dir" );
 492         }
 493
 494         /**
 495          * "Running test $desc..."
 496          * @access private
 497          */
 498         function showTesting( $desc ) {
 499                 print "Running test $desc... ";
 500         }
 501
 502         /**
 503          * Print a happy success message.
 504          *
 505          * @param string $desc The test name
 506          * @return bool
 507          * @access private
 508          */
 509         function showSuccess( $desc ) {
 510                 if( !$this->quiet ) {
 511                         print $this->termColor( '1;32' ) . 'PASSED' . $this->termReset() . "\n";
 512                 }
 513                 return true;
 514         }
 515
 516         /**
 517          * Print a failure message and provide some explanatory output
 518          * about what went wrong if so configured.
 519          *
 520          * @param string $desc The test name
 521          * @param string $result Expected HTML output
 522          * @param string $html Actual HTML output
 523          * @return bool
 524          * @access private
 525          */
 526         function showFailure( $desc, $result, $html ) {
 527                 if( $this->quiet ) {
 528                         # In quiet mode we didn't show the 'Testing' message before the
 529                         # test, in case it succeeded. Show it now:
 530                         $this->showTesting( $desc );
 531                 }
 532                 print $this->termColor( '1;31' ) . 'FAILED!' . $this->termReset() . "\n";
 533                 if( $this->showDiffs ) {
 534                         print $this->quickDiff( $result, $html );
 535                 }
 536                 if( !$this->wellFormed( $html ) ) {
 537                         print "XML error: $this->mXmlError\n";
 538                 }
 539                 return false;
 540         }
 541
 542         /**
 543          * Run given strings through a diff and return the (colorized) output.
 544          * Requires writable /tmp directory and a 'diff' command in the PATH.
 545          *
 546          * @param string $input
 547          * @param string $output
 548          * @param string $inFileTail Tailing for the input file name
 549          * @param string $outFileTail Tailing for the output file name
 550          * @return string
 551          * @access private
 552          */
 553         function quickDiff( $input, $output, $inFileTail='expected', $outFileTail='actual' ) {
 554                 $prefix = wfTempDir() . "/mwParser-" . mt_rand();
 555
 556                 $infile = "$prefix-$inFileTail";
 557                 $this->dumpToFile( $input, $infile );
 558
 559                 $outfile = "$prefix-$outFileTail";
 560                 $this->dumpToFile( $output, $outfile );
 561
 562                 $diff = `diff -au $infile $outfile`;
 563                 unlink( $infile );
 564                 unlink( $outfile );
 565
 566                 return $this->colorDiff( $diff );
 567         }
 568
 569         /**
 570          * Write the given string to a file, adding a final newline.
 571          *
 572          * @param string $data
 573          * @param string $filename
 574          * @access private
 575          */
 576         function dumpToFile( $data, $filename ) {
 577                 $file = fopen( $filename, "wt" );
 578                 fwrite( $file, $data . "\n" );
 579                 fclose( $file );
 580         }
 581
 582         /**
 583          * Return ANSI terminal escape code for changing text attribs/color,
 584          * or empty string if color output is disabled.
 585          *
 586          * @param string $color Semicolon-separated list of attribute/color codes
 587          * @return string
 588          * @access private
 589          */
 590         function termColor( $color ) {
 591                 if($this->lightcolor) {
 592                         return $this->color ? "\x1b[1;{$color}m" : '';
 593                 } else {
 594                         return $this->color ? "\x1b[{$color}m" : '';
 595                 }
 596         }
 597
 598         /**
 599          * Return ANSI terminal escape code for restoring default text attributes,
 600          * or empty string if color output is disabled.
 601          *
 602          * @return string
 603          * @access private
 604          */
 605         function termReset() {
 606                 return $this->color ? "\x1b[0m" : '';
 607         }
 608
 609         /**
 610          * Colorize unified diff output if set for ANSI color output.
 611          * Subtractions are colored blue, additions red.
 612          *
 613          * @param string $text
 614          * @return string
 615          * @access private
 616          */
 617         function colorDiff( $text ) {
 618                 return preg_replace(
 619                         array( '/^(-.*)$/m', '/^(\+.*)$/m' ),
 620                         array( $this->termColor( 34 ) . '$1' . $this->termReset(),
 621                                $this->termColor( 31 ) . '$1' . $this->termReset() ),
 622                         $text );
 623         }
 624
 625         /**
 626          * Insert a temporary test article
 627          * @param string $name the title, including any prefix
 628          * @param string $text the article text
 629          * @param int $line the input line number, for reporting errors
 630          * @static
 631          * @access private
 632          */
 633         function addArticle($name, $text, $line) {
 634                 $this->setupGlobals();
 635                 $title = Title::newFromText( $name );
 636                 if ( is_null($title) ) {
 637                         die( "invalid title at line $line\n" );
 638                 }
 639
 640                 $aid = $title->getArticleID( GAID_FOR_UPDATE );
 641                 if ($aid != 0) {
 642                         die( "duplicate article at line $line\n" );
 643                 }
 644
 645                 $art = new Article($title);
 646                 $art->insertNewArticle($text, '', false, false );
 647                 $this->teardownGlobals();
 648         }
 649
 650         /*
 651          * Run the "tidy" command on text if the $wgUseTidy
 652          * global is true
 653          *
 654          * @param string $text the text to tidy
 655          * @return string
 656          * @static
 657          * @access private
 658          */
 659         function tidy( $text ) {
 660                 global $wgUseTidy;
 661                 if ($wgUseTidy) {
 662                         $text = Parser::tidy($text);
 663                 }
 664                 return $text;
 665         }
 666
 667         /**
 668          * Hack up a private DOCTYPE with HTML's standard entity declarations.
 669          * PHP 4 seemed to know these if you gave it an HTML doctype, but
 670          * PHP 5.1 doesn't.
 671          * @return string
 672          * @access private
 673          */
 674         function hackDocType() {
 675                 global $wgHtmlEntities;
 676                 $out = "<!DOCTYPE html [\n";
 677                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
 678                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
 679                 }
 680                 $out .= "]>\n";
 681                 return $out;
 682         }
 683
 684         function wellFormed( $text ) {
 685                 $html =
 686                         $this->hackDocType() .
 687                         '<html>' .
 688                         $text .
 689                         '</html>';
 690
 691                 $parser = xml_parser_create( "UTF-8" );
 692
 693                 # case folding violates XML standard, turn it off
 694                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 695
 696                 if( !xml_parse( $parser, $html, true ) ) {
 697                         $err = xml_error_string( xml_get_error_code( $parser ) );
 698                         $position = xml_get_current_byte_index( $parser );
 699                         $fragment = $this->extractFragment( $html, $position );
 700                         $this->mXmlError = "$err at byte $position:\n$fragment";
 701                         xml_parser_free( $parser );
 702                         return false;
 703                 }
 704                 xml_parser_free( $parser );
 705                 return true;
 706         }
 707
 708         function extractFragment( $text, $position ) {
 709                 $start = max( 0, $position - 10 );
 710                 $before = $position - $start;
 711                 $fragment = '...' .
 712                         $this->termColor( 34 ) .
 713                         substr( $text, $start, $before ) .
 714                         $this->termColor( 0 ) .
 715                         $this->termColor( 31 ) .
 716                         $this->termColor( 1 ) .
 717                         substr( $text, $position, 1 ) .
 718                         $this->termColor( 0 ) .
 719                         $this->termColor( 34 ) .
 720                         substr( $text, $position + 1, 9 ) .
 721                         $this->termColor( 0 ) .
 722                         '...';
 723                 $display = str_replace( "\n", ' ', $fragment );
 724                 $caret = '   ' .
 725                         str_repeat( ' ', $before ) .
 726                         $this->termColor( 31 ) .
 727                         '^' .
 728                         $this->termColor( 0 );
 729                 return "$display\n$caret";
 730         }
 731
 732 }
 733
 734 ?>