tests/phpunit/includes/parser/NewParserTest.php

   1 <?php
   2
   3 /**
   4  * Although marked as a stub, can work independently.
   5  *
   6  * @group Database
   7  * @group Parser
   8  * @group Stub
   9  *
  10  * @todo covers tags
  11  */
  12 class NewParserTest extends MediaWikiTestCase {
  13         static protected $articles = []; // Array of test articles defined by the tests
  14         /* The data provider is run on a different instance than the test, so it must be static
  15          * When running tests from several files, all tests will see all articles.
  16          */
  17         static protected $backendToUse;
  18
  19         public $keepUploads = false;
  20         public $runDisabled = false;
  21         public $runParsoid = false;
  22         public $regex = '';
  23         public $showProgress = true;
  24         public $savedWeirdGlobals = [];
  25         public $savedGlobals = [];
  26         public $hooks = [];
  27         public $functionHooks = [];
  28         public $transparentHooks = [];
  29
  30         // Fuzz test
  31         public $maxFuzzTestLength = 300;
  32         public $fuzzSeed = 0;
  33         public $memoryLimit = 50;
  34
  35         /**
  36          * @var DjVuSupport
  37          */
  38         private $djVuSupport;
  39         /**
  40          * @var TidySupport
  41          */
  42         private $tidySupport;
  43
  44         protected $file = false;
  45
  46         public static function setUpBeforeClass() {
  47                 // Inject ParserTest well-known interwikis
  48                 ParserTest::setupInterwikis();
  49         }
  50
  51         protected function setUp() {
  52                 global $wgNamespaceAliases, $wgContLang;
  53                 global $wgHooks, $IP;
  54
  55                 parent::setUp();
  56
  57                 // Setup CLI arguments
  58                 if ( $this->getCliArg( 'regex' ) ) {
  59                         $this->regex = $this->getCliArg( 'regex' );
  60                 } else {
  61                         # Matches anything
  62                         $this->regex = '';
  63                 }
  64
  65                 $this->keepUploads = $this->getCliArg( 'keep-uploads' );
  66
  67                 $tmpGlobals = [];
  68
  69                 $tmpGlobals['wgLanguageCode'] = 'en';
  70                 $tmpGlobals['wgContLang'] = Language::factory( 'en' );
  71                 $tmpGlobals['wgSitename'] = 'MediaWiki';
  72                 $tmpGlobals['wgServer'] = 'http://example.org';
  73                 $tmpGlobals['wgServerName'] = 'example.org';
  74                 $tmpGlobals['wgScript'] = '/index.php';
  75                 $tmpGlobals['wgScriptPath'] = '/';
  76                 $tmpGlobals['wgArticlePath'] = '/wiki/$1';
  77                 $tmpGlobals['wgActionPaths'] = [];
  78                 $tmpGlobals['wgVariantArticlePath'] = false;
  79                 $tmpGlobals['wgExtensionAssetsPath'] = '/extensions';
  80                 $tmpGlobals['wgStylePath'] = '/skins';
  81                 $tmpGlobals['wgEnableUploads'] = true;
  82                 $tmpGlobals['wgUploadNavigationUrl'] = false;
  83                 $tmpGlobals['wgThumbnailScriptPath'] = false;
  84                 $tmpGlobals['wgLocalFileRepo'] = [
  85                         'class' => 'LocalRepo',
  86                         'name' => 'local',
  87                         'url' => 'http://example.com/images',
  88                         'hashLevels' => 2,
  89                         'transformVia404' => false,
  90                         'backend' => 'local-backend'
  91                 ];
  92                 $tmpGlobals['wgForeignFileRepos'] = [];
  93                 $tmpGlobals['wgDefaultExternalStore'] = [];
  94                 $tmpGlobals['wgParserCacheType'] = CACHE_NONE;
  95                 $tmpGlobals['wgCapitalLinks'] = true;
  96                 $tmpGlobals['wgNoFollowLinks'] = true;
  97                 $tmpGlobals['wgNoFollowDomainExceptions'] = [];
  98                 $tmpGlobals['wgExternalLinkTarget'] = false;
  99                 $tmpGlobals['wgThumbnailScriptPath'] = false;
 100                 $tmpGlobals['wgUseImageResize'] = true;
 101                 $tmpGlobals['wgAllowExternalImages'] = true;
 102                 $tmpGlobals['wgRawHtml'] = false;
 103                 $tmpGlobals['wgWellFormedXml'] = true;
 104                 $tmpGlobals['wgAllowMicrodataAttributes'] = true;
 105                 $tmpGlobals['wgExperimentalHtmlIds'] = false;
 106                 $tmpGlobals['wgAdaptiveMessageCache'] = true;
 107                 $tmpGlobals['wgUseDatabaseMessages'] = true;
 108                 $tmpGlobals['wgLocaltimezone'] = 'UTC';
 109                 $tmpGlobals['wgGroupPermissions'] = [
 110                         '*' => [
 111                                 'createaccount' => true,
 112                                 'read' => true,
 113                                 'edit' => true,
 114                                 'createpage' => true,
 115                                 'createtalk' => true,
 116                 ] ];
 117                 $tmpGlobals['wgNamespaceProtection'] = [ NS_MEDIAWIKI => 'editinterface' ];
 118
 119                 $tmpGlobals['wgParser'] = new StubObject(
 120                         'wgParser', $GLOBALS['wgParserConf']['class'],
 121                         [ $GLOBALS['wgParserConf'] ] );
 122
 123                 $tmpGlobals['wgFileExtensions'][] = 'svg';
 124                 $tmpGlobals['wgSVGConverter'] = 'rsvg';
 125                 $tmpGlobals['wgSVGConverters']['rsvg'] =
 126                         '$path/rsvg-convert -w $width -h $height -o $output $input';
 127
 128                 if ( $GLOBALS['wgStyleDirectory'] === false ) {
 129                         $tmpGlobals['wgStyleDirectory'] = "$IP/skins";
 130                 }
 131
 132                 # Replace all media handlers with a mock. We do not need to generate
 133                 # actual thumbnails to do parser testing, we only care about receiving
 134                 # a ThumbnailImage properly initialized.
 135                 global $wgMediaHandlers;
 136                 foreach ( $wgMediaHandlers as $type => $handler ) {
 137                         $tmpGlobals['wgMediaHandlers'][$type] = 'MockBitmapHandler';
 138                 }
 139                 // Vector images have to be handled slightly differently
 140                 $tmpGlobals['wgMediaHandlers']['image/svg+xml'] = 'MockSvgHandler';
 141
 142                 // DjVu images have to be handled slightly differently
 143                 $tmpGlobals['wgMediaHandlers']['image/vnd.djvu'] = 'MockDjVuHandler';
 144
 145                 // Ogg video/audio increasingly more differently
 146                 $tmpGlobals['wgMediaHandlers']['application/ogg'] = 'MockOggHandler';
 147
 148                 $tmpHooks = $wgHooks;
 149                 $tmpHooks['ParserTestParser'][] = 'ParserTestParserHook::setup';
 150                 $tmpHooks['ParserGetVariableValueTs'][] = 'ParserTest::getFakeTimestamp';
 151                 $tmpGlobals['wgHooks'] = $tmpHooks;
 152                 # add a namespace shadowing a interwiki link, to test
 153                 # proper precedence when resolving links. (bug 51680)
 154                 $tmpGlobals['wgExtraNamespaces'] = [ 100 => 'MemoryAlpha' ];
 155
 156                 $tmpGlobals['wgLocalInterwikis'] = [ 'local', 'mi' ];
 157                 # "extra language links"
 158                 # see https://gerrit.wikimedia.org/r/111390
 159                 $tmpGlobals['wgExtraInterlanguageLinkPrefixes'] = [ 'mul' ];
 160
 161                 // DjVu support
 162                 $this->djVuSupport = new DjVuSupport();
 163                 // Tidy support
 164                 $this->tidySupport = new TidySupport();
 165                 $tmpGlobals['wgTidyConfig'] = null;
 166                 $tmpGlobals['wgUseTidy'] = false;
 167                 $tmpGlobals['wgDebugTidy'] = false;
 168                 $tmpGlobals['wgTidyConf'] = $IP . '/includes/tidy/tidy.conf';
 169                 $tmpGlobals['wgTidyOpts'] = '';
 170                 $tmpGlobals['wgTidyInternal'] = $this->tidySupport->isInternal();
 171
 172                 $this->setMwGlobals( $tmpGlobals );
 173
 174                 $this->savedWeirdGlobals['image_alias'] = $wgNamespaceAliases['Image'];
 175                 $this->savedWeirdGlobals['image_talk_alias'] = $wgNamespaceAliases['Image_talk'];
 176
 177                 $wgNamespaceAliases['Image'] = NS_FILE;
 178                 $wgNamespaceAliases['Image_talk'] = NS_FILE_TALK;
 179
 180                 MWNamespace::getCanonicalNamespaces( true ); # reset namespace cache
 181                 $wgContLang->resetNamespaces(); # reset namespace cache
 182         }
 183
 184         protected function tearDown() {
 185                 global $wgNamespaceAliases, $wgContLang;
 186
 187                 $wgNamespaceAliases['Image'] = $this->savedWeirdGlobals['image_alias'];
 188                 $wgNamespaceAliases['Image_talk'] = $this->savedWeirdGlobals['image_talk_alias'];
 189
 190                 MWTidy::destroySingleton();
 191
 192                 // Restore backends
 193                 RepoGroup::destroySingleton();
 194                 FileBackendGroup::destroySingleton();
 195
 196                 // Remove temporary pages from the link cache
 197                 LinkCache::singleton()->clear();
 198
 199                 // Restore message cache (temporary pages and $wgUseDatabaseMessages)
 200                 MessageCache::destroyInstance();
 201
 202                 parent::tearDown();
 203
 204                 MWNamespace::getCanonicalNamespaces( true ); # reset namespace cache
 205                 $wgContLang->resetNamespaces(); # reset namespace cache
 206         }
 207
 208         public static function tearDownAfterClass() {
 209                 ParserTest::tearDownInterwikis();
 210                 parent::tearDownAfterClass();
 211         }
 212
 213         function addDBData() {
 214                 $this->tablesUsed[] = 'site_stats';
 215                 # disabled for performance
 216                 # $this->tablesUsed[] = 'image';
 217
 218                 # Update certain things in site_stats
 219                 $this->db->insert( 'site_stats',
 220                         [ 'ss_row_id' => 1, 'ss_images' => 2, 'ss_good_articles' => 1 ],
 221                         __METHOD__
 222                 );
 223
 224                 $user = User::newFromId( 0 );
 225                 LinkCache::singleton()->clear(); # Avoids the odd failure at creating the nullRevision
 226
 227                 # Upload DB table entries for files.
 228                 # We will upload the actual files later. Note that if anything causes LocalFile::load()
 229                 # to be triggered before then, it will break via maybeUpgrade() setting the fileExists
 230                 # member to false and storing it in cache.
 231                 # note that the size/width/height/bits/etc of the file
 232                 # are actually set by inspecting the file itself; the arguments
 233                 # to recordUpload2 have no effect.  That said, we try to make things
 234                 # match up so it is less confusing to readers of the code & tests.
 235                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Foobar.jpg' ) );
 236                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 237                         $image->recordUpload2(
 238                                 '', // archive name
 239                                 'Upload of some lame file',
 240                                 'Some lame file',
 241                                 [
 242                                         'size' => 7881,
 243                                         'width' => 1941,
 244                                         'height' => 220,
 245                                         'bits' => 8,
 246                                         'media_type' => MEDIATYPE_BITMAP,
 247                                         'mime' => 'image/jpeg',
 248                                         'metadata' => serialize( [] ),
 249                                         'sha1' => Wikimedia\base_convert( '1', 16, 36, 31 ),
 250                                         'fileExists' => true ],
 251                                 $this->db->timestamp( '20010115123500' ), $user
 252                         );
 253                 }
 254
 255                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Thumb.png' ) );
 256                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 257                         $image->recordUpload2(
 258                                 '', // archive name
 259                                 'Upload of some lame thumbnail',
 260                                 'Some lame thumbnail',
 261                                 [
 262                                         'size' => 22589,
 263                                         'width' => 135,
 264                                         'height' => 135,
 265                                         'bits' => 8,
 266                                         'media_type' => MEDIATYPE_BITMAP,
 267                                         'mime' => 'image/png',
 268                                         'metadata' => serialize( [] ),
 269                                         'sha1' => Wikimedia\base_convert( '2', 16, 36, 31 ),
 270                                         'fileExists' => true ],
 271                                 $this->db->timestamp( '20130225203040' ), $user
 272                         );
 273                 }
 274
 275                 # This image will be blacklisted in [[MediaWiki:Bad image list]]
 276                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Bad.jpg' ) );
 277                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 278                         $image->recordUpload2(
 279                                 '', // archive name
 280                                 'zomgnotcensored',
 281                                 'Borderline image',
 282                                 [
 283                                         'size' => 12345,
 284                                         'width' => 320,
 285                                         'height' => 240,
 286                                         'bits' => 24,
 287                                         'media_type' => MEDIATYPE_BITMAP,
 288                                         'mime' => 'image/jpeg',
 289                                         'metadata' => serialize( [] ),
 290                                         'sha1' => Wikimedia\base_convert( '3', 16, 36, 31 ),
 291                                         'fileExists' => true ],
 292                                 $this->db->timestamp( '20010115123500' ), $user
 293                         );
 294                 }
 295                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Foobar.svg' ) );
 296                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 297                         $image->recordUpload2( '', 'Upload of some lame SVG', 'Some lame SVG', [
 298                                         'size'        => 12345,
 299                                         'width'       => 240,
 300                                         'height'      => 180,
 301                                         'bits'        => 0,
 302                                         'media_type'  => MEDIATYPE_DRAWING,
 303                                         'mime'        => 'image/svg+xml',
 304                                         'metadata'    => serialize( [] ),
 305                                         'sha1'        => Wikimedia\base_convert( '', 16, 36, 31 ),
 306                                         'fileExists'  => true
 307                         ], $this->db->timestamp( '20010115123500' ), $user );
 308                 }
 309
 310                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Video.ogv' ) );
 311                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 312                         $image->recordUpload2( '', 'A pretty movie', 'Will it play', [
 313                                         'size'        => 12345,
 314                                         'width'       => 320,
 315                                         'height'      => 240,
 316                                         'bits'        => 0,
 317                                         'media_type'  => MEDIATYPE_VIDEO,
 318                                         'mime'        => 'application/ogg',
 319                                         'metadata'    => serialize( [] ),
 320                                         'sha1'        => Wikimedia\base_convert( '', 16, 36, 32 ),
 321                                         'fileExists'  => true
 322                         ], $this->db->timestamp( '20010115123500' ), $user );
 323                 }
 324
 325                 # A DjVu file
 326                 # A DjVu file
 327                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'LoremIpsum.djvu' ) );
 328                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 329                         $image->recordUpload2( '', 'Upload a DjVu', 'A DjVu', [
 330                                 'size' => 3249,
 331                                 'width' => 2480,
 332                                 'height' => 3508,
 333                                 'bits' => 0,
 334                                 'media_type' => MEDIATYPE_BITMAP,
 335                                 'mime' => 'image/vnd.djvu',
 336                                 'metadata' => '<?xml version="1.0" ?>
 337 <!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd">
 338 <DjVuXML>
 339 <HEAD></HEAD>
 340 <BODY><OBJECT height="3508" width="2480">
 341 <PARAM name="DPI" value="300" />
 342 <PARAM name="GAMMA" value="2.2" />
 343 </OBJECT>
 344 <OBJECT height="3508" width="2480">
 345 <PARAM name="DPI" value="300" />
 346 <PARAM name="GAMMA" value="2.2" />
 347 </OBJECT>
 348 <OBJECT height="3508" width="2480">
 349 <PARAM name="DPI" value="300" />
 350 <PARAM name="GAMMA" value="2.2" />
 351 </OBJECT>
 352 <OBJECT height="3508" width="2480">
 353 <PARAM name="DPI" value="300" />
 354 <PARAM name="GAMMA" value="2.2" />
 355 </OBJECT>
 356 <OBJECT height="3508" width="2480">
 357 <PARAM name="DPI" value="300" />
 358 <PARAM name="GAMMA" value="2.2" />
 359 </OBJECT>
 360 </BODY>
 361 </DjVuXML>',
 362                                 'sha1' => Wikimedia\base_convert( '', 16, 36, 31 ),
 363                                 'fileExists' => true
 364                         ], $this->db->timestamp( '20140115123600' ), $user );
 365                 }
 366         }
 367
 368         // ParserTest setup/teardown functions
 369
 370         /**
 371          * Set up the global variables for a consistent environment for each test.
 372          * Ideally this should replace the global configuration entirely.
 373          * @param array $opts
 374          * @param string $config
 375          * @return RequestContext
 376          */
 377         protected function setupGlobals( $opts = [], $config = '' ) {
 378                 global $wgFileBackends;
 379                 # Find out values for some special options.
 380                 $lang =
 381                         self::getOptionValue( 'language', $opts, 'en' );
 382                 $variant =
 383                         self::getOptionValue( 'variant', $opts, false );
 384                 $maxtoclevel =
 385                         self::getOptionValue( 'wgMaxTocLevel', $opts, 999 );
 386                 $linkHolderBatchSize =
 387                         self::getOptionValue( 'wgLinkHolderBatchSize', $opts, 1000 );
 388
 389                 $uploadDir = $this->getUploadDir();
 390                 if ( $this->getCliArg( 'use-filebackend' ) ) {
 391                         if ( self::$backendToUse ) {
 392                                 $backend = self::$backendToUse;
 393                         } else {
 394                                 $name = $this->getCliArg( 'use-filebackend' );
 395                                 $useConfig = [];
 396                                 foreach ( $wgFileBackends as $conf ) {
 397                                         if ( $conf['name'] == $name ) {
 398                                                 $useConfig = $conf;
 399                                         }
 400                                 }
 401                                 $useConfig['name'] = 'local-backend'; // swap name
 402                                 unset( $useConfig['lockManager'] );
 403                                 unset( $useConfig['fileJournal'] );
 404                                 $class = $useConfig['class'];
 405                                 self::$backendToUse = new $class( $useConfig );
 406                                 $backend = self::$backendToUse;
 407                         }
 408                 } else {
 409                         # Replace with a mock. We do not care about generating real
 410                         # files on the filesystem, just need to expose the file
 411                         # informations.
 412                         $backend = new MockFileBackend( [
 413                                 'name' => 'local-backend',
 414                                 'wikiId' => wfWikiId()
 415                         ] );
 416                 }
 417
 418                 $settings = [
 419                         'wgLocalFileRepo' => [
 420                                 'class' => 'LocalRepo',
 421                                 'name' => 'local',
 422                                 'url' => 'http://example.com/images',
 423                                 'hashLevels' => 2,
 424                                 'transformVia404' => false,
 425                                 'backend' => $backend
 426                         ],
 427                         'wgEnableUploads' => self::getOptionValue( 'wgEnableUploads', $opts, true ),
 428                         'wgLanguageCode' => $lang,
 429                         'wgDBprefix' => $this->db->getType() != 'oracle' ? 'unittest_' : 'ut_',
 430                         'wgRawHtml' => self::getOptionValue( 'wgRawHtml', $opts, false ),
 431                         'wgNamespacesWithSubpages' => [ NS_MAIN => isset( $opts['subpage'] ) ],
 432                         'wgAllowExternalImages' => self::getOptionValue( 'wgAllowExternalImages', $opts, true ),
 433                         'wgThumbLimits' => [ self::getOptionValue( 'thumbsize', $opts, 180 ) ],
 434                         'wgMaxTocLevel' => $maxtoclevel,
 435                         'wgUseTeX' => isset( $opts['math'] ) || isset( $opts['texvc'] ),
 436                         'wgMathDirectory' => $uploadDir . '/math',
 437                         'wgDefaultLanguageVariant' => $variant,
 438                         'wgLinkHolderBatchSize' => $linkHolderBatchSize,
 439                         'wgUseTidy' => isset( $opts['tidy'] ),
 440                 ];
 441
 442                 if ( $config ) {
 443                         $configLines = explode( "\n", $config );
 444
 445                         foreach ( $configLines as $line ) {
 446                                 list( $var, $value ) = explode( '=', $line, 2 );
 447
 448                                 $settings[$var] = eval( "return $value;" ); // ???
 449                         }
 450                 }
 451
 452                 $this->savedGlobals = [];
 453
 454                 /** @since 1.20 */
 455                 Hooks::run( 'ParserTestGlobals', [ &$settings ] );
 456
 457                 $langObj = Language::factory( $lang );
 458                 $settings['wgContLang'] = $langObj;
 459                 $settings['wgLang'] = $langObj;
 460
 461                 $context = new RequestContext();
 462                 $settings['wgOut'] = $context->getOutput();
 463                 $settings['wgUser'] = $context->getUser();
 464                 $settings['wgRequest'] = $context->getRequest();
 465
 466                 // We (re)set $wgThumbLimits to a single-element array above.
 467                 $context->getUser()->setOption( 'thumbsize', 0 );
 468
 469                 foreach ( $settings as $var => $val ) {
 470                         if ( array_key_exists( $var, $GLOBALS ) ) {
 471                                 $this->savedGlobals[$var] = $GLOBALS[$var];
 472                         }
 473
 474                         $GLOBALS[$var] = $val;
 475                 }
 476
 477                 MWTidy::destroySingleton();
 478                 MagicWord::clearCache();
 479
 480                 # The entries saved into RepoGroup cache with previous globals will be wrong.
 481                 RepoGroup::destroySingleton();
 482                 FileBackendGroup::destroySingleton();
 483
 484                 # Create dummy files in storage
 485                 $this->setupUploads();
 486
 487                 # Publish the articles after we have the final language set
 488                 $this->publishTestArticles();
 489
 490                 MessageCache::destroyInstance();
 491
 492                 return $context;
 493         }
 494
 495         /**
 496          * Get an FS upload directory (only applies to FSFileBackend)
 497          *
 498          * @return string The directory
 499          */
 500         protected function getUploadDir() {
 501                 if ( $this->keepUploads ) {
 502                         // Don't use getNewTempDirectory() as this is meant to persist
 503                         $dir = wfTempDir() . '/mwParser-images';
 504
 505                         if ( is_dir( $dir ) ) {
 506                                 return $dir;
 507                         }
 508                 } else {
 509                         $dir = $this->getNewTempDirectory();
 510                 }
 511
 512                 if ( file_exists( $dir ) ) {
 513                         wfDebug( "Already exists!\n" );
 514
 515                         return $dir;
 516                 }
 517
 518                 return $dir;
 519         }
 520
 521         /**
 522          * Create a dummy uploads directory which will contain a couple
 523          * of files in order to pass existence tests.
 524          *
 525          * @return string The directory
 526          */
 527         protected function setupUploads() {
 528                 global $IP;
 529
 530                 $base = $this->getBaseDir();
 531                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 532                 $backend->prepare( [ 'dir' => "$base/local-public/3/3a" ] );
 533                 $backend->store( [
 534                         'src' => "$IP/tests/phpunit/data/parser/headbg.jpg",
 535                         'dst' => "$base/local-public/3/3a/Foobar.jpg"
 536                 ] );
 537                 $backend->prepare( [ 'dir' => "$base/local-public/e/ea" ] );
 538                 $backend->store( [
 539                         'src' => "$IP/tests/phpunit/data/parser/wiki.png",
 540                         'dst' => "$base/local-public/e/ea/Thumb.png"
 541                 ] );
 542                 $backend->prepare( [ 'dir' => "$base/local-public/0/09" ] );
 543                 $backend->store( [
 544                         'src' => "$IP/tests/phpunit/data/parser/headbg.jpg",
 545                         'dst' => "$base/local-public/0/09/Bad.jpg"
 546                 ] );
 547                 $backend->prepare( [ 'dir' => "$base/local-public/5/5f" ] );
 548                 $backend->store( [
 549                         'src' => "$IP/tests/phpunit/data/parser/LoremIpsum.djvu",
 550                         'dst' => "$base/local-public/5/5f/LoremIpsum.djvu"
 551                 ] );
 552
 553                 // No helpful SVG file to copy, so make one ourselves
 554                 $data = '<?xml version="1.0" encoding="utf-8"?>' .
 555                         '<svg xmlns="http://www.w3.org/2000/svg"' .
 556                         ' version="1.1" width="240" height="180"/>';
 557
 558                 $backend->prepare( [ 'dir' => "$base/local-public/f/ff" ] );
 559                 $backend->quickCreate( [
 560                         'content' => $data, 'dst' => "$base/local-public/f/ff/Foobar.svg"
 561                 ] );
 562         }
 563
 564         /**
 565          * Restore default values and perform any necessary clean-up
 566          * after each test runs.
 567          */
 568         protected function teardownGlobals() {
 569                 $this->teardownUploads();
 570
 571                 foreach ( $this->savedGlobals as $var => $val ) {
 572                         $GLOBALS[$var] = $val;
 573                 }
 574         }
 575
 576         /**
 577          * Remove the dummy uploads directory
 578          */
 579         private function teardownUploads() {
 580                 if ( $this->keepUploads ) {
 581                         return;
 582                 }
 583
 584                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 585                 if ( $backend instanceof MockFileBackend ) {
 586                         # In memory backend, so dont bother cleaning them up.
 587                         return;
 588                 }
 589
 590                 $base = $this->getBaseDir();
 591                 // delete the files first, then the dirs.
 592                 self::deleteFiles(
 593                         [
 594                                 "$base/local-public/3/3a/Foobar.jpg",
 595                                 "$base/local-thumb/3/3a/Foobar.jpg/1000px-Foobar.jpg",
 596                                 "$base/local-thumb/3/3a/Foobar.jpg/100px-Foobar.jpg",
 597                                 "$base/local-thumb/3/3a/Foobar.jpg/120px-Foobar.jpg",
 598                                 "$base/local-thumb/3/3a/Foobar.jpg/1280px-Foobar.jpg",
 599                                 "$base/local-thumb/3/3a/Foobar.jpg/137px-Foobar.jpg",
 600                                 "$base/local-thumb/3/3a/Foobar.jpg/1500px-Foobar.jpg",
 601                                 "$base/local-thumb/3/3a/Foobar.jpg/177px-Foobar.jpg",
 602                                 "$base/local-thumb/3/3a/Foobar.jpg/180px-Foobar.jpg",
 603                                 "$base/local-thumb/3/3a/Foobar.jpg/200px-Foobar.jpg",
 604                                 "$base/local-thumb/3/3a/Foobar.jpg/206px-Foobar.jpg",
 605                                 "$base/local-thumb/3/3a/Foobar.jpg/20px-Foobar.jpg",
 606                                 "$base/local-thumb/3/3a/Foobar.jpg/220px-Foobar.jpg",
 607                                 "$base/local-thumb/3/3a/Foobar.jpg/265px-Foobar.jpg",
 608                                 "$base/local-thumb/3/3a/Foobar.jpg/270px-Foobar.jpg",
 609                                 "$base/local-thumb/3/3a/Foobar.jpg/274px-Foobar.jpg",
 610                                 "$base/local-thumb/3/3a/Foobar.jpg/300px-Foobar.jpg",
 611                                 "$base/local-thumb/3/3a/Foobar.jpg/30px-Foobar.jpg",
 612                                 "$base/local-thumb/3/3a/Foobar.jpg/330px-Foobar.jpg",
 613                                 "$base/local-thumb/3/3a/Foobar.jpg/353px-Foobar.jpg",
 614                                 "$base/local-thumb/3/3a/Foobar.jpg/360px-Foobar.jpg",
 615                                 "$base/local-thumb/3/3a/Foobar.jpg/400px-Foobar.jpg",
 616                                 "$base/local-thumb/3/3a/Foobar.jpg/40px-Foobar.jpg",
 617                                 "$base/local-thumb/3/3a/Foobar.jpg/440px-Foobar.jpg",
 618                                 "$base/local-thumb/3/3a/Foobar.jpg/442px-Foobar.jpg",
 619                                 "$base/local-thumb/3/3a/Foobar.jpg/450px-Foobar.jpg",
 620                                 "$base/local-thumb/3/3a/Foobar.jpg/50px-Foobar.jpg",
 621                                 "$base/local-thumb/3/3a/Foobar.jpg/600px-Foobar.jpg",
 622                                 "$base/local-thumb/3/3a/Foobar.jpg/640px-Foobar.jpg",
 623                                 "$base/local-thumb/3/3a/Foobar.jpg/70px-Foobar.jpg",
 624                                 "$base/local-thumb/3/3a/Foobar.jpg/75px-Foobar.jpg",
 625                                 "$base/local-thumb/3/3a/Foobar.jpg/960px-Foobar.jpg",
 626
 627                                 "$base/local-public/e/ea/Thumb.png",
 628
 629                                 "$base/local-public/0/09/Bad.jpg",
 630
 631                                 "$base/local-public/5/5f/LoremIpsum.djvu",
 632                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-2480px-LoremIpsum.djvu.jpg",
 633                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-3720px-LoremIpsum.djvu.jpg",
 634                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-4960px-LoremIpsum.djvu.jpg",
 635
 636                                 "$base/local-public/f/ff/Foobar.svg",
 637                                 "$base/local-thumb/f/ff/Foobar.svg/180px-Foobar.svg.png",
 638                                 "$base/local-thumb/f/ff/Foobar.svg/2000px-Foobar.svg.png",
 639                                 "$base/local-thumb/f/ff/Foobar.svg/270px-Foobar.svg.png",
 640                                 "$base/local-thumb/f/ff/Foobar.svg/3000px-Foobar.svg.png",
 641                                 "$base/local-thumb/f/ff/Foobar.svg/360px-Foobar.svg.png",
 642                                 "$base/local-thumb/f/ff/Foobar.svg/4000px-Foobar.svg.png",
 643                                 "$base/local-thumb/f/ff/Foobar.svg/langde-180px-Foobar.svg.png",
 644                                 "$base/local-thumb/f/ff/Foobar.svg/langde-270px-Foobar.svg.png",
 645                                 "$base/local-thumb/f/ff/Foobar.svg/langde-360px-Foobar.svg.png",
 646
 647                                 "$base/local-public/math/f/a/5/fa50b8b616463173474302ca3e63586b.png",
 648                         ]
 649                 );
 650         }
 651
 652         /**
 653          * Delete the specified files, if they exist.
 654          * @param array $files Full paths to files to delete.
 655          */
 656         private static function deleteFiles( $files ) {
 657                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 658                 foreach ( $files as $file ) {
 659                         $backend->delete( [ 'src' => $file ], [ 'force' => 1 ] );
 660                 }
 661                 foreach ( $files as $file ) {
 662                         $tmp = FileBackend::parentStoragePath( $file );
 663                         while ( $tmp ) {
 664                                 if ( !$backend->clean( [ 'dir' => $tmp ] )->isOK() ) {
 665                                         break;
 666                                 }
 667                                 $tmp = FileBackend::parentStoragePath( $tmp );
 668                         }
 669                 }
 670         }
 671
 672         protected function getBaseDir() {
 673                 return 'mwstore://local-backend';
 674         }
 675
 676         public function parserTestProvider() {
 677                 if ( $this->file === false ) {
 678                         global $wgParserTestFiles;
 679                         $this->file = $wgParserTestFiles[0];
 680                 }
 681
 682                 return new TestFileIterator( $this->file, $this );
 683         }
 684
 685         /**
 686          * Set the file from whose tests will be run by this instance
 687          * @param string $filename
 688          */
 689         public function setParserTestFile( $filename ) {
 690                 $this->file = $filename;
 691         }
 692
 693         /**
 694          * @group medium
 695          * @group ParserTests
 696          * @dataProvider parserTestProvider
 697          * @param string $desc
 698          * @param string $input
 699          * @param string $result
 700          * @param array $opts
 701          * @param array $config
 702          */
 703         public function testParserTest( $desc, $input, $result, $opts, $config ) {
 704                 if ( $this->regex != '' && !preg_match( '/' . $this->regex . '/', $desc ) ) {
 705                         $this->assertTrue( true ); // XXX: don't flood output with "test made no assertions"
 706                         // $this->markTestSkipped( 'Filtered out by the user' );
 707                         return;
 708                 }
 709
 710                 if ( !$this->isWikitextNS( NS_MAIN ) ) {
 711                         // parser tests frequently assume that the main namespace contains wikitext.
 712                         // @todo When setting up pages, force the content model. Only skip if
 713                         //        $wgtContentModelUseDB is false.
 714                         $this->markTestSkipped( "Main namespace does not support wikitext,"
 715                                 . "skipping parser test: $desc" );
 716                 }
 717
 718                 wfDebug( "Running parser test: $desc\n" );
 719
 720                 $opts = $this->parseOptions( $opts );
 721                 $context = $this->setupGlobals( $opts, $config );
 722
 723                 $user = $context->getUser();
 724                 $options = ParserOptions::newFromContext( $context );
 725
 726                 if ( isset( $opts['title'] ) ) {
 727                         $titleText = $opts['title'];
 728                 } else {
 729                         $titleText = 'Parser test';
 730                 }
 731
 732                 $local = isset( $opts['local'] );
 733                 $preprocessor = isset( $opts['preprocessor'] ) ? $opts['preprocessor'] : null;
 734                 $parser = $this->getParser( $preprocessor );
 735
 736                 $title = Title::newFromText( $titleText );
 737
 738                 # Parser test requiring math. Make sure texvc is executable
 739                 # or just skip such tests.
 740                 if ( isset( $opts['math'] ) || isset( $opts['texvc'] ) ) {
 741                         global $wgTexvc;
 742
 743                         if ( !isset( $wgTexvc ) ) {
 744                                 $this->markTestSkipped( "SKIPPED: \$wgTexvc is not set" );
 745                         } elseif ( !is_executable( $wgTexvc ) ) {
 746                                 $this->markTestSkipped( "SKIPPED: texvc binary does not exist"
 747                                         . " or is not executable.\n"
 748                                         . "Current configuration is:\n\$wgTexvc = '$wgTexvc'" );
 749                         }
 750                 }
 751
 752                 if ( isset( $opts['djvu'] ) ) {
 753                         if ( !$this->djVuSupport->isEnabled() ) {
 754                                 $this->markTestSkipped( "SKIPPED: djvu binaries do not exist or are not executable.\n" );
 755                         }
 756                 }
 757
 758                 if ( isset( $opts['tidy'] ) ) {
 759                         if ( !$this->tidySupport->isEnabled() ) {
 760                                 $this->markTestSkipped( "SKIPPED: tidy extension is not installed.\n" );
 761                         } else {
 762                                 $options->setTidy( true );
 763                         }
 764                 }
 765
 766                 if ( isset( $opts['pst'] ) ) {
 767                         $out = $parser->preSaveTransform( $input, $title, $user, $options );
 768                 } elseif ( isset( $opts['msg'] ) ) {
 769                         $out = $parser->transformMsg( $input, $options, $title );
 770                 } elseif ( isset( $opts['section'] ) ) {
 771                         $section = $opts['section'];
 772                         $out = $parser->getSection( $input, $section );
 773                 } elseif ( isset( $opts['replace'] ) ) {
 774                         $section = $opts['replace'][0];
 775                         $replace = $opts['replace'][1];
 776                         $out = $parser->replaceSection( $input, $section, $replace );
 777                 } elseif ( isset( $opts['comment'] ) ) {
 778                         $out = Linker::formatComment( $input, $title, $local );
 779                 } elseif ( isset( $opts['preload'] ) ) {
 780                         $out = $parser->getPreloadText( $input, $title, $options );
 781                 } else {
 782                         $output = $parser->parse( $input, $title, $options, true, true, 1337 );
 783                         $output->setTOCEnabled( !isset( $opts['notoc'] ) );
 784                         $out = $output->getText();
 785                         if ( isset( $opts['tidy'] ) ) {
 786                                 $out = preg_replace( '/\s+$/', '', $out );
 787                         }
 788
 789                         if ( isset( $opts['showtitle'] ) ) {
 790                                 if ( $output->getTitleText() ) {
 791                                         $title = $output->getTitleText();
 792                                 }
 793
 794                                 $out = "$title\n$out";
 795                         }
 796
 797                         if ( isset( $opts['showindicators'] ) ) {
 798                                 $indicators = '';
 799                                 foreach ( $output->getIndicators() as $id => $content ) {
 800                                         $indicators .= "$id=$content\n";
 801                                 }
 802                                 $out = $indicators . $out;
 803                         }
 804
 805                         if ( isset( $opts['ill'] ) ) {
 806                                 $out = implode( ' ', $output->getLanguageLinks() );
 807                         } elseif ( isset( $opts['cat'] ) ) {
 808                                 $outputPage = $context->getOutput();
 809                                 $outputPage->addCategoryLinks( $output->getCategories() );
 810                                 $cats = $outputPage->getCategoryLinks();
 811
 812                                 if ( isset( $cats['normal'] ) ) {
 813                                         $out = implode( ' ', $cats['normal'] );
 814                                 } else {
 815                                         $out = '';
 816                                 }
 817                         }
 818                         $parser->mPreprocessor = null;
 819                 }
 820
 821                 $this->teardownGlobals();
 822
 823                 $this->assertEquals( $result, $out, $desc );
 824         }
 825
 826         /**
 827          * Run a fuzz test series
 828          * Draw input from a set of test files
 829          *
 830          * @todo fixme Needs some work to not eat memory until the world explodes
 831          *
 832          * @group ParserFuzz
 833          */
 834         public function testFuzzTests() {
 835                 global $wgParserTestFiles;
 836
 837                 $files = $wgParserTestFiles;
 838
 839                 if ( $this->getCliArg( 'file' ) ) {
 840                         $files = [ $this->getCliArg( 'file' ) ];
 841                 }
 842
 843                 $dict = $this->getFuzzInput( $files );
 844                 $dictSize = strlen( $dict );
 845                 $logMaxLength = log( $this->maxFuzzTestLength );
 846
 847                 ini_set( 'memory_limit', $this->memoryLimit * 1048576 );
 848
 849                 $user = new User;
 850                 $opts = ParserOptions::newFromUser( $user );
 851                 $title = Title::makeTitle( NS_MAIN, 'Parser_test' );
 852
 853                 $id = 1;
 854
 855                 while ( true ) {
 856
 857                         // Generate test input
 858                         mt_srand( ++$this->fuzzSeed );
 859                         $totalLength = mt_rand( 1, $this->maxFuzzTestLength );
 860                         $input = '';
 861
 862                         while ( strlen( $input ) < $totalLength ) {
 863                                 $logHairLength = mt_rand( 0, 1000000 ) / 1000000 * $logMaxLength;
 864                                 $hairLength = min( intval( exp( $logHairLength ) ), $dictSize );
 865                                 $offset = mt_rand( 0, $dictSize - $hairLength );
 866                                 $input .= substr( $dict, $offset, $hairLength );
 867                         }
 868
 869                         $this->setupGlobals();
 870                         $parser = $this->getParser();
 871
 872                         // Run the test
 873                         try {
 874                                 $parser->parse( $input, $title, $opts );
 875                                 $this->assertTrue( true, "Test $id, fuzz seed {$this->fuzzSeed}" );
 876                         } catch ( Exception $exception ) {
 877                                 $input_dump = sprintf( "string(%d) \"%s\"\n", strlen( $input ), $input );
 878
 879                                 $this->assertTrue( false, "Test $id, fuzz seed {$this->fuzzSeed}. \n\n" .
 880                                         "Input: $input_dump\n\nError: {$exception->getMessage()}\n\n" .
 881                                         "Backtrace: {$exception->getTraceAsString()}" );
 882                         }
 883
 884                         $this->teardownGlobals();
 885                         $parser->__destruct();
 886
 887                         if ( $id % 100 == 0 ) {
 888                                 $usage = intval( memory_get_usage( true ) / $this->memoryLimit / 1048576 * 100 );
 889                                 // echo "{$this->fuzzSeed}: $numSuccess/$numTotal (mem: $usage%)\n";
 890                                 if ( $usage > 90 ) {
 891                                         $ret = "Out of memory:\n";
 892                                         $memStats = $this->getMemoryBreakdown();
 893
 894                                         foreach ( $memStats as $name => $usage ) {
 895                                                 $ret .= "$name: $usage\n";
 896                                         }
 897
 898                                         throw new MWException( $ret );
 899                                 }
 900                         }
 901
 902                         $id++;
 903                 }
 904         }
 905
 906         // Various getter functions
 907
 908         /**
 909          * Get an input dictionary from a set of parser test files
 910          * @param array $filenames
 911          * @return string
 912          */
 913         function getFuzzInput( $filenames ) {
 914                 $dict = '';
 915
 916                 foreach ( $filenames as $filename ) {
 917                         $contents = file_get_contents( $filename );
 918                         preg_match_all( '/!!\s*input\n(.*?)\n!!\s*result/s', $contents, $matches );
 919
 920                         foreach ( $matches[1] as $match ) {
 921                                 $dict .= $match . "\n";
 922                         }
 923                 }
 924
 925                 return $dict;
 926         }
 927
 928         /**
 929          * Get a memory usage breakdown
 930          * @return array
 931          */
 932         function getMemoryBreakdown() {
 933                 $memStats = [];
 934
 935                 foreach ( $GLOBALS as $name => $value ) {
 936                         $memStats['$' . $name] = strlen( serialize( $value ) );
 937                 }
 938
 939                 $classes = get_declared_classes();
 940
 941                 foreach ( $classes as $class ) {
 942                         $rc = new ReflectionClass( $class );
 943                         $props = $rc->getStaticProperties();
 944                         $memStats[$class] = strlen( serialize( $props ) );
 945                         $methods = $rc->getMethods();
 946
 947                         foreach ( $methods as $method ) {
 948                                 $memStats[$class] += strlen( serialize( $method->getStaticVariables() ) );
 949                         }
 950                 }
 951
 952                 $functions = get_defined_functions();
 953
 954                 foreach ( $functions['user'] as $function ) {
 955                         $rf = new ReflectionFunction( $function );
 956                         $memStats["$function()"] = strlen( serialize( $rf->getStaticVariables() ) );
 957                 }
 958
 959                 asort( $memStats );
 960
 961                 return $memStats;
 962         }
 963
 964         /**
 965          * Get a Parser object
 966          * @param Preprocessor $preprocessor
 967          * @return Parser
 968          */
 969         function getParser( $preprocessor = null ) {
 970                 global $wgParserConf;
 971
 972                 $class = $wgParserConf['class'];
 973                 $parser = new $class( [ 'preprocessorClass' => $preprocessor ] + $wgParserConf );
 974
 975                 Hooks::run( 'ParserTestParser', [ &$parser ] );
 976
 977                 return $parser;
 978         }
 979
 980         // Various action functions
 981
 982         public function addArticle( $name, $text, $line ) {
 983                 self::$articles[$name] = [ $text, $line ];
 984         }
 985
 986         public function publishTestArticles() {
 987                 if ( empty( self::$articles ) ) {
 988                         return;
 989                 }
 990
 991                 foreach ( self::$articles as $name => $info ) {
 992                         list( $text, $line ) = $info;
 993                         ParserTest::addArticle( $name, $text, $line, 'ignoreduplicate' );
 994                 }
 995         }
 996
 997         /**
 998          * Steal a callback function from the primary parser, save it for
 999          * application to our scary parser. If the hook is not installed,
1000          * abort processing of this file.
1001          *
1002          * @param string $name
1003          * @return bool True if tag hook is present
1004          */
1005         public function requireHook( $name ) {
1006                 global $wgParser;
1007                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1008                 return isset( $wgParser->mTagHooks[$name] );
1009         }
1010
1011         public function requireFunctionHook( $name ) {
1012                 global $wgParser;
1013                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1014                 return isset( $wgParser->mFunctionHooks[$name] );
1015         }
1016
1017         public function requireTransparentHook( $name ) {
1018                 global $wgParser;
1019                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1020                 return isset( $wgParser->mTransparentTagHooks[$name] );
1021         }
1022
1023         // Various "cleanup" functions
1024
1025         /**
1026          * Remove last character if it is a newline
1027          * @param string $s
1028          * @return string
1029          */
1030         public function removeEndingNewline( $s ) {
1031                 if ( substr( $s, -1 ) === "\n" ) {
1032                         return substr( $s, 0, -1 );
1033                 } else {
1034                         return $s;
1035                 }
1036         }
1037
1038         // Test options parser functions
1039
1040         protected function parseOptions( $instring ) {
1041                 $opts = [];
1042                 // foo
1043                 // foo=bar
1044                 // foo="bar baz"
1045                 // foo=[[bar baz]]
1046                 // foo=bar,"baz quux"
1047                 $regex = '/\b
1048                         ([\w-]+)                                                # Key
1049                         \b
1050                         (?:\s*
1051                                 =                                               # First sub-value
1052                                 \s*
1053                                 (
1054                                         "
1055                                                 [^"]*                   # Quoted val
1056                                         "
1057                                 |
1058                                         \[\[
1059                                                 [^]]*                   # Link target
1060                                         \]\]
1061                                 |
1062                                         [\w-]+                          # Plain word
1063                                 )
1064                                 (?:\s*
1065                                         ,                                       # Sub-vals 1..N
1066                                         \s*
1067                                         (
1068                                                 "[^"]*"                 # Quoted val
1069                                         |
1070                                                 \[\[[^]]*\]\]   # Link target
1071                                         |
1072                                                 [\w-]+                  # Plain word
1073                                         )
1074                                 )*
1075                         )?
1076                         /x';
1077
1078                 if ( preg_match_all( $regex, $instring, $matches, PREG_SET_ORDER ) ) {
1079                         foreach ( $matches as $bits ) {
1080                                 array_shift( $bits );
1081                                 $key = strtolower( array_shift( $bits ) );
1082                                 if ( count( $bits ) == 0 ) {
1083                                         $opts[$key] = true;
1084                                 } elseif ( count( $bits ) == 1 ) {
1085                                         $opts[$key] = $this->cleanupOption( array_shift( $bits ) );
1086                                 } else {
1087                                         // Array!
1088                                         $opts[$key] = array_map( [ $this, 'cleanupOption' ], $bits );
1089                                 }
1090                         }
1091                 }
1092
1093                 return $opts;
1094         }
1095
1096         protected function cleanupOption( $opt ) {
1097                 if ( substr( $opt, 0, 1 ) == '"' ) {
1098                         return substr( $opt, 1, -1 );
1099                 }
1100
1101                 if ( substr( $opt, 0, 2 ) == '[[' ) {
1102                         return substr( $opt, 2, -2 );
1103                 }
1104
1105                 return $opt;
1106         }
1107
1108         /**
1109          * Use a regex to find out the value of an option
1110          * @param string $key Name of option val to retrieve
1111          * @param array $opts Options array to look in
1112          * @param mixed $default Default value returned if not found
1113          * @return mixed
1114          */
1115         protected static function getOptionValue( $key, $opts, $default ) {
1116                 $key = strtolower( $key );
1117
1118                 if ( isset( $opts[$key] ) ) {
1119                         return $opts[$key];
1120                 } else {
1121                         return $default;
1122                 }
1123         }
1124 }