tests/phpunit/includes/parser/NewParserTest.php

   1 <?php
   2 /**
   3  * Although marked as a stub, can work independently.
   4  *
   5  * @group Database
   6  * @group Parser
   7  * @group Stub
   8  *
   9  * @todo covers tags
  10  */
  11 class NewParserTest extends MediaWikiTestCase {
  12         static protected $articles = []; // Array of test articles defined by the tests
  13         /* The data provider is run on a different instance than the test, so it must be static
  14          * When running tests from several files, all tests will see all articles.
  15          */
  16         static protected $backendToUse;
  17
  18         public $keepUploads = false;
  19         public $runDisabled = false;
  20         public $runParsoid = false;
  21         public $regex = '';
  22         public $showProgress = true;
  23         public $savedWeirdGlobals = [];
  24         public $savedGlobals = [];
  25         public $hooks = [];
  26         public $functionHooks = [];
  27         public $transparentHooks = [];
  28
  29         // Fuzz test
  30         public $maxFuzzTestLength = 300;
  31         public $fuzzSeed = 0;
  32         public $memoryLimit = 50;
  33
  34         /**
  35          * @var DjVuSupport
  36          */
  37         private $djVuSupport;
  38         /**
  39          * @var TidySupport
  40          */
  41         private $tidySupport;
  42
  43         protected $file = false;
  44
  45         public static function setUpBeforeClass() {
  46                 // Inject ParserTest well-known interwikis
  47                 ParserTest::setupInterwikis();
  48         }
  49
  50         protected function setUp() {
  51                 global $wgNamespaceAliases, $wgContLang;
  52                 global $wgHooks, $IP;
  53
  54                 parent::setUp();
  55
  56                 // Setup CLI arguments
  57                 if ( $this->getCliArg( 'regex' ) ) {
  58                         $this->regex = $this->getCliArg( 'regex' );
  59                 } else {
  60                         # Matches anything
  61                         $this->regex = '';
  62                 }
  63
  64                 $this->keepUploads = $this->getCliArg( 'keep-uploads' );
  65
  66                 $tmpGlobals = [];
  67
  68                 $tmpGlobals['wgLanguageCode'] = 'en';
  69                 $tmpGlobals['wgContLang'] = Language::factory( 'en' );
  70                 $tmpGlobals['wgSitename'] = 'MediaWiki';
  71                 $tmpGlobals['wgServer'] = 'http://example.org';
  72                 $tmpGlobals['wgServerName'] = 'example.org';
  73                 $tmpGlobals['wgScriptPath'] = '';
  74                 $tmpGlobals['wgScript'] = '/index.php';
  75                 $tmpGlobals['wgResourceBasePath'] = '';
  76                 $tmpGlobals['wgStylePath'] = '/skins';
  77                 $tmpGlobals['wgExtensionAssetsPath'] = '/extensions';
  78                 $tmpGlobals['wgArticlePath'] = '/wiki/$1';
  79                 $tmpGlobals['wgActionPaths'] = [];
  80                 $tmpGlobals['wgVariantArticlePath'] = false;
  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['wgExperimentalHtmlIds'] = false;
 104                 $tmpGlobals['wgAdaptiveMessageCache'] = true;
 105                 $tmpGlobals['wgUseDatabaseMessages'] = true;
 106                 $tmpGlobals['wgLocaltimezone'] = 'UTC';
 107                 $tmpGlobals['wgGroupPermissions'] = [
 108                         '*' => [
 109                                 'createaccount' => true,
 110                                 'read' => true,
 111                                 'edit' => true,
 112                                 'createpage' => true,
 113                                 'createtalk' => true,
 114                 ] ];
 115                 $tmpGlobals['wgNamespaceProtection'] = [ NS_MEDIAWIKI => 'editinterface' ];
 116
 117                 $tmpGlobals['wgParser'] = new StubObject(
 118                         'wgParser', $GLOBALS['wgParserConf']['class'],
 119                         [ $GLOBALS['wgParserConf'] ] );
 120
 121                 $tmpGlobals['wgFileExtensions'][] = 'svg';
 122                 $tmpGlobals['wgSVGConverter'] = 'rsvg';
 123                 $tmpGlobals['wgSVGConverters']['rsvg'] =
 124                         '$path/rsvg-convert -w $width -h $height -o $output $input';
 125
 126                 if ( $GLOBALS['wgStyleDirectory'] === false ) {
 127                         $tmpGlobals['wgStyleDirectory'] = "$IP/skins";
 128                 }
 129
 130                 # Replace all media handlers with a mock. We do not need to generate
 131                 # actual thumbnails to do parser testing, we only care about receiving
 132                 # a ThumbnailImage properly initialized.
 133                 global $wgMediaHandlers;
 134                 foreach ( $wgMediaHandlers as $type => $handler ) {
 135                         $tmpGlobals['wgMediaHandlers'][$type] = 'MockBitmapHandler';
 136                 }
 137                 // Vector images have to be handled slightly differently
 138                 $tmpGlobals['wgMediaHandlers']['image/svg+xml'] = 'MockSvgHandler';
 139
 140                 // DjVu images have to be handled slightly differently
 141                 $tmpGlobals['wgMediaHandlers']['image/vnd.djvu'] = 'MockDjVuHandler';
 142
 143                 // Ogg video/audio increasingly more differently
 144                 $tmpGlobals['wgMediaHandlers']['application/ogg'] = 'MockOggHandler';
 145
 146                 $tmpHooks = $wgHooks;
 147                 $tmpHooks['ParserTestParser'][] = 'ParserTestParserHook::setup';
 148                 $tmpHooks['ParserGetVariableValueTs'][] = 'ParserTest::getFakeTimestamp';
 149                 $tmpGlobals['wgHooks'] = $tmpHooks;
 150                 # add a namespace shadowing a interwiki link, to test
 151                 # proper precedence when resolving links. (bug 51680)
 152                 $tmpGlobals['wgExtraNamespaces'] = [
 153                         100 => 'MemoryAlpha',
 154                         101 => 'MemoryAlpha_talk'
 155                 ];
 156
 157                 $tmpGlobals['wgLocalInterwikis'] = [ 'local', 'mi' ];
 158                 # "extra language links"
 159                 # see https://gerrit.wikimedia.org/r/111390
 160                 $tmpGlobals['wgExtraInterlanguageLinkPrefixes'] = [ 'mul' ];
 161
 162                 // DjVu support
 163                 $this->djVuSupport = new DjVuSupport();
 164                 // Tidy support
 165                 $this->tidySupport = new TidySupport();
 166                 $tmpGlobals['wgTidyConfig'] = null;
 167                 $tmpGlobals['wgUseTidy'] = false;
 168                 $tmpGlobals['wgDebugTidy'] = false;
 169                 $tmpGlobals['wgTidyConf'] = $IP . '/includes/tidy/tidy.conf';
 170                 $tmpGlobals['wgTidyOpts'] = '';
 171                 $tmpGlobals['wgTidyInternal'] = $this->tidySupport->isInternal();
 172
 173                 $this->setMwGlobals( $tmpGlobals );
 174
 175                 $this->savedWeirdGlobals['image_alias'] = $wgNamespaceAliases['Image'];
 176                 $this->savedWeirdGlobals['image_talk_alias'] = $wgNamespaceAliases['Image_talk'];
 177
 178                 $wgNamespaceAliases['Image'] = NS_FILE;
 179                 $wgNamespaceAliases['Image_talk'] = NS_FILE_TALK;
 180
 181                 MWNamespace::getCanonicalNamespaces( true ); # reset namespace cache
 182                 $wgContLang->resetNamespaces(); # reset namespace cache
 183                 ParserTest::resetTitleServices();
 184         }
 185
 186         protected function tearDown() {
 187                 global $wgNamespaceAliases, $wgContLang;
 188
 189                 $wgNamespaceAliases['Image'] = $this->savedWeirdGlobals['image_alias'];
 190                 $wgNamespaceAliases['Image_talk'] = $this->savedWeirdGlobals['image_talk_alias'];
 191
 192                 MWTidy::destroySingleton();
 193
 194                 // Restore backends
 195                 RepoGroup::destroySingleton();
 196                 FileBackendGroup::destroySingleton();
 197
 198                 // Remove temporary pages from the link cache
 199                 LinkCache::singleton()->clear();
 200
 201                 // Restore message cache (temporary pages and $wgUseDatabaseMessages)
 202                 MessageCache::destroyInstance();
 203
 204                 parent::tearDown();
 205
 206                 MWNamespace::getCanonicalNamespaces( true ); # reset namespace cache
 207                 $wgContLang->resetNamespaces(); # reset namespace cache
 208         }
 209
 210         public static function tearDownAfterClass() {
 211                 ParserTest::tearDownInterwikis();
 212                 parent::tearDownAfterClass();
 213         }
 214
 215         function addDBDataOnce() {
 216                 # disabled for performance
 217                 # $this->tablesUsed[] = 'image';
 218
 219                 # Update certain things in site_stats
 220                 $this->db->insert( 'site_stats',
 221                         [ 'ss_row_id' => 1, 'ss_images' => 2, 'ss_good_articles' => 1 ],
 222                         __METHOD__,
 223                         [ 'IGNORE' ]
 224                 );
 225
 226                 $user = User::newFromId( 0 );
 227                 LinkCache::singleton()->clear(); # Avoids the odd failure at creating the nullRevision
 228
 229                 # Upload DB table entries for files.
 230                 # We will upload the actual files later. Note that if anything causes LocalFile::load()
 231                 # to be triggered before then, it will break via maybeUpgrade() setting the fileExists
 232                 # member to false and storing it in cache.
 233                 # note that the size/width/height/bits/etc of the file
 234                 # are actually set by inspecting the file itself; the arguments
 235                 # to recordUpload2 have no effect.  That said, we try to make things
 236                 # match up so it is less confusing to readers of the code & tests.
 237                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Foobar.jpg' ) );
 238                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 239                         $image->recordUpload2(
 240                                 '', // archive name
 241                                 'Upload of some lame file',
 242                                 'Some lame file',
 243                                 [
 244                                         'size' => 7881,
 245                                         'width' => 1941,
 246                                         'height' => 220,
 247                                         'bits' => 8,
 248                                         'media_type' => MEDIATYPE_BITMAP,
 249                                         'mime' => 'image/jpeg',
 250                                         'metadata' => serialize( [] ),
 251                                         'sha1' => Wikimedia\base_convert( '1', 16, 36, 31 ),
 252                                         'fileExists' => true ],
 253                                 $this->db->timestamp( '20010115123500' ), $user
 254                         );
 255                 }
 256
 257                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Thumb.png' ) );
 258                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 259                         $image->recordUpload2(
 260                                 '', // archive name
 261                                 'Upload of some lame thumbnail',
 262                                 'Some lame thumbnail',
 263                                 [
 264                                         'size' => 22589,
 265                                         'width' => 135,
 266                                         'height' => 135,
 267                                         'bits' => 8,
 268                                         'media_type' => MEDIATYPE_BITMAP,
 269                                         'mime' => 'image/png',
 270                                         'metadata' => serialize( [] ),
 271                                         'sha1' => Wikimedia\base_convert( '2', 16, 36, 31 ),
 272                                         'fileExists' => true ],
 273                                 $this->db->timestamp( '20130225203040' ), $user
 274                         );
 275                 }
 276
 277                 # This image will be blacklisted in [[MediaWiki:Bad image list]]
 278                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Bad.jpg' ) );
 279                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 280                         $image->recordUpload2(
 281                                 '', // archive name
 282                                 'zomgnotcensored',
 283                                 'Borderline image',
 284                                 [
 285                                         'size' => 12345,
 286                                         'width' => 320,
 287                                         'height' => 240,
 288                                         'bits' => 24,
 289                                         'media_type' => MEDIATYPE_BITMAP,
 290                                         'mime' => 'image/jpeg',
 291                                         'metadata' => serialize( [] ),
 292                                         'sha1' => Wikimedia\base_convert( '3', 16, 36, 31 ),
 293                                         'fileExists' => true ],
 294                                 $this->db->timestamp( '20010115123500' ), $user
 295                         );
 296                 }
 297                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Foobar.svg' ) );
 298                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 299                         $image->recordUpload2( '', 'Upload of some lame SVG', 'Some lame SVG', [
 300                                         'size'        => 12345,
 301                                         'width'       => 240,
 302                                         'height'      => 180,
 303                                         'bits'        => 0,
 304                                         'media_type'  => MEDIATYPE_DRAWING,
 305                                         'mime'        => 'image/svg+xml',
 306                                         'metadata'    => serialize( [] ),
 307                                         'sha1'        => Wikimedia\base_convert( '', 16, 36, 31 ),
 308                                         'fileExists'  => true
 309                         ], $this->db->timestamp( '20010115123500' ), $user );
 310                 }
 311
 312                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Video.ogv' ) );
 313                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 314                         $image->recordUpload2( '', 'A pretty movie', 'Will it play', [
 315                                         'size'        => 12345,
 316                                         'width'       => 320,
 317                                         'height'      => 240,
 318                                         'bits'        => 0,
 319                                         'media_type'  => MEDIATYPE_VIDEO,
 320                                         'mime'        => 'application/ogg',
 321                                         'metadata'    => serialize( [] ),
 322                                         'sha1'        => Wikimedia\base_convert( '', 16, 36, 32 ),
 323                                         'fileExists'  => true
 324                         ], $this->db->timestamp( '20010115123500' ), $user );
 325                 }
 326
 327                 # A DjVu file
 328                 # A DjVu file
 329                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'LoremIpsum.djvu' ) );
 330                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 331                         $image->recordUpload2( '', 'Upload a DjVu', 'A DjVu', [
 332                                 'size' => 3249,
 333                                 'width' => 2480,
 334                                 'height' => 3508,
 335                                 'bits' => 0,
 336                                 'media_type' => MEDIATYPE_BITMAP,
 337                                 'mime' => 'image/vnd.djvu',
 338                                 'metadata' => '<?xml version="1.0" ?>
 339 <!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd">
 340 <DjVuXML>
 341 <HEAD></HEAD>
 342 <BODY><OBJECT height="3508" width="2480">
 343 <PARAM name="DPI" value="300" />
 344 <PARAM name="GAMMA" value="2.2" />
 345 </OBJECT>
 346 <OBJECT height="3508" width="2480">
 347 <PARAM name="DPI" value="300" />
 348 <PARAM name="GAMMA" value="2.2" />
 349 </OBJECT>
 350 <OBJECT height="3508" width="2480">
 351 <PARAM name="DPI" value="300" />
 352 <PARAM name="GAMMA" value="2.2" />
 353 </OBJECT>
 354 <OBJECT height="3508" width="2480">
 355 <PARAM name="DPI" value="300" />
 356 <PARAM name="GAMMA" value="2.2" />
 357 </OBJECT>
 358 <OBJECT height="3508" width="2480">
 359 <PARAM name="DPI" value="300" />
 360 <PARAM name="GAMMA" value="2.2" />
 361 </OBJECT>
 362 </BODY>
 363 </DjVuXML>',
 364                                 'sha1' => Wikimedia\base_convert( '', 16, 36, 31 ),
 365                                 'fileExists' => true
 366                         ], $this->db->timestamp( '20140115123600' ), $user );
 367                 }
 368         }
 369
 370         // ParserTest setup/teardown functions
 371
 372         /**
 373          * Set up the global variables for a consistent environment for each test.
 374          * Ideally this should replace the global configuration entirely.
 375          * @param array $opts
 376          * @param string $config
 377          * @return RequestContext
 378          */
 379         protected function setupGlobals( $opts = [], $config = '' ) {
 380                 global $wgFileBackends;
 381                 # Find out values for some special options.
 382                 $lang =
 383                         self::getOptionValue( 'language', $opts, 'en' );
 384                 $variant =
 385                         self::getOptionValue( 'variant', $opts, false );
 386                 $maxtoclevel =
 387                         self::getOptionValue( 'wgMaxTocLevel', $opts, 999 );
 388                 $linkHolderBatchSize =
 389                         self::getOptionValue( 'wgLinkHolderBatchSize', $opts, 1000 );
 390
 391                 $uploadDir = $this->getUploadDir();
 392                 if ( $this->getCliArg( 'use-filebackend' ) ) {
 393                         if ( self::$backendToUse ) {
 394                                 $backend = self::$backendToUse;
 395                         } else {
 396                                 $name = $this->getCliArg( 'use-filebackend' );
 397                                 $useConfig = [];
 398                                 foreach ( $wgFileBackends as $conf ) {
 399                                         if ( $conf['name'] == $name ) {
 400                                                 $useConfig = $conf;
 401                                         }
 402                                 }
 403                                 $useConfig['name'] = 'local-backend'; // swap name
 404                                 unset( $useConfig['lockManager'] );
 405                                 unset( $useConfig['fileJournal'] );
 406                                 $class = $useConfig['class'];
 407                                 self::$backendToUse = new $class( $useConfig );
 408                                 $backend = self::$backendToUse;
 409                         }
 410                 } else {
 411                         # Replace with a mock. We do not care about generating real
 412                         # files on the filesystem, just need to expose the file
 413                         # informations.
 414                         $backend = new MockFileBackend( [
 415                                 'name' => 'local-backend',
 416                                 'wikiId' => wfWikiID()
 417                         ] );
 418                 }
 419
 420                 $settings = [
 421                         'wgLocalFileRepo' => [
 422                                 'class' => 'LocalRepo',
 423                                 'name' => 'local',
 424                                 'url' => 'http://example.com/images',
 425                                 'hashLevels' => 2,
 426                                 'transformVia404' => false,
 427                                 'backend' => $backend
 428                         ],
 429                         'wgEnableUploads' => self::getOptionValue( 'wgEnableUploads', $opts, true ),
 430                         'wgLanguageCode' => $lang,
 431                         'wgDBprefix' => $this->db->getType() != 'oracle' ? 'unittest_' : 'ut_',
 432                         'wgRawHtml' => self::getOptionValue( 'wgRawHtml', $opts, false ),
 433                         'wgNamespacesWithSubpages' => [ NS_MAIN => isset( $opts['subpage'] ) ],
 434                         'wgAllowExternalImages' => self::getOptionValue( 'wgAllowExternalImages', $opts, true ),
 435                         'wgThumbLimits' => [ self::getOptionValue( 'thumbsize', $opts, 180 ) ],
 436                         'wgMaxTocLevel' => $maxtoclevel,
 437                         'wgUseTeX' => isset( $opts['math'] ) || isset( $opts['texvc'] ),
 438                         'wgMathDirectory' => $uploadDir . '/math',
 439                         'wgDefaultLanguageVariant' => $variant,
 440                         'wgLinkHolderBatchSize' => $linkHolderBatchSize,
 441                         'wgUseTidy' => isset( $opts['tidy'] ),
 442                 ];
 443
 444                 if ( $config ) {
 445                         $configLines = explode( "\n", $config );
 446
 447                         foreach ( $configLines as $line ) {
 448                                 list( $var, $value ) = explode( '=', $line, 2 );
 449
 450                                 $settings[$var] = eval( "return $value;" ); // ???
 451                         }
 452                 }
 453
 454                 $this->savedGlobals = [];
 455
 456                 /** @since 1.20 */
 457                 Hooks::run( 'ParserTestGlobals', [ &$settings ] );
 458
 459                 $langObj = Language::factory( $lang );
 460                 $settings['wgContLang'] = $langObj;
 461                 $settings['wgLang'] = $langObj;
 462
 463                 $context = new RequestContext();
 464                 $settings['wgOut'] = $context->getOutput();
 465                 $settings['wgUser'] = $context->getUser();
 466                 $settings['wgRequest'] = $context->getRequest();
 467
 468                 // We (re)set $wgThumbLimits to a single-element array above.
 469                 $context->getUser()->setOption( 'thumbsize', 0 );
 470
 471                 foreach ( $settings as $var => $val ) {
 472                         if ( array_key_exists( $var, $GLOBALS ) ) {
 473                                 $this->savedGlobals[$var] = $GLOBALS[$var];
 474                         }
 475
 476                         $GLOBALS[$var] = $val;
 477                 }
 478
 479                 MWTidy::destroySingleton();
 480                 MagicWord::clearCache();
 481
 482                 # The entries saved into RepoGroup cache with previous globals will be wrong.
 483                 RepoGroup::destroySingleton();
 484                 FileBackendGroup::destroySingleton();
 485
 486                 # Create dummy files in storage
 487                 $this->setupUploads();
 488
 489                 # Publish the articles after we have the final language set
 490                 $this->publishTestArticles();
 491
 492                 MessageCache::destroyInstance();
 493
 494                 return $context;
 495         }
 496
 497         /**
 498          * Get an FS upload directory (only applies to FSFileBackend)
 499          *
 500          * @return string The directory
 501          */
 502         protected function getUploadDir() {
 503                 if ( $this->keepUploads ) {
 504                         // Don't use getNewTempDirectory() as this is meant to persist
 505                         $dir = wfTempDir() . '/mwParser-images';
 506
 507                         if ( is_dir( $dir ) ) {
 508                                 return $dir;
 509                         }
 510                 } else {
 511                         $dir = $this->getNewTempDirectory();
 512                 }
 513
 514                 if ( file_exists( $dir ) ) {
 515                         wfDebug( "Already exists!\n" );
 516
 517                         return $dir;
 518                 }
 519
 520                 return $dir;
 521         }
 522
 523         /**
 524          * Create a dummy uploads directory which will contain a couple
 525          * of files in order to pass existence tests.
 526          *
 527          * @return string The directory
 528          */
 529         protected function setupUploads() {
 530                 global $IP;
 531
 532                 $base = $this->getBaseDir();
 533                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 534                 $backend->prepare( [ 'dir' => "$base/local-public/3/3a" ] );
 535                 $backend->store( [
 536                         'src' => "$IP/tests/phpunit/data/parser/headbg.jpg",
 537                         'dst' => "$base/local-public/3/3a/Foobar.jpg"
 538                 ] );
 539                 $backend->prepare( [ 'dir' => "$base/local-public/e/ea" ] );
 540                 $backend->store( [
 541                         'src' => "$IP/tests/phpunit/data/parser/wiki.png",
 542                         'dst' => "$base/local-public/e/ea/Thumb.png"
 543                 ] );
 544                 $backend->prepare( [ 'dir' => "$base/local-public/0/09" ] );
 545                 $backend->store( [
 546                         'src' => "$IP/tests/phpunit/data/parser/headbg.jpg",
 547                         'dst' => "$base/local-public/0/09/Bad.jpg"
 548                 ] );
 549                 $backend->prepare( [ 'dir' => "$base/local-public/5/5f" ] );
 550                 $backend->store( [
 551                         'src' => "$IP/tests/phpunit/data/parser/LoremIpsum.djvu",
 552                         'dst' => "$base/local-public/5/5f/LoremIpsum.djvu"
 553                 ] );
 554
 555                 // No helpful SVG file to copy, so make one ourselves
 556                 $data = '<?xml version="1.0" encoding="utf-8"?>' .
 557                         '<svg xmlns="http://www.w3.org/2000/svg"' .
 558                         ' version="1.1" width="240" height="180"/>';
 559
 560                 $backend->prepare( [ 'dir' => "$base/local-public/f/ff" ] );
 561                 $backend->quickCreate( [
 562                         'content' => $data, 'dst' => "$base/local-public/f/ff/Foobar.svg"
 563                 ] );
 564         }
 565
 566         /**
 567          * Restore default values and perform any necessary clean-up
 568          * after each test runs.
 569          */
 570         protected function teardownGlobals() {
 571                 $this->teardownUploads();
 572
 573                 foreach ( $this->savedGlobals as $var => $val ) {
 574                         $GLOBALS[$var] = $val;
 575                 }
 576         }
 577
 578         /**
 579          * Remove the dummy uploads directory
 580          */
 581         private function teardownUploads() {
 582                 if ( $this->keepUploads ) {
 583                         return;
 584                 }
 585
 586                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 587                 if ( $backend instanceof MockFileBackend ) {
 588                         # In memory backend, so dont bother cleaning them up.
 589                         return;
 590                 }
 591
 592                 $base = $this->getBaseDir();
 593                 // delete the files first, then the dirs.
 594                 self::deleteFiles(
 595                         [
 596                                 "$base/local-public/3/3a/Foobar.jpg",
 597                                 "$base/local-thumb/3/3a/Foobar.jpg/1000px-Foobar.jpg",
 598                                 "$base/local-thumb/3/3a/Foobar.jpg/100px-Foobar.jpg",
 599                                 "$base/local-thumb/3/3a/Foobar.jpg/120px-Foobar.jpg",
 600                                 "$base/local-thumb/3/3a/Foobar.jpg/1280px-Foobar.jpg",
 601                                 "$base/local-thumb/3/3a/Foobar.jpg/137px-Foobar.jpg",
 602                                 "$base/local-thumb/3/3a/Foobar.jpg/1500px-Foobar.jpg",
 603                                 "$base/local-thumb/3/3a/Foobar.jpg/177px-Foobar.jpg",
 604                                 "$base/local-thumb/3/3a/Foobar.jpg/180px-Foobar.jpg",
 605                                 "$base/local-thumb/3/3a/Foobar.jpg/200px-Foobar.jpg",
 606                                 "$base/local-thumb/3/3a/Foobar.jpg/206px-Foobar.jpg",
 607                                 "$base/local-thumb/3/3a/Foobar.jpg/20px-Foobar.jpg",
 608                                 "$base/local-thumb/3/3a/Foobar.jpg/220px-Foobar.jpg",
 609                                 "$base/local-thumb/3/3a/Foobar.jpg/265px-Foobar.jpg",
 610                                 "$base/local-thumb/3/3a/Foobar.jpg/270px-Foobar.jpg",
 611                                 "$base/local-thumb/3/3a/Foobar.jpg/274px-Foobar.jpg",
 612                                 "$base/local-thumb/3/3a/Foobar.jpg/300px-Foobar.jpg",
 613                                 "$base/local-thumb/3/3a/Foobar.jpg/30px-Foobar.jpg",
 614                                 "$base/local-thumb/3/3a/Foobar.jpg/330px-Foobar.jpg",
 615                                 "$base/local-thumb/3/3a/Foobar.jpg/353px-Foobar.jpg",
 616                                 "$base/local-thumb/3/3a/Foobar.jpg/360px-Foobar.jpg",
 617                                 "$base/local-thumb/3/3a/Foobar.jpg/400px-Foobar.jpg",
 618                                 "$base/local-thumb/3/3a/Foobar.jpg/40px-Foobar.jpg",
 619                                 "$base/local-thumb/3/3a/Foobar.jpg/440px-Foobar.jpg",
 620                                 "$base/local-thumb/3/3a/Foobar.jpg/442px-Foobar.jpg",
 621                                 "$base/local-thumb/3/3a/Foobar.jpg/450px-Foobar.jpg",
 622                                 "$base/local-thumb/3/3a/Foobar.jpg/50px-Foobar.jpg",
 623                                 "$base/local-thumb/3/3a/Foobar.jpg/600px-Foobar.jpg",
 624                                 "$base/local-thumb/3/3a/Foobar.jpg/640px-Foobar.jpg",
 625                                 "$base/local-thumb/3/3a/Foobar.jpg/70px-Foobar.jpg",
 626                                 "$base/local-thumb/3/3a/Foobar.jpg/75px-Foobar.jpg",
 627                                 "$base/local-thumb/3/3a/Foobar.jpg/960px-Foobar.jpg",
 628
 629                                 "$base/local-public/e/ea/Thumb.png",
 630
 631                                 "$base/local-public/0/09/Bad.jpg",
 632
 633                                 "$base/local-public/5/5f/LoremIpsum.djvu",
 634                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-2480px-LoremIpsum.djvu.jpg",
 635                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-3720px-LoremIpsum.djvu.jpg",
 636                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-4960px-LoremIpsum.djvu.jpg",
 637
 638                                 "$base/local-public/f/ff/Foobar.svg",
 639                                 "$base/local-thumb/f/ff/Foobar.svg/180px-Foobar.svg.png",
 640                                 "$base/local-thumb/f/ff/Foobar.svg/2000px-Foobar.svg.png",
 641                                 "$base/local-thumb/f/ff/Foobar.svg/270px-Foobar.svg.png",
 642                                 "$base/local-thumb/f/ff/Foobar.svg/3000px-Foobar.svg.png",
 643                                 "$base/local-thumb/f/ff/Foobar.svg/360px-Foobar.svg.png",
 644                                 "$base/local-thumb/f/ff/Foobar.svg/4000px-Foobar.svg.png",
 645                                 "$base/local-thumb/f/ff/Foobar.svg/langde-180px-Foobar.svg.png",
 646                                 "$base/local-thumb/f/ff/Foobar.svg/langde-270px-Foobar.svg.png",
 647                                 "$base/local-thumb/f/ff/Foobar.svg/langde-360px-Foobar.svg.png",
 648
 649                                 "$base/local-public/math/f/a/5/fa50b8b616463173474302ca3e63586b.png",
 650                         ]
 651                 );
 652         }
 653
 654         /**
 655          * Delete the specified files, if they exist.
 656          * @param array $files Full paths to files to delete.
 657          */
 658         private static function deleteFiles( $files ) {
 659                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 660                 foreach ( $files as $file ) {
 661                         $backend->delete( [ 'src' => $file ], [ 'force' => 1 ] );
 662                 }
 663                 foreach ( $files as $file ) {
 664                         $tmp = FileBackend::parentStoragePath( $file );
 665                         while ( $tmp ) {
 666                                 if ( !$backend->clean( [ 'dir' => $tmp ] )->isOK() ) {
 667                                         break;
 668                                 }
 669                                 $tmp = FileBackend::parentStoragePath( $tmp );
 670                         }
 671                 }
 672         }
 673
 674         protected function getBaseDir() {
 675                 return 'mwstore://local-backend';
 676         }
 677
 678         public function parserTestProvider() {
 679                 if ( $this->file === false ) {
 680                         global $wgParserTestFiles;
 681                         $this->file = $wgParserTestFiles[0];
 682                 }
 683
 684                 return new TestFileDataProvider( $this->file, $this );
 685         }
 686
 687         /**
 688          * Set the file from whose tests will be run by this instance
 689          * @param string $filename
 690          */
 691         public function setParserTestFile( $filename ) {
 692                 $this->file = $filename;
 693         }
 694
 695         /**
 696          * @group medium
 697          * @group ParserTests
 698          * @dataProvider parserTestProvider
 699          * @param string $desc
 700          * @param string $input
 701          * @param string $result
 702          * @param array $opts
 703          * @param array $config
 704          */
 705         public function testParserTest( $desc, $input, $result, $opts, $config ) {
 706                 if ( $this->regex != '' && !preg_match( '/' . $this->regex . '/', $desc ) ) {
 707                         $this->assertTrue( true ); // XXX: don't flood output with "test made no assertions"
 708                         // $this->markTestSkipped( 'Filtered out by the user' );
 709                         return;
 710                 }
 711
 712                 if ( !$this->isWikitextNS( NS_MAIN ) ) {
 713                         // parser tests frequently assume that the main namespace contains wikitext.
 714                         // @todo When setting up pages, force the content model. Only skip if
 715                         //        $wgtContentModelUseDB is false.
 716                         $this->markTestSkipped( "Main namespace does not support wikitext,"
 717                                 . "skipping parser test: $desc" );
 718                 }
 719
 720                 wfDebug( "Running parser test: $desc\n" );
 721
 722                 $opts = $this->parseOptions( $opts );
 723                 $context = $this->setupGlobals( $opts, $config );
 724
 725                 $user = $context->getUser();
 726                 $options = ParserOptions::newFromContext( $context );
 727
 728                 if ( isset( $opts['title'] ) ) {
 729                         $titleText = $opts['title'];
 730                 } else {
 731                         $titleText = 'Parser test';
 732                 }
 733
 734                 $local = isset( $opts['local'] );
 735                 $preprocessor = isset( $opts['preprocessor'] ) ? $opts['preprocessor'] : null;
 736                 $parser = $this->getParser( $preprocessor );
 737
 738                 $title = Title::newFromText( $titleText );
 739
 740                 # Parser test requiring math. Make sure texvc is executable
 741                 # or just skip such tests.
 742                 if ( isset( $opts['math'] ) || isset( $opts['texvc'] ) ) {
 743                         global $wgTexvc;
 744
 745                         if ( !isset( $wgTexvc ) ) {
 746                                 $this->markTestSkipped( "SKIPPED: \$wgTexvc is not set" );
 747                         } elseif ( !is_executable( $wgTexvc ) ) {
 748                                 $this->markTestSkipped( "SKIPPED: texvc binary does not exist"
 749                                         . " or is not executable.\n"
 750                                         . "Current configuration is:\n\$wgTexvc = '$wgTexvc'" );
 751                         }
 752                 }
 753
 754                 if ( isset( $opts['djvu'] ) ) {
 755                         if ( !$this->djVuSupport->isEnabled() ) {
 756                                 $this->markTestSkipped( "SKIPPED: djvu binaries do not exist or are not executable.\n" );
 757                         }
 758                 }
 759
 760                 if ( isset( $opts['tidy'] ) ) {
 761                         if ( !$this->tidySupport->isEnabled() ) {
 762                                 $this->markTestSkipped( "SKIPPED: tidy extension is not installed.\n" );
 763                         } else {
 764                                 $options->setTidy( true );
 765                         }
 766                 }
 767
 768                 if ( isset( $opts['pst'] ) ) {
 769                         $out = $parser->preSaveTransform( $input, $title, $user, $options );
 770                 } elseif ( isset( $opts['msg'] ) ) {
 771                         $out = $parser->transformMsg( $input, $options, $title );
 772                 } elseif ( isset( $opts['section'] ) ) {
 773                         $section = $opts['section'];
 774                         $out = $parser->getSection( $input, $section );
 775                 } elseif ( isset( $opts['replace'] ) ) {
 776                         $section = $opts['replace'][0];
 777                         $replace = $opts['replace'][1];
 778                         $out = $parser->replaceSection( $input, $section, $replace );
 779                 } elseif ( isset( $opts['comment'] ) ) {
 780                         $out = Linker::formatComment( $input, $title, $local );
 781                 } elseif ( isset( $opts['preload'] ) ) {
 782                         $out = $parser->getPreloadText( $input, $title, $options );
 783                 } else {
 784                         $output = $parser->parse( $input, $title, $options, true, true, 1337 );
 785                         $output->setTOCEnabled( !isset( $opts['notoc'] ) );
 786                         $out = $output->getText();
 787                         if ( isset( $opts['tidy'] ) ) {
 788                                 $out = preg_replace( '/\s+$/', '', $out );
 789                         }
 790
 791                         if ( isset( $opts['showtitle'] ) ) {
 792                                 if ( $output->getTitleText() ) {
 793                                         $title = $output->getTitleText();
 794                                 }
 795
 796                                 $out = "$title\n$out";
 797                         }
 798
 799                         if ( isset( $opts['showindicators'] ) ) {
 800                                 $indicators = '';
 801                                 foreach ( $output->getIndicators() as $id => $content ) {
 802                                         $indicators .= "$id=$content\n";
 803                                 }
 804                                 $out = $indicators . $out;
 805                         }
 806
 807                         if ( isset( $opts['ill'] ) ) {
 808                                 $out = implode( ' ', $output->getLanguageLinks() );
 809                         } elseif ( isset( $opts['cat'] ) ) {
 810                                 $outputPage = $context->getOutput();
 811                                 $outputPage->addCategoryLinks( $output->getCategories() );
 812                                 $cats = $outputPage->getCategoryLinks();
 813
 814                                 if ( isset( $cats['normal'] ) ) {
 815                                         $out = implode( ' ', $cats['normal'] );
 816                                 } else {
 817                                         $out = '';
 818                                 }
 819                         }
 820                         $parser->mPreprocessor = null;
 821                 }
 822
 823                 $this->teardownGlobals();
 824
 825                 $this->assertEquals( $result, $out, $desc );
 826         }
 827
 828         /**
 829          * Run a fuzz test series
 830          * Draw input from a set of test files
 831          *
 832          * @todo fixme Needs some work to not eat memory until the world explodes
 833          *
 834          * @group ParserFuzz
 835          */
 836         public function testFuzzTests() {
 837                 global $wgParserTestFiles;
 838
 839                 $files = $wgParserTestFiles;
 840
 841                 if ( $this->getCliArg( 'file' ) ) {
 842                         $files = [ $this->getCliArg( 'file' ) ];
 843                 }
 844
 845                 $dict = $this->getFuzzInput( $files );
 846                 $dictSize = strlen( $dict );
 847                 $logMaxLength = log( $this->maxFuzzTestLength );
 848
 849                 ini_set( 'memory_limit', $this->memoryLimit * 1048576 );
 850
 851                 $user = new User;
 852                 $opts = ParserOptions::newFromUser( $user );
 853                 $title = Title::makeTitle( NS_MAIN, 'Parser_test' );
 854
 855                 $id = 1;
 856
 857                 while ( true ) {
 858
 859                         // Generate test input
 860                         mt_srand( ++$this->fuzzSeed );
 861                         $totalLength = mt_rand( 1, $this->maxFuzzTestLength );
 862                         $input = '';
 863
 864                         while ( strlen( $input ) < $totalLength ) {
 865                                 $logHairLength = mt_rand( 0, 1000000 ) / 1000000 * $logMaxLength;
 866                                 $hairLength = min( intval( exp( $logHairLength ) ), $dictSize );
 867                                 $offset = mt_rand( 0, $dictSize - $hairLength );
 868                                 $input .= substr( $dict, $offset, $hairLength );
 869                         }
 870
 871                         $this->setupGlobals();
 872                         $parser = $this->getParser();
 873
 874                         // Run the test
 875                         try {
 876                                 $parser->parse( $input, $title, $opts );
 877                                 $this->assertTrue( true, "Test $id, fuzz seed {$this->fuzzSeed}" );
 878                         } catch ( Exception $exception ) {
 879                                 $input_dump = sprintf( "string(%d) \"%s\"\n", strlen( $input ), $input );
 880
 881                                 $this->assertTrue( false, "Test $id, fuzz seed {$this->fuzzSeed}. \n\n" .
 882                                         "Input: $input_dump\n\nError: {$exception->getMessage()}\n\n" .
 883                                         "Backtrace: {$exception->getTraceAsString()}" );
 884                         }
 885
 886                         $this->teardownGlobals();
 887                         $parser->__destruct();
 888
 889                         if ( $id % 100 == 0 ) {
 890                                 $usage = intval( memory_get_usage( true ) / $this->memoryLimit / 1048576 * 100 );
 891                                 // echo "{$this->fuzzSeed}: $numSuccess/$numTotal (mem: $usage%)\n";
 892                                 if ( $usage > 90 ) {
 893                                         $ret = "Out of memory:\n";
 894                                         $memStats = $this->getMemoryBreakdown();
 895
 896                                         foreach ( $memStats as $name => $usage ) {
 897                                                 $ret .= "$name: $usage\n";
 898                                         }
 899
 900                                         throw new MWException( $ret );
 901                                 }
 902                         }
 903
 904                         $id++;
 905                 }
 906         }
 907
 908         // Various getter functions
 909
 910         /**
 911          * Get an input dictionary from a set of parser test files
 912          * @param array $filenames
 913          * @return string
 914          */
 915         function getFuzzInput( $filenames ) {
 916                 $dict = '';
 917
 918                 foreach ( $filenames as $filename ) {
 919                         $contents = file_get_contents( $filename );
 920                         preg_match_all( '/!!\s*input\n(.*?)\n!!\s*result/s', $contents, $matches );
 921
 922                         foreach ( $matches[1] as $match ) {
 923                                 $dict .= $match . "\n";
 924                         }
 925                 }
 926
 927                 return $dict;
 928         }
 929
 930         /**
 931          * Get a memory usage breakdown
 932          * @return array
 933          */
 934         function getMemoryBreakdown() {
 935                 $memStats = [];
 936
 937                 foreach ( $GLOBALS as $name => $value ) {
 938                         $memStats['$' . $name] = strlen( serialize( $value ) );
 939                 }
 940
 941                 $classes = get_declared_classes();
 942
 943                 foreach ( $classes as $class ) {
 944                         $rc = new ReflectionClass( $class );
 945                         $props = $rc->getStaticProperties();
 946                         $memStats[$class] = strlen( serialize( $props ) );
 947                         $methods = $rc->getMethods();
 948
 949                         foreach ( $methods as $method ) {
 950                                 $memStats[$class] += strlen( serialize( $method->getStaticVariables() ) );
 951                         }
 952                 }
 953
 954                 $functions = get_defined_functions();
 955
 956                 foreach ( $functions['user'] as $function ) {
 957                         $rf = new ReflectionFunction( $function );
 958                         $memStats["$function()"] = strlen( serialize( $rf->getStaticVariables() ) );
 959                 }
 960
 961                 asort( $memStats );
 962
 963                 return $memStats;
 964         }
 965
 966         /**
 967          * Get a Parser object
 968          * @param Preprocessor $preprocessor
 969          * @return Parser
 970          */
 971         function getParser( $preprocessor = null ) {
 972                 global $wgParserConf;
 973
 974                 $class = $wgParserConf['class'];
 975                 $parser = new $class( [ 'preprocessorClass' => $preprocessor ] + $wgParserConf );
 976
 977                 Hooks::run( 'ParserTestParser', [ &$parser ] );
 978
 979                 return $parser;
 980         }
 981
 982         // Various action functions
 983
 984         public function addArticle( $name, $text, $line ) {
 985                 self::$articles[$name] = [ $text, $line ];
 986         }
 987
 988         public function publishTestArticles() {
 989                 if ( empty( self::$articles ) ) {
 990                         return;
 991                 }
 992
 993                 foreach ( self::$articles as $name => $info ) {
 994                         list( $text, $line ) = $info;
 995                         ParserTest::addArticle( $name, $text, $line, 'ignoreduplicate' );
 996                 }
 997         }
 998
 999         /**
1000          * Steal a callback function from the primary parser, save it for
1001          * application to our scary parser. If the hook is not installed,
1002          * abort processing of this file.
1003          *
1004          * @param string $name
1005          * @return bool True if tag hook is present
1006          */
1007         public function requireHook( $name ) {
1008                 global $wgParser;
1009                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1010                 return isset( $wgParser->mTagHooks[$name] );
1011         }
1012
1013         public function requireFunctionHook( $name ) {
1014                 global $wgParser;
1015                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1016                 return isset( $wgParser->mFunctionHooks[$name] );
1017         }
1018
1019         public function requireTransparentHook( $name ) {
1020                 global $wgParser;
1021                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1022                 return isset( $wgParser->mTransparentTagHooks[$name] );
1023         }
1024
1025         // Various "cleanup" functions
1026
1027         /**
1028          * Remove last character if it is a newline
1029          * @param string $s
1030          * @return string
1031          */
1032         public function removeEndingNewline( $s ) {
1033                 if ( substr( $s, -1 ) === "\n" ) {
1034                         return substr( $s, 0, -1 );
1035                 } else {
1036                         return $s;
1037                 }
1038         }
1039
1040         // Test options parser functions
1041
1042         protected function parseOptions( $instring ) {
1043                 $opts = [];
1044                 // foo
1045                 // foo=bar
1046                 // foo="bar baz"
1047                 // foo=[[bar baz]]
1048                 // foo=bar,"baz quux"
1049                 $regex = '/\b
1050                         ([\w-]+)                                                # Key
1051                         \b
1052                         (?:\s*
1053                                 =                                               # First sub-value
1054                                 \s*
1055                                 (
1056                                         "
1057                                                 [^"]*                   # Quoted val
1058                                         "
1059                                 |
1060                                         \[\[
1061                                                 [^]]*                   # Link target
1062                                         \]\]
1063                                 |
1064                                         [\w-]+                          # Plain word
1065                                 )
1066                                 (?:\s*
1067                                         ,                                       # Sub-vals 1..N
1068                                         \s*
1069                                         (
1070                                                 "[^"]*"                 # Quoted val
1071                                         |
1072                                                 \[\[[^]]*\]\]   # Link target
1073                                         |
1074                                                 [\w-]+                  # Plain word
1075                                         )
1076                                 )*
1077                         )?
1078                         /x';
1079
1080                 if ( preg_match_all( $regex, $instring, $matches, PREG_SET_ORDER ) ) {
1081                         foreach ( $matches as $bits ) {
1082                                 array_shift( $bits );
1083                                 $key = strtolower( array_shift( $bits ) );
1084                                 if ( count( $bits ) == 0 ) {
1085                                         $opts[$key] = true;
1086                                 } elseif ( count( $bits ) == 1 ) {
1087                                         $opts[$key] = $this->cleanupOption( array_shift( $bits ) );
1088                                 } else {
1089                                         // Array!
1090                                         $opts[$key] = array_map( [ $this, 'cleanupOption' ], $bits );
1091                                 }
1092                         }
1093                 }
1094
1095                 return $opts;
1096         }
1097
1098         protected function cleanupOption( $opt ) {
1099                 if ( substr( $opt, 0, 1 ) == '"' ) {
1100                         return substr( $opt, 1, -1 );
1101                 }
1102
1103                 if ( substr( $opt, 0, 2 ) == '[[' ) {
1104                         return substr( $opt, 2, -2 );
1105                 }
1106
1107                 return $opt;
1108         }
1109
1110         /**
1111          * Use a regex to find out the value of an option
1112          * @param string $key Name of option val to retrieve
1113          * @param array $opts Options array to look in
1114          * @param mixed $default Default value returned if not found
1115          * @return mixed
1116          */
1117         protected static function getOptionValue( $key, $opts, $default ) {
1118                 $key = strtolower( $key );
1119
1120                 if ( isset( $opts[$key] ) ) {
1121                         return $opts[$key];
1122                 } else {
1123                         return $default;
1124                 }
1125         }
1126 }