tests/phpunit/includes/parser/NewParserTest.php

   1 <?php
   2
   3 use MediaWiki\MediaWikiServices;
   4
   5 /**
   6  * Although marked as a stub, can work independently.
   7  *
   8  * @group Database
   9  * @group Parser
  10  * @group Stub
  11  *
  12  * @todo covers tags
  13  */
  14 class NewParserTest extends MediaWikiTestCase {
  15         static protected $articles = []; // Array of test articles defined by the tests
  16         /* The data provider is run on a different instance than the test, so it must be static
  17          * When running tests from several files, all tests will see all articles.
  18          */
  19         static protected $backendToUse;
  20
  21         public $keepUploads = false;
  22         public $runDisabled = false;
  23         public $runParsoid = false;
  24         public $regex = '';
  25         public $showProgress = true;
  26         public $savedWeirdGlobals = [];
  27         public $savedGlobals = [];
  28         public $hooks = [];
  29         public $functionHooks = [];
  30         public $transparentHooks = [];
  31
  32         // Fuzz test
  33         public $maxFuzzTestLength = 300;
  34         public $fuzzSeed = 0;
  35         public $memoryLimit = 50;
  36
  37         /**
  38          * @var DjVuSupport
  39          */
  40         private $djVuSupport;
  41         /**
  42          * @var TidySupport
  43          */
  44         private $tidySupport;
  45
  46         protected $file = false;
  47
  48         public static function setUpBeforeClass() {
  49                 // Inject ParserTest well-known interwikis
  50                 ParserTest::setupInterwikis();
  51         }
  52
  53         protected function setUp() {
  54                 global $wgNamespaceAliases, $wgContLang;
  55                 global $wgHooks, $IP;
  56
  57                 parent::setUp();
  58
  59                 // Setup CLI arguments
  60                 if ( $this->getCliArg( 'regex' ) ) {
  61                         $this->regex = $this->getCliArg( 'regex' );
  62                 } else {
  63                         # Matches anything
  64                         $this->regex = '';
  65                 }
  66
  67                 $this->keepUploads = $this->getCliArg( 'keep-uploads' );
  68
  69                 $tmpGlobals = [];
  70
  71                 $tmpGlobals['wgLanguageCode'] = 'en';
  72                 $tmpGlobals['wgContLang'] = Language::factory( 'en' );
  73                 $tmpGlobals['wgSitename'] = 'MediaWiki';
  74                 $tmpGlobals['wgServer'] = 'http://example.org';
  75                 $tmpGlobals['wgServerName'] = 'example.org';
  76                 $tmpGlobals['wgScriptPath'] = '';
  77                 $tmpGlobals['wgScript'] = '/index.php';
  78                 $tmpGlobals['wgResourceBasePath'] = '';
  79                 $tmpGlobals['wgStylePath'] = '/skins';
  80                 $tmpGlobals['wgExtensionAssetsPath'] = '/extensions';
  81                 $tmpGlobals['wgArticlePath'] = '/wiki/$1';
  82                 $tmpGlobals['wgActionPaths'] = [];
  83                 $tmpGlobals['wgVariantArticlePath'] = false;
  84                 $tmpGlobals['wgEnableUploads'] = true;
  85                 $tmpGlobals['wgUploadNavigationUrl'] = false;
  86                 $tmpGlobals['wgThumbnailScriptPath'] = false;
  87                 $tmpGlobals['wgLocalFileRepo'] = [
  88                         'class' => 'LocalRepo',
  89                         'name' => 'local',
  90                         'url' => 'http://example.com/images',
  91                         'hashLevels' => 2,
  92                         'transformVia404' => false,
  93                         'backend' => 'local-backend'
  94                 ];
  95                 $tmpGlobals['wgForeignFileRepos'] = [];
  96                 $tmpGlobals['wgDefaultExternalStore'] = [];
  97                 $tmpGlobals['wgParserCacheType'] = CACHE_NONE;
  98                 $tmpGlobals['wgCapitalLinks'] = true;
  99                 $tmpGlobals['wgNoFollowLinks'] = true;
 100                 $tmpGlobals['wgNoFollowDomainExceptions'] = [ 'no-nofollow.org' ];
 101                 $tmpGlobals['wgExternalLinkTarget'] = false;
 102                 $tmpGlobals['wgThumbnailScriptPath'] = false;
 103                 $tmpGlobals['wgUseImageResize'] = true;
 104                 $tmpGlobals['wgAllowExternalImages'] = true;
 105                 $tmpGlobals['wgRawHtml'] = false;
 106                 $tmpGlobals['wgExperimentalHtmlIds'] = false;
 107                 $tmpGlobals['wgAdaptiveMessageCache'] = true;
 108                 $tmpGlobals['wgUseDatabaseMessages'] = true;
 109                 $tmpGlobals['wgLocaltimezone'] = 'UTC';
 110                 $tmpGlobals['wgGroupPermissions'] = [
 111                         '*' => [
 112                                 'createaccount' => true,
 113                                 'read' => true,
 114                                 'edit' => true,
 115                                 'createpage' => true,
 116                                 'createtalk' => true,
 117                 ] ];
 118                 $tmpGlobals['wgNamespaceProtection'] = [ NS_MEDIAWIKI => 'editinterface' ];
 119
 120                 $tmpGlobals['wgParser'] = new StubObject(
 121                         'wgParser', $GLOBALS['wgParserConf']['class'],
 122                         [ $GLOBALS['wgParserConf'] ] );
 123
 124                 $tmpGlobals['wgFileExtensions'][] = 'svg';
 125                 $tmpGlobals['wgSVGConverter'] = 'rsvg';
 126                 $tmpGlobals['wgSVGConverters']['rsvg'] =
 127                         '$path/rsvg-convert -w $width -h $height -o $output $input';
 128
 129                 if ( $GLOBALS['wgStyleDirectory'] === false ) {
 130                         $tmpGlobals['wgStyleDirectory'] = "$IP/skins";
 131                 }
 132
 133                 $tmpHooks = $wgHooks;
 134                 $tmpHooks['ParserTestParser'][] = 'ParserTestParserHook::setup';
 135                 $tmpHooks['ParserGetVariableValueTs'][] = 'ParserTest::getFakeTimestamp';
 136                 $tmpGlobals['wgHooks'] = $tmpHooks;
 137                 # add a namespace shadowing a interwiki link, to test
 138                 # proper precedence when resolving links. (bug 51680)
 139                 $tmpGlobals['wgExtraNamespaces'] = [
 140                         100 => 'MemoryAlpha',
 141                         101 => 'MemoryAlpha_talk'
 142                 ];
 143
 144                 $tmpGlobals['wgLocalInterwikis'] = [ 'local', 'mi' ];
 145                 # "extra language links"
 146                 # see https://gerrit.wikimedia.org/r/111390
 147                 $tmpGlobals['wgExtraInterlanguageLinkPrefixes'] = [ 'mul' ];
 148
 149                 // DjVu support
 150                 $this->djVuSupport = new DjVuSupport();
 151                 // Tidy support
 152                 $this->tidySupport = new TidySupport();
 153                 $tmpGlobals['wgTidyConfig'] = $this->tidySupport->getConfig();
 154                 $tmpGlobals['wgUseTidy'] = false;
 155
 156                 $this->setMwGlobals( $tmpGlobals );
 157
 158                 $this->savedWeirdGlobals['image_alias'] = $wgNamespaceAliases['Image'];
 159                 $this->savedWeirdGlobals['image_talk_alias'] = $wgNamespaceAliases['Image_talk'];
 160
 161                 $wgNamespaceAliases['Image'] = NS_FILE;
 162                 $wgNamespaceAliases['Image_talk'] = NS_FILE_TALK;
 163
 164                 MWNamespace::getCanonicalNamespaces( true ); # reset namespace cache
 165                 $wgContLang->resetNamespaces(); # reset namespace cache
 166                 ParserTest::resetTitleServices();
 167                 MediaWikiServices::getInstance()->disableService( 'MediaHandlerFactory' );
 168                 MediaWikiServices::getInstance()->redefineService(
 169                         'MediaHandlerFactory',
 170                         function() {
 171                                 return new MockMediaHandlerFactory();
 172                         }
 173                 );
 174         }
 175
 176         protected function tearDown() {
 177                 global $wgNamespaceAliases, $wgContLang;
 178
 179                 $wgNamespaceAliases['Image'] = $this->savedWeirdGlobals['image_alias'];
 180                 $wgNamespaceAliases['Image_talk'] = $this->savedWeirdGlobals['image_talk_alias'];
 181
 182                 MWTidy::destroySingleton();
 183
 184                 // Restore backends
 185                 RepoGroup::destroySingleton();
 186                 FileBackendGroup::destroySingleton();
 187
 188                 // Remove temporary pages from the link cache
 189                 LinkCache::singleton()->clear();
 190
 191                 // Restore message cache (temporary pages and $wgUseDatabaseMessages)
 192                 MessageCache::destroyInstance();
 193                 MediaWikiServices::getInstance()->resetServiceForTesting( 'MediaHandlerFactory' );
 194
 195                 parent::tearDown();
 196
 197                 MWNamespace::getCanonicalNamespaces( true ); # reset namespace cache
 198                 $wgContLang->resetNamespaces(); # reset namespace cache
 199         }
 200
 201         public static function tearDownAfterClass() {
 202                 ParserTest::tearDownInterwikis();
 203                 parent::tearDownAfterClass();
 204         }
 205
 206         function addDBDataOnce() {
 207                 # disabled for performance
 208                 # $this->tablesUsed[] = 'image';
 209
 210                 # Update certain things in site_stats
 211                 $this->db->insert( 'site_stats',
 212                         [ 'ss_row_id' => 1, 'ss_images' => 2, 'ss_good_articles' => 1 ],
 213                         __METHOD__,
 214                         [ 'IGNORE' ]
 215                 );
 216
 217                 $user = User::newFromId( 0 );
 218                 LinkCache::singleton()->clear(); # Avoids the odd failure at creating the nullRevision
 219
 220                 # Upload DB table entries for files.
 221                 # We will upload the actual files later. Note that if anything causes LocalFile::load()
 222                 # to be triggered before then, it will break via maybeUpgrade() setting the fileExists
 223                 # member to false and storing it in cache.
 224                 # note that the size/width/height/bits/etc of the file
 225                 # are actually set by inspecting the file itself; the arguments
 226                 # to recordUpload2 have no effect.  That said, we try to make things
 227                 # match up so it is less confusing to readers of the code & tests.
 228                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Foobar.jpg' ) );
 229                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 230                         $image->recordUpload2(
 231                                 '', // archive name
 232                                 'Upload of some lame file',
 233                                 'Some lame file',
 234                                 [
 235                                         'size' => 7881,
 236                                         'width' => 1941,
 237                                         'height' => 220,
 238                                         'bits' => 8,
 239                                         'media_type' => MEDIATYPE_BITMAP,
 240                                         'mime' => 'image/jpeg',
 241                                         'metadata' => serialize( [] ),
 242                                         'sha1' => Wikimedia\base_convert( '1', 16, 36, 31 ),
 243                                         'fileExists' => true ],
 244                                 $this->db->timestamp( '20010115123500' ), $user
 245                         );
 246                 }
 247
 248                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Thumb.png' ) );
 249                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 250                         $image->recordUpload2(
 251                                 '', // archive name
 252                                 'Upload of some lame thumbnail',
 253                                 'Some lame thumbnail',
 254                                 [
 255                                         'size' => 22589,
 256                                         'width' => 135,
 257                                         'height' => 135,
 258                                         'bits' => 8,
 259                                         'media_type' => MEDIATYPE_BITMAP,
 260                                         'mime' => 'image/png',
 261                                         'metadata' => serialize( [] ),
 262                                         'sha1' => Wikimedia\base_convert( '2', 16, 36, 31 ),
 263                                         'fileExists' => true ],
 264                                 $this->db->timestamp( '20130225203040' ), $user
 265                         );
 266                 }
 267
 268                 # This image will be blacklisted in [[MediaWiki:Bad image list]]
 269                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Bad.jpg' ) );
 270                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 271                         $image->recordUpload2(
 272                                 '', // archive name
 273                                 'zomgnotcensored',
 274                                 'Borderline image',
 275                                 [
 276                                         'size' => 12345,
 277                                         'width' => 320,
 278                                         'height' => 240,
 279                                         'bits' => 24,
 280                                         'media_type' => MEDIATYPE_BITMAP,
 281                                         'mime' => 'image/jpeg',
 282                                         'metadata' => serialize( [] ),
 283                                         'sha1' => Wikimedia\base_convert( '3', 16, 36, 31 ),
 284                                         'fileExists' => true ],
 285                                 $this->db->timestamp( '20010115123500' ), $user
 286                         );
 287                 }
 288                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Foobar.svg' ) );
 289                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 290                         $image->recordUpload2( '', 'Upload of some lame SVG', 'Some lame SVG', [
 291                                         'size'        => 12345,
 292                                         'width'       => 240,
 293                                         'height'      => 180,
 294                                         'bits'        => 0,
 295                                         'media_type'  => MEDIATYPE_DRAWING,
 296                                         'mime'        => 'image/svg+xml',
 297                                         'metadata'    => serialize( [] ),
 298                                         'sha1'        => Wikimedia\base_convert( '', 16, 36, 31 ),
 299                                         'fileExists'  => true
 300                         ], $this->db->timestamp( '20010115123500' ), $user );
 301                 }
 302
 303                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Video.ogv' ) );
 304                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 305                         $image->recordUpload2( '', 'A pretty movie', 'Will it play', [
 306                                         'size'        => 12345,
 307                                         'width'       => 320,
 308                                         'height'      => 240,
 309                                         'bits'        => 0,
 310                                         'media_type'  => MEDIATYPE_VIDEO,
 311                                         'mime'        => 'application/ogg',
 312                                         'metadata'    => serialize( [] ),
 313                                         'sha1'        => Wikimedia\base_convert( '', 16, 36, 32 ),
 314                                         'fileExists'  => true
 315                         ], $this->db->timestamp( '20010115123500' ), $user );
 316                 }
 317
 318                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'Audio.oga' ) );
 319                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 320                         $image->recordUpload2( '', 'An awesome hitsong ', 'Will it play', [
 321                                         'size'        => 12345,
 322                                         'width'       => 0,
 323                                         'height'      => 0,
 324                                         'bits'        => 0,
 325                                         'media_type'  => MEDIATYPE_AUDIO,
 326                                         'mime'        => 'application/ogg',
 327                                         'metadata'    => serialize( [] ),
 328                                         'sha1'        => Wikimedia\base_convert( '', 16, 36, 32 ),
 329                                         'fileExists'  => true
 330                         ], $this->db->timestamp( '20010115123500' ), $user );
 331                 }
 332
 333                 # A DjVu file
 334                 $image = wfLocalFile( Title::makeTitle( NS_FILE, 'LoremIpsum.djvu' ) );
 335                 if ( !$this->db->selectField( 'image', '1', [ 'img_name' => $image->getName() ] ) ) {
 336                         $image->recordUpload2( '', 'Upload a DjVu', 'A DjVu', [
 337                                 'size' => 3249,
 338                                 'width' => 2480,
 339                                 'height' => 3508,
 340                                 'bits' => 0,
 341                                 'media_type' => MEDIATYPE_BITMAP,
 342                                 'mime' => 'image/vnd.djvu',
 343                                 'metadata' => '<?xml version="1.0" ?>
 344 <!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd">
 345 <DjVuXML>
 346 <HEAD></HEAD>
 347 <BODY><OBJECT height="3508" width="2480">
 348 <PARAM name="DPI" value="300" />
 349 <PARAM name="GAMMA" value="2.2" />
 350 </OBJECT>
 351 <OBJECT height="3508" width="2480">
 352 <PARAM name="DPI" value="300" />
 353 <PARAM name="GAMMA" value="2.2" />
 354 </OBJECT>
 355 <OBJECT height="3508" width="2480">
 356 <PARAM name="DPI" value="300" />
 357 <PARAM name="GAMMA" value="2.2" />
 358 </OBJECT>
 359 <OBJECT height="3508" width="2480">
 360 <PARAM name="DPI" value="300" />
 361 <PARAM name="GAMMA" value="2.2" />
 362 </OBJECT>
 363 <OBJECT height="3508" width="2480">
 364 <PARAM name="DPI" value="300" />
 365 <PARAM name="GAMMA" value="2.2" />
 366 </OBJECT>
 367 </BODY>
 368 </DjVuXML>',
 369                                 'sha1' => Wikimedia\base_convert( '', 16, 36, 31 ),
 370                                 'fileExists' => true
 371                         ], $this->db->timestamp( '20140115123600' ), $user );
 372                 }
 373         }
 374
 375         // ParserTest setup/teardown functions
 376
 377         /**
 378          * Set up the global variables for a consistent environment for each test.
 379          * Ideally this should replace the global configuration entirely.
 380          * @param array $opts
 381          * @param string $config
 382          * @return RequestContext
 383          */
 384         protected function setupGlobals( $opts = [], $config = '' ) {
 385                 global $wgFileBackends;
 386                 # Find out values for some special options.
 387                 $lang =
 388                         self::getOptionValue( 'language', $opts, 'en' );
 389                 $variant =
 390                         self::getOptionValue( 'variant', $opts, false );
 391                 $maxtoclevel =
 392                         self::getOptionValue( 'wgMaxTocLevel', $opts, 999 );
 393                 $linkHolderBatchSize =
 394                         self::getOptionValue( 'wgLinkHolderBatchSize', $opts, 1000 );
 395
 396                 $uploadDir = $this->getUploadDir();
 397                 if ( $this->getCliArg( 'use-filebackend' ) ) {
 398                         if ( self::$backendToUse ) {
 399                                 $backend = self::$backendToUse;
 400                         } else {
 401                                 $name = $this->getCliArg( 'use-filebackend' );
 402                                 $useConfig = [];
 403                                 foreach ( $wgFileBackends as $conf ) {
 404                                         if ( $conf['name'] == $name ) {
 405                                                 $useConfig = $conf;
 406                                         }
 407                                 }
 408                                 $useConfig['name'] = 'local-backend'; // swap name
 409                                 unset( $useConfig['lockManager'] );
 410                                 unset( $useConfig['fileJournal'] );
 411                                 $class = $useConfig['class'];
 412                                 self::$backendToUse = new $class( $useConfig );
 413                                 $backend = self::$backendToUse;
 414                         }
 415                 } else {
 416                         # Replace with a mock. We do not care about generating real
 417                         # files on the filesystem, just need to expose the file
 418                         # informations.
 419                         $backend = new MockFileBackend( [
 420                                 'name' => 'local-backend',
 421                                 'wikiId' => wfWikiID()
 422                         ] );
 423                 }
 424
 425                 $settings = [
 426                         'wgLocalFileRepo' => [
 427                                 'class' => 'LocalRepo',
 428                                 'name' => 'local',
 429                                 'url' => 'http://example.com/images',
 430                                 'hashLevels' => 2,
 431                                 'transformVia404' => false,
 432                                 'backend' => $backend
 433                         ],
 434                         'wgEnableUploads' => self::getOptionValue( 'wgEnableUploads', $opts, true ),
 435                         'wgLanguageCode' => $lang,
 436                         'wgDBprefix' => $this->db->getType() != 'oracle' ? 'unittest_' : 'ut_',
 437                         'wgRawHtml' => self::getOptionValue( 'wgRawHtml', $opts, false ),
 438                         'wgNamespacesWithSubpages' => [ NS_MAIN => isset( $opts['subpage'] ) ],
 439                         'wgAllowExternalImages' => self::getOptionValue( 'wgAllowExternalImages', $opts, true ),
 440                         'wgThumbLimits' => [ self::getOptionValue( 'thumbsize', $opts, 180 ) ],
 441                         'wgMaxTocLevel' => $maxtoclevel,
 442                         'wgUseTeX' => isset( $opts['math'] ) || isset( $opts['texvc'] ),
 443                         'wgMathDirectory' => $uploadDir . '/math',
 444                         'wgDefaultLanguageVariant' => $variant,
 445                         'wgLinkHolderBatchSize' => $linkHolderBatchSize,
 446                         'wgUseTidy' => false,
 447                         'wgTidyConfig' => isset( $opts['tidy'] ) ? $this->tidySupport->getConfig() : null
 448                 ];
 449
 450                 if ( $config ) {
 451                         $configLines = explode( "\n", $config );
 452
 453                         foreach ( $configLines as $line ) {
 454                                 list( $var, $value ) = explode( '=', $line, 2 );
 455
 456                                 $settings[$var] = eval( "return $value;" ); // ???
 457                         }
 458                 }
 459
 460                 $this->savedGlobals = [];
 461
 462                 /** @since 1.20 */
 463                 Hooks::run( 'ParserTestGlobals', [ &$settings ] );
 464
 465                 $langObj = Language::factory( $lang );
 466                 $settings['wgContLang'] = $langObj;
 467                 $settings['wgLang'] = $langObj;
 468
 469                 $context = new RequestContext();
 470                 $settings['wgOut'] = $context->getOutput();
 471                 $settings['wgUser'] = $context->getUser();
 472                 $settings['wgRequest'] = $context->getRequest();
 473
 474                 // We (re)set $wgThumbLimits to a single-element array above.
 475                 $context->getUser()->setOption( 'thumbsize', 0 );
 476
 477                 foreach ( $settings as $var => $val ) {
 478                         if ( array_key_exists( $var, $GLOBALS ) ) {
 479                                 $this->savedGlobals[$var] = $GLOBALS[$var];
 480                         }
 481
 482                         $GLOBALS[$var] = $val;
 483                 }
 484
 485                 MWTidy::destroySingleton();
 486                 MagicWord::clearCache();
 487
 488                 # The entries saved into RepoGroup cache with previous globals will be wrong.
 489                 RepoGroup::destroySingleton();
 490                 FileBackendGroup::destroySingleton();
 491
 492                 # Create dummy files in storage
 493                 $this->setupUploads();
 494
 495                 # Publish the articles after we have the final language set
 496                 $this->publishTestArticles();
 497
 498                 MessageCache::destroyInstance();
 499
 500                 return $context;
 501         }
 502
 503         /**
 504          * Get an FS upload directory (only applies to FSFileBackend)
 505          *
 506          * @return string The directory
 507          */
 508         protected function getUploadDir() {
 509                 if ( $this->keepUploads ) {
 510                         // Don't use getNewTempDirectory() as this is meant to persist
 511                         $dir = wfTempDir() . '/mwParser-images';
 512
 513                         if ( is_dir( $dir ) ) {
 514                                 return $dir;
 515                         }
 516                 } else {
 517                         $dir = $this->getNewTempDirectory();
 518                 }
 519
 520                 if ( file_exists( $dir ) ) {
 521                         wfDebug( "Already exists!\n" );
 522
 523                         return $dir;
 524                 }
 525
 526                 return $dir;
 527         }
 528
 529         /**
 530          * Create a dummy uploads directory which will contain a couple
 531          * of files in order to pass existence tests.
 532          *
 533          * @return string The directory
 534          */
 535         protected function setupUploads() {
 536                 global $IP;
 537
 538                 $base = $this->getBaseDir();
 539                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 540                 $backend->prepare( [ 'dir' => "$base/local-public/3/3a" ] );
 541                 $backend->store( [
 542                         'src' => "$IP/tests/phpunit/data/parser/headbg.jpg",
 543                         'dst' => "$base/local-public/3/3a/Foobar.jpg"
 544                 ] );
 545                 $backend->prepare( [ 'dir' => "$base/local-public/e/ea" ] );
 546                 $backend->store( [
 547                         'src' => "$IP/tests/phpunit/data/parser/wiki.png",
 548                         'dst' => "$base/local-public/e/ea/Thumb.png"
 549                 ] );
 550                 $backend->prepare( [ 'dir' => "$base/local-public/0/09" ] );
 551                 $backend->store( [
 552                         'src' => "$IP/tests/phpunit/data/parser/headbg.jpg",
 553                         'dst' => "$base/local-public/0/09/Bad.jpg"
 554                 ] );
 555                 $backend->prepare( [ 'dir' => "$base/local-public/5/5f" ] );
 556                 $backend->store( [
 557                         'src' => "$IP/tests/phpunit/data/parser/LoremIpsum.djvu",
 558                         'dst' => "$base/local-public/5/5f/LoremIpsum.djvu"
 559                 ] );
 560
 561                 // No helpful SVG file to copy, so make one ourselves
 562                 $data = '<?xml version="1.0" encoding="utf-8"?>' .
 563                         '<svg xmlns="http://www.w3.org/2000/svg"' .
 564                         ' version="1.1" width="240" height="180"/>';
 565
 566                 $backend->prepare( [ 'dir' => "$base/local-public/f/ff" ] );
 567                 $backend->quickCreate( [
 568                         'content' => $data, 'dst' => "$base/local-public/f/ff/Foobar.svg"
 569                 ] );
 570         }
 571
 572         /**
 573          * Restore default values and perform any necessary clean-up
 574          * after each test runs.
 575          */
 576         protected function teardownGlobals() {
 577                 $this->teardownUploads();
 578
 579                 foreach ( $this->savedGlobals as $var => $val ) {
 580                         $GLOBALS[$var] = $val;
 581                 }
 582         }
 583
 584         /**
 585          * Remove the dummy uploads directory
 586          */
 587         private function teardownUploads() {
 588                 if ( $this->keepUploads ) {
 589                         return;
 590                 }
 591
 592                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 593                 if ( $backend instanceof MockFileBackend ) {
 594                         # In memory backend, so dont bother cleaning them up.
 595                         return;
 596                 }
 597
 598                 $base = $this->getBaseDir();
 599                 // delete the files first, then the dirs.
 600                 self::deleteFiles(
 601                         [
 602                                 "$base/local-public/3/3a/Foobar.jpg",
 603                                 "$base/local-thumb/3/3a/Foobar.jpg/1000px-Foobar.jpg",
 604                                 "$base/local-thumb/3/3a/Foobar.jpg/100px-Foobar.jpg",
 605                                 "$base/local-thumb/3/3a/Foobar.jpg/120px-Foobar.jpg",
 606                                 "$base/local-thumb/3/3a/Foobar.jpg/1280px-Foobar.jpg",
 607                                 "$base/local-thumb/3/3a/Foobar.jpg/137px-Foobar.jpg",
 608                                 "$base/local-thumb/3/3a/Foobar.jpg/1500px-Foobar.jpg",
 609                                 "$base/local-thumb/3/3a/Foobar.jpg/177px-Foobar.jpg",
 610                                 "$base/local-thumb/3/3a/Foobar.jpg/180px-Foobar.jpg",
 611                                 "$base/local-thumb/3/3a/Foobar.jpg/200px-Foobar.jpg",
 612                                 "$base/local-thumb/3/3a/Foobar.jpg/206px-Foobar.jpg",
 613                                 "$base/local-thumb/3/3a/Foobar.jpg/20px-Foobar.jpg",
 614                                 "$base/local-thumb/3/3a/Foobar.jpg/220px-Foobar.jpg",
 615                                 "$base/local-thumb/3/3a/Foobar.jpg/265px-Foobar.jpg",
 616                                 "$base/local-thumb/3/3a/Foobar.jpg/270px-Foobar.jpg",
 617                                 "$base/local-thumb/3/3a/Foobar.jpg/274px-Foobar.jpg",
 618                                 "$base/local-thumb/3/3a/Foobar.jpg/300px-Foobar.jpg",
 619                                 "$base/local-thumb/3/3a/Foobar.jpg/30px-Foobar.jpg",
 620                                 "$base/local-thumb/3/3a/Foobar.jpg/330px-Foobar.jpg",
 621                                 "$base/local-thumb/3/3a/Foobar.jpg/353px-Foobar.jpg",
 622                                 "$base/local-thumb/3/3a/Foobar.jpg/360px-Foobar.jpg",
 623                                 "$base/local-thumb/3/3a/Foobar.jpg/400px-Foobar.jpg",
 624                                 "$base/local-thumb/3/3a/Foobar.jpg/40px-Foobar.jpg",
 625                                 "$base/local-thumb/3/3a/Foobar.jpg/440px-Foobar.jpg",
 626                                 "$base/local-thumb/3/3a/Foobar.jpg/442px-Foobar.jpg",
 627                                 "$base/local-thumb/3/3a/Foobar.jpg/450px-Foobar.jpg",
 628                                 "$base/local-thumb/3/3a/Foobar.jpg/50px-Foobar.jpg",
 629                                 "$base/local-thumb/3/3a/Foobar.jpg/600px-Foobar.jpg",
 630                                 "$base/local-thumb/3/3a/Foobar.jpg/640px-Foobar.jpg",
 631                                 "$base/local-thumb/3/3a/Foobar.jpg/70px-Foobar.jpg",
 632                                 "$base/local-thumb/3/3a/Foobar.jpg/75px-Foobar.jpg",
 633                                 "$base/local-thumb/3/3a/Foobar.jpg/960px-Foobar.jpg",
 634
 635                                 "$base/local-public/e/ea/Thumb.png",
 636
 637                                 "$base/local-public/0/09/Bad.jpg",
 638
 639                                 "$base/local-public/5/5f/LoremIpsum.djvu",
 640                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-2480px-LoremIpsum.djvu.jpg",
 641                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-3720px-LoremIpsum.djvu.jpg",
 642                                 "$base/local-thumb/5/5f/LoremIpsum.djvu/page2-4960px-LoremIpsum.djvu.jpg",
 643
 644                                 "$base/local-public/f/ff/Foobar.svg",
 645                                 "$base/local-thumb/f/ff/Foobar.svg/180px-Foobar.svg.png",
 646                                 "$base/local-thumb/f/ff/Foobar.svg/2000px-Foobar.svg.png",
 647                                 "$base/local-thumb/f/ff/Foobar.svg/270px-Foobar.svg.png",
 648                                 "$base/local-thumb/f/ff/Foobar.svg/3000px-Foobar.svg.png",
 649                                 "$base/local-thumb/f/ff/Foobar.svg/360px-Foobar.svg.png",
 650                                 "$base/local-thumb/f/ff/Foobar.svg/4000px-Foobar.svg.png",
 651                                 "$base/local-thumb/f/ff/Foobar.svg/langde-180px-Foobar.svg.png",
 652                                 "$base/local-thumb/f/ff/Foobar.svg/langde-270px-Foobar.svg.png",
 653                                 "$base/local-thumb/f/ff/Foobar.svg/langde-360px-Foobar.svg.png",
 654
 655                                 "$base/local-public/math/f/a/5/fa50b8b616463173474302ca3e63586b.png",
 656                         ]
 657                 );
 658         }
 659
 660         /**
 661          * Delete the specified files, if they exist.
 662          * @param array $files Full paths to files to delete.
 663          */
 664         private static function deleteFiles( $files ) {
 665                 $backend = RepoGroup::singleton()->getLocalRepo()->getBackend();
 666                 foreach ( $files as $file ) {
 667                         $backend->delete( [ 'src' => $file ], [ 'force' => 1 ] );
 668                 }
 669                 foreach ( $files as $file ) {
 670                         $tmp = FileBackend::parentStoragePath( $file );
 671                         while ( $tmp ) {
 672                                 if ( !$backend->clean( [ 'dir' => $tmp ] )->isOK() ) {
 673                                         break;
 674                                 }
 675                                 $tmp = FileBackend::parentStoragePath( $tmp );
 676                         }
 677                 }
 678         }
 679
 680         protected function getBaseDir() {
 681                 return 'mwstore://local-backend';
 682         }
 683
 684         public function parserTestProvider() {
 685                 if ( $this->file === false ) {
 686                         global $wgParserTestFiles;
 687                         $this->file = $wgParserTestFiles[0];
 688                 }
 689
 690                 return new TestFileDataProvider( $this->file, $this );
 691         }
 692
 693         /**
 694          * Set the file from whose tests will be run by this instance
 695          * @param string $filename
 696          */
 697         public function setParserTestFile( $filename ) {
 698                 $this->file = $filename;
 699         }
 700
 701         /**
 702          * @group medium
 703          * @group ParserTests
 704          * @dataProvider parserTestProvider
 705          * @param string $desc
 706          * @param string $input
 707          * @param string $result
 708          * @param array $opts
 709          * @param array $config
 710          */
 711         public function testParserTest( $desc, $input, $result, $opts, $config ) {
 712                 if ( $this->regex != '' && !preg_match( '/' . $this->regex . '/', $desc ) ) {
 713                         $this->assertTrue( true ); // XXX: don't flood output with "test made no assertions"
 714                         // $this->markTestSkipped( 'Filtered out by the user' );
 715                         return;
 716                 }
 717
 718                 if ( !$this->isWikitextNS( NS_MAIN ) ) {
 719                         // parser tests frequently assume that the main namespace contains wikitext.
 720                         // @todo When setting up pages, force the content model. Only skip if
 721                         //        $wgtContentModelUseDB is false.
 722                         $this->markTestSkipped( "Main namespace does not support wikitext,"
 723                                 . "skipping parser test: $desc" );
 724                 }
 725
 726                 wfDebug( "Running parser test: $desc\n" );
 727
 728                 $opts = $this->parseOptions( $opts );
 729                 $context = $this->setupGlobals( $opts, $config );
 730
 731                 $user = $context->getUser();
 732                 $options = ParserOptions::newFromContext( $context );
 733
 734                 if ( isset( $opts['title'] ) ) {
 735                         $titleText = $opts['title'];
 736                 } else {
 737                         $titleText = 'Parser test';
 738                 }
 739
 740                 $local = isset( $opts['local'] );
 741                 $preprocessor = isset( $opts['preprocessor'] ) ? $opts['preprocessor'] : null;
 742                 $parser = $this->getParser( $preprocessor );
 743
 744                 $title = Title::newFromText( $titleText );
 745
 746                 # Parser test requiring math. Make sure texvc is executable
 747                 # or just skip such tests.
 748                 if ( isset( $opts['math'] ) || isset( $opts['texvc'] ) ) {
 749                         global $wgTexvc;
 750
 751                         if ( !isset( $wgTexvc ) ) {
 752                                 $this->markTestSkipped( "SKIPPED: \$wgTexvc is not set" );
 753                         } elseif ( !is_executable( $wgTexvc ) ) {
 754                                 $this->markTestSkipped( "SKIPPED: texvc binary does not exist"
 755                                         . " or is not executable.\n"
 756                                         . "Current configuration is:\n\$wgTexvc = '$wgTexvc'" );
 757                         }
 758                 }
 759
 760                 if ( isset( $opts['djvu'] ) ) {
 761                         if ( !$this->djVuSupport->isEnabled() ) {
 762                                 $this->markTestSkipped( "SKIPPED: djvu binaries do not exist or are not executable.\n" );
 763                         }
 764                 }
 765
 766                 if ( isset( $opts['tidy'] ) ) {
 767                         if ( !$this->tidySupport->isEnabled() ) {
 768                                 $this->markTestSkipped( "SKIPPED: tidy extension is not installed.\n" );
 769                         } else {
 770                                 $options->setTidy( true );
 771                         }
 772                 }
 773
 774                 if ( isset( $opts['pst'] ) ) {
 775                         $out = $parser->preSaveTransform( $input, $title, $user, $options );
 776                 } elseif ( isset( $opts['msg'] ) ) {
 777                         $out = $parser->transformMsg( $input, $options, $title );
 778                 } elseif ( isset( $opts['section'] ) ) {
 779                         $section = $opts['section'];
 780                         $out = $parser->getSection( $input, $section );
 781                 } elseif ( isset( $opts['replace'] ) ) {
 782                         $section = $opts['replace'][0];
 783                         $replace = $opts['replace'][1];
 784                         $out = $parser->replaceSection( $input, $section, $replace );
 785                 } elseif ( isset( $opts['comment'] ) ) {
 786                         $out = Linker::formatComment( $input, $title, $local );
 787                 } elseif ( isset( $opts['preload'] ) ) {
 788                         $out = $parser->getPreloadText( $input, $title, $options );
 789                 } else {
 790                         $output = $parser->parse( $input, $title, $options, true, true, 1337 );
 791                         $output->setTOCEnabled( !isset( $opts['notoc'] ) );
 792                         $out = $output->getText();
 793                         if ( isset( $opts['tidy'] ) ) {
 794                                 $out = preg_replace( '/\s+$/', '', $out );
 795                         }
 796
 797                         if ( isset( $opts['showtitle'] ) ) {
 798                                 if ( $output->getTitleText() ) {
 799                                         $title = $output->getTitleText();
 800                                 }
 801
 802                                 $out = "$title\n$out";
 803                         }
 804
 805                         if ( isset( $opts['showindicators'] ) ) {
 806                                 $indicators = '';
 807                                 foreach ( $output->getIndicators() as $id => $content ) {
 808                                         $indicators .= "$id=$content\n";
 809                                 }
 810                                 $out = $indicators . $out;
 811                         }
 812
 813                         if ( isset( $opts['ill'] ) ) {
 814                                 $out = implode( ' ', $output->getLanguageLinks() );
 815                         } elseif ( isset( $opts['cat'] ) ) {
 816                                 $outputPage = $context->getOutput();
 817                                 $outputPage->addCategoryLinks( $output->getCategories() );
 818                                 $cats = $outputPage->getCategoryLinks();
 819
 820                                 if ( isset( $cats['normal'] ) ) {
 821                                         $out = implode( ' ', $cats['normal'] );
 822                                 } else {
 823                                         $out = '';
 824                                 }
 825                         }
 826                         $parser->mPreprocessor = null;
 827                 }
 828
 829                 $this->teardownGlobals();
 830
 831                 $this->assertEquals( $result, $out, $desc );
 832         }
 833
 834         /**
 835          * Run a fuzz test series
 836          * Draw input from a set of test files
 837          *
 838          * @todo fixme Needs some work to not eat memory until the world explodes
 839          *
 840          * @group ParserFuzz
 841          */
 842         public function testFuzzTests() {
 843                 global $wgParserTestFiles;
 844
 845                 $files = $wgParserTestFiles;
 846
 847                 if ( $this->getCliArg( 'file' ) ) {
 848                         $files = [ $this->getCliArg( 'file' ) ];
 849                 }
 850
 851                 $dict = $this->getFuzzInput( $files );
 852                 $dictSize = strlen( $dict );
 853                 $logMaxLength = log( $this->maxFuzzTestLength );
 854
 855                 ini_set( 'memory_limit', $this->memoryLimit * 1048576 );
 856
 857                 $user = new User;
 858                 $opts = ParserOptions::newFromUser( $user );
 859                 $title = Title::makeTitle( NS_MAIN, 'Parser_test' );
 860
 861                 $id = 1;
 862
 863                 while ( true ) {
 864
 865                         // Generate test input
 866                         mt_srand( ++$this->fuzzSeed );
 867                         $totalLength = mt_rand( 1, $this->maxFuzzTestLength );
 868                         $input = '';
 869
 870                         while ( strlen( $input ) < $totalLength ) {
 871                                 $logHairLength = mt_rand( 0, 1000000 ) / 1000000 * $logMaxLength;
 872                                 $hairLength = min( intval( exp( $logHairLength ) ), $dictSize );
 873                                 $offset = mt_rand( 0, $dictSize - $hairLength );
 874                                 $input .= substr( $dict, $offset, $hairLength );
 875                         }
 876
 877                         $this->setupGlobals();
 878                         $parser = $this->getParser();
 879
 880                         // Run the test
 881                         try {
 882                                 $parser->parse( $input, $title, $opts );
 883                                 $this->assertTrue( true, "Test $id, fuzz seed {$this->fuzzSeed}" );
 884                         } catch ( Exception $exception ) {
 885                                 $input_dump = sprintf( "string(%d) \"%s\"\n", strlen( $input ), $input );
 886
 887                                 $this->assertTrue( false, "Test $id, fuzz seed {$this->fuzzSeed}. \n\n" .
 888                                         "Input: $input_dump\n\nError: {$exception->getMessage()}\n\n" .
 889                                         "Backtrace: {$exception->getTraceAsString()}" );
 890                         }
 891
 892                         $this->teardownGlobals();
 893                         $parser->__destruct();
 894
 895                         if ( $id % 100 == 0 ) {
 896                                 $usage = intval( memory_get_usage( true ) / $this->memoryLimit / 1048576 * 100 );
 897                                 // echo "{$this->fuzzSeed}: $numSuccess/$numTotal (mem: $usage%)\n";
 898                                 if ( $usage > 90 ) {
 899                                         $ret = "Out of memory:\n";
 900                                         $memStats = $this->getMemoryBreakdown();
 901
 902                                         foreach ( $memStats as $name => $usage ) {
 903                                                 $ret .= "$name: $usage\n";
 904                                         }
 905
 906                                         throw new MWException( $ret );
 907                                 }
 908                         }
 909
 910                         $id++;
 911                 }
 912         }
 913
 914         // Various getter functions
 915
 916         /**
 917          * Get an input dictionary from a set of parser test files
 918          * @param array $filenames
 919          * @return string
 920          */
 921         function getFuzzInput( $filenames ) {
 922                 $dict = '';
 923
 924                 foreach ( $filenames as $filename ) {
 925                         $contents = file_get_contents( $filename );
 926                         preg_match_all( '/!!\s*input\n(.*?)\n!!\s*result/s', $contents, $matches );
 927
 928                         foreach ( $matches[1] as $match ) {
 929                                 $dict .= $match . "\n";
 930                         }
 931                 }
 932
 933                 return $dict;
 934         }
 935
 936         /**
 937          * Get a memory usage breakdown
 938          * @return array
 939          */
 940         function getMemoryBreakdown() {
 941                 $memStats = [];
 942
 943                 foreach ( $GLOBALS as $name => $value ) {
 944                         $memStats['$' . $name] = strlen( serialize( $value ) );
 945                 }
 946
 947                 $classes = get_declared_classes();
 948
 949                 foreach ( $classes as $class ) {
 950                         $rc = new ReflectionClass( $class );
 951                         $props = $rc->getStaticProperties();
 952                         $memStats[$class] = strlen( serialize( $props ) );
 953                         $methods = $rc->getMethods();
 954
 955                         foreach ( $methods as $method ) {
 956                                 $memStats[$class] += strlen( serialize( $method->getStaticVariables() ) );
 957                         }
 958                 }
 959
 960                 $functions = get_defined_functions();
 961
 962                 foreach ( $functions['user'] as $function ) {
 963                         $rf = new ReflectionFunction( $function );
 964                         $memStats["$function()"] = strlen( serialize( $rf->getStaticVariables() ) );
 965                 }
 966
 967                 asort( $memStats );
 968
 969                 return $memStats;
 970         }
 971
 972         /**
 973          * Get a Parser object
 974          * @param Preprocessor $preprocessor
 975          * @return Parser
 976          */
 977         function getParser( $preprocessor = null ) {
 978                 global $wgParserConf;
 979
 980                 $class = $wgParserConf['class'];
 981                 $parser = new $class( [ 'preprocessorClass' => $preprocessor ] + $wgParserConf );
 982
 983                 Hooks::run( 'ParserTestParser', [ &$parser ] );
 984
 985                 return $parser;
 986         }
 987
 988         // Various action functions
 989
 990         public function addArticle( $name, $text, $line ) {
 991                 self::$articles[$name] = [ $text, $line ];
 992         }
 993
 994         public function publishTestArticles() {
 995                 if ( empty( self::$articles ) ) {
 996                         return;
 997                 }
 998
 999                 foreach ( self::$articles as $name => $info ) {
1000                         list( $text, $line ) = $info;
1001                         ParserTest::addArticle( $name, $text, $line, 'ignoreduplicate' );
1002                 }
1003         }
1004
1005         /**
1006          * Steal a callback function from the primary parser, save it for
1007          * application to our scary parser. If the hook is not installed,
1008          * abort processing of this file.
1009          *
1010          * @param string $name
1011          * @return bool True if tag hook is present
1012          */
1013         public function requireHook( $name ) {
1014                 global $wgParser;
1015                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1016                 return isset( $wgParser->mTagHooks[$name] );
1017         }
1018
1019         public function requireFunctionHook( $name ) {
1020                 global $wgParser;
1021                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1022                 return isset( $wgParser->mFunctionHooks[$name] );
1023         }
1024
1025         public function requireTransparentHook( $name ) {
1026                 global $wgParser;
1027                 $wgParser->firstCallInit(); // make sure hooks are loaded.
1028                 return isset( $wgParser->mTransparentTagHooks[$name] );
1029         }
1030
1031         // Various "cleanup" functions
1032
1033         /**
1034          * Remove last character if it is a newline
1035          * @param string $s
1036          * @return string
1037          */
1038         public function removeEndingNewline( $s ) {
1039                 if ( substr( $s, -1 ) === "\n" ) {
1040                         return substr( $s, 0, -1 );
1041                 } else {
1042                         return $s;
1043                 }
1044         }
1045
1046         // Test options parser functions
1047
1048         protected function parseOptions( $instring ) {
1049                 $opts = [];
1050                 // foo
1051                 // foo=bar
1052                 // foo="bar baz"
1053                 // foo=[[bar baz]]
1054                 // foo=bar,"baz quux"
1055                 $regex = '/\b
1056                         ([\w-]+)                                                # Key
1057                         \b
1058                         (?:\s*
1059                                 =                                               # First sub-value
1060                                 \s*
1061                                 (
1062                                         "
1063                                                 [^"]*                   # Quoted val
1064                                         "
1065                                 |
1066                                         \[\[
1067                                                 [^]]*                   # Link target
1068                                         \]\]
1069                                 |
1070                                         [\w-]+                          # Plain word
1071                                 )
1072                                 (?:\s*
1073                                         ,                                       # Sub-vals 1..N
1074                                         \s*
1075                                         (
1076                                                 "[^"]*"                 # Quoted val
1077                                         |
1078                                                 \[\[[^]]*\]\]   # Link target
1079                                         |
1080                                                 [\w-]+                  # Plain word
1081                                         )
1082                                 )*
1083                         )?
1084                         /x';
1085
1086                 if ( preg_match_all( $regex, $instring, $matches, PREG_SET_ORDER ) ) {
1087                         foreach ( $matches as $bits ) {
1088                                 array_shift( $bits );
1089                                 $key = strtolower( array_shift( $bits ) );
1090                                 if ( count( $bits ) == 0 ) {
1091                                         $opts[$key] = true;
1092                                 } elseif ( count( $bits ) == 1 ) {
1093                                         $opts[$key] = $this->cleanupOption( array_shift( $bits ) );
1094                                 } else {
1095                                         // Array!
1096                                         $opts[$key] = array_map( [ $this, 'cleanupOption' ], $bits );
1097                                 }
1098                         }
1099                 }
1100
1101                 return $opts;
1102         }
1103
1104         protected function cleanupOption( $opt ) {
1105                 if ( substr( $opt, 0, 1 ) == '"' ) {
1106                         return substr( $opt, 1, -1 );
1107                 }
1108
1109                 if ( substr( $opt, 0, 2 ) == '[[' ) {
1110                         return substr( $opt, 2, -2 );
1111                 }
1112
1113                 return $opt;
1114         }
1115
1116         /**
1117          * Use a regex to find out the value of an option
1118          * @param string $key Name of option val to retrieve
1119          * @param array $opts Options array to look in
1120          * @param mixed $default Default value returned if not found
1121          * @return mixed
1122          */
1123         protected static function getOptionValue( $key, $opts, $default ) {
1124                 $key = strtolower( $key );
1125
1126                 if ( isset( $opts[$key] ) ) {
1127                         return $opts[$key];
1128                 } else {
1129                         return $default;
1130                 }
1131         }
1132 }