Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 551 |
|
0.00% |
0 / 20 |
CRAP | |
0.00% |
0 / 1 |
| TestRunner | |
0.00% |
0 / 551 |
|
0.00% |
0 / 20 |
31862 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
6 | |||
| newEnv | |
0.00% |
0 / 19 |
|
0.00% |
0 / 1 |
6 | |||
| normalizeTitleKey | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addArticle | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
2 | |||
| buildTests | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
20 | |||
| convertWt2Html | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
6 | |||
| convertHtml2Wt | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
20 | |||
| runTest | |
0.00% |
0 / 69 |
|
0.00% |
0 / 1 |
756 | |||
| addParserOutputInfo | |
0.00% |
0 / 72 |
|
0.00% |
0 / 1 |
702 | |||
| getStandaloneMetadataSection | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| processParsedHTML | |
0.00% |
0 / 19 |
|
0.00% |
0 / 1 |
90 | |||
| processSerializedWT | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
42 | |||
| checkHTML | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
6 | |||
| checkMetadata | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
2 | |||
| filterDsr | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| filterNodeDsr | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| checkWikitext | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
56 | |||
| updateKnownFailures | |
0.00% |
0 / 78 |
|
0.00% |
0 / 1 |
930 | |||
| processTest | |
0.00% |
0 / 108 |
|
0.00% |
0 / 1 |
1482 | |||
| run | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
90 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\ParserTests; |
| 5 | |
| 6 | use Closure; |
| 7 | use Psr\Log\LoggerInterface; |
| 8 | use Wikimedia\Assert\Assert; |
| 9 | use Wikimedia\Bcp47Code\Bcp47CodeValue; |
| 10 | use Wikimedia\Parsoid\Config\Api\DataAccess; |
| 11 | use Wikimedia\Parsoid\Config\Api\PageConfig; |
| 12 | use Wikimedia\Parsoid\Config\Env; |
| 13 | use Wikimedia\Parsoid\Config\StubMetadataCollector; |
| 14 | use Wikimedia\Parsoid\Core\SelectiveUpdateData; |
| 15 | use Wikimedia\Parsoid\DOM\Document; |
| 16 | use Wikimedia\Parsoid\DOM\Element; |
| 17 | use Wikimedia\Parsoid\Ext\ParsoidExtensionAPI; |
| 18 | use Wikimedia\Parsoid\Mocks\MockPageConfig; |
| 19 | use Wikimedia\Parsoid\Mocks\MockPageContent; |
| 20 | use Wikimedia\Parsoid\Utils\ContentUtils; |
| 21 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 22 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 23 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 24 | use Wikimedia\Parsoid\Utils\ScriptUtils; |
| 25 | use Wikimedia\Parsoid\Utils\Title; |
| 26 | use Wikimedia\Parsoid\Utils\Utils; |
| 27 | use Wikimedia\Parsoid\Wt2Html\PageConfigFrame; |
| 28 | |
| 29 | /** |
| 30 | * Test runner for parser tests |
| 31 | */ |
| 32 | class TestRunner { |
| 33 | // Hard-code some interwiki prefixes, as is done |
| 34 | // in ParserTestRunner::appendInterwikiSetup() in core |
| 35 | // Note that ApiQuerySiteInfo will always expand the URL to include a |
| 36 | // protocol, but will set 'protorel' to indicate whether its internal |
| 37 | // form included a protocol or not. So in this file 'url' will always |
| 38 | // have a protocol and we'll include an explicit 'protorel' field; but |
| 39 | // in core there is no 'protorel' field and 'url' will not always have |
| 40 | // a protocol. |
| 41 | private const PARSER_TESTS_IWPS = [ |
| 42 | [ |
| 43 | 'prefix' => 'wikinvest', |
| 44 | 'local' => true, |
| 45 | // This url doesn't have a $1 to exercise the fix in |
| 46 | // ConfigUtils::computeInterwikiMap |
| 47 | 'url' => 'https://meta.wikimedia.org/wiki/Interwiki_map/discontinued#Wikinvest', |
| 48 | 'protorel' => false |
| 49 | ], |
| 50 | [ |
| 51 | 'prefix' => 'local', |
| 52 | 'url' => 'http://example.org/wiki/$1', |
| 53 | 'local' => true, |
| 54 | 'localinterwiki' => true |
| 55 | ], |
| 56 | [ |
| 57 | // Local interwiki that matches a namespace name (T228616) |
| 58 | 'prefix' => 'project', |
| 59 | 'url' => 'http://example.org/wiki/$1', |
| 60 | 'local' => true, |
| 61 | 'localinterwiki' => true |
| 62 | ], |
| 63 | [ |
| 64 | 'prefix' => 'wikipedia', |
| 65 | 'url' => 'http://en.wikipedia.org/wiki/$1' |
| 66 | ], |
| 67 | [ |
| 68 | 'prefix' => 'meatball', |
| 69 | // this has been updated in the live wikis, but the parser tests |
| 70 | // expect the old value (as set in parserTest.inc:setupInterwikis()) |
| 71 | 'url' => 'http://www.usemod.com/cgi-bin/mb.pl?$1' |
| 72 | ], |
| 73 | [ |
| 74 | 'prefix' => 'memoryalpha', |
| 75 | 'url' => 'http://www.memory-alpha.org/en/index.php/$1' |
| 76 | ], |
| 77 | [ |
| 78 | 'prefix' => 'zh', |
| 79 | 'url' => 'http://zh.wikipedia.org/wiki/$1', |
| 80 | 'language' => "中文", |
| 81 | 'local' => true |
| 82 | ], |
| 83 | [ |
| 84 | 'prefix' => 'es', |
| 85 | 'url' => 'http://es.wikipedia.org/wiki/$1', |
| 86 | 'language' => "español", |
| 87 | 'local' => true |
| 88 | ], |
| 89 | [ |
| 90 | 'prefix' => 'fr', |
| 91 | 'url' => 'http://fr.wikipedia.org/wiki/$1', |
| 92 | 'language' => "français", |
| 93 | 'local' => true |
| 94 | ], |
| 95 | [ |
| 96 | 'prefix' => 'ru', |
| 97 | 'url' => 'http://ru.wikipedia.org/wiki/$1', |
| 98 | 'language' => "русский", |
| 99 | 'local' => true |
| 100 | ], |
| 101 | [ |
| 102 | 'prefix' => 'mi', |
| 103 | 'url' => 'http://example.org/wiki/$1', |
| 104 | // better for testing if one of the |
| 105 | // localinterwiki prefixes is also a language |
| 106 | 'language' => 'Test', |
| 107 | 'local' => true, |
| 108 | 'localinterwiki' => true |
| 109 | ], |
| 110 | [ |
| 111 | 'prefix' => 'mul', |
| 112 | 'url' => 'http://wikisource.org/wiki/$1', |
| 113 | 'extralanglink' => true, |
| 114 | 'linktext' => 'Multilingual', |
| 115 | 'sitename' => 'WikiSource', |
| 116 | 'local' => true |
| 117 | ], |
| 118 | // added to core's ParserTestRunner::appendInterwikiSetup() to support |
| 119 | // Parsoid tests [T254181] |
| 120 | [ |
| 121 | 'prefix' => 'en', |
| 122 | 'url' => 'http://en.wikipedia.org/wiki/$1', |
| 123 | 'language' => 'English', |
| 124 | 'local' => true, |
| 125 | 'protorel' => true |
| 126 | ], |
| 127 | [ |
| 128 | 'prefix' => 'stats', |
| 129 | 'local' => true, |
| 130 | 'url' => 'https://stats.wikimedia.org/$1' |
| 131 | ], |
| 132 | [ |
| 133 | 'prefix' => 'gerrit', |
| 134 | 'local' => true, |
| 135 | 'url' => 'https://gerrit.wikimedia.org/$1' |
| 136 | ], |
| 137 | ]; |
| 138 | |
| 139 | /** @var bool */ |
| 140 | private $runDisabled; |
| 141 | |
| 142 | /** @var bool */ |
| 143 | private $runPHP; |
| 144 | |
| 145 | /** @var string */ |
| 146 | private $offsetType; |
| 147 | |
| 148 | /** @var string */ |
| 149 | private $testFileName; |
| 150 | |
| 151 | /** @var string */ |
| 152 | private $testFilePath; |
| 153 | |
| 154 | /** @var ?string */ |
| 155 | private $knownFailuresInfix; |
| 156 | |
| 157 | /** @var string */ |
| 158 | private $knownFailuresPath; |
| 159 | |
| 160 | /** @var array */ |
| 161 | private $articles; |
| 162 | |
| 163 | /** @var LoggerInterface */ |
| 164 | private $defaultLogger; |
| 165 | |
| 166 | /** |
| 167 | * Sets one of 'regex' or 'string' properties |
| 168 | * - $testFilter['raw'] is the value of the filter |
| 169 | * - if $testFilter['regex'] is true, $testFilter['raw'] is used as a regex filter. |
| 170 | * - If $testFilter['string'] is true, $testFilter['raw'] is used as a plain string filter. |
| 171 | * @var ?array |
| 172 | */ |
| 173 | private $testFilter; |
| 174 | |
| 175 | /** @var Test[] */ |
| 176 | private $testCases; |
| 177 | |
| 178 | /** @var Stats */ |
| 179 | private $stats; |
| 180 | |
| 181 | /** @var MockApiHelper */ |
| 182 | private $mockApi; |
| 183 | |
| 184 | /** @var SiteConfig */ |
| 185 | private $siteConfig; |
| 186 | |
| 187 | /** @var DataAccess */ |
| 188 | private $dataAccess; |
| 189 | |
| 190 | /** |
| 191 | * Global cross-test env object only to be used for title processing while |
| 192 | * reading the parserTests file. |
| 193 | * |
| 194 | * Every test constructs its own private $env object. |
| 195 | * |
| 196 | * @var Env |
| 197 | */ |
| 198 | private $dummyEnv; |
| 199 | |
| 200 | /** |
| 201 | * Options needed to construct the per-test private $env object |
| 202 | * @var array |
| 203 | */ |
| 204 | private $envOptions; |
| 205 | |
| 206 | /** |
| 207 | * @param string $testFilePath |
| 208 | * @param ?string $knownFailuresInfix |
| 209 | * @param string[] $modes |
| 210 | */ |
| 211 | public function __construct( string $testFilePath, ?string $knownFailuresInfix, array $modes ) { |
| 212 | $this->testFilePath = $testFilePath; |
| 213 | $this->knownFailuresInfix = $knownFailuresInfix; |
| 214 | |
| 215 | $testFilePathInfo = pathinfo( $testFilePath ); |
| 216 | $this->testFileName = $testFilePathInfo['basename']; |
| 217 | |
| 218 | $newModes = []; |
| 219 | $modes[] = 'metadata'; |
| 220 | foreach ( $modes as $mode ) { |
| 221 | $newModes[$mode] = new Stats(); |
| 222 | $newModes[$mode]->failList = []; |
| 223 | $newModes[$mode]->result = ''; // XML reporter uses this. |
| 224 | } |
| 225 | |
| 226 | $this->stats = new Stats(); |
| 227 | $this->stats->modes = $newModes; |
| 228 | |
| 229 | $this->mockApi = new MockApiHelper( null, fn ( $title )=>$this->normalizeTitleKey( $title ) ); |
| 230 | $this->siteConfig = new SiteConfig( $this->mockApi, [] ); |
| 231 | $this->dataAccess = new DataAccess( $this->mockApi, $this->siteConfig, [ 'stripProto' => false ] ); |
| 232 | $this->dummyEnv = new Env( |
| 233 | $this->siteConfig, |
| 234 | // Unused; needed to satisfy Env signature requirements |
| 235 | new MockPageConfig( $this->siteConfig, [], new MockPageContent( [ 'main' => '' ] ) ), |
| 236 | // Unused; needed to satisfy Env signature requirements |
| 237 | $this->dataAccess, |
| 238 | // Unused; needed to satisfy Env signature requirements |
| 239 | new StubMetadataCollector( $this->siteConfig ) |
| 240 | ); |
| 241 | |
| 242 | // Init interwiki map to parser tests info. |
| 243 | // This suppresses interwiki info from cached configs. |
| 244 | $this->siteConfig->setupInterwikiMap( self::PARSER_TESTS_IWPS ); |
| 245 | $this->siteConfig->reset(); |
| 246 | } |
| 247 | |
| 248 | private function newEnv( Test $test, string $wikitext ): Env { |
| 249 | $title = $this->dummyEnv->makeTitleFromURLDecodedStr( |
| 250 | $test->pageName() |
| 251 | ); |
| 252 | |
| 253 | $opts = [ |
| 254 | 'title' => $title, |
| 255 | 'pageContent' => $wikitext, |
| 256 | 'pageLanguage' => $this->siteConfig->langBcp47(), |
| 257 | 'pageLanguagedir' => $this->siteConfig->rtl() ? 'rtl' : 'ltr' |
| 258 | ]; |
| 259 | |
| 260 | $pageConfig = new PageConfig( null, $this->siteConfig, $opts ); |
| 261 | |
| 262 | $env = new Env( |
| 263 | $this->siteConfig, |
| 264 | $pageConfig, |
| 265 | $this->dataAccess, |
| 266 | new StubMetadataCollector( $this->siteConfig ), |
| 267 | $this->envOptions |
| 268 | ); |
| 269 | $env->pageCache = $this->articles; |
| 270 | |
| 271 | // Set parsing resource limits. |
| 272 | // $env->setResourceLimits(); |
| 273 | |
| 274 | return $env; |
| 275 | } |
| 276 | |
| 277 | private function normalizeTitleKey( string $title ): string { |
| 278 | return $this->dummyEnv->normalizedTitleKey( $title, false, true ); |
| 279 | } |
| 280 | |
| 281 | private function addArticle( Article $art ): array { |
| 282 | $key = $this->normalizeTitleKey( $art->title ); |
| 283 | $oldVal = $this->articles[$key] ?? null; |
| 284 | $this->articles[$key] = $art->text; |
| 285 | $teardown = [ |
| 286 | function () use ( $key, $oldVal ) { |
| 287 | $this->articles[$key] = $oldVal; |
| 288 | }, |
| 289 | $this->mockApi->addArticle( $key, $art ), |
| 290 | ]; |
| 291 | return $teardown; |
| 292 | } |
| 293 | |
| 294 | /** |
| 295 | * Parse the test file and set up articles and test cases |
| 296 | * @param array $options |
| 297 | */ |
| 298 | private function buildTests( array $options ): void { |
| 299 | // Startup by loading .txt test file |
| 300 | $warnFunc = static function ( string $warnMsg ): void { |
| 301 | error_log( $warnMsg ); |
| 302 | }; |
| 303 | $normFunc = function ( string $title ): string { |
| 304 | return $this->normalizeTitleKey( $title ); |
| 305 | }; |
| 306 | $testReader = TestFileReader::read( |
| 307 | $this->testFilePath, $warnFunc, $normFunc, $this->knownFailuresInfix |
| 308 | ); |
| 309 | $this->knownFailuresPath = $testReader->knownFailuresPath; |
| 310 | $this->testCases = $testReader->testCases; |
| 311 | $this->articles = []; |
| 312 | foreach ( $testReader->articles as $art ) { |
| 313 | $this->addArticle( $art ); |
| 314 | } |
| 315 | if ( !ScriptUtils::booleanOption( $options['quieter'] ?? '' ) ) { |
| 316 | if ( $this->knownFailuresPath ) { |
| 317 | error_log( 'Loaded known failures from ' . $this->knownFailuresPath ); |
| 318 | } else { |
| 319 | error_log( 'No known failures found.' ); |
| 320 | } |
| 321 | } |
| 322 | } |
| 323 | |
| 324 | /** |
| 325 | * Convert a wikitext string to an HTML Node |
| 326 | * |
| 327 | * @param Env $env |
| 328 | * @param Test $test |
| 329 | * @param string $mode |
| 330 | * @param string $wikitext |
| 331 | * @return Document |
| 332 | */ |
| 333 | private function convertWt2Html( |
| 334 | Env $env, Test $test, string $mode, string $wikitext |
| 335 | ): Document { |
| 336 | // FIXME: Ugly! Maybe we should switch to using the entrypoint to |
| 337 | // the library for parserTests instead of reusing the environment |
| 338 | // and touching these internals. |
| 339 | $content = $env->getPageConfig()->getRevisionContent(); |
| 340 | // @phan-suppress-next-line PhanUndeclaredProperty |
| 341 | $content->data['main']['content'] = $wikitext; |
| 342 | $env->topFrame = new PageConfigFrame( |
| 343 | $env, $env->getPageConfig(), $env->getSiteConfig() |
| 344 | ); |
| 345 | if ( $mode === 'html2html' ) { |
| 346 | // Since this was set when serializing we need to setup a new doc |
| 347 | $env->setupTopLevelDoc(); |
| 348 | } |
| 349 | $handler = $env->getContentHandler(); |
| 350 | $extApi = new ParsoidExtensionAPI( $env ); |
| 351 | $doc = $handler->toDOM( $extApi ); |
| 352 | DOMDataUtils::visitAndStoreDataAttribs( DOMCompat::getBody( $doc ), [ |
| 353 | 'storeInPageBundle' => false, |
| 354 | ] ); |
| 355 | return $doc; |
| 356 | } |
| 357 | |
| 358 | /** |
| 359 | * Convert a DOM to Wikitext. |
| 360 | * |
| 361 | * @param Env $env |
| 362 | * @param Test $test |
| 363 | * @param string $mode |
| 364 | * @param Document $doc |
| 365 | * @return string |
| 366 | */ |
| 367 | private function convertHtml2Wt( Env $env, Test $test, string $mode, Document $doc ): string { |
| 368 | $startsAtWikitext = $mode === 'wt2wt' || $mode === 'wt2html' || $mode === 'selser'; |
| 369 | if ( $mode === 'selser' ) { |
| 370 | $selserData = new SelectiveUpdateData( $test->wikitext, $test->cachedBODYstr ); |
| 371 | } else { |
| 372 | $selserData = null; |
| 373 | } |
| 374 | Assert::invariant( |
| 375 | !DOMDataUtils::isPreparedAndLoaded( $doc ), |
| 376 | "doc should not be prepared and loaded already" |
| 377 | ); |
| 378 | DOMDataUtils::prepareDoc( $doc ); |
| 379 | DOMDataUtils::visitAndLoadDataAttribs( |
| 380 | DOMCompat::getBody( $doc ), [ |
| 381 | 'markNew' => true, 'validateXMLNames' => true, |
| 382 | ] |
| 383 | ); |
| 384 | // Mark the document as loaded so we can try to catch errors which |
| 385 | // might try to reload this again later. |
| 386 | DOMDataUtils::getBag( $doc )->loaded = true; |
| 387 | |
| 388 | $env->setupTopLevelDoc( $doc ); |
| 389 | $extApi = new ParsoidExtensionAPI( $env ); |
| 390 | return $env->getContentHandler()->fromDOM( $extApi, $selserData ); |
| 391 | } |
| 392 | |
| 393 | /** |
| 394 | * Run test in the requested mode |
| 395 | * @param Test $test |
| 396 | * @param string $mode |
| 397 | * @param array $options |
| 398 | */ |
| 399 | private function runTest( Test $test, string $mode, array $options ): void { |
| 400 | $test->time = []; |
| 401 | $testOpts = $test->options; |
| 402 | |
| 403 | // These changes are for environment options that change between runs of |
| 404 | // different modes. See `processTest` for changes per test. |
| 405 | |
| 406 | // Page language matches "wiki language" (which is set by |
| 407 | // the item 'language' option). |
| 408 | |
| 409 | // Variant conversion is disabled by default |
| 410 | $this->envOptions['wtVariantLanguage'] = null; |
| 411 | $this->envOptions['htmlVariantLanguage'] = null; |
| 412 | // The test can explicitly opt-in to variant conversion with the |
| 413 | // 'langconv' option. |
| 414 | if ( $testOpts['langconv'] ?? null ) { |
| 415 | // These test option names are deprecated: |
| 416 | // (Note that test options names are lowercased by the reader.) |
| 417 | if ( $testOpts['sourcevariant'] ?? false ) { |
| 418 | $this->envOptions['wtVariantLanguage'] = Utils::mwCodeToBcp47( |
| 419 | $testOpts['sourcevariant'], true, $this->siteConfig->getLogger() |
| 420 | ); |
| 421 | } |
| 422 | if ( $testOpts['variant'] ?? false ) { |
| 423 | $this->envOptions['htmlVariantLanguage'] = Utils::mwCodeToBcp47( |
| 424 | $testOpts['variant'], true, $this->siteConfig->getLogger() |
| 425 | ); |
| 426 | } |
| 427 | // Preferred option names, which are also specified in bcp-47 codes |
| 428 | // (Note that test options names are lowercased by the reader.) |
| 429 | if ( $testOpts['wtvariantlanguage'] ?? false ) { |
| 430 | $this->envOptions['wtVariantLanguage'] = |
| 431 | new Bcp47CodeValue( $testOpts['wtvariantlanguage'] ); |
| 432 | } |
| 433 | if ( $testOpts['htmlvariantlanguage'] ?? false ) { |
| 434 | $this->envOptions['htmlVariantLanguage'] = |
| 435 | new Bcp47CodeValue( $testOpts['htmlvariantlanguage'] ); |
| 436 | } |
| 437 | } |
| 438 | |
| 439 | $env = $this->newEnv( $test, $test->wikitext ?? '' ); |
| 440 | |
| 441 | // Some useful booleans |
| 442 | $startsAtHtml = $mode === 'html2html' || $mode === 'html2wt'; |
| 443 | $endsAtHtml = $mode === 'wt2html' || $mode === 'html2html'; |
| 444 | |
| 445 | $parsoidOnly = isset( $test->sections['html/parsoid'] ) || |
| 446 | isset( $test->sections['html/parsoid+standalone'] ) || ( |
| 447 | !empty( $testOpts['parsoid'] ) && |
| 448 | !isset( $testOpts['parsoid']['normalizePhp'] ) |
| 449 | ); |
| 450 | $test->time['start'] = microtime( true ); |
| 451 | $doc = null; |
| 452 | $wt = null; |
| 453 | |
| 454 | if ( isset( $test->sections['html/parsoid+standalone'] ) ) { |
| 455 | $test->parsoidHtml = $test->sections['html/parsoid+standalone']; |
| 456 | } |
| 457 | |
| 458 | // Source preparation |
| 459 | if ( $startsAtHtml ) { |
| 460 | $html = $test->parsoidHtml ?? ''; |
| 461 | if ( !$parsoidOnly ) { |
| 462 | // Strip some php output that has no wikitext representation |
| 463 | // (like .mw-editsection) and won't html2html roundtrip and |
| 464 | // therefore causes false failures. |
| 465 | $html = TestUtils::normalizePhpOutput( $html ); |
| 466 | } |
| 467 | // FUTURE WORK: it would be better if we used |
| 468 | // ContentUtils::createAndLoadDocument() here -- and for all of the |
| 469 | // ::parseHTML() calls below this as well. |
| 470 | $doc = DOMUtils::parseHTML( $html ); |
| 471 | $wt = $this->convertHtml2Wt( $env, $test, $mode, $doc ); |
| 472 | } else { // startsAtWikitext |
| 473 | // Always serialize DOM to string and reparse before passing to wt2wt |
| 474 | if ( $test->cachedBODYstr === null ) { |
| 475 | $doc = $this->convertWt2Html( $env, $test, $mode, $test->wikitext ); |
| 476 | |
| 477 | // Cache parsed HTML |
| 478 | $test->cachedBODYstr = ContentUtils::toXML( DOMCompat::getBody( $doc ) ); |
| 479 | |
| 480 | // - In wt2html mode, pass through original DOM |
| 481 | // so that it is serialized just once. |
| 482 | // - In wt2wt and selser modes, pass through serialized and |
| 483 | // reparsed DOM so that fostering/normalization effects |
| 484 | // are reproduced. |
| 485 | if ( $mode === 'wt2html' ) { |
| 486 | // no-op |
| 487 | } else { |
| 488 | $doc = DOMUtils::parseHTML( $test->cachedBODYstr ); |
| 489 | } |
| 490 | } else { |
| 491 | $doc = DOMUtils::parseHTML( $test->cachedBODYstr ); |
| 492 | } |
| 493 | } |
| 494 | |
| 495 | // Generate and make changes for the selser test mode |
| 496 | $testManualChanges = $testOpts['parsoid']['changes'] ?? null; |
| 497 | if ( $mode === 'selser' ) { |
| 498 | if ( $testManualChanges && $test->changetree === [ 'manual' ] ) { |
| 499 | $test->applyManualChanges( $doc ); |
| 500 | } else { |
| 501 | $changetree = isset( $options['changetree'] ) ? |
| 502 | json_decode( $options['changetree'] ) : $test->changetree; |
| 503 | if ( !$changetree ) { |
| 504 | $changetree = $test->generateChanges( $doc ); |
| 505 | } |
| 506 | $dumpOpts = [ |
| 507 | 'dom:post-changes' => $env->hasDumpFlag( 'dom:post-changes' ), |
| 508 | 'logger' => $env->getSiteConfig()->getLogger() |
| 509 | ]; |
| 510 | $test->applyChanges( $dumpOpts, $doc, $changetree ); |
| 511 | } |
| 512 | // Save the modified DOM so we can re-test it later. |
| 513 | // Always serialize to string and reparse before passing to selser/wt2wt. |
| 514 | $test->changedHTMLStr = ContentUtils::toXML( DOMCompat::getBody( $doc ) ); |
| 515 | $doc = DOMUtils::parseHTML( $test->changedHTMLStr ); |
| 516 | } elseif ( $mode === 'wt2wt' ) { |
| 517 | // Handle a 'changes' option if present. |
| 518 | if ( $testManualChanges ) { |
| 519 | $test->applyManualChanges( $doc ); |
| 520 | } |
| 521 | } |
| 522 | |
| 523 | // Roundtrip stage |
| 524 | if ( $mode === 'wt2wt' || $mode === 'selser' ) { |
| 525 | $wt = $this->convertHtml2Wt( $env, $test, $mode, $doc ); |
| 526 | } elseif ( $mode === 'html2html' ) { |
| 527 | $doc = $this->convertWt2Html( $env, $test, $mode, $wt ); |
| 528 | } |
| 529 | |
| 530 | // Result verification stage |
| 531 | if ( $endsAtHtml ) { |
| 532 | $this->processParsedHTML( $env, $test, $options, $mode, $doc ); |
| 533 | } else { |
| 534 | $this->processSerializedWT( $env, $test, $options, $mode, $wt ); |
| 535 | } |
| 536 | } |
| 537 | |
| 538 | /** |
| 539 | * Process test options that impact output. |
| 540 | * These are almost always only pertinent in wt2html test modes. |
| 541 | * Returns: |
| 542 | * - null if there are no applicable output options. |
| 543 | * - true if the output matches expected output for the requested option(s). |
| 544 | * - false otherwise |
| 545 | * |
| 546 | * @param Env $env |
| 547 | * @param Test $test |
| 548 | * @param array $options |
| 549 | * @param string $mode |
| 550 | * @param Document $doc |
| 551 | * @param ?string $metadataExpected A metadata section from the test, |
| 552 | * or null if none present. If a metadata section is not present, |
| 553 | * the metadata output is added to $doc, otherwise it is returned |
| 554 | * in $metadataActual |
| 555 | * @param ?string &$metadataActual The "actual" metadata output for |
| 556 | * this test. |
| 557 | */ |
| 558 | private function addParserOutputInfo( |
| 559 | Env $env, Test $test, array $options, string $mode, Document $doc, |
| 560 | ?string $metadataExpected, ?string &$metadataActual |
| 561 | ): void { |
| 562 | $output = $env->getMetadata(); |
| 563 | $opts = $test->options; |
| 564 | '@phan-var StubMetadataCollector $output'; // @var StubMetadataCollector $metadata |
| 565 | // See ParserTestRunner::addParserOutputInfo() in core. |
| 566 | $before = []; |
| 567 | $after = []; |
| 568 | |
| 569 | // 'showtitle' not yet supported |
| 570 | |
| 571 | // unlike other link types, this dumps the 'sort' property as well |
| 572 | if ( isset( $opts['cat'] ) ) { |
| 573 | $defaultSortKey = $output->getPageProperty( 'defaultsort' ) ?? ''; |
| 574 | foreach ( |
| 575 | $output->getLinkList( StubMetadataCollector::LINKTYPE_CATEGORY ) |
| 576 | as [ 'link' => $link, 'sort' => $sort ] |
| 577 | ) { |
| 578 | $sortkey = $sort ?: $defaultSortKey; |
| 579 | $name = $link->getDBkey(); |
| 580 | $after[] = "cat=$name sort=$sortkey"; |
| 581 | } |
| 582 | } |
| 583 | |
| 584 | if ( isset( $opts['extlinks'] ) ) { |
| 585 | foreach ( $output->getExternalLinks() as $url => $ignore ) { |
| 586 | $after[] = "extlink=$url"; |
| 587 | } |
| 588 | } |
| 589 | |
| 590 | // Unlike other link types, this is stored as text, not dbkey |
| 591 | if ( isset( $opts['ill'] ) ) { |
| 592 | foreach ( |
| 593 | $output->getLinkList( StubMetadataCollector::LINKTYPE_LANGUAGE ) |
| 594 | as [ 'link' => $ll ] |
| 595 | ) { |
| 596 | $after[] = "ill=" . Title::newFromLinkTarget( $ll, $this->siteConfig )->getFullText(); |
| 597 | } |
| 598 | } |
| 599 | |
| 600 | $linkoptions = [ |
| 601 | [ 'iwl', 'iwl=', StubMetadataCollector::LINKTYPE_INTERWIKI ], |
| 602 | [ 'links', 'link=', StubMetadataCollector::LINKTYPE_LOCAL ], |
| 603 | [ 'special', 'special=', StubMetadataCollector::LINKTYPE_SPECIAL ], |
| 604 | [ 'templates', 'template=', StubMetadataCollector::LINKTYPE_TEMPLATE ], |
| 605 | ]; |
| 606 | foreach ( $linkoptions as [ $optName, $prefix, $type ] ) { |
| 607 | if ( isset( $opts[$optName] ) ) { |
| 608 | foreach ( $output->getLinkList( $type ) as [ 'link' => $ll ] ) { |
| 609 | $after[] = $prefix . Title::newFromLinkTarget( $ll, $this->siteConfig )->getPrefixedDBkey(); |
| 610 | } |
| 611 | } |
| 612 | } |
| 613 | |
| 614 | if ( isset( $opts['extension'] ) ) { |
| 615 | $extList = $opts['extension']; |
| 616 | if ( !is_array( $extList ) ) { |
| 617 | $extList = [ $extList ]; |
| 618 | } |
| 619 | foreach ( $extList as $ext ) { |
| 620 | $after[] = "extension[$ext]=" . |
| 621 | // XXX should use JsonCodec |
| 622 | json_encode( |
| 623 | $output->getExtensionData( $ext ), |
| 624 | JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT |
| 625 | ); |
| 626 | } |
| 627 | } |
| 628 | if ( isset( $opts['property'] ) ) { |
| 629 | $propList = $opts['property']; |
| 630 | if ( !is_array( $propList ) ) { |
| 631 | $propList = [ $propList ]; |
| 632 | } |
| 633 | foreach ( $propList as $prop ) { |
| 634 | $after[] = "property[$prop]=" . |
| 635 | ( $output->getPageProperty( $prop ) ?? '' ); |
| 636 | } |
| 637 | } |
| 638 | if ( isset( $opts['showflags'] ) ) { |
| 639 | $actualFlags = $output->getOutputFlags(); |
| 640 | sort( $actualFlags ); |
| 641 | $after[] = "flags=" . implode( ', ', $actualFlags ); |
| 642 | } |
| 643 | if ( isset( $opts['showtocdata'] ) ) { |
| 644 | $tocData = $output->getTOCData(); |
| 645 | if ( $tocData !== null ) { |
| 646 | $after[] = $tocData->prettyPrint(); |
| 647 | } |
| 648 | } |
| 649 | if ( isset( $opts['showindicators'] ) ) { |
| 650 | foreach ( $output->getIndicators() as $name => $content ) { |
| 651 | $after[] = "$name=$content"; |
| 652 | } |
| 653 | } |
| 654 | if ( isset( $opts['showmedia'] ) ) { |
| 655 | $images = array_map( |
| 656 | static fn ( $item ) => $item['link']->getDBkey(), |
| 657 | $output->getLinkList( StubMetadataCollector::LINKTYPE_MEDIA ) |
| 658 | ); |
| 659 | $after[] = 'images=' . implode( ', ', $images ); |
| 660 | } |
| 661 | if ( $metadataExpected === null ) { |
| 662 | // legacy format, add $before and $after to $doc |
| 663 | $body = DOMCompat::getBody( $doc ); |
| 664 | if ( count( $before ) ) { |
| 665 | $before = $doc->createTextNode( implode( "\n", $before ) ); |
| 666 | $body->insertBefore( $before, $body->firstChild ); |
| 667 | } |
| 668 | if ( count( $after ) ) { |
| 669 | $after = $doc->createTextNode( implode( "\n", $after ) ); |
| 670 | $body->appendChild( $after ); |
| 671 | } |
| 672 | } else { |
| 673 | $metadataActual = implode( "\n", array_merge( $before, $after ) ); |
| 674 | } |
| 675 | } |
| 676 | |
| 677 | /** |
| 678 | * Return the appropriate metadata section for this test, given that |
| 679 | * we are running in parsoid "standalone" mode, or 'null' if none is |
| 680 | * present. |
| 681 | * @param Test $test |
| 682 | * @return ?string The expected metadata for this test |
| 683 | */ |
| 684 | public static function getStandaloneMetadataSection( Test $test ): ?string { |
| 685 | return // specific results for parsoid standalone mode |
| 686 | $test->sections['metadata/parsoid+standalone'] ?? |
| 687 | // specific results for parsoid |
| 688 | $test->sections['metadata/parsoid'] ?? |
| 689 | // generic for all parsers (even standalone) |
| 690 | $test->sections['metadata'] ?? |
| 691 | // missing (== use legacy combined output format) |
| 692 | null; |
| 693 | } |
| 694 | |
| 695 | /** |
| 696 | * Check the given HTML result against the expected result, |
| 697 | * and throw an exception if necessary. |
| 698 | * |
| 699 | * @param Env $env |
| 700 | * @param Test $test |
| 701 | * @param array $options |
| 702 | * @param string $mode |
| 703 | * @param Document $doc |
| 704 | */ |
| 705 | private function processParsedHTML( |
| 706 | Env $env, Test $test, array $options, string $mode, Document $doc |
| 707 | ): void { |
| 708 | $modeObj = new TestMode( $mode ); |
| 709 | $test->time['end'] = microtime( true ); |
| 710 | $metadataExpected = self::getStandaloneMetadataSection( $test ); |
| 711 | $metadataActual = null; |
| 712 | if ( isset( $test->options['nohtml'] ) ) { |
| 713 | $body = DOMCompat::getBody( $doc ); |
| 714 | DOMCompat::replaceChildren( $body ); |
| 715 | } |
| 716 | $this->addParserOutputInfo( |
| 717 | $env, $test, $options, $mode, $doc, |
| 718 | $metadataExpected, $metadataActual |
| 719 | ); |
| 720 | if ( $test->parsoidHtml !== null ) { |
| 721 | $checkPassed = $this->checkHTML( $test, DOMCompat::getBody( $doc ), $options, $mode ); |
| 722 | } else { |
| 723 | // Running the test for metadata, presumably. |
| 724 | $checkPassed = true; |
| 725 | } |
| 726 | |
| 727 | // We could also check metadata in the html2html or wt2wt |
| 728 | // modes, but (a) we'd need a separate key for known failures |
| 729 | // to avoid overwriting the wt2html metadata results, and (b) |
| 730 | // any failures would probably be redundant with html2wt |
| 731 | // failures and not indicative of a "real" root cause bug. |
| 732 | if ( $metadataExpected !== null && !$modeObj->isCachingMode() && $mode === 'wt2html' ) { |
| 733 | $metadataResult = $this->checkMetadata( $test, $metadataExpected, $metadataActual ?? '', $options ); |
| 734 | $checkPassed = $checkPassed && $metadataResult; |
| 735 | } |
| 736 | |
| 737 | // Only throw an error if --exit-unexpected was set and there was an error |
| 738 | // Otherwise, continue running tests |
| 739 | if ( $options['exit-unexpected'] && !$checkPassed ) { |
| 740 | throw new UnexpectedException; |
| 741 | } |
| 742 | } |
| 743 | |
| 744 | /** |
| 745 | * Check the given wikitext result against the expected result, |
| 746 | * and throw an exception if necessary. |
| 747 | * |
| 748 | * @param Env $env |
| 749 | * @param Test $test |
| 750 | * @param array $options |
| 751 | * @param string $mode |
| 752 | * @param string $wikitext |
| 753 | */ |
| 754 | private function processSerializedWT( |
| 755 | Env $env, Test $test, array $options, string $mode, string $wikitext |
| 756 | ): void { |
| 757 | $test->time['end'] = microtime( true ); |
| 758 | |
| 759 | if ( $mode === 'selser' && $options['selser'] !== 'noauto' ) { |
| 760 | if ( $test->changetree === [ 5 ] ) { |
| 761 | $test->resultWT = $test->wikitext; |
| 762 | } else { |
| 763 | $doc = DOMUtils::parseHTML( $test->changedHTMLStr ); |
| 764 | $test->resultWT = $this->convertHtml2Wt( $env, $test, 'wt2wt', $doc ); |
| 765 | } |
| 766 | } |
| 767 | |
| 768 | $checkPassed = $this->checkWikitext( $test, $wikitext, $options, $mode ); |
| 769 | |
| 770 | // Only throw an error if --exit-unexpected was set and there was an error |
| 771 | // Otherwise, continue running tests |
| 772 | if ( $options['exit-unexpected'] && !$checkPassed ) { |
| 773 | throw new UnexpectedException; |
| 774 | } |
| 775 | } |
| 776 | |
| 777 | private function checkHTML( |
| 778 | Test $test, Element $out, array $options, string $mode |
| 779 | ): bool { |
| 780 | [ $normOut, $normExpected ] = $test->normalizeHTML( $out, $test->cachedNormalizedHTML ); |
| 781 | $expected = [ 'normal' => $normExpected, 'raw' => $test->parsoidHtml ]; |
| 782 | $actual = [ |
| 783 | 'normal' => $normOut, |
| 784 | 'raw' => ContentUtils::toXML( $out, [ 'innerXML' => true ] ), |
| 785 | 'input' => ( $mode === 'html2html' ) ? $test->parsoidHtml : $test->wikitext |
| 786 | ]; |
| 787 | |
| 788 | return $options['reportResult']( |
| 789 | $this->stats, $test, $options, $mode, $expected, $actual |
| 790 | ); |
| 791 | } |
| 792 | |
| 793 | private function checkMetadata( |
| 794 | Test $test, string $metadataExpected, string $metadataActual, array $options |
| 795 | ): bool { |
| 796 | $expected = [ 'normal' => $metadataExpected, 'raw' => $metadataExpected ]; |
| 797 | $actual = [ |
| 798 | 'normal' => $metadataActual, |
| 799 | 'raw' => $metadataActual, |
| 800 | 'input' => $test->wikitext, |
| 801 | ]; |
| 802 | $mode = 'metadata'; |
| 803 | |
| 804 | return $options['reportResult']( |
| 805 | $this->stats, $test, $options, $mode, $expected, $actual |
| 806 | ); |
| 807 | } |
| 808 | |
| 809 | /** |
| 810 | * Removes DSR from data-parsoid for test normalization of a complete document. If |
| 811 | * data-parsoid gets subsequently empty, removes it too. |
| 812 | * @param string $raw |
| 813 | * @return string |
| 814 | */ |
| 815 | private function filterDsr( string $raw ): string { |
| 816 | $doc = ContentUtils::createAndLoadDocument( $raw ); |
| 817 | foreach ( $doc->childNodes as $child ) { |
| 818 | if ( $child instanceof Element ) { |
| 819 | $this->filterNodeDsr( $child ); |
| 820 | } |
| 821 | } |
| 822 | $ret = ContentUtils::ppToXML( DOMCompat::getBody( $doc ), [ 'innerXML' => true ] ); |
| 823 | $ret = preg_replace( '/\sdata-parsoid="{}"/', '', $ret ); |
| 824 | return $ret; |
| 825 | } |
| 826 | |
| 827 | /** |
| 828 | * Removes DSR from data-parsoid for test normalization of an element. |
| 829 | */ |
| 830 | private function filterNodeDsr( Element $el ) { |
| 831 | $dp = DOMDataUtils::getDataParsoid( $el ); |
| 832 | unset( $dp->dsr ); |
| 833 | // XXX: could also set TempData::IS_NEW if !$dp->isModified(), |
| 834 | // rather than using the preg_replace above. |
| 835 | foreach ( $el->childNodes as $child ) { |
| 836 | if ( $child instanceof Element ) { |
| 837 | $this->filterNodeDsr( $child ); |
| 838 | } |
| 839 | } |
| 840 | } |
| 841 | |
| 842 | private function checkWikitext( |
| 843 | Test $test, string $out, array $options, string $mode |
| 844 | ): bool { |
| 845 | if ( $mode === 'html2wt' ) { |
| 846 | $input = $test->parsoidHtml; |
| 847 | $testWikitext = $test->wikitext; |
| 848 | } elseif ( $mode === 'wt2wt' ) { |
| 849 | if ( isset( $test->options['parsoid']['changes'] ) ) { |
| 850 | $input = $test->wikitext; |
| 851 | $testWikitext = $test->sections['wikitext/edited']; |
| 852 | } else { |
| 853 | $input = $testWikitext = $test->wikitext; |
| 854 | } |
| 855 | } else { /* selser */ |
| 856 | if ( $test->changetree === [ 5 ] ) { /* selser with oracle */ |
| 857 | $input = $test->changedHTMLStr; |
| 858 | $testWikitext = $test->wikitext; |
| 859 | $out = preg_replace( '/<!--' . Test::STATIC_RANDOM_STRING . '-->/', '', $out ); |
| 860 | } elseif ( $test->changetree === [ 'manual' ] && |
| 861 | isset( $test->options['parsoid']['changes'] ) |
| 862 | ) { /* manual changes */ |
| 863 | $input = $test->wikitext; |
| 864 | $testWikitext = $test->sections['wikitext/edited']; |
| 865 | } else { /* automated selser changes, no oracle */ |
| 866 | $input = $test->changedHTMLStr; |
| 867 | $testWikitext = $test->resultWT; |
| 868 | } |
| 869 | } |
| 870 | |
| 871 | [ $normalizedOut, $normalizedExpected ] = $test->normalizeWT( $out, $testWikitext ); |
| 872 | |
| 873 | $expected = [ 'normal' => $normalizedExpected, 'raw' => $testWikitext ]; |
| 874 | $actual = [ 'normal' => $normalizedOut, 'raw' => $out, 'input' => $input ]; |
| 875 | |
| 876 | return $options['reportResult']( |
| 877 | $this->stats, $test, $options, $mode, $expected, $actual ); |
| 878 | } |
| 879 | |
| 880 | private function updateKnownFailures( array $options ): array { |
| 881 | // Check in case any tests were removed but we didn't update |
| 882 | // the knownFailures |
| 883 | $knownFailuresChanged = false; |
| 884 | $allModes = $options['wt2html'] && $options['wt2wt'] && |
| 885 | $options['html2wt'] && $options['html2html'] && |
| 886 | isset( $options['selser'] ) && !( |
| 887 | isset( $options['filter'] ) || |
| 888 | isset( $options['regex'] ) || |
| 889 | isset( $options['maxtests'] ) |
| 890 | ); |
| 891 | $offsetType = $options['offsetType'] ?? 'byte'; |
| 892 | |
| 893 | // Update knownFailures, if requested |
| 894 | if ( $allModes || |
| 895 | ScriptUtils::booleanOption( $options['updateKnownFailures'] ?? null ) |
| 896 | ) { |
| 897 | if ( $this->knownFailuresPath !== null ) { |
| 898 | $old = file_get_contents( $this->knownFailuresPath ); |
| 899 | } else { |
| 900 | // If file doesn't exist, use the JSON representation of an |
| 901 | // empty array, so it compares equal in the case that we |
| 902 | // end up with an empty array of known failures below. |
| 903 | $old = '{}'; |
| 904 | } |
| 905 | $testKnownFailures = []; |
| 906 | $kfModes = array_merge( $options['modes'], [ 'metadata' ] ); |
| 907 | foreach ( $kfModes as $mode ) { |
| 908 | foreach ( $this->stats->modes[$mode]->failList as $fail ) { |
| 909 | $testKnownFailures[$fail['testName']] ??= []; |
| 910 | Assert::invariant( |
| 911 | !isset( $testKnownFailures[$fail['testName']][$mode . $fail['suffix']] ), |
| 912 | "Overwriting known failures result for " . $fail['testName'] . " " . $mode . $fail['suffix'] |
| 913 | ); |
| 914 | $testKnownFailures[$fail['testName']][$mode . $fail['suffix']] = $fail['raw']; |
| 915 | } |
| 916 | } |
| 917 | // Sort, otherwise, titles get added above based on the first |
| 918 | // failing mode, which can make diffs harder to verify when |
| 919 | // failing modes change. |
| 920 | ksort( $testKnownFailures ); |
| 921 | $contents = json_encode( |
| 922 | $testKnownFailures, |
| 923 | JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES | |
| 924 | JSON_FORCE_OBJECT | JSON_UNESCAPED_UNICODE |
| 925 | ) . "\n"; |
| 926 | if ( ScriptUtils::booleanOption( $options['updateKnownFailures'] ?? null ) ) { |
| 927 | if ( $this->knownFailuresPath !== null ) { |
| 928 | file_put_contents( $this->knownFailuresPath, $contents ); |
| 929 | } else { |
| 930 | // To be safe, we don't try to write a file that doesn't |
| 931 | // (yet) exist. Create an empty file if you need to, and |
| 932 | // then we'll happily update it for you. |
| 933 | throw new \RuntimeException( |
| 934 | "Known failures file for {$this->testFileName} does not exist, " . |
| 935 | "and so won't be updated." |
| 936 | ); |
| 937 | } |
| 938 | } elseif ( $allModes && $offsetType === 'byte' ) { |
| 939 | $knownFailuresChanged = $contents !== $old; |
| 940 | } |
| 941 | } |
| 942 | // Write updated tests from failed ones |
| 943 | if ( ScriptUtils::booleanOption( $options['update-tests'] ?? null ) || |
| 944 | ScriptUtils::booleanOption( $options['update-unexpected'] ?? null ) |
| 945 | ) { |
| 946 | $updateFormat = $options['update-format']; |
| 947 | if ( $updateFormat !== 'raw' && $updateFormat !== 'actualNormalized' ) { |
| 948 | $updateFormat = 'noDsr'; |
| 949 | } |
| 950 | |
| 951 | $fileContent = file_get_contents( $this->testFilePath ); |
| 952 | foreach ( [ 'wt2html', 'metadata' ] as $mode ) { |
| 953 | foreach ( $this->stats->modes[$mode]->failList as $fail ) { |
| 954 | if ( $options['update-tests'] || $fail['unexpected'] ) { |
| 955 | $exp = '/(!!\s*test\s*' . |
| 956 | preg_quote( $fail['testName'], '/' ) . |
| 957 | '(?:(?!!!\s*end)[\s\S])*' . |
| 958 | ')(' . preg_quote( $fail['expected'], '/' ) . |
| 959 | ')/m'; |
| 960 | $fail['noDsr'] = $fail['raw']; |
| 961 | if ( $updateFormat === 'noDsr' && $mode !== 'metadata' ) { |
| 962 | $fail['noDsr'] = $this->filterDsr( $fail['noDsr'] ); |
| 963 | } |
| 964 | $fileContent = preg_replace_callback( |
| 965 | $exp, |
| 966 | static function ( array $matches ) use ( $fail, $updateFormat ) { |
| 967 | return $matches[1] . $fail[$updateFormat]; |
| 968 | }, |
| 969 | $fileContent |
| 970 | ); |
| 971 | } |
| 972 | } |
| 973 | } |
| 974 | file_put_contents( $this->testFilePath, $fileContent ); |
| 975 | } |
| 976 | |
| 977 | // print out the summary |
| 978 | $options['reportSummary']( |
| 979 | $options['modes'], $this->stats, $this->testFileName, |
| 980 | $this->testFilter, $knownFailuresChanged, $options |
| 981 | ); |
| 982 | |
| 983 | // we're done! |
| 984 | // exit status 1 == uncaught exception |
| 985 | $failures = $this->stats->allFailures(); |
| 986 | $exitCode = ( $failures > 0 || $knownFailuresChanged ) ? 2 : 0; |
| 987 | if ( ScriptUtils::booleanOption( $options['exit-zero'] ?? null ) ) { |
| 988 | $exitCode = 0; |
| 989 | } |
| 990 | |
| 991 | return [ |
| 992 | 'exitCode' => $exitCode, |
| 993 | 'stats' => $this->stats, |
| 994 | 'file' => $this->testFileName, |
| 995 | 'knownFailuresChanged' => $knownFailuresChanged |
| 996 | ]; |
| 997 | } |
| 998 | |
| 999 | /** |
| 1000 | * Run the test in all requested modes. |
| 1001 | * |
| 1002 | * @param Test $test |
| 1003 | * @param array $options |
| 1004 | */ |
| 1005 | private function processTest( Test $test, array $options ): void { |
| 1006 | if ( !$test->options ) { |
| 1007 | $test->options = []; |
| 1008 | } |
| 1009 | |
| 1010 | $testOpts = $test->options; |
| 1011 | |
| 1012 | // ensure that test is not skipped if it has a wikitext/edited or |
| 1013 | // html/parsoid+langconv section (but not a parsoid html section) |
| 1014 | $haveHtml = ( $test->parsoidHtml !== null ) || |
| 1015 | isset( $test->sections['wikitext/edited'] ) || |
| 1016 | isset( $test->sections['html/parsoid+standalone'] ) || |
| 1017 | isset( $test->sections['html/parsoid+langconv'] ) || |
| 1018 | self::getStandaloneMetadataSection( $test ) !== null; |
| 1019 | $hasHtmlParsoid = |
| 1020 | isset( $test->sections['html/parsoid'] ) || |
| 1021 | isset( $test->sections['html/parsoid+standalone'] ); |
| 1022 | |
| 1023 | // Skip test whose title does not match --filter |
| 1024 | // or which is disabled or php-only |
| 1025 | if ( $test->wikitext === null || |
| 1026 | !$haveHtml || |
| 1027 | ( isset( $testOpts['disabled'] ) && !$this->runDisabled ) || |
| 1028 | ( isset( $testOpts['php'] ) && !( |
| 1029 | $hasHtmlParsoid || $this->runPHP ) |
| 1030 | ) || |
| 1031 | !$test->matchesFilter( $this->testFilter ) |
| 1032 | ) { |
| 1033 | return; |
| 1034 | } |
| 1035 | |
| 1036 | $suppressErrors = !empty( $testOpts['parsoid']['suppressErrors'] ); |
| 1037 | $this->siteConfig->setLogger( $suppressErrors ? |
| 1038 | $this->siteConfig->suppressLogger : $this->defaultLogger ); |
| 1039 | |
| 1040 | $targetModes = $test->computeTestModes( $options['modes'] ); |
| 1041 | |
| 1042 | // Filter out html2* tests if we don't have an HTML section |
| 1043 | // (Most likely there's either a metadata section or a html/php |
| 1044 | // section but not html/parsoid section.) |
| 1045 | if ( $test->parsoidHtml === null && !isset( $test->sections['html/parsoid+standalone'] ) ) { |
| 1046 | $targetModes = array_diff( $targetModes, [ 'html2wt', 'html2html' ] ); |
| 1047 | } |
| 1048 | |
| 1049 | if ( !count( $targetModes ) ) { |
| 1050 | return; |
| 1051 | } |
| 1052 | |
| 1053 | // Honor language option |
| 1054 | $prefix = $testOpts['language'] ?? 'enwiki'; |
| 1055 | if ( !str_contains( $prefix, 'wiki' ) ) { |
| 1056 | // Convert to our enwiki.. format |
| 1057 | $prefix .= 'wiki'; |
| 1058 | } |
| 1059 | |
| 1060 | // Switch to requested wiki |
| 1061 | $this->mockApi->setApiPrefix( $prefix ); |
| 1062 | $this->siteConfig->reset(); |
| 1063 | |
| 1064 | // Add the title associated with the current test as a known title to |
| 1065 | // be consistent with the test runner in the core repo. |
| 1066 | $teardown = $this->addArticle( new Article( [ |
| 1067 | 'title' => $test->pageName(), |
| 1068 | 'text' => $test->wikitext ?? '', |
| 1069 | // Fake it |
| 1070 | 'type' => 'article', |
| 1071 | 'filename' => 'fake', |
| 1072 | 'lineNumStart' => 0, |
| 1073 | 'lineNumEnd' => 0, |
| 1074 | ] ) ); |
| 1075 | |
| 1076 | // We don't do any sanity checking or type casting on $test->config |
| 1077 | // values here: if you set a bogus value in a parser test it *should* |
| 1078 | // blow things up, so that you fix your test case. |
| 1079 | |
| 1080 | // Update $wgInterwikiMagic flag |
| 1081 | // default (undefined) setting is true |
| 1082 | $this->siteConfig->setInterwikiMagic( |
| 1083 | $test->config['wgInterwikiMagic'] ?? true |
| 1084 | ); |
| 1085 | |
| 1086 | // Update $wgEnableMagicLinks flag |
| 1087 | // default (undefined) setting is true for all types |
| 1088 | foreach ( [ "RFC", "ISBN", "PMID" ] as $v ) { |
| 1089 | $this->siteConfig->setMagicLinkEnabled( |
| 1090 | $v, |
| 1091 | ( $test->config['wgEnableMagicLinks'] ?? [] )[$v] ?? true |
| 1092 | ); |
| 1093 | } |
| 1094 | if ( isset( $testOpts['pmid-interwiki'] ) ) { |
| 1095 | $this->siteConfig->setupInterwikiMap( array_merge( self::PARSER_TESTS_IWPS, [ |
| 1096 | // Added to support T145590#8608455 |
| 1097 | [ |
| 1098 | 'prefix' => 'pmid', |
| 1099 | 'local' => true, |
| 1100 | 'url' => '//www.ncbi.nlm.nih.gov/pubmed/$1?dopt=Abstract', |
| 1101 | ] |
| 1102 | ] ) ); |
| 1103 | $teardown[] = fn () => $this->siteConfig->setupInterwikiMap( self::PARSER_TESTS_IWPS ); |
| 1104 | } |
| 1105 | |
| 1106 | // FIXME: Cite-specific hack |
| 1107 | $this->siteConfig->responsiveReferences = [ |
| 1108 | 'enabled' => $test->config['wgCiteResponsiveReferences'] ?? |
| 1109 | $this->siteConfig->responsiveReferences['enabled'], |
| 1110 | 'threshold' => $test->config['wgCiteResponsiveReferencesThreshold'] ?? |
| 1111 | $this->siteConfig->responsiveReferences['threshold'], |
| 1112 | ]; |
| 1113 | |
| 1114 | if ( isset( $test->config['wgNoFollowLinks'] ) ) { |
| 1115 | $this->siteConfig->setNoFollowConfig( |
| 1116 | 'nofollow', $test->config['wgNoFollowLinks'] |
| 1117 | ); |
| 1118 | } |
| 1119 | |
| 1120 | if ( isset( $test->config['wgNoFollowDomainExceptions'] ) ) { |
| 1121 | $this->siteConfig->setNoFollowConfig( |
| 1122 | 'domainexceptions', |
| 1123 | $test->config['wgNoFollowDomainExceptions'] |
| 1124 | ); |
| 1125 | } |
| 1126 | |
| 1127 | // FIXME: Redundant with $testOpts['externallinktarget'] below |
| 1128 | if ( isset( $test->config['wgExternalLinkTarget'] ) ) { |
| 1129 | $this->siteConfig->setExternalLinkTarget( |
| 1130 | $test->config['wgExternalLinkTarget'] |
| 1131 | ); |
| 1132 | } |
| 1133 | |
| 1134 | $this->siteConfig->v3pf = $test->config['wgParsoidExperimentalParserFunctionSupport'] ?? false; |
| 1135 | |
| 1136 | // Process test-specific options |
| 1137 | if ( $testOpts ) { |
| 1138 | Assert::invariant( !isset( $testOpts['extensions'] ), |
| 1139 | 'Cannot configure extensions in tests' ); |
| 1140 | |
| 1141 | $availableParsoidTestOpts = [ 'wrapSections' ]; |
| 1142 | foreach ( $availableParsoidTestOpts as $opt ) { |
| 1143 | if ( isset( $testOpts['parsoid'][$opt] ) ) { |
| 1144 | $this->envOptions[$opt] = $testOpts['parsoid'][$opt]; |
| 1145 | } |
| 1146 | } |
| 1147 | |
| 1148 | $this->siteConfig->disableSubpagesForNS( 0 ); |
| 1149 | if ( isset( $testOpts['subpage'] ) ) { |
| 1150 | $this->siteConfig->enableSubpagesForNS( 0 ); |
| 1151 | } |
| 1152 | |
| 1153 | $allowedPrefixes = [ '' ]; // all allowed |
| 1154 | if ( isset( $testOpts['wgallowexternalimages'] ) && |
| 1155 | !preg_match( '/^(1|true|)$/D', $testOpts['wgallowexternalimages'] ) |
| 1156 | ) { |
| 1157 | $allowedPrefixes = []; |
| 1158 | } |
| 1159 | $this->siteConfig->allowedExternalImagePrefixes = $allowedPrefixes; |
| 1160 | |
| 1161 | // Emulate PHP parser's tag hook to tunnel content past the sanitizer |
| 1162 | if ( isset( $testOpts['styletag'] ) ) { |
| 1163 | $teardown[] = $this->siteConfig->registerParserTestExtension( StyleTag::class ); |
| 1164 | } |
| 1165 | |
| 1166 | if ( ( $testOpts['wgrawhtml'] ?? null ) === '1' ) { |
| 1167 | $teardown[] = $this->siteConfig->registerParserTestExtension( RawHTML::class ); |
| 1168 | } |
| 1169 | |
| 1170 | if ( isset( $testOpts['thumbsize'] ) ) { |
| 1171 | $this->siteConfig->thumbsize = (int)$testOpts['thumbsize']; |
| 1172 | } |
| 1173 | if ( isset( $testOpts['annotations'] ) ) { |
| 1174 | $teardown[] = $this->siteConfig->registerParserTestExtension( DummyAnnotation::class ); |
| 1175 | } |
| 1176 | if ( isset( $testOpts['i18next'] ) ) { |
| 1177 | $teardown[] = $this->siteConfig->registerParserTestExtension( I18nTag::class ); |
| 1178 | } |
| 1179 | if ( isset( $testOpts['externallinktarget'] ) ) { |
| 1180 | $this->siteConfig->setExternalLinkTarget( $testOpts['externallinktarget'] ); |
| 1181 | } |
| 1182 | } |
| 1183 | |
| 1184 | // Ensure ParserHook is always registered! |
| 1185 | $teardown[] = $this->siteConfig->registerParserTestExtension( ParserHook::class ); |
| 1186 | |
| 1187 | $runner = $this; |
| 1188 | $test->testAllModes( $targetModes, $options, Closure::fromCallable( [ $this, 'runTest' ] ) ); |
| 1189 | |
| 1190 | foreach ( array_reverse( $teardown ) as $t ) { |
| 1191 | $t(); |
| 1192 | } |
| 1193 | } |
| 1194 | |
| 1195 | /** |
| 1196 | * Run parser tests for the file with the provided options |
| 1197 | * |
| 1198 | * @param array $options |
| 1199 | * @return array |
| 1200 | */ |
| 1201 | public function run( array $options ): array { |
| 1202 | $this->runDisabled = ScriptUtils::booleanOption( $options['run-disabled'] ?? null ); |
| 1203 | $this->runPHP = ScriptUtils::booleanOption( $options['run-php'] ?? null ); |
| 1204 | $this->offsetType = $options['offsetType'] ?? 'byte'; |
| 1205 | |
| 1206 | // Test case filtering |
| 1207 | $this->testFilter = null; |
| 1208 | if ( isset( $options['filter'] ) || isset( $options['regex'] ) ) { |
| 1209 | $this->testFilter = [ |
| 1210 | 'raw' => $options['regex'] ?? $options['filter'], |
| 1211 | 'regex' => isset( $options['regex'] ), |
| 1212 | 'string' => isset( $options['filter'] ) |
| 1213 | ]; |
| 1214 | } |
| 1215 | |
| 1216 | $this->buildTests( $options ); |
| 1217 | |
| 1218 | // Trim test cases to the desired amount |
| 1219 | if ( isset( $options['maxtests'] ) ) { |
| 1220 | $n = $options['maxtests']; |
| 1221 | if ( $n > 0 ) { |
| 1222 | $this->testCases = array_slice( $this->testCases, 0, $n ); |
| 1223 | } |
| 1224 | } |
| 1225 | |
| 1226 | $defaultOpts = [ |
| 1227 | 'wrapSections' => false, |
| 1228 | 'nativeTemplateExpansion' => true, |
| 1229 | 'offsetType' => $this->offsetType, |
| 1230 | ]; |
| 1231 | ScriptUtils::setDebuggingFlags( $defaultOpts, $options ); |
| 1232 | ScriptUtils::setTemplatingAndProcessingFlags( $defaultOpts, $options ); |
| 1233 | |
| 1234 | if ( |
| 1235 | ScriptUtils::booleanOption( $options['quiet'] ?? null ) || |
| 1236 | ScriptUtils::booleanOption( $options['quieter'] ?? null ) |
| 1237 | ) { |
| 1238 | $defaultOpts['logLevels'] = [ 'fatal', 'error' ]; |
| 1239 | } |
| 1240 | |
| 1241 | // Save default logger so we can be reset it after temporarily |
| 1242 | // switching to the suppressLogger to suppress expected error messages. |
| 1243 | $this->defaultLogger = $this->siteConfig->getLogger(); |
| 1244 | |
| 1245 | /** |
| 1246 | * PORT-FIXME(T238722) |
| 1247 | * // Enable sampling to assert it's working while testing. |
| 1248 | * $parsoidConfig->loggerSampling = [ [ '/^warn(\/|$)/', 100 ] ]; |
| 1249 | * |
| 1250 | * // Override env's `setLogger` to record if we see `fatal` or `error` |
| 1251 | * // while running parser tests. (Keep it clean, folks! Use |
| 1252 | * // "suppressError" option on the test if error is expected.) |
| 1253 | * $env->setLogger = ( ( function ( $parserTests, $superSetLogger ) { |
| 1254 | * return function ( $_logger ) use ( &$parserTests ) { |
| 1255 | * superSetLogger( $_logger ); |
| 1256 | * $this->log = function ( $level ) use ( &$_logger, &$parserTests ) { |
| 1257 | * if ( $_logger !== $parserTests->suppressLogger && |
| 1258 | * preg_match( '/^(fatal|error)\b/', $level ) |
| 1259 | * ) { |
| 1260 | * $parserTests->stats->loggedErrorCount++; |
| 1261 | * } |
| 1262 | * return $_logger->log( ...$arguments ); |
| 1263 | * }; |
| 1264 | * }; |
| 1265 | * } ) ); |
| 1266 | */ |
| 1267 | |
| 1268 | $options['reportStart'](); |
| 1269 | |
| 1270 | // Run tests |
| 1271 | foreach ( $this->testCases as $test ) { |
| 1272 | try { |
| 1273 | $this->envOptions = $defaultOpts; |
| 1274 | $this->processTest( $test, $options ); |
| 1275 | } catch ( UnexpectedException $e ) { |
| 1276 | // Exit unexpected |
| 1277 | break; |
| 1278 | } |
| 1279 | } |
| 1280 | |
| 1281 | // Update knownFailures |
| 1282 | return $this->updateKnownFailures( $options ); |
| 1283 | } |
| 1284 | } |