Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
43.61% |
222 / 509 |
|
16.00% |
4 / 25 |
CRAP | |
0.00% |
0 / 1 |
| ParsoidHandler | |
43.61% |
222 / 509 |
|
16.00% |
4 / 25 |
3065.06 | |
0.00% |
0 / 1 |
| factory | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| __construct | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| getSupportedRequestTypes | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| assertDomainIsCorrect | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
42 | |||
| getParsedBody | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
42 | |||
| getOpts | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| getRequestAttributes | |
0.00% |
0 / 87 |
|
0.00% |
0 / 1 |
462 | |||
| getHtmlOutputRendererHelper | |
95.00% |
19 / 20 |
|
0.00% |
0 / 1 |
6 | |||
| getHtmlInputTransformHelper | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
1 | |||
| acceptable | |
0.00% |
0 / 31 |
|
0.00% |
0 / 1 |
306 | |||
| tryToCreatePageConfig | |
100.00% |
41 / 41 |
|
100.00% |
1 / 1 |
7 | |||
| tryToCreatePageIdentity | |
55.56% |
5 / 9 |
|
0.00% |
0 / 1 |
3.79 | |||
| getTransformEndpoint | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getPageContentEndpoint | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getRevisionContentEndpoint | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| wtLint | |
53.85% |
7 / 13 |
|
0.00% |
0 / 1 |
3.88 | |||
| wt2html | |
84.62% |
77 / 91 |
|
0.00% |
0 / 1 |
28.46 | |||
| newParsoid | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| parseHTML | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| html2wt | |
87.50% |
7 / 8 |
|
0.00% |
0 / 1 |
3.02 | |||
| pb2pb | |
0.00% |
0 / 46 |
|
0.00% |
0 / 1 |
90 | |||
| updateRedLinks | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
2 | |||
| languageConversion | |
85.42% |
41 / 48 |
|
0.00% |
0 / 1 |
4.05 | |||
| execute | n/a |
0 / 0 |
n/a |
0 / 0 |
0 | |||||
| validatePb | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| pageConfigToPageIdentity | |
44.44% |
4 / 9 |
|
0.00% |
0 / 1 |
2.69 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Copyright (C) 2011-2020 Wikimedia Foundation and others. |
| 4 | * |
| 5 | * @license GPL-2.0-or-later |
| 6 | */ |
| 7 | |
| 8 | namespace MediaWiki\Rest\Handler; |
| 9 | |
| 10 | use Composer\Semver\Semver; |
| 11 | use InvalidArgumentException; |
| 12 | use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface; |
| 13 | use LogicException; |
| 14 | use MediaWiki\Content\WikitextContent; |
| 15 | use MediaWiki\Context\RequestContext; |
| 16 | use MediaWiki\Language\LanguageCode; |
| 17 | use MediaWiki\Logger\LoggerFactory; |
| 18 | use MediaWiki\MainConfigNames; |
| 19 | use MediaWiki\MediaWikiServices; |
| 20 | use MediaWiki\Page\PageIdentity; |
| 21 | use MediaWiki\Page\ProperPageIdentity; |
| 22 | use MediaWiki\Parser\Parser; |
| 23 | use MediaWiki\Parser\ParserOptions; |
| 24 | use MediaWiki\Parser\ParserOutput; |
| 25 | use MediaWiki\Parser\Parsoid\Config\SiteConfig; |
| 26 | use MediaWiki\Parser\Parsoid\PageBundleParserOutputConverter; |
| 27 | use MediaWiki\Registration\ExtensionRegistry; |
| 28 | use MediaWiki\Rest\Handler; |
| 29 | use MediaWiki\Rest\Handler\Helper\HtmlInputTransformHelper; |
| 30 | use MediaWiki\Rest\Handler\Helper\HtmlOutputRendererHelper; |
| 31 | use MediaWiki\Rest\Handler\Helper\ParsoidFormatHelper; |
| 32 | use MediaWiki\Rest\HttpException; |
| 33 | use MediaWiki\Rest\LocalizedHttpException; |
| 34 | use MediaWiki\Rest\RequestInterface; |
| 35 | use MediaWiki\Rest\Response; |
| 36 | use MediaWiki\Rest\ResponseHeaders; |
| 37 | use MediaWiki\Revision\MutableRevisionRecord; |
| 38 | use MediaWiki\Revision\RevisionAccessException; |
| 39 | use MediaWiki\Revision\RevisionLookup; |
| 40 | use MediaWiki\Revision\SlotRecord; |
| 41 | use MediaWiki\Revision\SuppressedDataException; |
| 42 | use MediaWiki\Title\MalformedTitleException; |
| 43 | use MediaWiki\Title\Title; |
| 44 | use MediaWiki\WikiMap\WikiMap; |
| 45 | use MobileContext; |
| 46 | use Wikimedia\Http\HttpAcceptParser; |
| 47 | use Wikimedia\Message\DataMessageValue; |
| 48 | use Wikimedia\Message\MessageValue; |
| 49 | use Wikimedia\Parsoid\Config\DataAccess; |
| 50 | use Wikimedia\Parsoid\Config\PageConfig; |
| 51 | use Wikimedia\Parsoid\Config\PageConfigFactory; |
| 52 | use Wikimedia\Parsoid\Core\ClientError; |
| 53 | use Wikimedia\Parsoid\Core\HtmlPageBundle; |
| 54 | use Wikimedia\Parsoid\Core\ResourceLimitExceededException; |
| 55 | use Wikimedia\Parsoid\DOM\Document; |
| 56 | use Wikimedia\Parsoid\Parsoid; |
| 57 | use Wikimedia\Parsoid\Utils\ContentUtils; |
| 58 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 59 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 60 | use Wikimedia\Parsoid\Utils\Timing; |
| 61 | |
| 62 | // TODO logging, timeouts(?), CORS |
| 63 | // TODO content negotiation (routes.js routes.acceptable) |
| 64 | // TODO handle MaxConcurrentCallsError (pool counter?) |
| 65 | |
| 66 | /** |
| 67 | * Base class for Parsoid handlers. |
| 68 | * @internal For use by the Parsoid extension |
| 69 | */ |
| 70 | abstract class ParsoidHandler extends Handler { |
| 71 | |
| 72 | private RevisionLookup $revisionLookup; |
| 73 | protected SiteConfig $siteConfig; |
| 74 | protected PageConfigFactory $pageConfigFactory; |
| 75 | protected DataAccess $dataAccess; |
| 76 | |
| 77 | /** @var ExtensionRegistry */ |
| 78 | protected $extensionRegistry; |
| 79 | |
| 80 | /** @var ?StatsdDataFactoryInterface A statistics aggregator */ |
| 81 | protected $metrics; |
| 82 | |
| 83 | /** @var array */ |
| 84 | private $requestAttributes; |
| 85 | |
| 86 | public static function factory(): static { |
| 87 | $services = MediaWikiServices::getInstance(); |
| 88 | // @phan-suppress-next-line PhanTypeInstantiateAbstractStatic |
| 89 | return new static( |
| 90 | $services->getRevisionLookup(), |
| 91 | $services->getParsoidSiteConfig(), |
| 92 | $services->getParsoidPageConfigFactory(), |
| 93 | $services->getParsoidDataAccess() |
| 94 | ); |
| 95 | } |
| 96 | |
| 97 | public function __construct( |
| 98 | RevisionLookup $revisionLookup, |
| 99 | SiteConfig $siteConfig, |
| 100 | PageConfigFactory $pageConfigFactory, |
| 101 | DataAccess $dataAccess |
| 102 | ) { |
| 103 | $this->revisionLookup = $revisionLookup; |
| 104 | $this->siteConfig = $siteConfig; |
| 105 | $this->pageConfigFactory = $pageConfigFactory; |
| 106 | $this->dataAccess = $dataAccess; |
| 107 | $this->extensionRegistry = ExtensionRegistry::getInstance(); |
| 108 | $this->metrics = $siteConfig->metrics(); |
| 109 | } |
| 110 | |
| 111 | public function getSupportedRequestTypes(): array { |
| 112 | return array_merge( parent::getSupportedRequestTypes(), [ |
| 113 | 'application/x-www-form-urlencoded', |
| 114 | 'multipart/form-data' |
| 115 | ] ); |
| 116 | } |
| 117 | |
| 118 | /** |
| 119 | * Verify that the {domain} path parameter matches the actual domain. |
| 120 | * @todo Remove this when we no longer need to support the {domain} |
| 121 | * parameter with backwards compatibility with the parsoid |
| 122 | * extension. |
| 123 | * @param string $domain Domain name parameter to validate |
| 124 | */ |
| 125 | protected function assertDomainIsCorrect( $domain ): void { |
| 126 | // We are cutting some corners here (IDN, non-ASCII casing) |
| 127 | // since domain name support is provisional. |
| 128 | // TODO use a proper validator instead |
| 129 | $server = RequestContext::getMain()->getConfig()->get( MainConfigNames::Server ); |
| 130 | $expectedDomain = parse_url( $server, PHP_URL_HOST ); |
| 131 | if ( !$expectedDomain ) { |
| 132 | throw new LogicException( 'Cannot parse $wgServer' ); |
| 133 | } |
| 134 | if ( strcasecmp( $expectedDomain, $domain ) === 0 ) { |
| 135 | return; |
| 136 | } |
| 137 | |
| 138 | // TODO: This should really go away! It's only acceptable because |
| 139 | // this entire method is going to be removed once we no longer |
| 140 | // need the parsoid extension endpoints with the {domain} parameter. |
| 141 | if ( $this->extensionRegistry->isLoaded( 'MobileFrontend' ) ) { |
| 142 | // @phan-suppress-next-line PhanUndeclaredClassMethod |
| 143 | $mobileServer = MobileContext::singleton()->getMobileUrl( $server ); |
| 144 | $expectedMobileDomain = parse_url( $mobileServer, PHP_URL_HOST ); |
| 145 | if ( $expectedMobileDomain && strcasecmp( $expectedMobileDomain, $domain ) === 0 ) { |
| 146 | return; |
| 147 | } |
| 148 | } |
| 149 | |
| 150 | $msg = new DataMessageValue( |
| 151 | 'mwparsoid-invalid-domain', |
| 152 | [], |
| 153 | 'invalid-domain', |
| 154 | [ 'expected' => $expectedDomain, 'actual' => $domain, ] |
| 155 | ); |
| 156 | |
| 157 | throw new LocalizedHttpException( $msg, 400, [ |
| 158 | 'error' => 'parameter-validation-failed', |
| 159 | 'name' => 'domain', |
| 160 | 'value' => $domain, |
| 161 | 'failureCode' => $msg->getCode(), |
| 162 | 'failureData' => $msg->getData(), |
| 163 | ] ); |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * Get the parsed body by content-type |
| 168 | */ |
| 169 | protected function getParsedBody(): array { |
| 170 | $request = $this->getRequest(); |
| 171 | [ $contentType ] = explode( ';', $request->getHeader( 'Content-Type' )[0] ?? '', 2 ); |
| 172 | switch ( $contentType ) { |
| 173 | case 'application/x-www-form-urlencoded': |
| 174 | case 'multipart/form-data': |
| 175 | return $request->getPostParams(); |
| 176 | case 'application/json': |
| 177 | $json = json_decode( $request->getBody()->getContents(), true ); |
| 178 | if ( !is_array( $json ) ) { |
| 179 | throw new LocalizedHttpException( |
| 180 | new MessageValue( "rest-json-body-parse-error", [ 'not a valid JSON object' ] ), 400 ); |
| 181 | } |
| 182 | return $json; |
| 183 | default: |
| 184 | throw new LocalizedHttpException( |
| 185 | new MessageValue( "rest-unsupported-content-type", [ $contentType ?? '(null)' ] ), |
| 186 | 415 |
| 187 | ); |
| 188 | } |
| 189 | } |
| 190 | |
| 191 | protected function getOpts( array $body, RequestInterface $request ): array { |
| 192 | return array_merge( |
| 193 | $body, |
| 194 | array_intersect_key( $request->getPathParams(), [ 'from' => true, 'format' => true ] ) |
| 195 | ); |
| 196 | } |
| 197 | |
| 198 | /** |
| 199 | * Rough equivalent of req.local from Parsoid-JS. |
| 200 | * FIXME most of these should be replaced with more native ways of handling the request. |
| 201 | * @return array |
| 202 | */ |
| 203 | protected function &getRequestAttributes(): array { |
| 204 | if ( $this->requestAttributes ) { |
| 205 | return $this->requestAttributes; |
| 206 | } |
| 207 | |
| 208 | $request = $this->getRequest(); |
| 209 | $body = ( $request->getMethod() === 'POST' ) ? $this->getParsedBody() : []; |
| 210 | $opts = $this->getOpts( $body, $request ); |
| 211 | '@phan-var array<string,array|bool|string> $opts'; // @var array<string,array|bool|string> $opts |
| 212 | $contentLanguage = $request->getHeaderLine( 'Content-Language' ) ?: null; |
| 213 | if ( $contentLanguage ) { |
| 214 | $contentLanguage = LanguageCode::normalizeNonstandardCodeAndWarn( |
| 215 | $contentLanguage |
| 216 | ); |
| 217 | } |
| 218 | $attribs = [ |
| 219 | 'pageName' => $request->getPathParam( 'title' ) ?? '', |
| 220 | 'oldid' => $request->getPathParam( 'revision' ), |
| 221 | // "body_only" flag to return just the body (instead of the entire HTML doc) |
| 222 | // We would like to deprecate use of this flag: T181657 |
| 223 | 'body_only' => $request->getQueryParams()['body_only'] ?? $body['body_only'] ?? null, |
| 224 | 'errorEnc' => ParsoidFormatHelper::ERROR_ENCODING[$opts['format']] ?? 'plain', |
| 225 | 'iwp' => WikiMap::getCurrentWikiId(), // PORT-FIXME verify |
| 226 | 'offsetType' => $body['offsetType'] |
| 227 | ?? $request->getQueryParams()['offsetType'] |
| 228 | // Lint requests should return UCS2 offsets by default |
| 229 | ?? ( $opts['format'] === ParsoidFormatHelper::FORMAT_LINT ? 'ucs2' : 'byte' ), |
| 230 | 'pagelanguage' => $contentLanguage, |
| 231 | ]; |
| 232 | |
| 233 | // For use in getHtmlOutputRendererHelper |
| 234 | $opts['stash'] = $request->getQueryParams()['stash'] ?? false; |
| 235 | |
| 236 | if ( $request->getMethod() === 'POST' ) { |
| 237 | if ( isset( $opts['original']['revid'] ) ) { |
| 238 | $attribs['oldid'] = $opts['original']['revid']; |
| 239 | } |
| 240 | if ( isset( $opts['original']['title'] ) ) { |
| 241 | $attribs['pageName'] = $opts['original']['title']; |
| 242 | } |
| 243 | } |
| 244 | if ( $attribs['oldid'] !== null ) { |
| 245 | if ( $attribs['oldid'] === '' ) { |
| 246 | $attribs['oldid'] = null; |
| 247 | } else { |
| 248 | $attribs['oldid'] = (int)$attribs['oldid']; |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | // For use in getHtmlOutputRendererHelper |
| 253 | $opts['accept-language'] = $request->getHeaderLine( 'Accept-Language' ) ?: null; |
| 254 | |
| 255 | $acceptLanguage = null; |
| 256 | if ( $opts['accept-language'] !== null ) { |
| 257 | $acceptLanguage = LanguageCode::normalizeNonstandardCodeAndWarn( |
| 258 | HtmlOutputRendererHelper::getAcceptedTargetLanguage( |
| 259 | $opts['accept-language'] |
| 260 | ) |
| 261 | ); |
| 262 | } |
| 263 | |
| 264 | // Init pageName if oldid is provided and is a valid revision |
| 265 | if ( ( $attribs['pageName'] === '' ) && $attribs['oldid'] ) { |
| 266 | $rev = $this->revisionLookup->getRevisionById( $attribs['oldid'] ); |
| 267 | if ( $rev ) { |
| 268 | $attribs['pageName'] = $rev->getPage()->getDBkey(); |
| 269 | } |
| 270 | } |
| 271 | |
| 272 | $attribs['envOptions'] = [ |
| 273 | // We use `prefix` but ought to use `domain` (T206764) |
| 274 | 'prefix' => $attribs['iwp'], |
| 275 | // For the legacy "domain" path parameter used by the endpoints exposed |
| 276 | // by the parsoid extension. Will be null for core endpoints. |
| 277 | 'domain' => $request->getPathParam( 'domain' ), |
| 278 | 'pageName' => $attribs['pageName'], |
| 279 | 'cookie' => $request->getHeaderLine( 'Cookie' ), |
| 280 | 'reqId' => $request->getHeaderLine( 'X-Request-Id' ), |
| 281 | 'userAgent' => $request->getHeaderLine( 'User-Agent' ), |
| 282 | // Used in pb2pb variant updates and wtLint |
| 283 | 'htmlVariantLanguage' => $acceptLanguage, |
| 284 | // Semver::satisfies checks below expect a valid outputContentVersion value. |
| 285 | // Better to set it here instead of adding the default value at every check. |
| 286 | 'outputContentVersion' => Parsoid::defaultHTMLVersion(), |
| 287 | ]; |
| 288 | |
| 289 | # Convert language codes in $opts['updates']['variant'] if present |
| 290 | $sourceVariant = $opts['updates']['variant']['wikitext'] ?? |
| 291 | $opts['updates']['variant']['source'] ?? null; |
| 292 | if ( $sourceVariant ) { |
| 293 | $sourceVariant = LanguageCode::normalizeNonstandardCodeAndWarn( |
| 294 | $sourceVariant |
| 295 | ); |
| 296 | unset( $opts['updates']['variant']['source'] ); |
| 297 | $opts['updates']['variant']['wikitext'] = $sourceVariant; |
| 298 | } |
| 299 | $targetVariant = $opts['updates']['variant']['html'] ?? |
| 300 | $opts['updates']['variant']['target'] ?? null; |
| 301 | if ( $targetVariant ) { |
| 302 | $targetVariant = LanguageCode::normalizeNonstandardCodeAndWarn( |
| 303 | $targetVariant |
| 304 | ); |
| 305 | unset( $opts['updates']['variant']['target'] ); |
| 306 | $opts['updates']['variant']['html'] = $targetVariant; |
| 307 | } |
| 308 | if ( isset( $opts['wikitext']['headers']['content-language'] ) ) { |
| 309 | $contentLanguage = $opts['wikitext']['headers']['content-language']; |
| 310 | $contentLanguage = LanguageCode::normalizeNonstandardCodeAndWarn( |
| 311 | $contentLanguage |
| 312 | ); |
| 313 | $opts['wikitext']['headers']['content-language'] = $contentLanguage; |
| 314 | } |
| 315 | if ( isset( $opts['original']['wikitext']['headers']['content-language'] ) ) { |
| 316 | $contentLanguage = $opts['original']['wikitext']['headers']['content-language']; |
| 317 | $contentLanguage = LanguageCode::normalizeNonstandardCodeAndWarn( |
| 318 | $contentLanguage |
| 319 | ); |
| 320 | $opts['original']['wikitext']['headers']['content-language'] = $contentLanguage; |
| 321 | } |
| 322 | |
| 323 | $attribs['opts'] = $opts; |
| 324 | |
| 325 | // TODO: Remove assertDomainIsCorrect() once we no longer need to support the {domain} |
| 326 | // parameter for the endpoints exposed by the parsoid extension. |
| 327 | if ( $attribs['envOptions']['domain'] !== null ) { |
| 328 | $this->assertDomainIsCorrect( $attribs['envOptions']['domain'] ); |
| 329 | } |
| 330 | |
| 331 | $this->requestAttributes = $attribs; |
| 332 | return $this->requestAttributes; |
| 333 | } |
| 334 | |
| 335 | /** |
| 336 | * @param array $attribs |
| 337 | * @param ?string $source |
| 338 | * @param PageIdentity $page |
| 339 | * @param ?int $revId |
| 340 | * |
| 341 | * @return HtmlOutputRendererHelper |
| 342 | */ |
| 343 | private function getHtmlOutputRendererHelper( |
| 344 | array $attribs, |
| 345 | ?string $source, |
| 346 | PageIdentity $page, |
| 347 | ?int $revId |
| 348 | ): HtmlOutputRendererHelper { |
| 349 | $services = MediaWikiServices::getInstance(); |
| 350 | |
| 351 | // Request lenient rev handling |
| 352 | $lenientRevHandling = true; |
| 353 | |
| 354 | $authority = $this->getAuthority(); |
| 355 | |
| 356 | $params = []; |
| 357 | $helper = $services->getPageRestHelperFactory()->newHtmlOutputRendererHelper( |
| 358 | $page, $params, $authority, $revId, $lenientRevHandling |
| 359 | ); |
| 360 | |
| 361 | // XXX: should default to the page's content model? |
| 362 | $model = $attribs['opts']['contentmodel'] |
| 363 | ?? ( $attribs['envOptions']['contentmodel'] ?? CONTENT_MODEL_WIKITEXT ); |
| 364 | |
| 365 | if ( $source !== null ) { |
| 366 | $helper->setContentSource( $source, $model ); |
| 367 | } |
| 368 | |
| 369 | if ( isset( $attribs['opts']['stash'] ) ) { |
| 370 | $helper->setStashingEnabled( $attribs['opts']['stash'] ); |
| 371 | } |
| 372 | |
| 373 | if ( isset( $attribs['envOptions']['outputContentVersion'] ) ) { |
| 374 | $helper->setOutputProfileVersion( $attribs['envOptions']['outputContentVersion'] ); |
| 375 | } |
| 376 | |
| 377 | if ( isset( $attribs['pagelanguage'] ) ) { |
| 378 | $helper->setPageLanguage( $attribs['pagelanguage'] ); |
| 379 | } |
| 380 | |
| 381 | if ( isset( $attribs['opts']['accept-language'] ) ) { |
| 382 | $helper->setVariantConversionLanguage( $attribs['opts']['accept-language'] ); |
| 383 | } |
| 384 | |
| 385 | return $helper; |
| 386 | } |
| 387 | |
| 388 | /** |
| 389 | * @param array $attribs |
| 390 | * @param string $html |
| 391 | * @param PageIdentity $page |
| 392 | * |
| 393 | * @return HtmlInputTransformHelper |
| 394 | */ |
| 395 | protected function getHtmlInputTransformHelper( |
| 396 | array $attribs, |
| 397 | string $html, |
| 398 | PageIdentity $page |
| 399 | ): HtmlInputTransformHelper { |
| 400 | $services = MediaWikiServices::getInstance(); |
| 401 | |
| 402 | $parameters = $attribs['opts'] + $attribs; |
| 403 | $body = $attribs['opts']; |
| 404 | |
| 405 | $body['html'] = $html; |
| 406 | |
| 407 | $helper = $services->getPageRestHelperFactory()->newHtmlInputTransformHelper( |
| 408 | $attribs['envOptions'] + [ |
| 409 | 'offsetType' => $attribs['offsetType'], |
| 410 | ], |
| 411 | $page, |
| 412 | $body, |
| 413 | $parameters |
| 414 | ); |
| 415 | |
| 416 | $helper->setMetrics( $this->siteConfig->prefixedStatsFactory() ); |
| 417 | |
| 418 | return $helper; |
| 419 | } |
| 420 | |
| 421 | /** |
| 422 | * FIXME: Combine with ParsoidFormatHelper::parseContentTypeHeader |
| 423 | */ |
| 424 | private const NEW_SPEC = |
| 425 | '#^https://www.mediawiki.org/wiki/Specs/(HTML|pagebundle)/(\d+\.\d+\.\d+)$#D'; |
| 426 | |
| 427 | /** |
| 428 | * This method checks if we support the requested content formats |
| 429 | * As a side-effect, it updates $attribs to set outputContentVersion |
| 430 | * that Parsoid should generate based on request headers. |
| 431 | * |
| 432 | * @param array &$attribs Request attributes from getRequestAttributes() |
| 433 | * @return bool |
| 434 | */ |
| 435 | protected function acceptable( array &$attribs ): bool { |
| 436 | $request = $this->getRequest(); |
| 437 | $format = $attribs['opts']['format']; |
| 438 | |
| 439 | if ( |
| 440 | $format === ParsoidFormatHelper::FORMAT_WIKITEXT || |
| 441 | $format === ParsoidFormatHelper::FORMAT_LINT |
| 442 | ) { |
| 443 | return true; |
| 444 | } |
| 445 | |
| 446 | $acceptHeader = $request->getHeader( 'Accept' ); |
| 447 | if ( !$acceptHeader ) { |
| 448 | return true; |
| 449 | } |
| 450 | |
| 451 | $parser = new HttpAcceptParser(); |
| 452 | $acceptableTypes = $parser->parseAccept( $acceptHeader[0] ); // FIXME: Multiple headers valid? |
| 453 | if ( !$acceptableTypes ) { |
| 454 | return true; |
| 455 | } |
| 456 | |
| 457 | // `acceptableTypes` is already sorted by quality. |
| 458 | foreach ( $acceptableTypes as $t ) { |
| 459 | $type = "{$t['type']}/{$t['subtype']}"; |
| 460 | $profile = $t['params']['profile'] ?? null; |
| 461 | if ( |
| 462 | ( $format === ParsoidFormatHelper::FORMAT_HTML && $type === 'text/html' ) || |
| 463 | ( $format === ParsoidFormatHelper::FORMAT_PAGEBUNDLE && $type === 'application/json' ) |
| 464 | ) { |
| 465 | if ( $profile ) { |
| 466 | preg_match( self::NEW_SPEC, $profile, $matches ); |
| 467 | if ( $matches && strtolower( $matches[1] ) === $format ) { |
| 468 | $contentVersion = Parsoid::resolveContentVersion( $matches[2] ); |
| 469 | if ( $contentVersion ) { |
| 470 | // $attribs mutated here! |
| 471 | $attribs['envOptions']['outputContentVersion'] = $contentVersion; |
| 472 | return true; |
| 473 | } else { |
| 474 | continue; |
| 475 | } |
| 476 | } else { |
| 477 | continue; |
| 478 | } |
| 479 | } else { |
| 480 | return true; |
| 481 | } |
| 482 | } elseif ( |
| 483 | ( $type === '*/*' ) || |
| 484 | ( $format === ParsoidFormatHelper::FORMAT_HTML && $type === 'text/*' ) |
| 485 | ) { |
| 486 | return true; |
| 487 | } |
| 488 | } |
| 489 | |
| 490 | return false; |
| 491 | } |
| 492 | |
| 493 | /** |
| 494 | * Try to create a PageConfig object. If we get an exception (because content |
| 495 | * may be missing or inaccessible), throw an appropriate HTTP response object |
| 496 | * for callers to handle. |
| 497 | * |
| 498 | * @param array $attribs |
| 499 | * @param ?string $wikitextOverride |
| 500 | * Custom wikitext to use instead of the real content of the page. |
| 501 | * @param bool $html2WtMode |
| 502 | * @return PageConfig |
| 503 | * @throws HttpException |
| 504 | */ |
| 505 | protected function tryToCreatePageConfig( |
| 506 | array $attribs, ?string $wikitextOverride = null, bool $html2WtMode = false |
| 507 | ): PageConfig { |
| 508 | $revId = $attribs['oldid']; |
| 509 | $pagelanguageOverride = $attribs['pagelanguage']; |
| 510 | $title = $attribs['pageName']; |
| 511 | |
| 512 | $title = ( $title !== '' ) ? Title::newFromText( $title ) : Title::newMainPage(); |
| 513 | if ( !$title ) { |
| 514 | throw new LocalizedHttpException( |
| 515 | new MessageValue( "rest-invalid-title", [ 'pageName' ] ), 400 |
| 516 | ); |
| 517 | } |
| 518 | $user = RequestContext::getMain()->getUser(); |
| 519 | |
| 520 | if ( $wikitextOverride === null ) { |
| 521 | $revisionRecord = null; |
| 522 | } else { |
| 523 | // Create a mutable revision record point to the same revision |
| 524 | // and set to the desired wikitext. |
| 525 | $revisionRecord = new MutableRevisionRecord( $title ); |
| 526 | // Don't set id to $revId if we have $wikitextOverride |
| 527 | // A revision corresponds to specific wikitext, which $wikitextOverride |
| 528 | // might not be. |
| 529 | $revisionRecord->setId( 0 ); |
| 530 | $revisionRecord->setSlot( |
| 531 | SlotRecord::newUnsaved( |
| 532 | SlotRecord::MAIN, |
| 533 | new WikitextContent( $wikitextOverride ) |
| 534 | ) |
| 535 | ); |
| 536 | } |
| 537 | |
| 538 | $hasOldId = ( $revId !== null ); |
| 539 | $ensureAccessibleContent = !$html2WtMode || $hasOldId; |
| 540 | |
| 541 | // When transforming for lint, we aren't going to go through a ParserOutputAccess |
| 542 | // to checkPreconditions |
| 543 | $checkAuthority = ( |
| 544 | ( $attribs['opts']['format'] ?? '' ) === ParsoidFormatHelper::FORMAT_LINT |
| 545 | ); |
| 546 | |
| 547 | try { |
| 548 | // Note: Parsoid by design isn't supposed to use the user |
| 549 | // context right now, and all user state is expected to be |
| 550 | // introduced as a post-parse transform. So although we pass a |
| 551 | // User here, it only currently affects the output in obscure |
| 552 | // corner cases; see PageConfigFactory::create() for more. |
| 553 | // @phan-suppress-next-line PhanUndeclaredMethod method defined in subtype |
| 554 | $pageConfig = $this->pageConfigFactory->createFromParserOptions( |
| 555 | ParserOptions::newFromUser( $user ), |
| 556 | $title, |
| 557 | $revisionRecord ?? $revId, |
| 558 | $pagelanguageOverride, |
| 559 | $ensureAccessibleContent, |
| 560 | $checkAuthority |
| 561 | ); |
| 562 | } catch ( SuppressedDataException $e ) { |
| 563 | throw new LocalizedHttpException( |
| 564 | new MessageValue( "rest-permission-denied-revision", [ $e->getMessage() ] ), 403 |
| 565 | ); |
| 566 | } catch ( RevisionAccessException $e ) { |
| 567 | throw new LocalizedHttpException( |
| 568 | new MessageValue( "rest-specified-revision-unavailable", [ $e->getMessage() ] ), 404 |
| 569 | ); |
| 570 | } |
| 571 | |
| 572 | // All good! |
| 573 | return $pageConfig; |
| 574 | } |
| 575 | |
| 576 | /** |
| 577 | * Try to create a PageIdentity object. |
| 578 | * If no page is specified in the request, this will return the wiki's main page. |
| 579 | * If an invalid page is requested, this throws an appropriate HTTPException. |
| 580 | * |
| 581 | * @param array $attribs |
| 582 | * @return PageIdentity |
| 583 | * @throws HttpException |
| 584 | */ |
| 585 | protected function tryToCreatePageIdentity( array $attribs ): PageIdentity { |
| 586 | if ( $attribs['pageName'] === '' ) { |
| 587 | return Title::newMainPage(); |
| 588 | } |
| 589 | |
| 590 | // XXX: Should be injected, but the Parsoid extension relies on the |
| 591 | // constructor signature. Also, ParsoidHandler should go away soon anyway. |
| 592 | $pageStore = MediaWikiServices::getInstance()->getPageStore(); |
| 593 | |
| 594 | $page = $pageStore->getPageByText( $attribs['pageName'] ); |
| 595 | |
| 596 | if ( !$page ) { |
| 597 | throw new LocalizedHttpException( |
| 598 | new MessageValue( "rest-invalid-title", [ 'pageName' ] ), 400 |
| 599 | ); |
| 600 | } |
| 601 | |
| 602 | return $page; |
| 603 | } |
| 604 | |
| 605 | /** |
| 606 | * Get the path for the transform endpoint. May be overwritten to override the path. |
| 607 | * |
| 608 | * This is done in the parsoid extension, for backwards compatibility |
| 609 | * with the old endpoint URLs. |
| 610 | * |
| 611 | * @stable to override |
| 612 | * |
| 613 | * @param string $format The format the endpoint is expected to return. |
| 614 | * |
| 615 | * @return string |
| 616 | */ |
| 617 | protected function getTransformEndpoint( string $format = ParsoidFormatHelper::FORMAT_HTML ): string { |
| 618 | return '/coredev/v0/transform/{from}/to/{format}/{title}/{revision}'; |
| 619 | } |
| 620 | |
| 621 | /** |
| 622 | * Get the path for the page content endpoint. May be overwritten to override the path. |
| 623 | * |
| 624 | * This is done in the parsoid extension, for backwards compatibility |
| 625 | * with the old endpoint URLs. |
| 626 | * |
| 627 | * @stable to override |
| 628 | * |
| 629 | * @param string $format The format the endpoint is expected to return. |
| 630 | * |
| 631 | * @return string |
| 632 | */ |
| 633 | protected function getPageContentEndpoint( string $format = ParsoidFormatHelper::FORMAT_HTML ): string { |
| 634 | if ( $format !== ParsoidFormatHelper::FORMAT_HTML ) { |
| 635 | throw new InvalidArgumentException( 'Unsupported page content format: ' . $format ); |
| 636 | } |
| 637 | return '/v1/page/{title}/html'; |
| 638 | } |
| 639 | |
| 640 | /** |
| 641 | * Get the path for the page content endpoint. May be overwritten to override the path. |
| 642 | * |
| 643 | * This is done in the parsoid extension, for backwards compatibility |
| 644 | * with the old endpoint URLs. |
| 645 | * |
| 646 | * @stable to override |
| 647 | * |
| 648 | * @param string $format The format the endpoint is expected to return. |
| 649 | * |
| 650 | * @return string |
| 651 | */ |
| 652 | protected function getRevisionContentEndpoint( string $format = ParsoidFormatHelper::FORMAT_HTML ): string { |
| 653 | if ( $format !== ParsoidFormatHelper::FORMAT_HTML ) { |
| 654 | throw new InvalidArgumentException( 'Unsupported revision content format: ' . $format ); |
| 655 | } |
| 656 | return '/v1/revision/{revision}/html'; |
| 657 | } |
| 658 | |
| 659 | private function wtLint( |
| 660 | PageConfig $pageConfig, array $attribs, ?array $linterOverrides = [] |
| 661 | ): array { |
| 662 | $envOptions = $attribs['envOptions'] + [ |
| 663 | 'linterOverrides' => $linterOverrides, |
| 664 | 'offsetType' => $attribs['offsetType'], |
| 665 | ]; |
| 666 | try { |
| 667 | $parsoid = $this->newParsoid(); |
| 668 | $parserOutput = new ParserOutput(); |
| 669 | return $parsoid->wikitext2lint( $pageConfig, $envOptions, $parserOutput ); |
| 670 | } catch ( ClientError $e ) { |
| 671 | throw new LocalizedHttpException( new MessageValue( "rest-parsoid-error", [ $e->getMessage() ] ), 400 ); |
| 672 | } catch ( ResourceLimitExceededException $e ) { |
| 673 | throw new LocalizedHttpException( |
| 674 | new MessageValue( "rest-parsoid-resource-exceeded", [ $e->getMessage() ] ), 413 |
| 675 | ); |
| 676 | } |
| 677 | } |
| 678 | |
| 679 | /** |
| 680 | * Wikitext -> HTML helper. |
| 681 | * Spec'd in https://phabricator.wikimedia.org/T75955 and the API tests. |
| 682 | * |
| 683 | * @param PageConfig $pageConfig |
| 684 | * @param array $attribs Request attributes from getRequestAttributes() |
| 685 | * @param ?string $wikitext Wikitext to transform (or null to use the |
| 686 | * page specified in the request attributes). |
| 687 | * |
| 688 | * @return Response |
| 689 | */ |
| 690 | protected function wt2html( |
| 691 | PageConfig $pageConfig, array $attribs, ?string $wikitext = null |
| 692 | ) { |
| 693 | $request = $this->getRequest(); |
| 694 | $opts = $attribs['opts']; |
| 695 | $format = $opts['format']; |
| 696 | $oldid = $attribs['oldid']; |
| 697 | $stash = $opts['stash'] ?? false; |
| 698 | |
| 699 | if ( $format === ParsoidFormatHelper::FORMAT_LINT ) { |
| 700 | $linterOverrides = []; |
| 701 | if ( $this->extensionRegistry->isLoaded( 'Linter' ) ) { // T360809 |
| 702 | $disabled = []; |
| 703 | $services = MediaWikiServices::getInstance(); |
| 704 | $linterCategories = $services->getMainConfig()->get( 'LinterCategories' ); |
| 705 | foreach ( $linterCategories as $name => $cat ) { |
| 706 | if ( $cat['priority'] === 'none' ) { |
| 707 | $disabled[] = $name; |
| 708 | } |
| 709 | } |
| 710 | $linterOverrides['disabled'] = $disabled; |
| 711 | } |
| 712 | $lints = $this->wtLint( $pageConfig, $attribs, $linterOverrides ); |
| 713 | $response = $this->getResponseFactory()->createJson( $lints ); |
| 714 | return $response; |
| 715 | } |
| 716 | |
| 717 | // TODO: This method should take a PageIdentity + revId, |
| 718 | // to reduce the usage of PageConfig in MW core. |
| 719 | $helper = $this->getHtmlOutputRendererHelper( |
| 720 | $attribs, |
| 721 | $wikitext, |
| 722 | $this->pageConfigToPageIdentity( $pageConfig ), |
| 723 | // Id will be 0 if we have $wikitext but that isn't valid |
| 724 | // to call $helper->setRevision with. In any case, the revision |
| 725 | // will be reset when $helper->setContent is called with $wikitext. |
| 726 | // Ideally, the revision would be pass through here instead of |
| 727 | // the id and wikitext. |
| 728 | $pageConfig->getRevisionId() ?: null |
| 729 | ); |
| 730 | |
| 731 | $needsPageBundle = ( $format === ParsoidFormatHelper::FORMAT_PAGEBUNDLE ); |
| 732 | |
| 733 | if ( $attribs['body_only'] ) { |
| 734 | $helper->setFlavor( 'fragment' ); |
| 735 | } elseif ( !$needsPageBundle ) { |
| 736 | // Inline data-parsoid. This will happen when no special params are set. |
| 737 | $helper->setFlavor( 'edit' ); |
| 738 | } |
| 739 | |
| 740 | if ( $wikitext === null && $oldid !== null ) { |
| 741 | $mstr = 'pageWithOldid'; |
| 742 | } else { |
| 743 | $mstr = 'wt'; |
| 744 | } |
| 745 | |
| 746 | $parseTiming = Timing::start(); |
| 747 | |
| 748 | if ( $needsPageBundle ) { |
| 749 | $pb = $helper->getPageBundle(); |
| 750 | |
| 751 | // Handle custom offset requests as a pb2pb transform |
| 752 | if ( |
| 753 | $helper->isParsoidContent() && |
| 754 | ( $attribs['offsetType'] !== 'byte' ) |
| 755 | ) { |
| 756 | $parsoid = $this->newParsoid(); |
| 757 | $pb = $parsoid->pb2pb( |
| 758 | $pageConfig, |
| 759 | 'convertoffsets', |
| 760 | $pb, |
| 761 | [ |
| 762 | 'inputOffsetType' => 'byte', |
| 763 | 'outputOffsetType' => $attribs['offsetType'] |
| 764 | ] |
| 765 | ); |
| 766 | } |
| 767 | |
| 768 | if ( $attribs['body_only'] ) { |
| 769 | $pb->html = Parser::extractBody( $pb->html ); |
| 770 | } |
| 771 | |
| 772 | $response = $this->getResponseFactory()->createJson( $pb->responseData() ); |
| 773 | $helper->putHeaders( $response, false ); |
| 774 | |
| 775 | ParsoidFormatHelper::setContentType( |
| 776 | $response, |
| 777 | ParsoidFormatHelper::FORMAT_PAGEBUNDLE, |
| 778 | $pb->version |
| 779 | ); |
| 780 | } else { |
| 781 | $out = $helper->getHtml(); |
| 782 | |
| 783 | // TODO: offsetType conversion isn't supported right now for non-pagebundle endpoints |
| 784 | // Once the OutputTransform framework lands, we might revisit this. |
| 785 | |
| 786 | $response = $this->getResponseFactory()->create(); |
| 787 | if ( $attribs['body_only'] ) { |
| 788 | // body_only must yield body-only output. getContentHolderText() |
| 789 | // lazily strips a full-document wrapper, so it both honors |
| 790 | // body_only and keeps fixing the variant-conversion case (which |
| 791 | // re-wraps the fragment into a full document). |
| 792 | $response->getBody()->write( $out->getContentHolderText() ); |
| 793 | } else { |
| 794 | // The 'edit' flavor is a full document (with inline data-parsoid |
| 795 | // attributes); emit it from the page bundle. |
| 796 | $response->getBody()->write( $helper->getPageBundle()->html ); |
| 797 | } |
| 798 | |
| 799 | $helper->putHeaders( $response, true ); |
| 800 | |
| 801 | // Emit an ETag only if stashing is enabled. It's not reliably useful otherwise. |
| 802 | if ( $stash ) { |
| 803 | $eTag = $helper->getETag(); |
| 804 | if ( $eTag ) { |
| 805 | $response->setHeader( 'ETag', $eTag ); |
| 806 | } |
| 807 | } |
| 808 | } |
| 809 | |
| 810 | // XXX: For pagebundle requests, this can be somewhat inflated |
| 811 | // because of pagebundle json-encoding overheads |
| 812 | $outSize = $response->getBody()->getSize(); |
| 813 | $parseTime = $parseTiming->end(); |
| 814 | |
| 815 | // Ignore slow parse metrics for non-oldid parses |
| 816 | if ( $mstr === 'pageWithOldid' ) { |
| 817 | if ( $parseTime > 3000 ) { |
| 818 | LoggerFactory::getInstance( 'slow-parsoid' ) |
| 819 | ->info( 'Parsing {title} was slow, took {time} seconds', [ |
| 820 | 'time' => number_format( $parseTime / 1000, 2 ), |
| 821 | 'title' => Title::newFromLinkTarget( $pageConfig->getLinkTarget() )->getPrefixedText(), |
| 822 | ] ); |
| 823 | } |
| 824 | |
| 825 | if ( $parseTime > 10 && $outSize > 100 ) { |
| 826 | // * Don't bother with this metric for really small parse times |
| 827 | // p99 for initialization time is ~7ms according to grafana. |
| 828 | // So, 10ms ensures that startup overheads don't skew the metrics |
| 829 | // * For body_only=false requests, <head> section isn't generated |
| 830 | // and if the output is small, per-request overheads can skew |
| 831 | // the timePerKB metrics. |
| 832 | |
| 833 | // NOTE: This is slightly misleading since there are fixed costs |
| 834 | // for generating output like the <head> section and should be factored in, |
| 835 | // but this is good enough for now as a useful first degree of approximation. |
| 836 | $timePerKB = $parseTime * 1024 / $outSize; |
| 837 | if ( $timePerKB > 500 ) { |
| 838 | // At 100ms/KB, even a 100KB page which isn't that large will take 10s. |
| 839 | // So, we probably want to shoot for a threshold under 100ms. |
| 840 | // But, let's start with 500ms+ outliers first and see what we uncover. |
| 841 | LoggerFactory::getInstance( 'slow-parsoid' ) |
| 842 | ->info( 'Parsing {title} was slow, timePerKB took {timePerKB} ms, total: {time} seconds', [ |
| 843 | 'time' => number_format( $parseTime / 1000, 2 ), |
| 844 | 'timePerKB' => number_format( $timePerKB, 1 ), |
| 845 | 'title' => Title::newFromLinkTarget( $pageConfig->getLinkTarget() )->getPrefixedText(), |
| 846 | ] ); |
| 847 | } |
| 848 | } |
| 849 | } |
| 850 | |
| 851 | if ( $wikitext !== null ) { |
| 852 | // Don't cache requests when wt is set in case somebody uses |
| 853 | // GET for wikitext parsing |
| 854 | // XXX: can we just refuse to do wikitext parsing in a GET request? |
| 855 | $response->setHeader( ResponseHeaders::CACHE_CONTROL, 'private,no-cache,s-maxage=0' ); |
| 856 | } elseif ( $oldid !== null ) { |
| 857 | // XXX: can this go away? Parsoid's PageContent class doesn't expose suppressed revision content. |
| 858 | if ( $request->getHeaderLine( 'Cookie' ) || |
| 859 | $request->getHeaderLine( 'Authorization' ) ) { |
| 860 | // Don't cache requests with a session. |
| 861 | $response->setHeader( ResponseHeaders::CACHE_CONTROL, 'private,no-cache,s-maxage=0' ); |
| 862 | } |
| 863 | } |
| 864 | return $response; |
| 865 | } |
| 866 | |
| 867 | protected function newParsoid(): Parsoid { |
| 868 | return new Parsoid( $this->siteConfig, $this->dataAccess ); |
| 869 | } |
| 870 | |
| 871 | protected function parseHTML( string $html, bool $validateXMLNames = false ): Document { |
| 872 | return DOMUtils::parseHTML( $html, $validateXMLNames ); |
| 873 | } |
| 874 | |
| 875 | /** |
| 876 | * @param PageConfig|PageIdentity $page |
| 877 | * @param array $attribs Attributes gotten from requests |
| 878 | * @param string $html Original HTML |
| 879 | * |
| 880 | * @return Response |
| 881 | * @throws HttpException |
| 882 | */ |
| 883 | protected function html2wt( |
| 884 | $page, array $attribs, string $html |
| 885 | ) { |
| 886 | if ( $page instanceof PageConfig ) { |
| 887 | // TODO: Deprecate passing a PageConfig. |
| 888 | // Ideally, callers would use HtmlToContentTransform directly. |
| 889 | $page = Title::newFromLinkTarget( $page->getLinkTarget() ); |
| 890 | } |
| 891 | |
| 892 | try { |
| 893 | $transform = $this->getHtmlInputTransformHelper( $attribs, $html, $page ); |
| 894 | |
| 895 | $response = $this->getResponseFactory()->create(); |
| 896 | $transform->putContent( $response ); |
| 897 | |
| 898 | return $response; |
| 899 | } catch ( ClientError $e ) { |
| 900 | throw new LocalizedHttpException( new MessageValue( "rest-parsoid-error", [ $e->getMessage() ] ), 400 ); |
| 901 | } |
| 902 | } |
| 903 | |
| 904 | /** |
| 905 | * Pagebundle -> pagebundle helper. |
| 906 | * |
| 907 | * @param array<string,array|string> $attribs |
| 908 | * @return Response |
| 909 | * @throws HttpException |
| 910 | */ |
| 911 | protected function pb2pb( array $attribs ) { |
| 912 | $opts = $attribs['opts']; |
| 913 | |
| 914 | $revision = $opts['previous'] ?? $opts['original'] ?? null; |
| 915 | if ( !isset( $revision['html'] ) ) { |
| 916 | throw new LocalizedHttpException( new MessageValue( "rest-missing-revision-html" ), 400 ); |
| 917 | } |
| 918 | |
| 919 | $vOriginal = ParsoidFormatHelper::parseContentTypeHeader( |
| 920 | $revision['html']['headers']['content-type'] ?? '' ); |
| 921 | if ( $vOriginal === null ) { |
| 922 | throw new LocalizedHttpException( new MessageValue( "rest-missing-revision-html-content-type" ), 400 ); |
| 923 | } |
| 924 | $attribs['envOptions']['inputContentVersion'] = $vOriginal; |
| 925 | '@phan-var array<string,array|string> $attribs'; // @var array<string,array|string> $attribs |
| 926 | |
| 927 | $this->metrics->increment( |
| 928 | 'pb2pb.original.version.' . $attribs['envOptions']['inputContentVersion'] |
| 929 | ); |
| 930 | |
| 931 | if ( !empty( $opts['updates'] ) ) { |
| 932 | // FIXME: Handling missing revisions uniformly for all update types |
| 933 | // is not probably the right thing to do but probably okay for now. |
| 934 | // This might need revisiting as we add newer types. |
| 935 | $pageConfig = $this->tryToCreatePageConfig( $attribs, null, true ); |
| 936 | // If we're only updating parts of the original version, it should |
| 937 | // satisfy the requested content version, since we'll be returning |
| 938 | // that same one. |
| 939 | // FIXME: Since this endpoint applies the acceptable middleware, |
| 940 | // `getOutputContentVersion` is not what's been passed in, but what |
| 941 | // can be produced. Maybe that should be selectively applied so |
| 942 | // that we can update older versions where it makes sense? |
| 943 | // Uncommenting below implies that we can only update the latest |
| 944 | // version, since carrot semantics is applied in both directions. |
| 945 | // if ( !Semver::satisfies( |
| 946 | // $attribs['envOptions']['inputContentVersion'], |
| 947 | // "^{$attribs['envOptions']['outputContentVersion']}" |
| 948 | // ) ) { |
| 949 | // throw new HttpException( |
| 950 | // 'We do not know how to do this conversion.', 415 |
| 951 | // ); |
| 952 | // } |
| 953 | if ( !empty( $opts['updates']['redlinks'] ) ) { |
| 954 | // Q(arlolra): Should redlinks be more complex than a bool? |
| 955 | // See gwicke's proposal at T114413#2240381 |
| 956 | return $this->updateRedLinks( $pageConfig, $attribs, $revision ); |
| 957 | } elseif ( isset( $opts['updates']['variant'] ) ) { |
| 958 | return $this->languageConversion( $pageConfig, $attribs, $revision ); |
| 959 | } else { |
| 960 | throw new LocalizedHttpException( new MessageValue( "rest-unknown-parsoid-transformation" ), 400 ); |
| 961 | } |
| 962 | } |
| 963 | |
| 964 | // TODO(arlolra): subbu has some sage advice in T114413#2365456 that |
| 965 | // we should probably be more explicit about the pb2pb conversion |
| 966 | // requested rather than this increasingly complex fallback logic. |
| 967 | $downgrade = Parsoid::findDowngrade( |
| 968 | $attribs['envOptions']['inputContentVersion'], |
| 969 | $attribs['envOptions']['outputContentVersion'] |
| 970 | ); |
| 971 | if ( $downgrade ) { |
| 972 | $pb = HtmlPageBundle::newFromJsonArray( [ |
| 973 | 'html' => $revision['html']['body'], |
| 974 | 'parsoid' => $revision['data-parsoid']['body'] ?? null, |
| 975 | 'mw' => $revision['data-mw']['body'] ?? null, |
| 976 | 'counters' => $revision['counters']['body'] ?? null, |
| 977 | ] ); |
| 978 | $this->validatePb( $pb, $attribs['envOptions']['inputContentVersion'] ); |
| 979 | Parsoid::downgrade( $downgrade, $pb, $this->siteConfig ); |
| 980 | |
| 981 | if ( !empty( $attribs['body_only'] ) ) { |
| 982 | $doc = $this->parseHTML( $pb->html ); |
| 983 | $body = DOMCompat::getBody( $doc ); |
| 984 | $pb->html = ContentUtils::toXML( $body, [ 'innerXML' => true ] ); |
| 985 | } |
| 986 | |
| 987 | $response = $this->getResponseFactory()->createJson( $pb->responseData() ); |
| 988 | ParsoidFormatHelper::setContentType( |
| 989 | $response, ParsoidFormatHelper::FORMAT_PAGEBUNDLE, $pb->version |
| 990 | ); |
| 991 | return $response; |
| 992 | // Ensure we only reuse from semantically similar content versions. |
| 993 | } elseif ( Semver::satisfies( $attribs['envOptions']['outputContentVersion'], |
| 994 | '^' . $attribs['envOptions']['inputContentVersion'] ) ) { |
| 995 | $pageConfig = $this->tryToCreatePageConfig( $attribs ); |
| 996 | return $this->wt2html( $pageConfig, $attribs ); |
| 997 | } else { |
| 998 | throw new LocalizedHttpException( new MessageValue( "rest-unsupported-profile-conversion" ), 415 ); |
| 999 | } |
| 1000 | } |
| 1001 | |
| 1002 | /** |
| 1003 | * Update red links on a document. |
| 1004 | * |
| 1005 | * @param PageConfig $pageConfig |
| 1006 | * @param array $attribs |
| 1007 | * @param array $revision |
| 1008 | * @return Response |
| 1009 | */ |
| 1010 | protected function updateRedLinks( |
| 1011 | PageConfig $pageConfig, array $attribs, array $revision |
| 1012 | ) { |
| 1013 | $parsoid = $this->newParsoid(); |
| 1014 | |
| 1015 | $pb = HtmlPageBundle::newFromJsonArray( [ |
| 1016 | 'html' => $revision['html']['body'], |
| 1017 | 'parsoid' => $revision['data-parsoid']['body'] ?? null, |
| 1018 | 'mw' => $revision['data-mw']['body'] ?? null, |
| 1019 | 'counters' => $revision['counters']['body'] ?? null, |
| 1020 | 'version' => $attribs['envOptions']['inputContentVersion'], |
| 1021 | 'headers' => $revision['html']['headers'] ?? null, |
| 1022 | 'contentmodel' => $revision['contentmodel'] ?? null, |
| 1023 | ] ); |
| 1024 | |
| 1025 | $out = $parsoid->pb2pb( $pageConfig, 'redlinks', $pb, [] ); |
| 1026 | |
| 1027 | $this->validatePb( $out, $attribs['envOptions']['inputContentVersion'] ); |
| 1028 | |
| 1029 | $response = $this->getResponseFactory()->createJson( $out->responseData() ); |
| 1030 | ParsoidFormatHelper::setContentType( |
| 1031 | $response, ParsoidFormatHelper::FORMAT_PAGEBUNDLE, $out->version |
| 1032 | ); |
| 1033 | return $response; |
| 1034 | } |
| 1035 | |
| 1036 | /** |
| 1037 | * Do variant conversion on a document. |
| 1038 | * |
| 1039 | * @param PageConfig $pageConfig |
| 1040 | * @param array $attribs |
| 1041 | * @param array $revision |
| 1042 | * @return Response |
| 1043 | * @throws HttpException |
| 1044 | */ |
| 1045 | protected function languageConversion( |
| 1046 | PageConfig $pageConfig, array $attribs, array $revision |
| 1047 | ) { |
| 1048 | $opts = $attribs['opts']; |
| 1049 | $target = $opts['updates']['variant']['html'] ?? |
| 1050 | $opts['updates']['variant']['target'] ?? |
| 1051 | $attribs['envOptions']['htmlVariantLanguage']; |
| 1052 | $source = $opts['updates']['variant']['wikitext'] ?? |
| 1053 | $opts['updates']['variant']['source'] ?? null; |
| 1054 | |
| 1055 | if ( !$target ) { |
| 1056 | throw new LocalizedHttpException( new MessageValue( "rest-target-variant-required" ), 400 ); |
| 1057 | } |
| 1058 | |
| 1059 | $pageIdentity = $this->tryToCreatePageIdentity( $attribs ); |
| 1060 | |
| 1061 | $pb = HtmlPageBundle::newFromJsonArray( [ |
| 1062 | 'html' => $revision['html']['body'], |
| 1063 | 'parsoid' => $revision['data-parsoid']['body'] ?? null, |
| 1064 | 'mw' => $revision['data-mw']['body'] ?? null, |
| 1065 | 'counters' => $revision['counters']['body'] ?? null, |
| 1066 | 'version' => $attribs['envOptions']['inputContentVersion'], |
| 1067 | 'headers' => $revision['html']['headers'] ?? null, |
| 1068 | 'contentmodel' => $revision['contentmodel'] ?? null, |
| 1069 | ] ); |
| 1070 | |
| 1071 | // XXX: DI should inject HtmlTransformFactory |
| 1072 | $languageVariantConverter = MediaWikiServices::getInstance() |
| 1073 | ->getHtmlTransformFactory() |
| 1074 | ->getLanguageVariantConverter( $pageIdentity ); |
| 1075 | $httpContentLanguage = $attribs['pagelanguage' ] ?? null; |
| 1076 | if ( $httpContentLanguage ) { |
| 1077 | $languageVariantConverter->setPageLanguageOverride( $httpContentLanguage ); |
| 1078 | } |
| 1079 | // Convert PageBundle to ParserOutput |
| 1080 | $parserOutput = PageBundleParserOutputConverter::parserOutputFromPageBundle( |
| 1081 | $pb, |
| 1082 | title: $pageIdentity, |
| 1083 | siteConfig: $this->siteConfig |
| 1084 | ); |
| 1085 | try { |
| 1086 | $parserOutput = $languageVariantConverter->convertParserOutputVariant( |
| 1087 | $parserOutput, $target, $source, |
| 1088 | ); |
| 1089 | } catch ( InvalidArgumentException $e ) { |
| 1090 | throw new LocalizedHttpException( |
| 1091 | new MessageValue( "rest-unsupported-language-conversion", [ $source ?? '(unspecified)', $target ] ), |
| 1092 | 400, |
| 1093 | [ 'reason' => $e->getMessage() ] |
| 1094 | ); |
| 1095 | } |
| 1096 | $out = PageBundleParserOutputConverter::htmlPageBundleFromParserOutput( |
| 1097 | $parserOutput, siteConfig: $this->siteConfig, bodyOnly: false, |
| 1098 | ); |
| 1099 | $out->headers['vary'] ??= 'Accept-Language'; |
| 1100 | |
| 1101 | $response = $this->getResponseFactory()->createJson( $out->responseData() ); |
| 1102 | $response->addHeader( 'Vary', 'Accept-Language' ); |
| 1103 | ParsoidFormatHelper::setContentType( |
| 1104 | $response, ParsoidFormatHelper::FORMAT_PAGEBUNDLE, $out->version |
| 1105 | ); |
| 1106 | return $response; |
| 1107 | } |
| 1108 | |
| 1109 | /** @inheritDoc */ |
| 1110 | abstract public function execute(): Response; |
| 1111 | |
| 1112 | /** |
| 1113 | * Validate a HtmlPageBundle against the given contentVersion, and throw |
| 1114 | * an HttpException if it does not match. |
| 1115 | * @param HtmlPageBundle $pb |
| 1116 | * @param string $contentVersion |
| 1117 | * @throws HttpException |
| 1118 | */ |
| 1119 | private function validatePb( HtmlPageBundle $pb, string $contentVersion ): void { |
| 1120 | $errorMessage = ''; |
| 1121 | if ( !$pb->validate( $contentVersion, $errorMessage ) ) { |
| 1122 | throw new LocalizedHttpException( |
| 1123 | new MessageValue( "rest-page-bundle-validation-error", [ $errorMessage ] ), |
| 1124 | 400 |
| 1125 | ); |
| 1126 | } |
| 1127 | } |
| 1128 | |
| 1129 | /** |
| 1130 | * @param PageConfig $page |
| 1131 | * |
| 1132 | * @return ProperPageIdentity |
| 1133 | * @throws HttpException |
| 1134 | */ |
| 1135 | private function pageConfigToPageIdentity( PageConfig $page ): ProperPageIdentity { |
| 1136 | $services = MediaWikiServices::getInstance(); |
| 1137 | |
| 1138 | $title = $page->getLinkTarget(); |
| 1139 | try { |
| 1140 | $page = $services->getPageStore()->getPageForLink( $title ); |
| 1141 | } catch ( MalformedTitleException | InvalidArgumentException ) { |
| 1142 | // Note that even some well-formed links are still invalid |
| 1143 | // parameters for getPageForLink(), e.g. interwiki links or special pages. |
| 1144 | throw new HttpException( |
| 1145 | "Bad title: $title", # uses LinkTarget::__toString() |
| 1146 | 400 |
| 1147 | ); |
| 1148 | } |
| 1149 | |
| 1150 | return $page; |
| 1151 | } |
| 1152 | } |