Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
65.99% |
229 / 347 |
|
15.38% |
2 / 13 |
CRAP | |
0.00% |
0 / 1 |
| HtmlInputTransformHelper | |
65.99% |
229 / 347 |
|
15.38% |
2 / 13 |
416.64 | |
0.00% |
0 / 1 |
| __construct | |
92.31% |
12 / 13 |
|
0.00% |
0 / 1 |
2.00 | |||
| getParamSettings | |
0.00% |
0 / 44 |
|
0.00% |
0 / 1 |
2 | |||
| normalizeParameters | |
72.73% |
16 / 22 |
|
0.00% |
0 / 1 |
21.19 | |||
| init | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| initInternal | |
77.78% |
49 / 63 |
|
0.00% |
0 / 1 |
24.39 | |||
| getTransform | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setMetrics | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| setOriginal | |
78.38% |
58 / 74 |
|
0.00% |
0 / 1 |
28.35 | |||
| getContent | |
72.22% |
13 / 18 |
|
0.00% |
0 / 1 |
4.34 | |||
| putContent | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
2 | |||
| fetchParserOutputFromParsoid | |
73.68% |
28 / 38 |
|
0.00% |
0 / 1 |
13.21 | |||
| fetchSelserContextFromStash | |
89.36% |
42 / 47 |
|
0.00% |
0 / 1 |
5.03 | |||
| throwHttpExceptionForStatus | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * This program is free software; you can redistribute it and/or modify |
| 4 | * it under the terms of the GNU General Public License as published by |
| 5 | * the Free Software Foundation; either version 2 of the License, or |
| 6 | * (at your option) any later version. |
| 7 | * |
| 8 | * This program is distributed in the hope that it will be useful, |
| 9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 11 | * GNU General Public License for more details. |
| 12 | * |
| 13 | * You should have received a copy of the GNU General Public License along |
| 14 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 15 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 16 | * http://www.gnu.org/copyleft/gpl.html |
| 17 | * |
| 18 | * @file |
| 19 | */ |
| 20 | namespace MediaWiki\Rest\Handler\Helper; |
| 21 | |
| 22 | use InvalidArgumentException; |
| 23 | use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface; |
| 24 | use MediaWiki\Content\Content; |
| 25 | use MediaWiki\Edit\ParsoidOutputStash; |
| 26 | use MediaWiki\Edit\ParsoidRenderID; |
| 27 | use MediaWiki\Edit\SelserContext; |
| 28 | use MediaWiki\Language\LanguageCode; |
| 29 | use MediaWiki\MainConfigNames; |
| 30 | use MediaWiki\Page\PageIdentity; |
| 31 | use MediaWiki\Page\PageLookup; |
| 32 | use MediaWiki\Page\PageRecord; |
| 33 | use MediaWiki\Page\ParserOutputAccess; |
| 34 | use MediaWiki\Parser\ParserOptions; |
| 35 | use MediaWiki\Parser\ParserOutput; |
| 36 | use MediaWiki\Parser\Parsoid\HtmlToContentTransform; |
| 37 | use MediaWiki\Parser\Parsoid\HtmlTransformFactory; |
| 38 | use MediaWiki\Parser\Parsoid\PageBundleParserOutputConverter; |
| 39 | use MediaWiki\Rest\Handler; |
| 40 | use MediaWiki\Rest\HttpException; |
| 41 | use MediaWiki\Rest\LocalizedHttpException; |
| 42 | use MediaWiki\Rest\ResponseInterface; |
| 43 | use MediaWiki\Revision\RevisionAccessException; |
| 44 | use MediaWiki\Revision\RevisionLookup; |
| 45 | use MediaWiki\Revision\RevisionRecord; |
| 46 | use MediaWiki\Status\Status; |
| 47 | use MWUnknownContentModelException; |
| 48 | use Wikimedia\Bcp47Code\Bcp47Code; |
| 49 | use Wikimedia\Message\MessageValue; |
| 50 | use Wikimedia\ParamValidator\ParamValidator; |
| 51 | use Wikimedia\Parsoid\Core\ClientError; |
| 52 | use Wikimedia\Parsoid\Core\PageBundle; |
| 53 | use Wikimedia\Parsoid\Core\ResourceLimitExceededException; |
| 54 | use Wikimedia\Parsoid\Parsoid; |
| 55 | use Wikimedia\Stats\StatsFactory; |
| 56 | |
| 57 | /** |
| 58 | * REST helper for converting HTML to page content source (e.g. wikitext). |
| 59 | * |
| 60 | * @since 1.40 |
| 61 | * |
| 62 | * @unstable Pending consolidation of the Parsoid extension with core code. |
| 63 | */ |
| 64 | class HtmlInputTransformHelper { |
| 65 | /** |
| 66 | * @internal |
| 67 | */ |
| 68 | public const CONSTRUCTOR_OPTIONS = [ |
| 69 | MainConfigNames::ParsoidCacheConfig |
| 70 | ]; |
| 71 | |
| 72 | /** @var PageIdentity|null */ |
| 73 | private $page = null; |
| 74 | |
| 75 | /** |
| 76 | * @var HtmlToContentTransform |
| 77 | */ |
| 78 | private $transform; |
| 79 | |
| 80 | /** |
| 81 | * @var array |
| 82 | */ |
| 83 | private $envOptions; |
| 84 | |
| 85 | private StatsFactory $statsFactory; |
| 86 | private HtmlTransformFactory $htmlTransformFactory; |
| 87 | private ParsoidOutputStash $parsoidOutputStash; |
| 88 | private ParserOutputAccess $parserOutputAccess; |
| 89 | private PageLookup $pageLookup; |
| 90 | private RevisionLookup $revisionLookup; |
| 91 | |
| 92 | /** |
| 93 | * @param StatsFactory $statsFactory |
| 94 | * @param HtmlTransformFactory $htmlTransformFactory |
| 95 | * @param ParsoidOutputStash $parsoidOutputStash |
| 96 | * @param ParserOutputAccess $parserOutputAccess |
| 97 | * @param PageLookup $pageLookup |
| 98 | * @param RevisionLookup $revisionLookup |
| 99 | * @param array $envOptions |
| 100 | * @param ?PageIdentity $page |
| 101 | * @param array|string $body Body structure, or an HTML string |
| 102 | * @param array $parameters |
| 103 | * @param RevisionRecord|null $originalRevision |
| 104 | * @param Bcp47Code|null $pageLanguage |
| 105 | */ |
| 106 | public function __construct( |
| 107 | StatsFactory $statsFactory, |
| 108 | HtmlTransformFactory $htmlTransformFactory, |
| 109 | ParsoidOutputStash $parsoidOutputStash, |
| 110 | ParserOutputAccess $parserOutputAccess, |
| 111 | PageLookup $pageLookup, |
| 112 | RevisionLookup $revisionLookup, |
| 113 | array $envOptions = [], |
| 114 | ?PageIdentity $page = null, |
| 115 | $body = '', |
| 116 | array $parameters = [], |
| 117 | ?RevisionRecord $originalRevision = null, |
| 118 | ?Bcp47Code $pageLanguage = null |
| 119 | ) { |
| 120 | $this->statsFactory = $statsFactory; |
| 121 | $this->htmlTransformFactory = $htmlTransformFactory; |
| 122 | $this->parsoidOutputStash = $parsoidOutputStash; |
| 123 | $this->envOptions = $envOptions + [ |
| 124 | 'outputContentVersion' => Parsoid::defaultHTMLVersion(), |
| 125 | 'offsetType' => 'byte', |
| 126 | ]; |
| 127 | $this->parserOutputAccess = $parserOutputAccess; |
| 128 | $this->pageLookup = $pageLookup; |
| 129 | $this->revisionLookup = $revisionLookup; |
| 130 | if ( $page === null ) { |
| 131 | wfDeprecated( __METHOD__ . ' without $page', '1.43' ); |
| 132 | } else { |
| 133 | $this->initInternal( $page, $body, $parameters, $originalRevision, $pageLanguage ); |
| 134 | } |
| 135 | } |
| 136 | |
| 137 | /** |
| 138 | * @return array |
| 139 | */ |
| 140 | public function getParamSettings(): array { |
| 141 | // JSON body schema: |
| 142 | /* |
| 143 | doc: |
| 144 | properties: |
| 145 | headers: |
| 146 | type: array |
| 147 | items: |
| 148 | type: string |
| 149 | body: |
| 150 | type: [ string, object ] |
| 151 | required: [ body ] |
| 152 | |
| 153 | body: |
| 154 | properties: |
| 155 | offsetType: |
| 156 | type: string |
| 157 | revid: |
| 158 | type: integer |
| 159 | renderid: |
| 160 | type: string |
| 161 | etag: |
| 162 | type: string |
| 163 | html: |
| 164 | type: [ doc, string ] |
| 165 | data-mw: |
| 166 | type: doc |
| 167 | original: |
| 168 | properties: |
| 169 | html: |
| 170 | type: doc |
| 171 | source: |
| 172 | type: doc |
| 173 | data-mw: |
| 174 | type: doc |
| 175 | data-parsoid: |
| 176 | type: doc |
| 177 | required: [ html ] |
| 178 | */ |
| 179 | |
| 180 | // FUTURE: more params |
| 181 | // - slot (for loading the base content) |
| 182 | |
| 183 | return [ |
| 184 | // XXX: should we really declare this here? Or should end endpoint do this? |
| 185 | // We are not reading this property... |
| 186 | 'title' => [ |
| 187 | Handler::PARAM_SOURCE => 'path', |
| 188 | ParamValidator::PARAM_TYPE => 'string', |
| 189 | ParamValidator::PARAM_DEFAULT => '', |
| 190 | ParamValidator::PARAM_REQUIRED => false, |
| 191 | Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-html-input-title' ) |
| 192 | ], |
| 193 | // XXX: Needed for compatibility with the parsoid transform endpoint. |
| 194 | // But revid should just be part of the info about the original data |
| 195 | // in the body. |
| 196 | 'oldid' => [ |
| 197 | Handler::PARAM_SOURCE => 'path', |
| 198 | ParamValidator::PARAM_TYPE => 'int', |
| 199 | ParamValidator::PARAM_DEFAULT => 0, |
| 200 | ParamValidator::PARAM_REQUIRED => false, |
| 201 | Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-html-input-oldid' ) |
| 202 | ], |
| 203 | // XXX: Supported for compatibility with the parsoid transform endpoint. |
| 204 | // If given, it should be 'html' or 'pagebundle'. |
| 205 | 'from' => [ |
| 206 | Handler::PARAM_SOURCE => 'path', |
| 207 | ParamValidator::PARAM_TYPE => 'string', |
| 208 | ParamValidator::PARAM_DEFAULT => '', |
| 209 | ParamValidator::PARAM_REQUIRED => false, |
| 210 | Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-html-input-from' ) |
| 211 | ], |
| 212 | // XXX: Supported for compatibility with the parsoid transform endpoint. |
| 213 | // Ignored. |
| 214 | 'format' => [ |
| 215 | Handler::PARAM_SOURCE => 'path', |
| 216 | ParamValidator::PARAM_TYPE => 'string', |
| 217 | ParamValidator::PARAM_DEFAULT => '', |
| 218 | ParamValidator::PARAM_REQUIRED => false, |
| 219 | Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-html-input-format' ) |
| 220 | ], |
| 221 | 'contentmodel' => [ // XXX: get this from the Accept header? |
| 222 | Handler::PARAM_SOURCE => 'query', |
| 223 | ParamValidator::PARAM_TYPE => 'string', |
| 224 | ParamValidator::PARAM_DEFAULT => '', |
| 225 | ParamValidator::PARAM_REQUIRED => false, |
| 226 | Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-html-input-contentmodel' ) |
| 227 | ], |
| 228 | 'language' => [ // TODO: get this from Accept-Language header?! |
| 229 | Handler::PARAM_SOURCE => 'query', |
| 230 | ParamValidator::PARAM_TYPE => 'string', |
| 231 | ParamValidator::PARAM_DEFAULT => '', |
| 232 | ParamValidator::PARAM_REQUIRED => false, |
| 233 | Handler::PARAM_DESCRIPTION => new MessageValue( 'rest-param-desc-html-input-language' ) |
| 234 | ] |
| 235 | ]; |
| 236 | } |
| 237 | |
| 238 | /** |
| 239 | * Modify body and parameters to provide compatibility with legacy endpoints. |
| 240 | * |
| 241 | * @see ParsoidHandler::getRequestAttributes |
| 242 | * |
| 243 | * @param array<string,mixed> &$body |
| 244 | * @param array<string,mixed> &$parameters |
| 245 | * |
| 246 | * @throws HttpException |
| 247 | * |
| 248 | * @return void |
| 249 | */ |
| 250 | private static function normalizeParameters( array &$body, array &$parameters ) { |
| 251 | // If the revision ID is given in the path, pretend it was given in the body. |
| 252 | if ( isset( $parameters['oldid'] ) && (int)$parameters['oldid'] > 0 ) { |
| 253 | $body['original']['revid'] = (int)$parameters['oldid']; |
| 254 | } |
| 255 | |
| 256 | // If an etag is given in the body, use it as the render ID. |
| 257 | // Note that we support ETag format in the renderid field. |
| 258 | if ( !empty( $body['original']['etag'] ) ) { |
| 259 | // @phan-suppress-next-line PhanTypeInvalidDimOffset false positive |
| 260 | $body['original']['renderid'] = $body['original']['etag']; |
| 261 | } |
| 262 | |
| 263 | // Accept 'wikitext' as an alias for 'source'. |
| 264 | if ( isset( $body['original']['wikitext'] ) ) { |
| 265 | // @phan-suppress-next-line PhanTypeInvalidDimOffset false positive |
| 266 | $body['original']['source'] = $body['original']['wikitext']; |
| 267 | unset( $body['original']['wikitext'] ); |
| 268 | } |
| 269 | |
| 270 | // If 'from' is not set, we accept page bundle style input as well as full HTML. |
| 271 | // If 'from' is set, we only accept page bundle style input if it is set to FORMAT_PAGEBUNDLE. |
| 272 | if ( |
| 273 | isset( $parameters['from'] ) && $parameters['from'] !== '' && |
| 274 | $parameters['from'] !== ParsoidFormatHelper::FORMAT_PAGEBUNDLE |
| 275 | ) { |
| 276 | unset( $body['original']['data-parsoid']['body'] ); |
| 277 | unset( $body['original']['data-mw']['body'] ); |
| 278 | unset( $body['data-mw']['body'] ); |
| 279 | } |
| 280 | |
| 281 | // If 'from' is given, it must be html or pagebundle. |
| 282 | if ( |
| 283 | isset( $parameters['from'] ) && $parameters['from'] !== '' && |
| 284 | $parameters['from'] !== ParsoidFormatHelper::FORMAT_HTML && |
| 285 | $parameters['from'] !== ParsoidFormatHelper::FORMAT_PAGEBUNDLE |
| 286 | ) { |
| 287 | throw new LocalizedHttpException( |
| 288 | new MessageValue( "rest-unsupported-transform-input", [ $parameters['from'] ] ), 400 |
| 289 | ); |
| 290 | } |
| 291 | |
| 292 | if ( isset( $body['contentmodel'] ) && $body['contentmodel'] !== '' ) { |
| 293 | $parameters['contentmodel'] = $body['contentmodel']; |
| 294 | } elseif ( isset( $parameters['format'] ) && $parameters['format'] !== '' ) { |
| 295 | $parameters['contentmodel'] = $parameters['format']; |
| 296 | } |
| 297 | } |
| 298 | |
| 299 | /** |
| 300 | * @param PageIdentity $page |
| 301 | * @param array|string $body Body structure, or an HTML string |
| 302 | * @param array $parameters |
| 303 | * @param RevisionRecord|null $originalRevision |
| 304 | * @param Bcp47Code|null $pageLanguage |
| 305 | * |
| 306 | * @throws HttpException |
| 307 | * @deprecated since 1.43; pass arguments to constructor instead |
| 308 | */ |
| 309 | public function init( |
| 310 | PageIdentity $page, |
| 311 | $body, |
| 312 | array $parameters, |
| 313 | ?RevisionRecord $originalRevision = null, |
| 314 | ?Bcp47Code $pageLanguage = null |
| 315 | ) { |
| 316 | wfDeprecated( __METHOD__, '1.43' ); |
| 317 | $this->initInternal( $page, $body, $parameters, $originalRevision, $pageLanguage ); |
| 318 | } |
| 319 | |
| 320 | /** |
| 321 | * @param PageIdentity $page |
| 322 | * @param array|string $body Body structure, or an HTML string |
| 323 | * @param array $parameters |
| 324 | * @param RevisionRecord|null $originalRevision |
| 325 | * @param Bcp47Code|null $pageLanguage |
| 326 | * |
| 327 | * @throws HttpException |
| 328 | */ |
| 329 | private function initInternal( |
| 330 | PageIdentity $page, |
| 331 | $body, |
| 332 | array $parameters, |
| 333 | ?RevisionRecord $originalRevision = null, |
| 334 | ?Bcp47Code $pageLanguage = null |
| 335 | ) { |
| 336 | if ( is_string( $body ) ) { |
| 337 | $body = [ 'html' => $body ]; |
| 338 | } |
| 339 | |
| 340 | self::normalizeParameters( $body, $parameters ); |
| 341 | |
| 342 | $this->page = $page; |
| 343 | |
| 344 | if ( !isset( $body['html'] ) ) { |
| 345 | throw new LocalizedHttpException( new MessageValue( "rest-missing-body-field", [ 'html' ] ) ); |
| 346 | } |
| 347 | |
| 348 | $html = is_array( $body['html'] ) ? $body['html']['body'] : $body['html']; |
| 349 | |
| 350 | // TODO: validate $body against a proper schema. |
| 351 | $this->transform = $this->htmlTransformFactory->getHtmlToContentTransform( |
| 352 | $html, |
| 353 | $this->page |
| 354 | ); |
| 355 | |
| 356 | $this->transform->setMetrics( $this->statsFactory ); |
| 357 | |
| 358 | // NOTE: Env::getContentModel will fall back to the page's recorded content model |
| 359 | // if none is set here. |
| 360 | $this->transform->setOptions( [ |
| 361 | 'contentmodel' => $parameters['contentmodel'] ?? null, |
| 362 | 'offsetType' => $body['offsetType'] ?? $this->envOptions['offsetType'], |
| 363 | ] ); |
| 364 | |
| 365 | $original = $body['original'] ?? []; |
| 366 | $originalRendering = null; |
| 367 | |
| 368 | if ( !isset( $original['html'] ) && !empty( $original['renderid'] ) ) { |
| 369 | $key = $original['renderid']; |
| 370 | if ( preg_match( '!^(W/)?".*"$!', $key ) ) { |
| 371 | $originalRendering = ParsoidRenderID::newFromETag( $key ); |
| 372 | |
| 373 | if ( !$originalRendering ) { |
| 374 | throw new LocalizedHttpException( new MessageValue( "rest-bad-etag", [ $key ] ), 400 ); |
| 375 | } |
| 376 | } else { |
| 377 | $originalRendering = ParsoidRenderID::newFromKey( $key ); |
| 378 | } |
| 379 | } elseif ( !empty( $original['html'] ) || !empty( $original['data-parsoid'] ) ) { |
| 380 | // NOTE: We might have an incomplete PageBundle here, with no HTML but with data-parsoid! |
| 381 | // XXX: Do we need to support that, or can that just be a 400? |
| 382 | $originalRendering = new PageBundle( |
| 383 | $original['html']['body'] ?? '', |
| 384 | $original['data-parsoid']['body'] ?? null, |
| 385 | $original['data-mw']['body'] ?? null, |
| 386 | null, // will be derived from $original['html']['headers']['content-type'] |
| 387 | $original['html']['headers'] ?? [] |
| 388 | ); |
| 389 | } |
| 390 | |
| 391 | if ( !$originalRevision && !empty( $original['revid'] ) ) { |
| 392 | $originalRevision = (int)$original['revid']; |
| 393 | } |
| 394 | |
| 395 | if ( $originalRevision || $originalRendering ) { |
| 396 | $this->setOriginal( $originalRevision, $originalRendering ); |
| 397 | } else { |
| 398 | if ( $this->page->exists() ) { |
| 399 | $this->statsFactory |
| 400 | ->getCounter( 'html_input_transform_total' ) |
| 401 | ->setLabel( 'original_html_given', 'false' ) |
| 402 | ->setLabel( 'page_exists', 'true' ) |
| 403 | ->setLabel( 'status', 'unknown' ) |
| 404 | ->copyToStatsdAt( 'html_input_transform.original_html.not_given.page_exists' ) |
| 405 | ->increment(); |
| 406 | } else { |
| 407 | $this->statsFactory |
| 408 | ->getCounter( 'html_input_transform_total' ) |
| 409 | ->setLabel( 'original_html_given', 'false' ) |
| 410 | ->setLabel( 'page_exists', 'false' ) |
| 411 | ->setLabel( 'status', 'unknown' ) |
| 412 | ->copyToStatsdAt( 'html_input_transform.original_html.not_given.page_not_exist' ) |
| 413 | ->increment(); |
| 414 | } |
| 415 | } |
| 416 | |
| 417 | if ( isset( $body['data-mw']['body'] ) ) { |
| 418 | $this->transform->setModifiedDataMW( $body['data-mw']['body'] ); |
| 419 | } |
| 420 | |
| 421 | if ( $pageLanguage ) { |
| 422 | $this->transform->setContentLanguage( $pageLanguage ); |
| 423 | } elseif ( isset( $parameters['language'] ) && $parameters['language'] !== '' ) { |
| 424 | $pageLanguage = LanguageCode::normalizeNonstandardCodeAndWarn( |
| 425 | $parameters['language'] |
| 426 | ); |
| 427 | $this->transform->setContentLanguage( $pageLanguage ); |
| 428 | } |
| 429 | |
| 430 | if ( isset( $original['source']['body'] ) ) { |
| 431 | // XXX: do we really have to support wikitext overrides? |
| 432 | $this->transform->setOriginalText( $original['source']['body'] ); |
| 433 | } |
| 434 | } |
| 435 | |
| 436 | /** |
| 437 | * Return HTMLTransform object, so additional context can be provided by calling setters on it. |
| 438 | * @return HtmlToContentTransform |
| 439 | */ |
| 440 | public function getTransform(): HtmlToContentTransform { |
| 441 | return $this->transform; |
| 442 | } |
| 443 | |
| 444 | /** |
| 445 | * Set metrics sink. |
| 446 | * |
| 447 | * @note Passing a StatsdDataFactoryInterface here has been deprecated |
| 448 | * since 1.43. |
| 449 | * |
| 450 | * @param StatsFactory|StatsdDataFactoryInterface $statsFactory |
| 451 | */ |
| 452 | public function setMetrics( $statsFactory ) { |
| 453 | if ( $statsFactory instanceof StatsdDataFactoryInterface ) { |
| 454 | // Uncomment this once all WMF code has been transitioned, but |
| 455 | // leave it in for the 1.43 release. |
| 456 | wfDeprecated( __METHOD__ . ' with StatsdDataFactoryInterface', '1.43' ); |
| 457 | return; |
| 458 | } |
| 459 | $this->statsFactory = $statsFactory; |
| 460 | |
| 461 | if ( $this->transform ) { |
| 462 | $this->transform->setMetrics( $statsFactory ); |
| 463 | } |
| 464 | } |
| 465 | |
| 466 | /** |
| 467 | * Supply information about the revision and rendering that was the original basis of |
| 468 | * the input HTML. This is used to apply selective serialization (selser), if possible. |
| 469 | * |
| 470 | * @param RevisionRecord|int|null $rev |
| 471 | * @param ParsoidRenderID|PageBundle|ParserOutput|null $originalRendering |
| 472 | */ |
| 473 | public function setOriginal( $rev, $originalRendering ) { |
| 474 | if ( $originalRendering instanceof ParsoidRenderID ) { |
| 475 | $renderId = $originalRendering; |
| 476 | |
| 477 | // If the client asked for a render ID, load original data from stash |
| 478 | try { |
| 479 | $selserContext = $this->fetchSelserContextFromStash( $renderId ); |
| 480 | } catch ( InvalidArgumentException $ex ) { |
| 481 | $this->statsFactory |
| 482 | ->getCounter( 'html_input_transform_total' ) |
| 483 | ->setLabel( 'original_html_given', 'as_renderid' ) |
| 484 | ->setLabel( 'page_exists', 'unknown' ) |
| 485 | ->setLabel( 'status', 'bad_renderid' ) |
| 486 | ->copyToStatsdAt( 'html_input_transform.original_html.given.as_renderid.bad' ) |
| 487 | ->increment(); |
| 488 | throw new LocalizedHttpException( new MessageValue( "rest-bad-stash-key" ), |
| 489 | 400, |
| 490 | [ |
| 491 | 'reason' => $ex->getMessage(), |
| 492 | 'key' => "$renderId" |
| 493 | ] |
| 494 | ); |
| 495 | } |
| 496 | |
| 497 | if ( !$selserContext ) { |
| 498 | // NOTE: When the client asked for a specific stash key (resp. etag), |
| 499 | // we should fail with a 412 if we don't have the specific rendering. |
| 500 | // On the other hand, of the client only provided a base revision ID, |
| 501 | // we can re-parse and hope for the best. |
| 502 | |
| 503 | throw new LocalizedHttpException( |
| 504 | new MessageValue( "rest-no-stashed-content", [ $renderId->getKey() ] ), 412 |
| 505 | ); |
| 506 | |
| 507 | // TODO: This class should provide getETag and getLastModified methods for use by |
| 508 | // the REST endpoint, to provide proper support for conditionals. |
| 509 | // However, that requires some refactoring of how HTTP conditional checks |
| 510 | // work in the Handler base class. |
| 511 | } |
| 512 | |
| 513 | if ( !$rev ) { |
| 514 | $rev = $renderId->getRevisionID(); |
| 515 | } |
| 516 | |
| 517 | $originalRendering = $selserContext->getPageBundle(); |
| 518 | $content = $selserContext->getContent(); |
| 519 | |
| 520 | if ( $content ) { |
| 521 | $this->transform->setOriginalContent( $content ); |
| 522 | } |
| 523 | } elseif ( !$originalRendering && $rev ) { |
| 524 | // The client provided a revision ID, but not stash key. |
| 525 | // Try to get a rendering for the given revision, and use it as the basis for selser. |
| 526 | // Chances are good that the resulting diff will be reasonably clean. |
| 527 | // NOTE: If we don't have a revision ID, we should not attempt selser! |
| 528 | $originalRendering = $this->fetchParserOutputFromParsoid( $this->page, $rev, true ); |
| 529 | |
| 530 | if ( $originalRendering ) { |
| 531 | $this->statsFactory->getCounter( 'html_input_transform_total' ) |
| 532 | ->setLabel( 'original_html_given', 'as_revid' ) |
| 533 | ->setLabel( 'page_exists', 'unknown' ) |
| 534 | ->setLabel( 'status', 'found' ) |
| 535 | ->copyToStatsdAt( 'html_input_transform.original_html.given.as_revid.found' ) |
| 536 | ->increment(); |
| 537 | } else { |
| 538 | $this->statsFactory->getCounter( 'html_input_transform_total' ) |
| 539 | ->setLabel( 'original_html_given', 'as_revid' ) |
| 540 | ->setLabel( 'page_exists', 'unknown' ) |
| 541 | ->setLabel( 'status', 'not_found' ) |
| 542 | ->copyToStatsdAt( 'html_input_transform.original_html.given.as_revid.not_found' ) |
| 543 | ->increment(); |
| 544 | } |
| 545 | } elseif ( $originalRendering ) { |
| 546 | $this->statsFactory->getCounter( 'html_input_transform_total' ) |
| 547 | ->setLabel( 'original_html_given', 'true' ) |
| 548 | ->setLabel( 'page_exists', 'unknown' ) |
| 549 | ->setLabel( 'status', 'verbatim' ) |
| 550 | ->copyToStatsdAt( 'html_input_transform.original_html.given.verbatim' ) |
| 551 | ->increment(); |
| 552 | } |
| 553 | |
| 554 | if ( $originalRendering instanceof ParserOutput ) { |
| 555 | $originalRendering = PageBundleParserOutputConverter::pageBundleFromParserOutput( $originalRendering ); |
| 556 | |
| 557 | // NOTE: Use the default if we got a ParserOutput object. |
| 558 | // Don't apply the default if we got passed a PageBundle, |
| 559 | // in that case, we want to require the version to be explicit. |
| 560 | if ( $originalRendering->version === null && !isset( $originalRendering->headers['content-type'] ) ) { |
| 561 | $originalRendering->version = Parsoid::defaultHTMLVersion(); |
| 562 | } |
| 563 | } |
| 564 | |
| 565 | if ( !$originalRendering instanceof PageBundle ) { |
| 566 | return; |
| 567 | } |
| 568 | |
| 569 | if ( $originalRendering->version !== null ) { |
| 570 | $this->transform->setOriginalSchemaVersion( $originalRendering->version ); |
| 571 | } elseif ( !empty( $originalRendering->headers['content-type'] ) ) { |
| 572 | $vOriginal = ParsoidFormatHelper::parseContentTypeHeader( |
| 573 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable Silly Phan, we just checked. |
| 574 | $originalRendering->headers['content-type'] |
| 575 | ); |
| 576 | |
| 577 | if ( $vOriginal ) { |
| 578 | $this->transform->setOriginalSchemaVersion( $vOriginal ); |
| 579 | } |
| 580 | } |
| 581 | |
| 582 | if ( $rev instanceof RevisionRecord ) { |
| 583 | $this->transform->setOriginalRevision( $rev ); |
| 584 | } elseif ( $rev && is_int( $rev ) ) { |
| 585 | $this->transform->setOriginalRevisionId( $rev ); |
| 586 | } |
| 587 | |
| 588 | // NOTE: We might have an incomplete PageBundle here, with no HTML. |
| 589 | // PageBundle::$html is declared to not be nullable, so it would be set to the empty |
| 590 | // string if not given. |
| 591 | if ( $originalRendering->html !== '' ) { |
| 592 | $this->transform->setOriginalHtml( $originalRendering->html ); |
| 593 | } |
| 594 | |
| 595 | if ( $originalRendering->parsoid !== null ) { |
| 596 | $this->transform->setOriginalDataParsoid( $originalRendering->parsoid ); |
| 597 | } |
| 598 | |
| 599 | if ( $originalRendering->mw !== null ) { |
| 600 | $this->transform->setOriginalDataMW( $originalRendering->mw ); |
| 601 | } |
| 602 | } |
| 603 | |
| 604 | /** |
| 605 | * @return Content the content derived from the input HTML. |
| 606 | * @throws HttpException |
| 607 | */ |
| 608 | public function getContent(): Content { |
| 609 | try { |
| 610 | return $this->transform->htmlToContent(); |
| 611 | } catch ( ClientError $e ) { |
| 612 | throw new LocalizedHttpException( |
| 613 | new MessageValue( 'rest-html-backend-error', [ $e->getMessage() ] ), |
| 614 | 400, |
| 615 | [ 'reason' => $e->getMessage() ] |
| 616 | ); |
| 617 | } catch ( ResourceLimitExceededException $e ) { |
| 618 | throw new LocalizedHttpException( |
| 619 | new MessageValue( 'rest-resource-limit-exceeded' ), |
| 620 | 413, |
| 621 | [ 'reason' => $e->getMessage() ] |
| 622 | ); |
| 623 | } catch ( MWUnknownContentModelException $e ) { |
| 624 | throw new LocalizedHttpException( |
| 625 | new MessageValue( "rest-unknown-content-model", [ $e->getModelId() ] ), |
| 626 | 400 |
| 627 | ); |
| 628 | } |
| 629 | } |
| 630 | |
| 631 | /** |
| 632 | * Creates a response containing the content derived from the input HTML. |
| 633 | * This will set the appropriate Content-Type header. |
| 634 | * |
| 635 | * @param ResponseInterface $response |
| 636 | */ |
| 637 | public function putContent( ResponseInterface $response ) { |
| 638 | $content = $this->getContent(); |
| 639 | $data = $content->serialize(); |
| 640 | |
| 641 | try { |
| 642 | $contentType = ParsoidFormatHelper::getContentType( |
| 643 | $content->getModel(), |
| 644 | $this->envOptions['outputContentVersion'] |
| 645 | ); |
| 646 | } catch ( InvalidArgumentException $e ) { |
| 647 | // If Parsoid doesn't know the content type, |
| 648 | // ask the ContentHandler! |
| 649 | $contentType = $content->getDefaultFormat(); |
| 650 | } |
| 651 | |
| 652 | $response->setHeader( 'Content-Type', $contentType ); |
| 653 | $response->getBody()->write( $data ); |
| 654 | } |
| 655 | |
| 656 | /** |
| 657 | * @param PageIdentity $page |
| 658 | * @param RevisionRecord|int $revision |
| 659 | * @param bool $mayParse |
| 660 | * |
| 661 | * @return ParserOutput|null |
| 662 | * @throws HttpException |
| 663 | */ |
| 664 | private function fetchParserOutputFromParsoid( PageIdentity $page, $revision, bool $mayParse ): ?ParserOutput { |
| 665 | $parserOptions = ParserOptions::newFromAnon(); |
| 666 | $parserOptions->setUseParsoid(); |
| 667 | |
| 668 | try { |
| 669 | if ( !$page instanceof PageRecord ) { |
| 670 | $name = "$page"; |
| 671 | $page = $this->pageLookup->getPageByReference( $page ); |
| 672 | if ( !$page ) { |
| 673 | throw new RevisionAccessException( 'Page {name} not found', |
| 674 | [ 'name' => $name ] ); |
| 675 | } |
| 676 | } |
| 677 | |
| 678 | if ( is_int( $revision ) ) { |
| 679 | $revId = $revision; |
| 680 | $revision = $this->revisionLookup->getRevisionById( $revId, 0, $page ); |
| 681 | |
| 682 | if ( !$revision ) { |
| 683 | throw new RevisionAccessException( 'Revision {revId} not found', |
| 684 | [ 'revId' => $revId ] ); |
| 685 | } |
| 686 | } |
| 687 | |
| 688 | if ( $page->getId() !== $revision->getPageId() ) { |
| 689 | throw new RevisionAccessException( 'Revision {revId} does not belong to page {name}', |
| 690 | [ 'name' => $page->getDBkey(), |
| 691 | 'revId' => $revision->getId() ] ); |
| 692 | } |
| 693 | |
| 694 | if ( $mayParse ) { |
| 695 | try { |
| 696 | $status = $this->parserOutputAccess->getParserOutput( |
| 697 | $page, $parserOptions, $revision |
| 698 | ); |
| 699 | } catch ( ClientError $e ) { |
| 700 | $status = Status::newFatal( 'parsoid-client-error', $e->getMessage() ); |
| 701 | } catch ( ResourceLimitExceededException $e ) { |
| 702 | $status = Status::newFatal( 'parsoid-resource-limit-exceeded', $e->getMessage() ); |
| 703 | } |
| 704 | |
| 705 | if ( !$status->isOK() ) { |
| 706 | $this->throwHttpExceptionForStatus( $status ); |
| 707 | } |
| 708 | |
| 709 | $parserOutput = $status->getValue(); |
| 710 | } else { |
| 711 | $parserOutput = $this->parserOutputAccess->getCachedParserOutput( |
| 712 | $page, $parserOptions, $revision |
| 713 | ); |
| 714 | } |
| 715 | } catch ( RevisionAccessException $e ) { |
| 716 | // The client supplied bad revision ID, or the revision was deleted or suppressed. |
| 717 | throw new LocalizedHttpException( new MessageValue( "rest-specified-revision-unavailable" ), |
| 718 | 404, |
| 719 | [ 'reason' => $e->getMessage() ] |
| 720 | ); |
| 721 | } |
| 722 | |
| 723 | return $parserOutput; |
| 724 | } |
| 725 | |
| 726 | /** |
| 727 | * @param ParsoidRenderID $renderID |
| 728 | * |
| 729 | * @return SelserContext|null |
| 730 | */ |
| 731 | private function fetchSelserContextFromStash( $renderID ): ?SelserContext { |
| 732 | $selserContext = $this->parsoidOutputStash->get( $renderID ); |
| 733 | $labels = [ |
| 734 | 'original_html_given' => 'as_renderid', |
| 735 | 'page_exists' => 'unknown', |
| 736 | 'status' => 'hit-stashed' |
| 737 | ]; |
| 738 | $counter = $this->statsFactory->getCounter( 'html_input_transform_total' ); |
| 739 | if ( $selserContext ) { |
| 740 | $counter->setLabels( $labels ) |
| 741 | ->copyToStatsdAt( 'html_input_transform.original_html.given.as_renderid.stash_hit.found.hit' ) |
| 742 | ->increment(); |
| 743 | return $selserContext; |
| 744 | } else { |
| 745 | // Looks like the rendering is gone from stash (or the client send us a bogus key). |
| 746 | // Try to load it from the parser cache instead. |
| 747 | // On a wiki with low edit frequency, there is a good chance that it's still there. |
| 748 | try { |
| 749 | $parserOutput = $this->fetchParserOutputFromParsoid( $this->page, $renderID->getRevisionID(), false ); |
| 750 | |
| 751 | if ( !$parserOutput ) { |
| 752 | $labels[ 'status' ] = 'miss-fallback_not_found'; |
| 753 | $counter->setLabels( $labels )->copyToStatsdAt( |
| 754 | 'html_input_transform.original_html.given.as_renderid.stash_miss_pc_fallback.not_found.miss' |
| 755 | )->increment(); |
| 756 | return null; |
| 757 | } |
| 758 | |
| 759 | $cachedRenderID = ParsoidRenderID::newFromParserOutput( $parserOutput ); |
| 760 | if ( $cachedRenderID->getKey() !== $renderID->getKey() ) { |
| 761 | $labels[ 'status' ] = 'mismatch-fallback_not_found'; |
| 762 | $counter->setLabels( $labels ) |
| 763 | ->copyToStatsdAt( |
| 764 | 'html_input_transform.original_html.given.as_renderid.' . |
| 765 | 'stash_miss_pc_fallback.not_found.mismatch' |
| 766 | ) |
| 767 | ->increment(); |
| 768 | |
| 769 | // It's not the correct rendering. |
| 770 | return null; |
| 771 | } |
| 772 | $labels[ 'status' ] = 'hit-fallback_found'; |
| 773 | $counter->setLabels( $labels ) |
| 774 | ->copyToStatsdAt( |
| 775 | 'html_input_transform.original_html.given.as_renderid.' . |
| 776 | 'stash_miss_pc_fallback.found.hit' |
| 777 | ) |
| 778 | ->increment(); |
| 779 | |
| 780 | $pb = PageBundleParserOutputConverter::pageBundleFromParserOutput( $parserOutput ); |
| 781 | return new SelserContext( $pb, $renderID->getRevisionID() ); |
| 782 | } catch ( HttpException $e ) { |
| 783 | $labels[ 'status' ] = 'failed-fallback_not_found'; |
| 784 | $counter->setLabels( $labels ) |
| 785 | ->copyToStatsdAt( |
| 786 | 'html_input_transform.original_html.given.as_renderid.' . |
| 787 | 'stash_miss_pc_fallback.not_found.failed' |
| 788 | ) |
| 789 | ->increment(); |
| 790 | |
| 791 | // If the revision isn't found, don't trigger a 404. Return null to trigger a 412. |
| 792 | return null; |
| 793 | } |
| 794 | } |
| 795 | } |
| 796 | |
| 797 | /** |
| 798 | * @param Status $status |
| 799 | * |
| 800 | * @return never |
| 801 | * @throws HttpException |
| 802 | */ |
| 803 | private function throwHttpExceptionForStatus( Status $status ) { |
| 804 | // TODO: make this nicer. |
| 805 | if ( $status->hasMessage( 'parsoid-resource-limit-exceeded' ) ) { |
| 806 | throw new LocalizedHttpException( new MessageValue( "rest-parsoid-resource-exceeded" ), |
| 807 | 413, |
| 808 | [ 'reason' => $status->getHTML() ] |
| 809 | ); |
| 810 | } else { |
| 811 | throw new LocalizedHttpException( new MessageValue( "rest-parsoid-error" ), |
| 812 | 400, |
| 813 | [ 'reason' => $status->getHTML() ] |
| 814 | ); |
| 815 | } |
| 816 | } |
| 817 | |
| 818 | } |