Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
48.37% |
89 / 184 |
|
17.65% |
3 / 17 |
CRAP | |
0.00% |
0 / 1 |
| Parsoid | |
48.37% |
89 / 184 |
|
17.65% |
3 / 17 |
379.45 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| defaultHTMLVersion | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| resolveContentVersion | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| supportsLanguageConversion | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setupCommonOptions | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
72 | |||
| parseWikitext | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
42 | |||
| wikitext2html | |
95.65% |
22 / 23 |
|
0.00% |
0 / 1 |
5 | |||
| wikitext2lint | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| dom2wikitext | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
6 | |||
| html2wikitext | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| pb2pb | |
98.33% |
59 / 60 |
|
0.00% |
0 / 1 |
6 | |||
| substTopLevelTemplates | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| findDowngrade | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| downgrade | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
30 | |||
| implementsLanguageConversion | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| implementsLanguageConversionBcp47 | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| downgrade999to2 | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid; |
| 5 | |
| 6 | use Composer\Semver\Comparator; |
| 7 | use Composer\Semver\Semver; |
| 8 | use InvalidArgumentException; |
| 9 | use LogicException; |
| 10 | use Wikimedia\Bcp47Code\Bcp47Code; |
| 11 | use Wikimedia\Parsoid\Config\DataAccess; |
| 12 | use Wikimedia\Parsoid\Config\Env; |
| 13 | use Wikimedia\Parsoid\Config\PageConfig; |
| 14 | use Wikimedia\Parsoid\Config\SiteConfig; |
| 15 | use Wikimedia\Parsoid\Config\StubMetadataCollector; |
| 16 | use Wikimedia\Parsoid\Core\ContentMetadataCollector; |
| 17 | use Wikimedia\Parsoid\Core\PageBundle; |
| 18 | use Wikimedia\Parsoid\Core\ResourceLimitExceededException; |
| 19 | use Wikimedia\Parsoid\Core\SelserData; |
| 20 | use Wikimedia\Parsoid\DOM\Document; |
| 21 | use Wikimedia\Parsoid\Ext\ParsoidExtensionAPI; |
| 22 | use Wikimedia\Parsoid\Language\LanguageConverter; |
| 23 | use Wikimedia\Parsoid\Logger\LintLogger; |
| 24 | use Wikimedia\Parsoid\Utils\ContentUtils; |
| 25 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 26 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 27 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 28 | use Wikimedia\Parsoid\Utils\Utils; |
| 29 | use Wikimedia\Parsoid\Wikitext\Wikitext; |
| 30 | use Wikimedia\Parsoid\Wt2Html\PP\Processors\AddRedLinks; |
| 31 | use Wikimedia\Parsoid\Wt2Html\PP\Processors\ConvertOffsets; |
| 32 | |
| 33 | class Parsoid { |
| 34 | |
| 35 | /** |
| 36 | * Available HTML content versions. |
| 37 | * @see https://www.mediawiki.org/wiki/Parsoid/API#Content_Negotiation |
| 38 | * @see https://www.mediawiki.org/wiki/Specs/HTML#Versioning |
| 39 | */ |
| 40 | public const AVAILABLE_VERSIONS = [ '2.7.0', '999.0.0' ]; |
| 41 | |
| 42 | private const DOWNGRADES = [ |
| 43 | [ 'from' => '999.0.0', 'to' => '2.0.0', 'func' => 'downgrade999to2' ], |
| 44 | ]; |
| 45 | |
| 46 | /** @var SiteConfig */ |
| 47 | private $siteConfig; |
| 48 | |
| 49 | /** @var DataAccess */ |
| 50 | private $dataAccess; |
| 51 | |
| 52 | /** |
| 53 | * @param SiteConfig $siteConfig |
| 54 | * @param DataAccess $dataAccess |
| 55 | */ |
| 56 | public function __construct( |
| 57 | SiteConfig $siteConfig, DataAccess $dataAccess |
| 58 | ) { |
| 59 | $this->siteConfig = $siteConfig; |
| 60 | $this->dataAccess = $dataAccess; |
| 61 | } |
| 62 | |
| 63 | /** |
| 64 | * Returns the default HTML content version |
| 65 | * @return string |
| 66 | */ |
| 67 | public static function defaultHTMLVersion(): string { |
| 68 | return self::AVAILABLE_VERSIONS[0]; |
| 69 | } |
| 70 | |
| 71 | /** |
| 72 | * See if any content version Parsoid knows how to produce satisfies the |
| 73 | * the supplied version, when interpreted with semver caret semantics. |
| 74 | * This will allow us to make backwards compatible changes, without the need |
| 75 | * for clients to bump the version in their headers all the time. |
| 76 | * |
| 77 | * @param string $version |
| 78 | * @return string|null |
| 79 | */ |
| 80 | public static function resolveContentVersion( string $version ) { |
| 81 | foreach ( self::AVAILABLE_VERSIONS as $i => $a ) { |
| 82 | if ( Semver::satisfies( $a, "^{$version}" ) && |
| 83 | // The section wrapping in 1.6.x should have induced a major |
| 84 | // version bump, since it requires upgrading clients to |
| 85 | // handle it. We therefore hardcode this in so that we can |
| 86 | // fail hard. |
| 87 | Comparator::greaterThanOrEqualTo( $version, '1.6.0' ) |
| 88 | ) { |
| 89 | return $a; |
| 90 | } |
| 91 | } |
| 92 | return null; |
| 93 | } |
| 94 | |
| 95 | /** |
| 96 | * Determine if language conversion is enabled, aka if the optional |
| 97 | * wikimedia/langconv library is installed. |
| 98 | * @return bool True if the wikimedia/langconv library is available |
| 99 | */ |
| 100 | public static function supportsLanguageConversion(): bool { |
| 101 | return class_exists( '\Wikimedia\LangConv\ReplacementMachine' ); |
| 102 | } |
| 103 | |
| 104 | /** |
| 105 | * @param array $options |
| 106 | * @return array |
| 107 | */ |
| 108 | private function setupCommonOptions( array $options ): array { |
| 109 | $envOptions = []; |
| 110 | if ( isset( $options['offsetType'] ) ) { |
| 111 | $envOptions['offsetType'] = $options['offsetType']; |
| 112 | } |
| 113 | if ( isset( $options['traceFlags'] ) ) { |
| 114 | $envOptions['traceFlags'] = $options['traceFlags']; |
| 115 | } |
| 116 | if ( isset( $options['dumpFlags'] ) ) { |
| 117 | $envOptions['dumpFlags'] = $options['dumpFlags']; |
| 118 | } |
| 119 | if ( isset( $options['debugFlags'] ) ) { |
| 120 | $envOptions['debugFlags'] = $options['debugFlags']; |
| 121 | } |
| 122 | if ( !empty( $options['htmlVariantLanguage'] ) ) { |
| 123 | $envOptions['htmlVariantLanguage'] = $options['htmlVariantLanguage']; |
| 124 | } |
| 125 | if ( !empty( $options['wtVariantLanguage'] ) ) { |
| 126 | $envOptions['wtVariantLanguage'] = $options['wtVariantLanguage']; |
| 127 | } |
| 128 | if ( isset( $options['logLevels'] ) ) { |
| 129 | $envOptions['logLevels'] = $options['logLevels']; |
| 130 | } |
| 131 | return $envOptions; |
| 132 | } |
| 133 | |
| 134 | /** |
| 135 | * Parsing code shared between the next two methods. |
| 136 | * |
| 137 | * @param PageConfig $pageConfig |
| 138 | * @param ContentMetadataCollector $metadata |
| 139 | * @param array $options See wikitext2html. |
| 140 | * @return array |
| 141 | */ |
| 142 | private function parseWikitext( |
| 143 | PageConfig $pageConfig, |
| 144 | ContentMetadataCollector $metadata, |
| 145 | array $options = [] |
| 146 | ): array { |
| 147 | $envOptions = $this->setupCommonOptions( $options ); |
| 148 | if ( isset( $options['outputContentVersion'] ) ) { |
| 149 | $envOptions['outputContentVersion'] = $options['outputContentVersion']; |
| 150 | } |
| 151 | $envOptions['discardDataParsoid'] = !empty( $options['discardDataParsoid'] ); |
| 152 | if ( isset( $options['wrapSections'] ) ) { |
| 153 | $envOptions['wrapSections'] = !empty( $options['wrapSections'] ); |
| 154 | } |
| 155 | if ( isset( $options['pageBundle'] ) ) { |
| 156 | $envOptions['pageBundle'] = !empty( $options['pageBundle'] ); |
| 157 | } |
| 158 | if ( isset( $options['logLinterData'] ) ) { |
| 159 | $envOptions['logLinterData'] = !empty( $options['logLinterData'] ); |
| 160 | } |
| 161 | $env = new Env( |
| 162 | $this->siteConfig, $pageConfig, $this->dataAccess, $metadata, $envOptions |
| 163 | ); |
| 164 | if ( !$env->compareWt2HtmlLimit( |
| 165 | 'wikitextSize', strlen( $env->topFrame->getSrcText() ) |
| 166 | ) ) { |
| 167 | throw new ResourceLimitExceededException( |
| 168 | "wt2html: wikitextSize limit exceeded" |
| 169 | ); |
| 170 | } |
| 171 | $contentmodel = $options['contentmodel'] ?? null; |
| 172 | $handler = $env->getContentHandler( $contentmodel ); |
| 173 | $extApi = new ParsoidExtensionAPI( $env ); |
| 174 | return [ $env, $handler->toDOM( $extApi ), $contentmodel ]; |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * Parse the wikitext supplied in a `PageConfig` to HTML. |
| 179 | * |
| 180 | * @param PageConfig $pageConfig |
| 181 | * @param array $options [ |
| 182 | * 'wrapSections' => (bool) Whether `<section>` wrappers should be added. |
| 183 | * 'pageBundle' => (bool) Sets ids on nodes and stores |
| 184 | * data-* attributes in a JSON blob. |
| 185 | * 'body_only' => (bool|null) Only return the <body> children (T181657) |
| 186 | * 'outputContentVersion' => (string|null) Version of HTML to output. |
| 187 | * `null` returns the default version. |
| 188 | * 'contentmodel' => (string|null) The content model of the input. |
| 189 | * 'discardDataParsoid' => (bool) Drop all data-parsoid annotations. |
| 190 | * 'offsetType' => (string) ucs2, char, byte are valid values |
| 191 | * what kind of source offsets should be emitted? |
| 192 | * 'htmlVariantLanguage' => (string|Bcp47Code) If non-null, the language variant used for Parsoid HTML. |
| 193 | * A MediaWiki-internal language code string (deprecated), |
| 194 | * or a Bcp47Code object. |
| 195 | * 'wtVariantLanguage' => (string|Bcp47Code) If non-null, the language variant used for wikitext. |
| 196 | * A MediaWiki-internal language code string (deprecated), |
| 197 | * or a Bcp47Code object. |
| 198 | * 'logLinterData' => (bool) Should we log linter data if linting is enabled? |
| 199 | * 'traceFlags' => (array) associative array with tracing options |
| 200 | * 'dumpFlags' => (array) associative array with dump options |
| 201 | * 'debugFlags' => (array) associative array with debug options |
| 202 | * 'logLevels' => (string[]) Levels to log |
| 203 | * ] |
| 204 | * @param ?array &$headers |
| 205 | * @param ?ContentMetadataCollector $metadata Pass in a CMC in order to |
| 206 | * collect and retrieve metadata about the parse. |
| 207 | * @return PageBundle|string |
| 208 | */ |
| 209 | public function wikitext2html( |
| 210 | PageConfig $pageConfig, array $options = [], ?array &$headers = null, |
| 211 | ?ContentMetadataCollector $metadata = null |
| 212 | ) { |
| 213 | if ( $metadata === null ) { |
| 214 | $metadata = new StubMetadataCollector( $this->siteConfig->getLogger() ); |
| 215 | } |
| 216 | [ $env, $doc, $contentmodel ] = $this->parseWikitext( $pageConfig, $metadata, $options ); |
| 217 | // FIXME: Does this belong in parseWikitext so that the other endpoint |
| 218 | // is covered as well? It probably depends on expectations of the |
| 219 | // Rest API. If callers of /page/lint/ assume that will update the |
| 220 | // results on the Special page. |
| 221 | if ( $env->getSiteConfig()->linting() ) { |
| 222 | ( new LintLogger( $env ) )->logLintOutput(); |
| 223 | } |
| 224 | $headers = DOMUtils::findHttpEquivHeaders( $doc ); |
| 225 | $body_only = !empty( $options['body_only'] ); |
| 226 | $node = $body_only ? DOMCompat::getBody( $doc ) : $doc; |
| 227 | if ( $env->pageBundle ) { |
| 228 | $out = ContentUtils::extractDpAndSerialize( $node, [ |
| 229 | 'innerXML' => $body_only, |
| 230 | ] ); |
| 231 | return new PageBundle( |
| 232 | $out['html'], |
| 233 | $out['pb']->parsoid, $out['pb']->mw ?? null, |
| 234 | $env->getOutputContentVersion(), |
| 235 | $headers, |
| 236 | $contentmodel |
| 237 | ); |
| 238 | } else { |
| 239 | $xml = ContentUtils::toXML( $node, [ |
| 240 | 'innerXML' => $body_only, |
| 241 | ] ); |
| 242 | return $xml; |
| 243 | } |
| 244 | } |
| 245 | |
| 246 | /** |
| 247 | * Lint the wikitext supplied in a `PageConfig`. |
| 248 | * |
| 249 | * @param PageConfig $pageConfig |
| 250 | * @param array $options See wikitext2html. |
| 251 | * @return array |
| 252 | */ |
| 253 | public function wikitext2lint( |
| 254 | PageConfig $pageConfig, array $options = [] |
| 255 | ): array { |
| 256 | $metadata = new StubMetadataCollector( $this->siteConfig->getLogger() ); |
| 257 | [ $env, ] = $this->parseWikitext( $pageConfig, $metadata, $options ); |
| 258 | return $env->getLints(); |
| 259 | } |
| 260 | |
| 261 | /** |
| 262 | * Serialize DOM to wikitext. |
| 263 | * |
| 264 | * @param PageConfig $pageConfig |
| 265 | * @param Document $doc Data attributes are expected to have been applied |
| 266 | * already. Loading them will happen once the environment is created. |
| 267 | * @param array $options [ |
| 268 | * 'inputContentVersion' => (string) The content version of the input. |
| 269 | * Necessary if it differs from the current default in order to |
| 270 | * account for any serialization differences. |
| 271 | * 'offsetType' => (string) ucs2, char, byte are valid values |
| 272 | * what kind of source offsets are present in the HTML? |
| 273 | * 'contentmodel' => (string|null) The content model of the input. |
| 274 | * 'htmlVariantLanguage' => (string|Bcp47Code) If non-null, the language variant used for Parsoid HTML. |
| 275 | * A MediaWiki-internal language code string (deprecated), |
| 276 | * or a Bcp47Code object. |
| 277 | * 'wtVariantLanguage' => (string|Bcp47Code) If non-null, the language variant used for wikitext. |
| 278 | * A MediaWiki-internal language code string (deprecated), |
| 279 | * or a Bcp47Code object. |
| 280 | * 'traceFlags' => (array) associative array with tracing options |
| 281 | * 'dumpFlags' => (array) associative array with dump options |
| 282 | * 'debugFlags' => (array) associative array with debug options |
| 283 | * 'logLevels' => (string[]) Levels to log |
| 284 | * 'htmlSize' => (int) Size of the HTML that generated $doc |
| 285 | * ] |
| 286 | * @param ?SelserData $selserData |
| 287 | * @return string |
| 288 | */ |
| 289 | public function dom2wikitext( |
| 290 | PageConfig $pageConfig, Document $doc, array $options = [], |
| 291 | ?SelserData $selserData = null |
| 292 | ): string { |
| 293 | $envOptions = $this->setupCommonOptions( $options ); |
| 294 | if ( isset( $options['inputContentVersion'] ) ) { |
| 295 | $envOptions['inputContentVersion'] = $options['inputContentVersion']; |
| 296 | } |
| 297 | $envOptions['topLevelDoc'] = $doc; |
| 298 | $metadata = new StubMetadataCollector( $this->siteConfig->getLogger() ); |
| 299 | $env = new Env( |
| 300 | $this->siteConfig, $pageConfig, $this->dataAccess, $metadata, $envOptions |
| 301 | ); |
| 302 | $env->bumpHtml2WtResourceUse( 'htmlSize', $options['htmlSize'] ?? 0 ); |
| 303 | $contentmodel = $options['contentmodel'] ?? null; |
| 304 | $handler = $env->getContentHandler( $contentmodel ); |
| 305 | $extApi = new ParsoidExtensionAPI( $env ); |
| 306 | return $handler->fromDOM( $extApi, $selserData ); |
| 307 | } |
| 308 | |
| 309 | /** |
| 310 | * Serialize HTML to wikitext. Convenience method for dom2wikitext. |
| 311 | * |
| 312 | * @param PageConfig $pageConfig |
| 313 | * @param string $html |
| 314 | * @param array $options |
| 315 | * @param ?SelserData $selserData |
| 316 | * @return string |
| 317 | */ |
| 318 | public function html2wikitext( |
| 319 | PageConfig $pageConfig, string $html, array $options = [], |
| 320 | ?SelserData $selserData = null |
| 321 | ): string { |
| 322 | $doc = DOMUtils::parseHTML( $html, true ); |
| 323 | return $this->dom2wikitext( $pageConfig, $doc, $options, $selserData ); |
| 324 | } |
| 325 | |
| 326 | /** |
| 327 | * Update the supplied PageBundle based on the `$update` type. |
| 328 | * |
| 329 | * 'redlinks': Refreshes the classes of known, missing, etc. links. |
| 330 | * 'variant': Converts the HTML based on the supplied variant. |
| 331 | * |
| 332 | * Note that these are DOM transforms, and not roundtrips through wikitext. |
| 333 | * |
| 334 | * @param PageConfig $pageConfig |
| 335 | * @param string $update 'redlinks'|'variant' |
| 336 | * @param PageBundle $pb |
| 337 | * @param array $options |
| 338 | * @return PageBundle |
| 339 | */ |
| 340 | public function pb2pb( |
| 341 | PageConfig $pageConfig, string $update, PageBundle $pb, |
| 342 | array $options = [] |
| 343 | ): PageBundle { |
| 344 | $envOptions = [ |
| 345 | 'pageBundle' => true, |
| 346 | 'topLevelDoc' => DOMUtils::parseHTML( $pb->toHtml(), true ), |
| 347 | ]; |
| 348 | $metadata = new StubMetadataCollector( $this->siteConfig->getLogger() ); |
| 349 | $env = new Env( |
| 350 | $this->siteConfig, $pageConfig, $this->dataAccess, $metadata, $envOptions |
| 351 | ); |
| 352 | $doc = $env->topLevelDoc; |
| 353 | DOMDataUtils::visitAndLoadDataAttribs( |
| 354 | DOMCompat::getBody( $doc ), [ 'markNew' => true ] |
| 355 | ); |
| 356 | ContentUtils::convertOffsets( |
| 357 | $env, $doc, $env->getRequestOffsetType(), 'byte' |
| 358 | ); |
| 359 | if ( $update === 'redlinks' ) { |
| 360 | ( new AddRedLinks() )->run( $env, DOMCompat::getBody( $doc ) ); |
| 361 | } elseif ( $update === 'variant' ) { |
| 362 | // Note that `maybeConvert` could still be a no-op, in case the |
| 363 | // __NOCONTENTCONVERT__ magic word is present, or the targetVariant |
| 364 | // is a base language code or otherwise invalid. |
| 365 | LanguageConverter::maybeConvert( |
| 366 | $env, $doc, |
| 367 | Utils::mwCodeToBcp47( $options['variant']['target'] ), |
| 368 | $options['variant']['source'] ? |
| 369 | Utils::mwCodeToBcp47( $options['variant']['source'] ) : null |
| 370 | ); |
| 371 | // Update content-language and vary headers. |
| 372 | // This also ensures there is a <head> element. |
| 373 | $ensureHeader = static function ( string $h ) use ( $doc ) { |
| 374 | $el = DOMCompat::querySelector( $doc, "meta[http-equiv=\"{$h}\"i]" ); |
| 375 | if ( !$el ) { |
| 376 | $el = DOMUtils::appendToHead( $doc, 'meta', [ |
| 377 | 'http-equiv' => $h, |
| 378 | ] ); |
| 379 | } |
| 380 | return $el; |
| 381 | }; |
| 382 | ( $ensureHeader( 'content-language' ) )->setAttribute( |
| 383 | 'content', $env->htmlContentLanguageBcp47()->toBcp47Code() |
| 384 | ); |
| 385 | ( $ensureHeader( 'vary' ) )->setAttribute( |
| 386 | 'content', $env->htmlVary() |
| 387 | ); |
| 388 | } else { |
| 389 | throw new LogicException( 'Unknown transformation.' ); |
| 390 | } |
| 391 | ( new ConvertOffsets() )->run( $env, DOMCompat::getBody( $doc ), [], true ); |
| 392 | DOMDataUtils::visitAndStoreDataAttribs( |
| 393 | DOMCompat::getBody( $doc ), [ |
| 394 | 'discardDataParsoid' => $env->discardDataParsoid, |
| 395 | 'storeInPageBundle' => $env->pageBundle, |
| 396 | 'env' => $env, |
| 397 | ] |
| 398 | ); |
| 399 | $body_only = !empty( $options['body_only'] ); |
| 400 | $node = $body_only ? DOMCompat::getBody( $doc ) : $doc; |
| 401 | DOMDataUtils::injectPageBundle( $doc, DOMDataUtils::getPageBundle( $doc ) ); |
| 402 | $out = ContentUtils::extractDpAndSerialize( $node, [ |
| 403 | 'innerXML' => $body_only, |
| 404 | ] ); |
| 405 | return new PageBundle( |
| 406 | $out['html'], |
| 407 | $out['pb']->parsoid, $out['pb']->mw ?? null, |
| 408 | // Prefer the passed in version, since this was just a transformation |
| 409 | $pb->version ?? $env->getOutputContentVersion(), |
| 410 | DOMUtils::findHttpEquivHeaders( $doc ), |
| 411 | // Prefer the passed in content model |
| 412 | $pb->contentmodel ?? $pageConfig->getContentModel() |
| 413 | ); |
| 414 | } |
| 415 | |
| 416 | /** |
| 417 | * Perform pre-save transformations with top-level templates subst'd. |
| 418 | * |
| 419 | * @param PageConfig $pageConfig |
| 420 | * @param string $wikitext |
| 421 | * @return string |
| 422 | */ |
| 423 | public function substTopLevelTemplates( |
| 424 | PageConfig $pageConfig, string $wikitext |
| 425 | ): string { |
| 426 | $metadata = new StubMetadataCollector( $this->siteConfig->getLogger() ); |
| 427 | $env = new Env( $this->siteConfig, $pageConfig, $this->dataAccess, $metadata ); |
| 428 | return Wikitext::pst( $env, $wikitext, true /* $substTLTemplates */ ); |
| 429 | } |
| 430 | |
| 431 | /** |
| 432 | * Check whether a given content version can be downgraded to the requested |
| 433 | * content version. |
| 434 | * |
| 435 | * @param string $from Current content version |
| 436 | * @param string $to Requested content version |
| 437 | * @return string[]|null The downgrade that will fulfill the request, as |
| 438 | * [ 'from' => <old version>, 'to' => <new version> ], or null if it |
| 439 | * can't be fulfilled. |
| 440 | */ |
| 441 | public static function findDowngrade( string $from, string $to ): ?array { |
| 442 | foreach ( self::DOWNGRADES as list( 'from' => $dgFrom, 'to' => $dgTo ) ) { |
| 443 | if ( |
| 444 | Semver::satisfies( $from, "^$dgFrom" ) && |
| 445 | Semver::satisfies( $to, "^$dgTo" ) |
| 446 | ) { |
| 447 | // FIXME: Make this a class? |
| 448 | return [ 'from' => $dgFrom, 'to' => $dgTo ]; |
| 449 | } |
| 450 | } |
| 451 | return null; |
| 452 | } |
| 453 | |
| 454 | /** |
| 455 | * Downgrade a document to an older content version. |
| 456 | * |
| 457 | * @param string[] $dg Value returned by findDowngrade(). |
| 458 | * @param PageBundle $pageBundle |
| 459 | */ |
| 460 | public static function downgrade( |
| 461 | array $dg, PageBundle $pageBundle |
| 462 | ): void { |
| 463 | foreach ( self::DOWNGRADES as list( 'from' => $dgFrom, 'to' => $dgTo, 'func' => $dgFunc ) ) { |
| 464 | if ( $dg['from'] === $dgFrom && $dg['to'] === $dgTo ) { |
| 465 | call_user_func( [ self::class, $dgFunc ], $pageBundle ); |
| 466 | |
| 467 | // FIXME: Maybe this resolve should just be part of the $dg |
| 468 | $pageBundle->version = self::resolveContentVersion( $dg['to'] ); |
| 469 | |
| 470 | // FIXME: Maybe this should be a helper to avoid the rt |
| 471 | $doc = DOMUtils::parseHTML( $pageBundle->html ); |
| 472 | // Match the http-equiv meta to the content-type header |
| 473 | $meta = DOMCompat::querySelector( $doc, |
| 474 | 'meta[property="mw:htmlVersion"], meta[property="mw:html:version"]' ); |
| 475 | if ( $meta ) { |
| 476 | $meta->setAttribute( 'content', $pageBundle->version ); |
| 477 | $pageBundle->html = ContentUtils::toXML( $doc ); |
| 478 | } |
| 479 | |
| 480 | return; |
| 481 | } |
| 482 | } |
| 483 | throw new InvalidArgumentException( |
| 484 | "Unsupported downgrade: {$dg['from']} -> {$dg['to']}" |
| 485 | ); |
| 486 | } |
| 487 | |
| 488 | /** |
| 489 | * Check if language variant conversion is implemented for a language |
| 490 | * |
| 491 | * @internal FIXME: Remove once Parsoid's language variant work is completed |
| 492 | * @param PageConfig $pageConfig |
| 493 | * @param string $targetVariantCode Variant code to check |
| 494 | * @return bool |
| 495 | * @deprecated Use ::implementsLanguageConversionBcp47() |
| 496 | */ |
| 497 | public function implementsLanguageConversion( PageConfig $pageConfig, string $targetVariantCode ): bool { |
| 498 | // argh, another interface that doesn't use Bcp47Code :( |
| 499 | return $this->implementsLanguageConversionBcp47( |
| 500 | $pageConfig, Utils::mwCodeToBcp47( $targetVariantCode ) |
| 501 | ); |
| 502 | } |
| 503 | |
| 504 | /** |
| 505 | * Check if language variant conversion is implemented for a language |
| 506 | * |
| 507 | * @internal FIXME: Remove once Parsoid's language variant work is completed |
| 508 | * @param PageConfig $pageConfig |
| 509 | * @param Bcp47Code $targetVariant Variant language to check |
| 510 | * @return bool |
| 511 | */ |
| 512 | public function implementsLanguageConversionBcp47( PageConfig $pageConfig, Bcp47Code $targetVariant ): bool { |
| 513 | $metadata = new StubMetadataCollector( $this->siteConfig->getLogger() ); |
| 514 | $env = new Env( $this->siteConfig, $pageConfig, $this->dataAccess, $metadata ); |
| 515 | |
| 516 | return LanguageConverter::implementsLanguageConversion( $env, $targetVariant ); |
| 517 | } |
| 518 | |
| 519 | /** |
| 520 | * Downgrade the given document and pagebundle from 999.x to 2.x. |
| 521 | * |
| 522 | * @param PageBundle $pageBundle |
| 523 | */ |
| 524 | private static function downgrade999to2( PageBundle $pageBundle ) { |
| 525 | // Effectively, skip applying data-parsoid. Note that if we were to |
| 526 | // support a pb2html downgrade, we'd need to apply the full thing, |
| 527 | // but that would create complications where ids would be left behind. |
| 528 | // See the comment in around `DOMDataUtils::applyPageBundle` |
| 529 | $newPageBundle = new PageBundle( |
| 530 | $pageBundle->html, |
| 531 | [ 'ids' => [] ], |
| 532 | $pageBundle->mw |
| 533 | ); |
| 534 | $pageBundle->html = $newPageBundle->toHtml(); |
| 535 | // Now, modify the pagebundle to the expected form. This is important |
| 536 | // since, at least in the serialization path, the original pb will be |
| 537 | // applied to the modified content and its presence could cause lost |
| 538 | // deletions. |
| 539 | $pageBundle->mw = [ 'ids' => [] ]; |
| 540 | } |
| 541 | } |