Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
4.33% |
34 / 785 |
|
0.00% |
0 / 25 |
CRAP | |
0.00% |
0 / 1 |
| WikiLinkHandler | |
4.33% |
34 / 785 |
|
0.00% |
0 / 25 |
48590.58 | |
0.00% |
0 / 1 |
| hrefParts | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| __construct | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| getWikiLinkTargetInfo | |
55.74% |
34 / 61 |
|
0.00% |
0 / 1 |
38.20 | |||
| onRedirect | |
0.00% |
0 / 34 |
|
0.00% |
0 / 1 |
30 | |||
| bailTokens | |
0.00% |
0 / 27 |
|
0.00% |
0 / 1 |
6 | |||
| onWikiLink | |
0.00% |
0 / 37 |
|
0.00% |
0 / 1 |
182 | |||
| wikiLinkHandler | |
0.00% |
0 / 26 |
|
0.00% |
0 / 1 |
156 | |||
| buildLinkAttrs | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
156 | |||
| addLinkAttributesAndGetContent | |
0.00% |
0 / 74 |
|
0.00% |
0 / 1 |
930 | |||
| renderWikiLink | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| renderCategory | |
0.00% |
0 / 35 |
|
0.00% |
0 / 1 |
42 | |||
| renderLanguageLink | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
30 | |||
| renderInterwikiLink | |
0.00% |
0 / 25 |
|
0.00% |
0 / 1 |
30 | |||
| getWrapperInfo | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
72 | |||
| getOptionInfo | |
0.00% |
0 / 25 |
|
0.00% |
0 / 1 |
42 | |||
| isWikitextOpt | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| stringifyOptionTokens | |
0.00% |
0 / 61 |
|
0.00% |
0 / 1 |
702 | |||
| getFormat | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getUsed | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
6 | |||
| hasTransclusion | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| renderFile | |
0.00% |
0 / 241 |
|
0.00% |
0 / 1 |
3906 | |||
| specialFilePath | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| linkToMedia | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
30 | |||
| renderMedia | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
12 | |||
| onTag | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | /** |
| 5 | * Simple link handler. |
| 6 | * |
| 7 | * TODO: keep round-trip information in meta tag or the like |
| 8 | */ |
| 9 | |
| 10 | namespace Wikimedia\Parsoid\Wt2Html\TT; |
| 11 | |
| 12 | use stdClass; |
| 13 | use Wikimedia\Assert\Assert; |
| 14 | use Wikimedia\Parsoid\Config\Env; |
| 15 | use Wikimedia\Parsoid\Core\DOMCompat; |
| 16 | use Wikimedia\Parsoid\Core\DomSourceRange; |
| 17 | use Wikimedia\Parsoid\Core\InternalException; |
| 18 | use Wikimedia\Parsoid\Core\Sanitizer; |
| 19 | use Wikimedia\Parsoid\Core\SourceRange; |
| 20 | use Wikimedia\Parsoid\Language\Language; |
| 21 | use Wikimedia\Parsoid\NodeData\DataMw; |
| 22 | use Wikimedia\Parsoid\NodeData\DataMwAttrib; |
| 23 | use Wikimedia\Parsoid\NodeData\DataMwError; |
| 24 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 25 | use Wikimedia\Parsoid\NodeData\TempData; |
| 26 | use Wikimedia\Parsoid\Tokens\EndTagTk; |
| 27 | use Wikimedia\Parsoid\Tokens\EOFTk; |
| 28 | use Wikimedia\Parsoid\Tokens\KV; |
| 29 | use Wikimedia\Parsoid\Tokens\SelfclosingTagTk; |
| 30 | use Wikimedia\Parsoid\Tokens\TagTk; |
| 31 | use Wikimedia\Parsoid\Tokens\Token; |
| 32 | use Wikimedia\Parsoid\Tokens\XMLTagTk; |
| 33 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 34 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 35 | use Wikimedia\Parsoid\Utils\PipelineUtils; |
| 36 | use Wikimedia\Parsoid\Utils\Title; |
| 37 | use Wikimedia\Parsoid\Utils\TitleException; |
| 38 | use Wikimedia\Parsoid\Utils\TokenUtils; |
| 39 | use Wikimedia\Parsoid\Utils\Utils; |
| 40 | use Wikimedia\Parsoid\Wikitext\Consts; |
| 41 | use Wikimedia\Parsoid\Wt2Html\PipelineContentCache; |
| 42 | use Wikimedia\Parsoid\Wt2Html\TokenHandlerPipeline; |
| 43 | |
| 44 | class WikiLinkHandler extends XMLTagBasedHandler { |
| 45 | private static bool $cachingEnabled = true; |
| 46 | |
| 47 | /** |
| 48 | * @return ?array{prefix: string, title: string} |
| 49 | */ |
| 50 | private static function hrefParts( string $str ): ?array { |
| 51 | if ( preg_match( '/^([^:]+):(.*)$/D', $str, $matches ) ) { |
| 52 | return [ 'prefix' => $matches[1], 'title' => $matches[2] ]; |
| 53 | } else { |
| 54 | return null; |
| 55 | } |
| 56 | } |
| 57 | |
| 58 | private ?PipelineContentCache $wikilinkCache = null; |
| 59 | |
| 60 | /** @inheritDoc */ |
| 61 | public function __construct( TokenHandlerPipeline $manager, array $options ) { |
| 62 | parent::__construct( $manager, $options ); |
| 63 | |
| 64 | // Cache only on seeing the same source the fourth time. |
| 65 | // This minimizes cache bloat & token cloning penalties |
| 66 | // and reserving benefits for links seen at least 5 times. |
| 67 | if ( self::$cachingEnabled ) { |
| 68 | $this->wikilinkCache = $this->manager->getEnv()->getCache( |
| 69 | "wikilink", |
| 70 | [ "repeatThreshold" => 3, "cloneValue" => true ] |
| 71 | ); |
| 72 | } |
| 73 | } |
| 74 | |
| 75 | /** |
| 76 | * Normalize and analyze a wikilink target. |
| 77 | * |
| 78 | * Returns an object containing |
| 79 | * - href: The expanded target string |
| 80 | * - hrefSrc: The original target wikitext |
| 81 | * - title: A title object *or* |
| 82 | * - language: An interwikiInfo object *or* |
| 83 | * - interwiki: An interwikiInfo object. |
| 84 | * - localprefix: Set if the link had a localinterwiki prefix (or prefixes) |
| 85 | * - fromColonEscapedText: Target was colon-escaped ([[:en:foo]]) |
| 86 | * - prefix: The original namespace or language/interwiki prefix without a |
| 87 | * colon escape. |
| 88 | * |
| 89 | * @param Token $token |
| 90 | * @param string $href |
| 91 | * @param string $hrefSrc |
| 92 | * @return stdClass The target info. |
| 93 | * @throws InternalException |
| 94 | */ |
| 95 | private function getWikiLinkTargetInfo( Token $token, string $href, string $hrefSrc ): stdClass { |
| 96 | $env = $this->env; |
| 97 | $siteConfig = $env->getSiteConfig(); |
| 98 | $info = (object)[ |
| 99 | 'href' => $href, |
| 100 | 'hrefSrc' => $hrefSrc, |
| 101 | // Initialize these properties to avoid isset checks |
| 102 | 'interwiki' => null, |
| 103 | 'language' => null, |
| 104 | 'localprefix' => null, |
| 105 | 'fromColonEscapedText' => null |
| 106 | ]; |
| 107 | |
| 108 | if ( ( ltrim( $info->href )[0] ?? '' ) === ':' ) { |
| 109 | $info->fromColonEscapedText = true; |
| 110 | // Remove the colon escape |
| 111 | $info->href = substr( ltrim( $info->href ), 1 ); |
| 112 | } |
| 113 | if ( ( $info->href[0] ?? '' ) === ':' ) { |
| 114 | if ( $env->linting( 'multi-colon-escape' ) ) { |
| 115 | $lint = [ |
| 116 | 'dsr' => DomSourceRange::fromTsr( $token->dataParsoid->tsr ), |
| 117 | 'params' => [ 'href' => ':' . $info->href ], |
| 118 | 'templateInfo' => null |
| 119 | ]; |
| 120 | if ( $this->options['inTemplate'] ) { |
| 121 | // Match Linter.findEnclosingTemplateName(), by first |
| 122 | // converting the title to an href using env.makeLink |
| 123 | $name = PHPUtils::stripPrefix( |
| 124 | $env->makeLink( $this->manager->getFrame()->getTitle() ), |
| 125 | './' |
| 126 | ); |
| 127 | $lint['templateInfo'] = [ 'name' => $name ]; |
| 128 | // TODO(arlolra): Pass tsr info to the frame (T405759) |
| 129 | $lint['dsr'] = new DomSourceRange( 0, 0, null, null, source: null ); |
| 130 | } |
| 131 | $env->recordLint( 'multi-colon-escape', $lint ); |
| 132 | } |
| 133 | // This will get caught by the caller, and mark the target as invalid |
| 134 | throw new InternalException( 'Multiple colons prefixing href.' ); |
| 135 | } |
| 136 | |
| 137 | $title = $env->resolveTitle( Utils::decodeURIComponent( $info->href ) ); |
| 138 | $hrefBits = self::hrefParts( $info->href ); |
| 139 | if ( $hrefBits ) { |
| 140 | $nsPrefix = $hrefBits['prefix']; |
| 141 | $info->prefix = $nsPrefix; |
| 142 | $nnn = Utils::normalizeNamespaceName( trim( $nsPrefix ) ); |
| 143 | $interwikiInfo = $siteConfig->interwikiMapNoNamespaces()[$nnn] ?? null; |
| 144 | // check for interwiki / language links |
| 145 | $ns = $siteConfig->namespaceId( $nnn ); |
| 146 | // also check for url to protect against [[constructor:foo]] |
| 147 | if ( $ns !== null ) { |
| 148 | $info->title = $env->makeTitleFromURLDecodedStr( $title ); |
| 149 | } elseif ( isset( $interwikiInfo['localinterwiki'] ) ) { |
| 150 | if ( $hrefBits['title'] === '' ) { |
| 151 | // Empty title => main page (T66167) |
| 152 | $info->title = Title::newFromLinkTarget( |
| 153 | $siteConfig->mainPageLinkTarget(), $siteConfig |
| 154 | ); |
| 155 | } else { |
| 156 | $info->href = str_contains( $hrefBits['title'], ':' ) |
| 157 | ? ':' . $hrefBits['title'] : $hrefBits['title']; |
| 158 | // Recurse! |
| 159 | $info = $this->getWikiLinkTargetInfo( $token, $info->href, $info->hrefSrc ); |
| 160 | $info->localprefix = $nsPrefix . |
| 161 | ( $info->localprefix ? ( ':' . $info->localprefix ) : '' ); |
| 162 | } |
| 163 | } elseif ( !empty( $interwikiInfo['url'] ) ) { |
| 164 | $info->href = $hrefBits['title']; |
| 165 | // Ensure a valid title and store it for later use. |
| 166 | // (don't store as $info->title because that signals a wikilink) |
| 167 | $interwikiInfo['title'] = $env->makeTitleFromURLDecodedStr( $title ); |
| 168 | // Interwiki or language link? If no language info, or if it starts |
| 169 | // with an explicit ':' (like [[:en:Foo]]), it's not a language link. |
| 170 | if ( $info->fromColonEscapedText || |
| 171 | ( !isset( $interwikiInfo['language'] ) && !isset( $interwikiInfo['extralanglink'] ) ) |
| 172 | ) { |
| 173 | // An interwiki link. |
| 174 | $info->interwiki = $interwikiInfo; |
| 175 | // Remove the colon escape after an interwiki prefix |
| 176 | if ( ( ltrim( $info->href )[0] ?? '' ) === ':' ) { |
| 177 | $info->href = substr( ltrim( $info->href ), 1 ); |
| 178 | } |
| 179 | } else { |
| 180 | // A language link. |
| 181 | $info->language = $interwikiInfo; |
| 182 | } |
| 183 | } else { |
| 184 | $info->title = $env->makeTitleFromURLDecodedStr( $title ); |
| 185 | } |
| 186 | } else { |
| 187 | $info->title = $env->makeTitleFromURLDecodedStr( $title ); |
| 188 | } |
| 189 | |
| 190 | return $info; |
| 191 | } |
| 192 | |
| 193 | /** |
| 194 | * Handle mw:redirect tokens |
| 195 | * |
| 196 | * @param Token $token |
| 197 | * @return array<string|Token> |
| 198 | * @throws InternalException |
| 199 | */ |
| 200 | private function onRedirect( Token $token ): array { |
| 201 | // Avoid duplicating the link-processing code by invoking the |
| 202 | // standard onWikiLink handler on the embedded link, intercepting |
| 203 | // the generated tokens using the callback mechanism, reading |
| 204 | // the href from the result, and then creating a |
| 205 | // <link rel="mw:PageProp/redirect"> token from it. |
| 206 | |
| 207 | $rlink = new SelfclosingTagTk( 'link', |
| 208 | Utils::cloneArray( $token->attribs ?? [] ), |
| 209 | clone $token->dataParsoid, |
| 210 | $token->dataMw ? clone $token->dataMw : null ); |
| 211 | $wikiLinkTk = $rlink->dataParsoid->linkTk; |
| 212 | $rlink->setAttribute( 'rel', 'mw:PageProp/redirect' ); |
| 213 | |
| 214 | // Remove the nested wikiLinkTk token and the cloned href attribute |
| 215 | unset( $rlink->dataParsoid->linkTk ); |
| 216 | $rlink->removeAttribute( 'href' ); |
| 217 | |
| 218 | // Transfer href attribute back to wikiLinkTk, since it may have been |
| 219 | // template-expanded in the pipeline prior to this point. |
| 220 | $wikiLinkTk->attribs = Utils::cloneArray( $token->attribs ?? [] ); |
| 221 | |
| 222 | // Set "redirect" attribute on the wikilink token to indicate that |
| 223 | // image and category links should be handled as plain links. |
| 224 | $wikiLinkTk->setAttribute( 'redirect', 'true' ); |
| 225 | |
| 226 | // Render the wikilink (including interwiki links, etc) then collect |
| 227 | // the resulting href and transfer it to rlink. |
| 228 | $r = $this->onWikiLink( $wikiLinkTk ); |
| 229 | $firstToken = ( $r[0] ?? null ); |
| 230 | $isValid = $firstToken instanceof XMLTagTk && |
| 231 | in_array( $firstToken->getName(), [ 'a', 'link' ], true ); |
| 232 | if ( $isValid ) { |
| 233 | $da = $r[0]->dataParsoid; |
| 234 | $rlink->addNormalizedAttribute( 'href', $da->a['href'], $da->sa['href'] ); |
| 235 | return [ $rlink ]; |
| 236 | } else { |
| 237 | // Bail! Emit tokens as if they were parsed as a list item: |
| 238 | // #REDIRECT.... |
| 239 | $src = $rlink->dataParsoid->src; |
| 240 | $tsr = $rlink->dataParsoid->tsr; |
| 241 | preg_match( '/^([^#]*)(#)/', $src, $srcMatch ); |
| 242 | $ntokens = strlen( $srcMatch[1] ) ? [ $srcMatch[1] ] : []; |
| 243 | $hashPos = $tsr->start + strlen( $srcMatch[1] ); |
| 244 | $tsr0 = new SourceRange( $hashPos, $hashPos + 1, $tsr->source ); |
| 245 | $dp = new DataParsoid; |
| 246 | $dp->tsr = $tsr0; |
| 247 | $li = new TagTk( |
| 248 | 'listItem', |
| 249 | [ new KV( 'bullets', [ '#' ], $tsr0->expandTsrV() ) ], |
| 250 | $dp ); |
| 251 | $ntokens[] = $li; |
| 252 | $ntokens[] = substr( $src, strlen( $srcMatch[0] ) ); |
| 253 | PHPUtils::pushArray( $ntokens, $r ); |
| 254 | return $ntokens; |
| 255 | } |
| 256 | } |
| 257 | |
| 258 | /** |
| 259 | * @return array<string|Token> |
| 260 | */ |
| 261 | public static function bailTokens( TokenHandlerPipeline $manager, Token $token ): array { |
| 262 | $frame = $manager->getFrame(); |
| 263 | $tsr = $token->dataParsoid->tsr; |
| 264 | $frameSrc = $frame->getSource(); |
| 265 | $linkSrc = $tsr->substr( $frameSrc ); |
| 266 | $src = substr( $linkSrc, 1 ); |
| 267 | if ( $src === '' ) { |
| 268 | $manager->getEnv()->log( |
| 269 | 'error', 'Unable to determine link source.', |
| 270 | "frame: ", $frameSrc->getSrcText(), |
| 271 | 'tsr: ', $tsr, |
| 272 | "link: $linkSrc" |
| 273 | ); |
| 274 | return [ $linkSrc ]; // Forget about trying to tokenize this |
| 275 | } |
| 276 | $startOffset = $tsr->start + 1; |
| 277 | $toks = PipelineUtils::processContentInPipeline( |
| 278 | $manager->getEnv(), $frame, $src, [ |
| 279 | // FIXME: Set toplevel when bailing |
| 280 | // 'toplevel' => $atTopLevel ?? false, |
| 281 | 'sol' => false, |
| 282 | 'pipelineType' => 'wikitext-to-expanded-tokens', |
| 283 | 'srcOffsets' => new SourceRange( $startOffset, $startOffset + strlen( $src ), $tsr->source ), |
| 284 | 'pipelineOpts' => [ |
| 285 | 'expandTemplates' => $manager->getOptions()['expandTemplates'], |
| 286 | 'inTemplate' => $manager->getOptions()['inTemplate'], |
| 287 | ], |
| 288 | ] |
| 289 | ); |
| 290 | TokenUtils::stripEOFTkFromTokens( $toks ); |
| 291 | return array_merge( [ '[' ], $toks ); |
| 292 | } |
| 293 | |
| 294 | /** |
| 295 | * Handle a mw:WikiLink token. |
| 296 | * |
| 297 | * @param Token $token |
| 298 | * @return array<string|Token> |
| 299 | * @throws InternalException |
| 300 | */ |
| 301 | private function onWikiLink( Token $token ): array { |
| 302 | $env = $this->env; |
| 303 | $tsr = $token->dataParsoid->tsr ?? null; |
| 304 | $tsrSource = $tsr->source ?? null; |
| 305 | $tsrStart = $tsr->start ?? null; |
| 306 | |
| 307 | // Check if we have cached output for this wikilink source. |
| 308 | // Given wikilink-syntax source, token output is deterministic |
| 309 | // and so we can benefit from caching. |
| 310 | $src = $token->dataParsoid->src ?? ''; |
| 311 | $isCacheable = ( $this->wikilinkCache && $tsrStart !== null && strlen( $src ) > 0 ); |
| 312 | if ( $isCacheable ) { |
| 313 | $cachedOutput = $this->wikilinkCache->lookup( $src ); |
| 314 | if ( $cachedOutput !== null ) { |
| 315 | $toks = $cachedOutput['value']['tokens']; |
| 316 | TokenUtils::shiftTokenTSR( |
| 317 | $toks, |
| 318 | $tsrStart - $cachedOutput['value']['start'], |
| 319 | $tsrSource === $cachedOutput['source'] ? null : $tsrSource |
| 320 | ); |
| 321 | TokenUtils::dedupeAboutIds( $env, $toks ); |
| 322 | return $toks; |
| 323 | } |
| 324 | } |
| 325 | |
| 326 | $hrefKV = $token->getAttributeKV( 'href' ); |
| 327 | $hrefTokenStr = TokenUtils::tokensToString( $hrefKV->v ); |
| 328 | |
| 329 | // Don't allow internal links to pages containing PROTO: |
| 330 | // See Parser::handleInternalLinks2() |
| 331 | if ( $env->getSiteConfig()->hasValidProtocol( $hrefTokenStr ) ) { |
| 332 | return self::bailTokens( $this->manager, $token ); |
| 333 | } |
| 334 | |
| 335 | // Xmlish tags in title position are invalid. Not according to the |
| 336 | // preprocessor ABNF but at later stages in the legacy parser, |
| 337 | // namely handleInternalLinks. |
| 338 | if ( is_array( $hrefKV->v ) ) { |
| 339 | // Use the expanded attr instead of trying to unpackDOMFragments |
| 340 | // since the fragment will have been released when expanding to DOM |
| 341 | $expandedDom = $token->fetchExpandedAttrValue( 'href' ) |
| 342 | ?? $env->getTopLevelDoc()->createDocumentFragment(); |
| 343 | foreach ( DOMCompat::querySelectorAll( $expandedDom, '[typeof]' ) as $el ) { |
| 344 | if ( DOMUtils::matchTypeOf( $el, '#^mw:(Nowiki|Extension|DOMFragment/sealed)#' ) !== null ) { |
| 345 | return self::bailTokens( $this->manager, $token ); |
| 346 | } |
| 347 | } |
| 348 | } |
| 349 | |
| 350 | // First check if the expanded href contains a pipe. |
| 351 | if ( str_contains( $hrefTokenStr, '|' ) ) { |
| 352 | // It does. This 'href' was templated and also returned other |
| 353 | // parameters separated by a pipe. We don't have any sensible way to |
| 354 | // handle such a construct currently, so prevent people from editing |
| 355 | // it. See T226523 |
| 356 | // TODO: add useful debugging info for editors ('if you would like to |
| 357 | // make this content editable, then fix template X..') |
| 358 | // TODO: also check other parameters for pipes! |
| 359 | // NOTE: We'd need to clear firstPipeSrc if this case gets supported |
| 360 | return self::bailTokens( $this->manager, $token ); |
| 361 | } |
| 362 | |
| 363 | try { |
| 364 | $target = $this->getWikiLinkTargetInfo( $token, $hrefTokenStr, $hrefKV->vsrc ); |
| 365 | } catch ( TitleException | InternalException ) { |
| 366 | // Invalid title |
| 367 | return self::bailTokens( $this->manager, $token ); |
| 368 | } |
| 369 | |
| 370 | // Ok, it looks like we have a sensible href. Figure out which handler to use. |
| 371 | $isRedirect = (bool)$token->getAttributeV( 'redirect' ); |
| 372 | $toks = $this->wikiLinkHandler( $token, $target, $isRedirect ); |
| 373 | if ( $isCacheable ) { |
| 374 | $this->wikilinkCache->cache( $src, [ 'start' => $tsrStart, 'tokens' => $toks ], $tsrSource ); |
| 375 | } |
| 376 | |
| 377 | return $toks; |
| 378 | } |
| 379 | |
| 380 | /** |
| 381 | * Figure out which handler to use to render a given WikiLink token. Override |
| 382 | * this method to add new handlers or swap out existing handlers based on the |
| 383 | * target structure. |
| 384 | * |
| 385 | * @param Token $token |
| 386 | * @param stdClass $target |
| 387 | * @param bool $isRedirect |
| 388 | * @return array<string|Token> |
| 389 | * @throws InternalException |
| 390 | */ |
| 391 | private function wikiLinkHandler( |
| 392 | Token $token, stdClass $target, bool $isRedirect |
| 393 | ): array { |
| 394 | $title = $target->title ?? null; |
| 395 | if ( $title ) { |
| 396 | if ( $isRedirect ) { |
| 397 | return $this->renderWikiLink( $token, $target ); |
| 398 | } |
| 399 | $siteConfig = $this->env->getSiteConfig(); |
| 400 | $nsId = $title->getNamespace(); |
| 401 | if ( $nsId === $siteConfig->canonicalNamespaceId( 'media' ) ) { |
| 402 | // Render as a media link. |
| 403 | return $this->renderMedia( $token, $target ); |
| 404 | } |
| 405 | if ( |
| 406 | !$target->fromColonEscapedText && |
| 407 | // Protect from purely fragment links on pages in these namespaces |
| 408 | ( $target->href[0] ?? '' ) !== '#' |
| 409 | ) { |
| 410 | if ( $nsId === $siteConfig->canonicalNamespaceId( 'file' ) ) { |
| 411 | // Render as a file. |
| 412 | return $this->renderFile( $token, $target ); |
| 413 | } |
| 414 | if ( $nsId === $siteConfig->canonicalNamespaceId( 'category' ) ) { |
| 415 | // Render as a category membership. |
| 416 | return $this->renderCategory( $token, $target ); |
| 417 | } |
| 418 | } |
| 419 | |
| 420 | // Render as plain wiki links. |
| 421 | return $this->renderWikiLink( $token, $target ); |
| 422 | } |
| 423 | |
| 424 | // language and interwiki links |
| 425 | if ( $target->interwiki ) { |
| 426 | return $this->renderInterwikiLink( $token, $target ); |
| 427 | } |
| 428 | if ( $target->language ) { |
| 429 | $ns = $this->env->getContextTitle()->getNamespace(); |
| 430 | $noLanguageLinks = $this->env->getSiteConfig()->namespaceIsTalk( $ns ) || |
| 431 | !$this->env->getSiteConfig()->interwikiMagic(); |
| 432 | if ( $noLanguageLinks ) { |
| 433 | $target->interwiki = $target->language; |
| 434 | return $this->renderInterwikiLink( $token, $target ); |
| 435 | } |
| 436 | |
| 437 | return $this->renderLanguageLink( $token, $target ); |
| 438 | } |
| 439 | |
| 440 | // Neither a title, nor a language or interwiki. Should not happen. |
| 441 | throw new InternalException( 'Unknown link type' ); |
| 442 | } |
| 443 | |
| 444 | /** ------------------------------------------------------------ |
| 445 | * This (overloaded) function does three different things: |
| 446 | * - Extracts link text from attrs (when k === "mw:maybeContent"). |
| 447 | * As a performance micro-opt, only does if asked to (getLinkText) |
| 448 | * - Updates existing rdfa type with an additional rdf-type, |
| 449 | * if one is provided (rdfaType) |
| 450 | * - Collates about, typeof, and linkAttrs into a new attr. array |
| 451 | * |
| 452 | * @param list<KV> $attrs |
| 453 | * @param bool $getLinkText |
| 454 | * @param ?string $rdfaType |
| 455 | * @param ?list<KV> $linkAttrs |
| 456 | * |
| 457 | * @return array{attribs: list<KV>, contentKVs: list<KV>, hasRdfaType: bool} |
| 458 | */ |
| 459 | public static function buildLinkAttrs( |
| 460 | array $attrs, bool $getLinkText, ?string $rdfaType, |
| 461 | ?array $linkAttrs |
| 462 | ): array { |
| 463 | $newAttrs = []; |
| 464 | $linkTextKVs = []; |
| 465 | $about = null; |
| 466 | |
| 467 | // In one pass through the attribute array, fetch about, typeof, and linkText |
| 468 | // |
| 469 | // about && typeof are usually at the end of the array if at all present |
| 470 | foreach ( $attrs as $kv ) { |
| 471 | $k = $kv->k; |
| 472 | $v = $kv->v; |
| 473 | |
| 474 | // link-text attrs have the key "maybeContent" |
| 475 | if ( $getLinkText && $k === 'mw:maybeContent' ) { |
| 476 | $linkTextKVs[] = $kv; |
| 477 | } elseif ( is_string( $k ) && $k ) { |
| 478 | if ( trim( $k ) === 'typeof' ) { |
| 479 | $rdfaType = $rdfaType ? $rdfaType . ' ' . $v : $v; |
| 480 | } elseif ( trim( $k ) === 'about' ) { |
| 481 | $about = $v; |
| 482 | } |
| 483 | } |
| 484 | } |
| 485 | |
| 486 | if ( $rdfaType ) { |
| 487 | $newAttrs[] = new KV( 'typeof', $rdfaType ); |
| 488 | } |
| 489 | |
| 490 | if ( $about ) { |
| 491 | $newAttrs[] = new KV( 'about', $about ); |
| 492 | } |
| 493 | |
| 494 | if ( $linkAttrs ) { |
| 495 | PHPUtils::pushArray( $newAttrs, $linkAttrs ); |
| 496 | } |
| 497 | |
| 498 | return [ |
| 499 | 'attribs' => $newAttrs, |
| 500 | 'contentKVs' => $linkTextKVs, |
| 501 | 'hasRdfaType' => $rdfaType !== null |
| 502 | ]; |
| 503 | } |
| 504 | |
| 505 | /** |
| 506 | * Generic wiki link attribute setup on a passed-in new token based on the |
| 507 | * wikilink token and target. As a side effect, this method also extracts the |
| 508 | * link content tokens and returns them. |
| 509 | * |
| 510 | * @param Token $newTk |
| 511 | * @param Token $token |
| 512 | * @param stdClass $target |
| 513 | * @param bool $buildDOMFragment |
| 514 | * @return array<string|Token> |
| 515 | * @throws InternalException |
| 516 | */ |
| 517 | private function addLinkAttributesAndGetContent( |
| 518 | Token $newTk, Token $token, stdClass $target, bool $buildDOMFragment = false |
| 519 | ): array { |
| 520 | $attribs = $token->attribs; |
| 521 | $dataParsoid = $token->dataParsoid; |
| 522 | $dataMw = $token->dataMw; |
| 523 | $newAttrData = self::buildLinkAttrs( $attribs, true, null, [ new KV( 'rel', 'mw:WikiLink' ) ] ); |
| 524 | $content = $newAttrData['contentKVs']; |
| 525 | $env = $this->env; |
| 526 | |
| 527 | // Set attribs and dataParsoid |
| 528 | $newTk->attribs = $newAttrData['attribs']; |
| 529 | $newTk->dataParsoid = clone $dataParsoid; |
| 530 | $newTk->dataMw = $dataMw !== null ? clone $dataMw : null; |
| 531 | unset( $newTk->dataParsoid->src ); // clear src string since we can serialize this |
| 532 | |
| 533 | // Note: Link tails are handled on the DOM in handleLinkNeighbours, so no |
| 534 | // need to handle them here. |
| 535 | $l = count( $content ); |
| 536 | if ( $l > 0 ) { |
| 537 | $newTk->dataParsoid->stx = 'piped'; |
| 538 | $out = []; |
| 539 | // re-join content bits |
| 540 | foreach ( $content as $i => $kv ) { |
| 541 | $toks = $kv->v; |
| 542 | // since this is already a link, strip autolinks from content |
| 543 | // FIXME: Maybe add a stop in the grammar so that autolinks |
| 544 | // aren't tokenized in link content to begin with? |
| 545 | if ( !is_array( $toks ) ) { |
| 546 | $toks = [ $toks ]; |
| 547 | } |
| 548 | |
| 549 | $toks = array_values( array_filter( $toks, static function ( $t ) { |
| 550 | return $t !== ''; |
| 551 | } ) ); |
| 552 | $n = count( $toks ); |
| 553 | foreach ( $toks as $j => $t ) { |
| 554 | // Bail on media-syntax in wikilink-syntax scenarios, |
| 555 | // since the legacy parser explodes on [[, last one wins. |
| 556 | // Note that without this, anchors tags in media output |
| 557 | // will be stripped and we won't have the right structure |
| 558 | // when we get to the dom pass to add media info. |
| 559 | if ( |
| 560 | $t instanceof TagTk && |
| 561 | ( $t->getName() === 'figure' || $t->getName() === 'span' ) && |
| 562 | TokenUtils::matchTypeOf( $t, '#^mw:File($|/)#D' ) !== null |
| 563 | ) { |
| 564 | throw new InternalException( 'Media-in-link' ); |
| 565 | } |
| 566 | |
| 567 | if ( $t instanceof TagTk && $t->getName() === 'a' ) { |
| 568 | // Bail on wikilink-syntax in wiklink-syntax scenarios, |
| 569 | // since the legacy parser explodes on [[, last one wins |
| 570 | if ( |
| 571 | preg_match( |
| 572 | '#^mw:WikiLink(/Interwiki)?$#D', |
| 573 | $t->getAttributeV( 'rel' ) ?? '' |
| 574 | ) && |
| 575 | // ISBN links don't use wikilink-syntax but still |
| 576 | // get the same "rel", so should be ignored |
| 577 | ( $t->dataParsoid->stx ?? '' ) !== 'magiclink' |
| 578 | ) { |
| 579 | throw new InternalException( 'Link-in-link' ); |
| 580 | } |
| 581 | if ( $j + 1 < $n && $toks[$j + 1] instanceof EndTagTk && |
| 582 | $toks[$j + 1]->getName() === 'a' |
| 583 | ) { |
| 584 | // autonumbered links in the stream get rendered |
| 585 | // as an <a> tag with no content -- but these ought |
| 586 | // to be treated as plaintext since we don't allow |
| 587 | // nested links. |
| 588 | $out[] = '[' . $t->getAttributeV( 'href' ) . ']'; |
| 589 | } |
| 590 | // suppress <a> |
| 591 | continue; |
| 592 | } |
| 593 | |
| 594 | // Categories also use wikilink syntax so we bail to match |
| 595 | // legacy output. However, this isn't an a-in-a scenario |
| 596 | // so maybe should be permitted in the future. |
| 597 | if ( |
| 598 | $t instanceof SelfclosingTagTk && $t->getName() === 'link' && |
| 599 | preg_match( |
| 600 | '#^mw:PageProp/Category$#D', |
| 601 | $t->getAttributeV( 'rel' ) ?? '' |
| 602 | ) |
| 603 | ) { |
| 604 | throw new InternalException( 'Category-in-link' ); |
| 605 | } |
| 606 | |
| 607 | if ( $t instanceof EndTagTk && $t->getName() === 'a' ) { |
| 608 | continue; // suppress </a> |
| 609 | } |
| 610 | |
| 611 | $out[] = $t; |
| 612 | } |
| 613 | if ( $i < $l - 1 ) { |
| 614 | $out[] = '|'; |
| 615 | } |
| 616 | } |
| 617 | |
| 618 | if ( $buildDOMFragment ) { |
| 619 | // content = [part 0, .. part l-1] |
| 620 | // offsets = [start(part-0), end(part l-1)] |
| 621 | $offsets = isset( $dataParsoid->tsr ) ? |
| 622 | new SourceRange( |
| 623 | $content[0]->srcOffsets->value->start, |
| 624 | $content[$l - 1]->srcOffsets->value->end, |
| 625 | $dataParsoid->tsr->source ) : |
| 626 | null; |
| 627 | $content = [ PipelineUtils::getDOMFragmentToken( $out, $offsets, |
| 628 | [ 'inlineContext' => true, 'token' => $token ] ) ]; |
| 629 | } else { |
| 630 | $content = $out; |
| 631 | } |
| 632 | } else { |
| 633 | $newTk->dataParsoid->stx = 'simple'; |
| 634 | $morecontent = Utils::decodeURIComponent( $target->href ); |
| 635 | |
| 636 | // Try to match labeling in core |
| 637 | if ( $env->getSiteConfig()->namespaceHasSubpages( |
| 638 | $env->getContextTitle()->getNamespace() |
| 639 | ) ) { |
| 640 | // subpage links with a trailing slash get the trailing slashes stripped. |
| 641 | // See https://gerrit.wikimedia.org/r/173431 |
| 642 | if ( preg_match( '#^((\.\./)+|/)(?!\.\./)(.*?[^/])/+$#D', $morecontent, $match ) ) { |
| 643 | $morecontent = $match[3]; |
| 644 | } elseif ( str_starts_with( $morecontent, '../' ) ) { |
| 645 | // Subpages on interwiki / language links aren't valid, |
| 646 | // so $target->title should always be present here |
| 647 | $morecontent = $target->title->getPrefixedText(); |
| 648 | } |
| 649 | } |
| 650 | |
| 651 | // for interwiki links, include the interwiki prefix in the link text |
| 652 | if ( $target->interwiki ) { |
| 653 | $morecontent = $target->prefix . ':' . $morecontent; |
| 654 | } |
| 655 | |
| 656 | // for local links, include the local prefix in the link text |
| 657 | if ( $target->localprefix ) { |
| 658 | $morecontent = $target->localprefix . ':' . $morecontent; |
| 659 | } |
| 660 | |
| 661 | $content = [ $morecontent ]; |
| 662 | } |
| 663 | return $content; |
| 664 | } |
| 665 | |
| 666 | /** |
| 667 | * Render a plain wiki link. |
| 668 | * |
| 669 | * @param Token $token |
| 670 | * @param stdClass $target |
| 671 | * @return array<string|Token> |
| 672 | */ |
| 673 | private function renderWikiLink( Token $token, stdClass $target ): array { |
| 674 | $newTk = new TagTk( 'a' ); |
| 675 | try { |
| 676 | $content = $this->addLinkAttributesAndGetContent( $newTk, $token, $target, true ); |
| 677 | } catch ( InternalException ) { |
| 678 | return self::bailTokens( $this->manager, $token ); |
| 679 | } |
| 680 | |
| 681 | $newTk->addNormalizedAttribute( 'href', $this->env->makeLink( $target->title ), |
| 682 | $target->hrefSrc ); |
| 683 | |
| 684 | $newTk->setAttribute( 'title', $target->title->getPrefixedText() ); |
| 685 | |
| 686 | return array_merge( [ $newTk ], $content, [ new EndTagTk( 'a' ) ] ); |
| 687 | } |
| 688 | |
| 689 | /** |
| 690 | * Render a category 'link'. Categories are really page properties, and are |
| 691 | * normally rendered in a box at the bottom of an article. |
| 692 | * |
| 693 | * @param Token $token |
| 694 | * @param stdClass $target |
| 695 | * @return array<string|Token> |
| 696 | */ |
| 697 | private function renderCategory( Token $token, stdClass $target ): array { |
| 698 | $newTk = new SelfclosingTagTk( 'link' ); |
| 699 | try { |
| 700 | $content = $this->addLinkAttributesAndGetContent( $newTk, $token, $target ); |
| 701 | } catch ( InternalException ) { |
| 702 | return self::bailTokens( $this->manager, $token ); |
| 703 | } |
| 704 | $env = $this->env; |
| 705 | |
| 706 | // Change the rel to be mw:PageProp/Category |
| 707 | $newTk->getAttributeKV( 'rel' )->v = 'mw:PageProp/Category'; |
| 708 | |
| 709 | $newTk->addNormalizedAttribute( 'href', $env->makeLink( $target->title ), $target->hrefSrc ); |
| 710 | |
| 711 | // Change the href to include the sort key, if any (but don't update the rt info) |
| 712 | // Fallback to empty string for default sorting |
| 713 | $categorySort = ''; |
| 714 | $strContent = str_replace( "\n", '', TokenUtils::tokensToString( $content ) ); |
| 715 | if ( $strContent !== '' && $strContent !== $target->href ) { |
| 716 | $categorySort = $strContent; |
| 717 | $hrefkv = $newTk->getAttributeKV( 'href' ); |
| 718 | $hrefkv->v .= '#'; |
| 719 | $hrefkv->v .= str_replace( '#', '%23', Sanitizer::sanitizeTitleURI( $categorySort, false ) ); |
| 720 | } |
| 721 | |
| 722 | if ( count( $content ) !== 1 ) { |
| 723 | // Deal with sort keys that come from generated content (transclusions, etc.) |
| 724 | $key = [ 'txt' => 'mw:sortKey' ]; |
| 725 | $contentKV = $token->getAttributeKV( 'mw:maybeContent' ); |
| 726 | $so = $contentKV->valueOffset(); |
| 727 | $val = PipelineUtils::expandAttrValueToDOM( |
| 728 | $this->env, |
| 729 | $this->manager->getFrame(), |
| 730 | [ 'html' => $content, 'srcOffsets' => $so ], |
| 731 | $this->options['expandTemplates'], |
| 732 | $this->options['inTemplate'] |
| 733 | ); |
| 734 | $attr = new DataMwAttrib( $key, $val ); |
| 735 | $dataMw = $newTk->dataMw; |
| 736 | if ( $dataMw ) { |
| 737 | $dataMw->attribs[] = $attr; |
| 738 | } else { |
| 739 | $dataMw = new DataMw( [ 'attribs' => [ $attr ] ] ); |
| 740 | } |
| 741 | |
| 742 | // Mark token as having expanded attrs |
| 743 | $newTk->addAttribute( 'about', $env->newAboutId() ); |
| 744 | $newTk->addSpaceSeparatedAttribute( 'typeof', 'mw:ExpandedAttrs' ); |
| 745 | $newTk->dataMw = $dataMw; |
| 746 | } |
| 747 | $this->env->getMetadata()->addCategory( $target->title, $categorySort ); |
| 748 | return [ $newTk ]; |
| 749 | } |
| 750 | |
| 751 | /** |
| 752 | * Render a language link. Those normally appear in the list of alternate |
| 753 | * languages for an article in the sidebar, so are really a page property. |
| 754 | * |
| 755 | * @param Token $token |
| 756 | * @param stdClass $target |
| 757 | * @return array<string|Token> |
| 758 | */ |
| 759 | private function renderLanguageLink( Token $token, stdClass $target ): array { |
| 760 | // The prefix is listed in the interwiki map |
| 761 | |
| 762 | // TODO: If $target->language['deprecated'] is set and |
| 763 | // $target->language['extralanglink'] is *not* set, then we |
| 764 | // should use the normalized language name/prefix (from |
| 765 | // 'deprecated') when calling |
| 766 | // ContentMetadataCollector::addLanguageLink() here (which |
| 767 | // we should eventualy be doing) |
| 768 | |
| 769 | // TODO: might also want to add the language *code* here, |
| 770 | // which would be the language['bcp47'] property (added in |
| 771 | // change I82465261bc66f0b0cd30d361c299f08066494762) for an |
| 772 | // extralanglink, or the interwiki prefix otherwise; the |
| 773 | // latter is mediawiki-internal and maybe not BCP-47 compliant. |
| 774 | // This is for clients of the MediaWiki DOM spec HTML: the |
| 775 | // WMF domain prefix, the MediaWiki internal language code, |
| 776 | // and the actual *language* (ie bcp-47 code) can all differ |
| 777 | // from each other, due to various historical infelicities. |
| 778 | // Perhaps a `lang` attribute on the `link` would be appropriate. |
| 779 | |
| 780 | $newTk = new SelfclosingTagTk( 'link', [], $token->dataParsoid ); |
| 781 | try { |
| 782 | $this->addLinkAttributesAndGetContent( $newTk, $token, $target ); |
| 783 | } catch ( InternalException ) { |
| 784 | return self::bailTokens( $this->manager, $token ); |
| 785 | } |
| 786 | |
| 787 | // add title attribute giving the presentation name of the |
| 788 | // "extra language link" |
| 789 | // T329303: the 'linktext' comes from the system message |
| 790 | // `interlanguage-link-$prefix` and should be set in integrated mode |
| 791 | // using the localization features; the integrated-mode SiteConfig |
| 792 | // currently never sets the `linktext` property in |
| 793 | // SiteConfig::interwikiMap(). |
| 794 | // I52d50e2f75942a849908c6be7fc5169f00a5983a has some partial work |
| 795 | // on this. |
| 796 | if ( isset( $target->language['extralanglink'] ) && |
| 797 | !empty( $target->language['linktext'] ) |
| 798 | ) { |
| 799 | // XXX in standalone mode, this is user-interface-language text, |
| 800 | // not "content language" text. |
| 801 | $newTk->addNormalizedAttribute( 'title', $target->language['linktext'], null ); |
| 802 | } |
| 803 | |
| 804 | // We set an absolute link to the article in the other wiki/language |
| 805 | $title = Sanitizer::sanitizeTitleURI( Utils::decodeURIComponent( $target->href ), false ); |
| 806 | $absHref = str_replace( '$1', $title, $target->language['url'] ); |
| 807 | if ( isset( $target->language['protorel'] ) ) { |
| 808 | $absHref = preg_replace( '/^https?:/', '', $absHref, 1 ); |
| 809 | } |
| 810 | $newTk->addNormalizedAttribute( 'href', $absHref, $target->hrefSrc ); |
| 811 | |
| 812 | // Change the rel to be mw:PageProp/Language |
| 813 | $newTk->getAttributeKV( 'rel' )->v = 'mw:PageProp/Language'; |
| 814 | |
| 815 | // Add language link(s) to metadata |
| 816 | $this->env->getMetadata()->addLanguageLink( $target->language['title'] ); |
| 817 | |
| 818 | return [ $newTk ]; |
| 819 | } |
| 820 | |
| 821 | /** |
| 822 | * Render an interwiki link. |
| 823 | * |
| 824 | * @param Token $token |
| 825 | * @param stdClass $target |
| 826 | * @return array<string|Token> |
| 827 | */ |
| 828 | private function renderInterwikiLink( Token $token, stdClass $target ): array { |
| 829 | // The prefix is listed in the interwiki map |
| 830 | |
| 831 | $tokens = []; |
| 832 | $newTk = new TagTk( 'a', [], $token->dataParsoid ); |
| 833 | try { |
| 834 | $content = $this->addLinkAttributesAndGetContent( $newTk, $token, $target, true ); |
| 835 | } catch ( InternalException ) { |
| 836 | return self::bailTokens( $this->manager, $token ); |
| 837 | } |
| 838 | |
| 839 | // We set an absolute link to the article in the other wiki/language |
| 840 | $isLocal = !empty( $target->interwiki['local'] ); |
| 841 | $trimmedHref = trim( $target->href ); |
| 842 | $title = Sanitizer::sanitizeTitleURI( |
| 843 | Utils::decodeURIComponent( $trimmedHref ), |
| 844 | !$isLocal |
| 845 | ); |
| 846 | $absHref = str_replace( '$1', $title, $target->interwiki['url'] ); |
| 847 | if ( isset( $target->interwiki['protorel'] ) ) { |
| 848 | $absHref = preg_replace( '/^https?:/', '', $absHref, 1 ); |
| 849 | } |
| 850 | $newTk->addNormalizedAttribute( 'href', $absHref, $target->hrefSrc ); |
| 851 | |
| 852 | $newTk->getAttributeKV( 'rel' )->v = 'mw:WikiLink/Interwiki'; |
| 853 | |
| 854 | // Add title unless it's just a fragment (and trim off fragment) |
| 855 | // (The normalization here is similar to what Title#getPrefixedDBKey() does.) |
| 856 | if ( $target->href === '' || $target->href[0] !== '#' ) { |
| 857 | $titleAttr = $target->interwiki['prefix'] . ':' . |
| 858 | Utils::decodeURIComponent( str_replace( '_', ' ', |
| 859 | preg_replace( '/#.*/s', '', $trimmedHref, 1 ) ) ); |
| 860 | $newTk->setAttribute( 'title', $titleAttr ); |
| 861 | } |
| 862 | $tokens[] = $newTk; |
| 863 | |
| 864 | PHPUtils::pushArray( $tokens, $content ); |
| 865 | $tokens[] = new EndTagTk( 'a' ); |
| 866 | return $tokens; |
| 867 | } |
| 868 | |
| 869 | private const HORIZONTAL_ALIGNS = [ |
| 870 | // PHP parser wraps in <div class="floatnone"> |
| 871 | 'left', |
| 872 | // PHP parser wraps in <div class="center"><div class="floatnone"> |
| 873 | 'right', |
| 874 | // PHP parser wraps in <div class="floatleft"> |
| 875 | 'center', |
| 876 | // PHP parser wraps in <div class="floatright"> |
| 877 | 'none', |
| 878 | ]; |
| 879 | private const VERTICAL_ALIGNS = [ 'baseline', 'sub', 'super', 'top', 'text_top', 'middle', |
| 880 | 'bottom', 'text_bottom' ]; |
| 881 | |
| 882 | /** |
| 883 | * Get the style and class lists for an image's wrapper element. |
| 884 | * |
| 885 | * @param array $opts The option hash from renderFile. |
| 886 | * |
| 887 | * @return array{classes: list<string>, isInline: bool} |
| 888 | * - isInline: Whether the image is inline after handling options |
| 889 | * - classes: The list of classes for the wrapper. |
| 890 | */ |
| 891 | private static function getWrapperInfo( array $opts ): array { |
| 892 | $format = self::getFormat( $opts ); |
| 893 | $isInline = !in_array( $format, [ 'thumbnail', 'manualthumb', 'framed' ], true ); |
| 894 | $classes = []; |
| 895 | |
| 896 | if ( |
| 897 | !isset( $opts['size']['src'] ) && |
| 898 | // Framed and manualthumb images aren't scaled |
| 899 | !in_array( $format, [ 'manualthumb', 'framed' ], true ) |
| 900 | ) { |
| 901 | $classes[] = 'mw-default-size'; |
| 902 | } |
| 903 | |
| 904 | // Border isn't applicable to 'thumbnail', 'manualthumb', or 'framed' formats |
| 905 | // Using $isInline as a shorthand for that here (see above), |
| 906 | // but this isn't about being *inline* per se |
| 907 | if ( $isInline && isset( $opts['border'] ) ) { |
| 908 | $classes[] = 'mw-image-border'; |
| 909 | } |
| 910 | |
| 911 | $halign = $opts['halign']['v'] ?? null; |
| 912 | if ( in_array( $halign, self::HORIZONTAL_ALIGNS, true ) ) { |
| 913 | $isInline = false; |
| 914 | $classes[] = "mw-halign-$halign"; |
| 915 | } |
| 916 | |
| 917 | if ( $isInline ) { |
| 918 | $valignOpt = $opts['valign']['v'] ?? null; |
| 919 | if ( in_array( $valignOpt, self::VERTICAL_ALIGNS, true ) ) { |
| 920 | $classes[] = str_replace( '_', '-', "mw-valign-$valignOpt" ); |
| 921 | } |
| 922 | } |
| 923 | |
| 924 | return [ 'classes' => $classes, 'isInline' => $isInline ]; |
| 925 | } |
| 926 | |
| 927 | /** |
| 928 | * Determine the name of an option. |
| 929 | * |
| 930 | * @param string $optStr |
| 931 | * @param Env $env |
| 932 | * @return ?array{ck: string, v: string, ak: string, s: bool} |
| 933 | * - ck: Canonical key for the image option. |
| 934 | * - v: Value of the option. |
| 935 | * - ak: Aliased key for the image option; includes `"$1"` for placeholder. |
| 936 | * - s: Whether it's a simple option or one with a value. |
| 937 | */ |
| 938 | private static function getOptionInfo( string $optStr, Env $env ): ?array { |
| 939 | $oText = trim( $optStr ); |
| 940 | $siteConfig = $env->getSiteConfig(); |
| 941 | $getOption = $siteConfig->getMediaPrefixParameterizedAliasMatcher(); |
| 942 | // oText contains the localized name of this option. the |
| 943 | // canonical option names (from mediawiki upstream) are in |
| 944 | // English and contain an '(img|timedmedia)_' prefix. We drop the |
| 945 | // prefix before stuffing them in data-parsoid in order to |
| 946 | // save space (that's shortCanonicalOption) |
| 947 | $canonicalOption = $siteConfig->getMagicWordForMediaOption( $oText ) ?? ''; |
| 948 | $shortCanonicalOption = preg_replace( '/^(img|timedmedia)_/', '', $canonicalOption, 1 ); |
| 949 | // 'imgOption' is the key we'd put in opts; it names the 'group' |
| 950 | // for the option, and doesn't have an img_ prefix. |
| 951 | $imgOption = Consts::$Media['SimpleOptions'][$canonicalOption] ?? null; |
| 952 | if ( $imgOption !== null ) { |
| 953 | return [ |
| 954 | 'ck' => $imgOption, |
| 955 | 'v' => $shortCanonicalOption, |
| 956 | 'ak' => $optStr, |
| 957 | 's' => true |
| 958 | ]; |
| 959 | } |
| 960 | // If there isn't a literal match for the option, look for a |
| 961 | // prefix match (ie, img_width => `$1px`) |
| 962 | |
| 963 | // *Note* that the legacy parser doesn't have a "principled" |
| 964 | // precedence here (T372935), it just so happens that members |
| 965 | // of Consts::PrefixOptions like |
| 966 | // img_width/img_page/img_lang/timedmedia_* are added last (as |
| 967 | // handler parameters), and other prefixed options like |
| 968 | // img_link/img_alt/img_class *happen* to be last in the |
| 969 | // $internalParamMap. But the possibility for conflicts |
| 970 | // between prefixed parameters and literal options still |
| 971 | // exists in the legacy parser. |
| 972 | $bits = $getOption( $oText ); |
| 973 | $normalizedBit0 = $bits ? mb_strtolower( trim( $bits['k'] ) ) : null; |
| 974 | $key = $bits ? ( Consts::$Media['PrefixOptions'][$normalizedBit0] ?? null ) : null; |
| 975 | |
| 976 | // bits.a *used to have* the localized name for the prefix option |
| 977 | // (see SiteConfig::getMediaPrefixParameterizedAliasMatcher, this was |
| 978 | // dropped in the port from JS.) |
| 979 | // with $1 as a placeholder for the value, which is in bits.v |
| 980 | // 'normalizedBit0' is the canonical English option name |
| 981 | // (from mediawiki upstream) with a prefix. |
| 982 | // 'key' is the parsoid 'group' for the option; it doesn't |
| 983 | // have a prefix (it's the key we'd put in opts) |
| 984 | if ( $bits && $key ) { |
| 985 | $shortCanonicalOption = preg_replace( '/^(img|timedmedia)_/', '', $normalizedBit0, 1 ); |
| 986 | // map short canonical name to the localized version used |
| 987 | |
| 988 | // Note that we deliberately do entity decoding |
| 989 | // *after* splitting so that HTML-encoded pipes don't |
| 990 | // separate options. This matches PHP, whether or |
| 991 | // not it's a good idea. |
| 992 | return [ |
| 993 | 'ck' => $shortCanonicalOption, |
| 994 | 'v' => Utils::decodeWtEntities( $bits['v'] ), |
| 995 | 'ak' => $optStr, |
| 996 | 's' => false |
| 997 | ]; |
| 998 | } |
| 999 | |
| 1000 | return null; |
| 1001 | } |
| 1002 | |
| 1003 | private static function isWikitextOpt( |
| 1004 | Env $env, ?array &$optInfo, string $prefix, string $resultStr |
| 1005 | ): bool { |
| 1006 | // link and alt options are allowed to contain arbitrary |
| 1007 | // wikitext (even though only strings are supported in reality) |
| 1008 | // FIXME(SSS): Is this actually true of all options rather than |
| 1009 | // just link and alt? |
| 1010 | if ( $optInfo === null ) { |
| 1011 | $optInfo = self::getOptionInfo( $prefix . $resultStr, $env ); |
| 1012 | } |
| 1013 | return $optInfo !== null && in_array( $optInfo['ck'], [ 'link', 'alt' ], true ); |
| 1014 | } |
| 1015 | |
| 1016 | /** |
| 1017 | * Make option token streams into a stringy thing that we can recognize. |
| 1018 | * |
| 1019 | * @param array $tstream |
| 1020 | * @param string $prefix Anything that came before this part of the recursive call stack. |
| 1021 | * @param Env $env |
| 1022 | * @return ?string |
| 1023 | */ |
| 1024 | private static function stringifyOptionTokens( array $tstream, string $prefix, Env $env ): ?string { |
| 1025 | // Seems like this should be a more general "stripTags"-like function? |
| 1026 | $skipToEndOf = null; |
| 1027 | $optInfo = null; |
| 1028 | $resultStr = ''; |
| 1029 | |
| 1030 | for ( $i = 0; $i < count( $tstream ); $i++ ) { |
| 1031 | $currentToken = $tstream[$i]; |
| 1032 | |
| 1033 | if ( $skipToEndOf ) { |
| 1034 | if ( $currentToken instanceof EndTagTk && $currentToken->getName() === $skipToEndOf ) { |
| 1035 | $skipToEndOf = null; |
| 1036 | } |
| 1037 | continue; |
| 1038 | } |
| 1039 | |
| 1040 | if ( is_string( $currentToken ) ) { |
| 1041 | $resultStr .= $currentToken; |
| 1042 | } elseif ( is_array( $currentToken ) ) { |
| 1043 | $nextResult = self::stringifyOptionTokens( $currentToken, $prefix . $resultStr, $env ); |
| 1044 | |
| 1045 | if ( $nextResult === null ) { |
| 1046 | return null; |
| 1047 | } |
| 1048 | |
| 1049 | $resultStr .= $nextResult; |
| 1050 | } elseif ( |
| 1051 | ( $currentToken instanceof XMLTagTk ) && |
| 1052 | !( $currentToken instanceof EndTagTk ) |
| 1053 | ) { |
| 1054 | // This is actually a token |
| 1055 | if ( TokenUtils::hasDOMFragmentType( $currentToken ) ) { |
| 1056 | if ( self::isWikitextOpt( $env, $optInfo, $prefix, $resultStr ) ) { |
| 1057 | $str = TokenUtils::tokensToString( [ $currentToken ], false, [ |
| 1058 | // These tokens haven't been expanded to DOM yet |
| 1059 | // so unpacking them here is justifiable |
| 1060 | // FIXME: It's a little convoluted to figure out |
| 1061 | // that this is actually the case in the |
| 1062 | // AttributeExpander, but it seems like only |
| 1063 | // target/href ever gets expanded to DOM and |
| 1064 | // the rest of the wikilink_content/options |
| 1065 | // become mw:maybeContent that gets expanded |
| 1066 | // below where $hasExpandableOpt is set. |
| 1067 | 'unpackDOMFragments' => true, |
| 1068 | ] |
| 1069 | ); |
| 1070 | // Entity encode pipes since we wouldn't have split on |
| 1071 | // them from fragments and we're about to attempt to |
| 1072 | // when this function returns. |
| 1073 | $resultStr .= str_replace( '|', '|', $str ); |
| 1074 | $optInfo = null; // might change the nature of opt |
| 1075 | continue; |
| 1076 | } else { |
| 1077 | // if this is a nowiki, we must be in a caption |
| 1078 | return null; |
| 1079 | } |
| 1080 | } |
| 1081 | if ( $currentToken->getName() === 'mw-quote' ) { |
| 1082 | if ( self::isWikitextOpt( $env, $optInfo, $prefix, $resultStr ) ) { |
| 1083 | // just recurse inside |
| 1084 | $optInfo = null; // might change the nature of opt |
| 1085 | continue; |
| 1086 | } |
| 1087 | return null; |
| 1088 | } |
| 1089 | // Similar to TokenUtils.tokensToString()'s includeEntities |
| 1090 | if ( TokenUtils::isEntitySpanToken( $currentToken ) ) { |
| 1091 | $resultStr .= $currentToken->dataParsoid->src; |
| 1092 | $skipToEndOf = 'span'; |
| 1093 | continue; |
| 1094 | } |
| 1095 | if ( $currentToken->getName() === 'a' ) { |
| 1096 | if ( $optInfo === null ) { |
| 1097 | $optInfo = self::getOptionInfo( $prefix . $resultStr, $env ); |
| 1098 | if ( $optInfo === null ) { |
| 1099 | // An <a> tag before a valid option? |
| 1100 | // This is most likely a caption. |
| 1101 | return null; |
| 1102 | } |
| 1103 | } |
| 1104 | |
| 1105 | if ( self::isWikitextOpt( $env, $optInfo, $prefix, $resultStr ) ) { |
| 1106 | $tokenType = $currentToken->getAttributeV( 'rel' ); |
| 1107 | $isLink = $optInfo && $optInfo['ck'] === 'link'; |
| 1108 | // Reset the optInfo since we're changing the nature of it |
| 1109 | $optInfo = null; |
| 1110 | // Figure out the proper string to put here and break. |
| 1111 | if ( |
| 1112 | $tokenType === 'mw:ExtLink' && |
| 1113 | ( $currentToken->dataParsoid->stx ?? '' ) === 'url' |
| 1114 | ) { |
| 1115 | // Add the URL and entity encode any pipes |
| 1116 | // If the pipes are from entity decoding in the href, |
| 1117 | // we don't want to split media options on them |
| 1118 | // If the pipes are from template expansion, legacy |
| 1119 | // would have considered them delineating a media option |
| 1120 | // but let's not support that combination with the url |
| 1121 | $resultStr .= str_replace( |
| 1122 | '|', '|', $currentToken->getAttributeV( 'href' ) |
| 1123 | ); |
| 1124 | // Tell our loop to skip to the end of this tag |
| 1125 | $skipToEndOf = 'a'; |
| 1126 | } elseif ( $tokenType === 'mw:WikiLink/Interwiki' ) { |
| 1127 | if ( $isLink ) { |
| 1128 | $resultStr .= $currentToken->getAttributeV( 'href' ); |
| 1129 | $i += 2; |
| 1130 | continue; |
| 1131 | } |
| 1132 | // Nothing to do -- the link content will be |
| 1133 | // captured by walking the rest of the tokens. |
| 1134 | } elseif ( $tokenType === 'mw:WikiLink' || $tokenType === 'mw:MediaLink' ) { |
| 1135 | |
| 1136 | // Nothing to do -- the link content will be |
| 1137 | // captured by walking the rest of the tokens. |
| 1138 | } else { |
| 1139 | // There shouldn't be any other kind of link... |
| 1140 | // This is likely a caption. |
| 1141 | return null; |
| 1142 | } |
| 1143 | } else { |
| 1144 | // Why would there be an a tag without a link? |
| 1145 | return null; |
| 1146 | } |
| 1147 | } |
| 1148 | } |
| 1149 | } |
| 1150 | |
| 1151 | return $resultStr; |
| 1152 | } |
| 1153 | |
| 1154 | /** |
| 1155 | * Get the format for media. |
| 1156 | * |
| 1157 | * @param array $opts |
| 1158 | * @return string|null |
| 1159 | */ |
| 1160 | private static function getFormat( array $opts ): ?string { |
| 1161 | if ( $opts['manualthumb'] ) { |
| 1162 | return 'manualthumb'; |
| 1163 | } |
| 1164 | return $opts['format']['v'] ?? null; |
| 1165 | } |
| 1166 | |
| 1167 | private array $used = []; |
| 1168 | |
| 1169 | /** |
| 1170 | * This is the set of file options that apply to the container, rather |
| 1171 | * than the media element itself (or, apply generically to a span). |
| 1172 | * Other options depend on the fetched media type and won't necessary be |
| 1173 | * applied. |
| 1174 | * |
| 1175 | * @return array |
| 1176 | */ |
| 1177 | private function getUsed(): array { |
| 1178 | if ( $this->used ) { |
| 1179 | return $this->used; |
| 1180 | } |
| 1181 | $this->used = PHPUtils::makeSet( |
| 1182 | array_merge( |
| 1183 | [ |
| 1184 | 'lang', 'width', 'class', 'upright', |
| 1185 | 'border', 'frameless', 'framed', 'thumbnail', |
| 1186 | ], |
| 1187 | self::HORIZONTAL_ALIGNS, |
| 1188 | self::VERTICAL_ALIGNS |
| 1189 | ) |
| 1190 | ); |
| 1191 | return $this->used; |
| 1192 | } |
| 1193 | |
| 1194 | private function hasTransclusion( array $toks ): bool { |
| 1195 | foreach ( $toks as $t ) { |
| 1196 | if ( |
| 1197 | $t instanceof SelfclosingTagTk && |
| 1198 | TokenUtils::hasTypeOf( $t, 'mw:Transclusion' ) |
| 1199 | ) { |
| 1200 | return true; |
| 1201 | } |
| 1202 | } |
| 1203 | return false; |
| 1204 | } |
| 1205 | |
| 1206 | /** |
| 1207 | * Render a file. This can be an image, a sound, a PDF etc. |
| 1208 | * |
| 1209 | * @param Token $token |
| 1210 | * @param stdClass $target |
| 1211 | * @return array<string|Token> |
| 1212 | */ |
| 1213 | private function renderFile( Token $token, stdClass $target ): array { |
| 1214 | $manager = $this->manager; |
| 1215 | $env = $this->env; |
| 1216 | |
| 1217 | // FIXME: Re-enable use of media cache and figure out how that fits |
| 1218 | // into this new processing model. See T98995 |
| 1219 | |
| 1220 | $dataParsoid = clone $token->dataParsoid; |
| 1221 | $dataParsoid->optList = []; |
| 1222 | |
| 1223 | // Account for the possibility of an expanded target |
| 1224 | $dataMw = $token->dataMw ?? new DataMw(); |
| 1225 | |
| 1226 | $opts = [ |
| 1227 | 'title' => [ |
| 1228 | 'v' => $env->makeLink( $target->title ), |
| 1229 | 'src' => $token->getAttributeKV( 'href' )->vsrc |
| 1230 | ], |
| 1231 | 'size' => [ |
| 1232 | 'v' => [ |
| 1233 | 'height' => null, |
| 1234 | 'width' => null |
| 1235 | ] |
| 1236 | ], |
| 1237 | // Initialize these properties to avoid isset checks |
| 1238 | 'caption' => null, |
| 1239 | 'format' => null, |
| 1240 | 'manualthumb' => null, |
| 1241 | 'class' => null |
| 1242 | ]; |
| 1243 | |
| 1244 | $hasExpandableOpt = false; |
| 1245 | |
| 1246 | $optKVs = self::buildLinkAttrs( $token->attribs, true, null, null )['contentKVs']; |
| 1247 | while ( count( $optKVs ) > 0 ) { |
| 1248 | $oContent = array_shift( $optKVs ); |
| 1249 | Assert::invariant( $oContent instanceof KV, 'bad type' ); |
| 1250 | |
| 1251 | $origOptSrc = $oContent->v; |
| 1252 | if ( is_array( $origOptSrc ) && count( $origOptSrc ) === 1 ) { |
| 1253 | $origOptSrc = $origOptSrc[0]; |
| 1254 | } |
| 1255 | |
| 1256 | $oText = TokenUtils::tokensToString( $origOptSrc, true, [ 'includeEntities' => true ] ); |
| 1257 | |
| 1258 | if ( !is_string( $oText ) ) { |
| 1259 | // Might be that this is a valid option whose value is just |
| 1260 | // complicated. Try to figure it out, step through all tokens. |
| 1261 | $maybeOText = self::stringifyOptionTokens( $oText, '', $env ); |
| 1262 | if ( $maybeOText !== null ) { |
| 1263 | $oText = $maybeOText; |
| 1264 | } |
| 1265 | } |
| 1266 | |
| 1267 | $optInfo = null; |
| 1268 | if ( is_string( $oText ) ) { |
| 1269 | if ( str_contains( $oText, '|' ) ) { |
| 1270 | // Split the pipe-separated string into pieces |
| 1271 | // and convert each one into a KV obj and add them |
| 1272 | // to the beginning of the array. Note that this is |
| 1273 | // a hack to support templates that provide multiple |
| 1274 | // image options as a pipe-separated string. We aren't |
| 1275 | // really providing editing support for this yet, or |
| 1276 | // ever, maybe. |
| 1277 | // |
| 1278 | // TODO(arlolra): Tables in captions suppress breaking on |
| 1279 | // "linkdesc" pipes so `stringifyOptionTokens` should account |
| 1280 | // for pipes in table cell content. For the moment, breaking |
| 1281 | // here is acceptable since it matches the php implementation |
| 1282 | // bug for bug. |
| 1283 | $pieces = array_map( static function ( $s ) { |
| 1284 | return new KV( 'mw:maybeContent', $s ); |
| 1285 | }, explode( '|', $oText ) ); |
| 1286 | $optKVs = array_merge( $pieces, $optKVs ); |
| 1287 | |
| 1288 | // Record the fact that we won't provide editing support for this. |
| 1289 | $dataParsoid->uneditable = true; |
| 1290 | continue; |
| 1291 | } else { |
| 1292 | // We're being overly accepting of media options at this point, |
| 1293 | // since we don't know the type yet. After the info request, |
| 1294 | // we'll filter out those that aren't appropriate. |
| 1295 | $optInfo = self::getOptionInfo( $oText, $env ); |
| 1296 | } |
| 1297 | } |
| 1298 | |
| 1299 | $recordCaption = static function () use ( $oContent, $oText, $dataParsoid, &$opts ): void { |
| 1300 | $optsCaption = [ |
| 1301 | 'v' => $oContent->v, |
| 1302 | 'src' => $oContent->vsrc ?? $oText, |
| 1303 | 'srcOffsets' => $oContent->valueOffset(), |
| 1304 | // remember the position |
| 1305 | 'pos' => count( $dataParsoid->optList ) |
| 1306 | ]; |
| 1307 | // if there was a 'caption' previously, round-trip it as a |
| 1308 | // "bogus option". |
| 1309 | if ( !empty( $opts['caption'] ) ) { |
| 1310 | // Wrap the caption opt in an array since the option itself is an array! |
| 1311 | // Without the wrapping, the splicing will flatten the value. |
| 1312 | array_splice( $dataParsoid->optList, $opts['caption']['pos'], 0, [ [ |
| 1313 | 'ck' => 'bogus', |
| 1314 | 'ak' => $opts['caption']['src'] |
| 1315 | ] ] |
| 1316 | ); |
| 1317 | $optsCaption['pos']++; |
| 1318 | } |
| 1319 | $opts['caption'] = $optsCaption; |
| 1320 | }; |
| 1321 | |
| 1322 | // For the values of the caption and options, see |
| 1323 | // getOptionInfo's documentation above. |
| 1324 | // |
| 1325 | // If there are multiple captions, this code always |
| 1326 | // picks the last entry. This is the spec; see |
| 1327 | // "Image with multiple captions" parserTest. |
| 1328 | if ( !is_string( $oText ) || $optInfo === null || |
| 1329 | // Deprecated options |
| 1330 | in_array( $optInfo['ck'], [ 'disablecontrols' ], true ) |
| 1331 | ) { |
| 1332 | // No valid option found!? |
| 1333 | // Record for RT-ing |
| 1334 | $recordCaption(); |
| 1335 | continue; |
| 1336 | } |
| 1337 | |
| 1338 | // First option wins, the rest are 'bogus' |
| 1339 | // FIXME: For now, see T305628 |
| 1340 | if ( isset( $opts[$optInfo['ck']] ) || ( |
| 1341 | // All the formats are simple options with the key "format" |
| 1342 | // except for "manualthumb", so check if the format has been set |
| 1343 | in_array( $optInfo['ck'], [ 'format', 'manualthumb' ], true ) && ( |
| 1344 | self::getFormat( $opts ) || |
| 1345 | ( $this->options['extTagOpts']['suppressMediaFormats'] ?? false ) |
| 1346 | ) |
| 1347 | ) ) { |
| 1348 | $dataParsoid->optList[] = [ |
| 1349 | 'ck' => 'bogus', |
| 1350 | 'ak' => $optInfo['ak'] |
| 1351 | ]; |
| 1352 | continue; |
| 1353 | } |
| 1354 | |
| 1355 | $opt = [ |
| 1356 | 'ck' => $optInfo['v'], |
| 1357 | 'ak' => $oContent->vsrc ?? $optInfo['ak'] |
| 1358 | ]; |
| 1359 | |
| 1360 | if ( $optInfo['s'] === true ) { |
| 1361 | // Default: Simple image option |
| 1362 | $opts[$optInfo['ck']] = [ 'v' => $optInfo['v'] ]; |
| 1363 | } else { |
| 1364 | // Map short canonical name to the localized version used. |
| 1365 | $opt['ck'] = $optInfo['ck']; |
| 1366 | |
| 1367 | // The MediaWiki magic word for image dimensions is called 'width' |
| 1368 | // for historical reasons |
| 1369 | // Unlike other options, use last-specified width. |
| 1370 | if ( $optInfo['ck'] === 'width' ) { |
| 1371 | // We support a trailing 'px' here for historical reasons |
| 1372 | // (T15500, T53628, T207032) |
| 1373 | $maybeDim = Utils::parseMediaDimensions( |
| 1374 | $env->getSiteConfig(), $optInfo['v'], false, true |
| 1375 | ); |
| 1376 | if ( $maybeDim !== null ) { |
| 1377 | if ( $maybeDim['bogusPx'] ) { |
| 1378 | // Lint away redundant unit (T207032) |
| 1379 | $dataParsoid->setTempFlag( TempData::BOGUS_PX ); |
| 1380 | } |
| 1381 | $opts['size']['v'] = [ |
| 1382 | 'width' => Utils::validateMediaParam( $maybeDim['x'] ) ? $maybeDim['x'] : null, |
| 1383 | 'height' => array_key_exists( 'y', $maybeDim ) && |
| 1384 | Utils::validateMediaParam( $maybeDim['y'] ) ? $maybeDim['y'] : null |
| 1385 | ]; |
| 1386 | // Only round-trip a valid size |
| 1387 | $opts['size']['src'] = $oContent->vsrc ?? $optInfo['ak']; |
| 1388 | // check for duplicated options |
| 1389 | foreach ( $dataParsoid->optList as &$value ) { |
| 1390 | if ( $value['ck'] === 'width' ) { |
| 1391 | $value['ck'] = 'bogus'; // mark the previous definition as bogus, last one wins |
| 1392 | break; |
| 1393 | } |
| 1394 | } |
| 1395 | } else { |
| 1396 | $recordCaption(); |
| 1397 | continue; |
| 1398 | } |
| 1399 | // Lang is a global attribute and can be applied to all media elements |
| 1400 | // for editing and roundtripping. However, not all file handlers will |
| 1401 | // make use of it. This param validation is from the SVG handler but |
| 1402 | // seems generally applicable. |
| 1403 | } elseif ( $optInfo['ck'] === 'lang' && !Language::isValidInternalCode( $optInfo['v'] ) ) { |
| 1404 | $opt['ck'] = 'bogus'; |
| 1405 | } elseif ( |
| 1406 | $optInfo['ck'] === 'upright' && |
| 1407 | ( !is_numeric( $optInfo['v'] ) || $optInfo['v'] <= 0 ) |
| 1408 | ) { |
| 1409 | $opt['ck'] = 'bogus'; |
| 1410 | } else { |
| 1411 | $opts[$optInfo['ck']] = [ |
| 1412 | 'v' => $optInfo['v'], |
| 1413 | 'src' => $oContent->vsrc ?? $optInfo['ak'], |
| 1414 | 'srcOffsets' => $oContent->valueOffset(), |
| 1415 | ]; |
| 1416 | } |
| 1417 | } |
| 1418 | |
| 1419 | // Collect option in dataParsoid (becomes data-parsoid later on) |
| 1420 | // for faithful serialization. |
| 1421 | $dataParsoid->optList[] = $opt; |
| 1422 | |
| 1423 | // Collect source wikitext for image options for possible template expansion. |
| 1424 | $maybeOpt = !isset( self::getUsed()[$opt['ck']] ); |
| 1425 | // Links more often than not show up as arrays here because they're |
| 1426 | // tokenized as `autourl`. To avoid unnecessarily considering them |
| 1427 | // expanded, we'll use a more restrictive test, at the cost of |
| 1428 | // perhaps missing some edgy behaviour. |
| 1429 | if ( $opt['ck'] === 'link' ) { |
| 1430 | $expOpt = is_array( $origOptSrc ) && |
| 1431 | $this->hasTransclusion( $origOptSrc ); |
| 1432 | } else { |
| 1433 | $expOpt = is_array( $origOptSrc ); |
| 1434 | } |
| 1435 | if ( $maybeOpt || $expOpt ) { |
| 1436 | $val = []; |
| 1437 | if ( $expOpt ) { |
| 1438 | $hasExpandableOpt = true; |
| 1439 | $val['html'] = $origOptSrc; |
| 1440 | $val['srcOffsets'] = $oContent->valueOffset(); |
| 1441 | $val = PipelineUtils::expandAttrValueToDOM( |
| 1442 | $env, $manager->getFrame(), $val, |
| 1443 | $this->options['expandTemplates'], |
| 1444 | $this->options['inTemplate'] |
| 1445 | ); |
| 1446 | } |
| 1447 | |
| 1448 | // This is a bit of an abuse of the "txt" property since |
| 1449 | // `optInfo.v` isn't necessarily wikitext from source. |
| 1450 | // It's a result of the specialized stringifying above, which |
| 1451 | // if interpreted as wikitext upon serialization will result |
| 1452 | // in some (acceptable) normalization. |
| 1453 | // |
| 1454 | // We're storing these options in data-mw because they aren't |
| 1455 | // guaranteed to apply to all media types and we'd like to |
| 1456 | // avoid the need to back them out later. |
| 1457 | // |
| 1458 | // Note that the caption in the legacy parser depends on the |
| 1459 | // exact set of options parsed, which we aren't attempting to |
| 1460 | // try and replicate after fetching the media info, since we |
| 1461 | // consider that more of bug than a feature. It prevent anyone |
| 1462 | // from ever safely adding media options in the future. |
| 1463 | // |
| 1464 | // See T163582 |
| 1465 | if ( $maybeOpt ) { |
| 1466 | $val['txt'] = $optInfo['v']; |
| 1467 | } |
| 1468 | $dataMw->attribs ??= []; |
| 1469 | $dataMw->attribs[] = new DataMwAttrib( $opt['ck'], $val ); |
| 1470 | } |
| 1471 | } |
| 1472 | |
| 1473 | // Add the last caption in the right position if there is one |
| 1474 | if ( isset( $opts['caption'] ) ) { |
| 1475 | // Wrap the caption opt in an array since the option itself is an array! |
| 1476 | // Without the wrapping, the splicing will flatten the value. |
| 1477 | array_splice( $dataParsoid->optList, $opts['caption']['pos'], 0, [ [ |
| 1478 | 'ck' => 'caption', |
| 1479 | 'ak' => $opts['caption']['src'] |
| 1480 | ] ] |
| 1481 | ); |
| 1482 | } |
| 1483 | |
| 1484 | $format = self::getFormat( $opts ); |
| 1485 | |
| 1486 | // Handle image default sizes and upright option after extracting all |
| 1487 | // options |
| 1488 | $uprightFactor = null; |
| 1489 | if ( $format === 'framed' || $format === 'manualthumb' ) { |
| 1490 | // width and height is ignored for framed and manualthumb images |
| 1491 | // https://phabricator.wikimedia.org/T64258 |
| 1492 | $opts['size']['v'] = [ 'width' => null, 'height' => null ]; |
| 1493 | // Mark any definitions as bogus |
| 1494 | foreach ( $dataParsoid->optList as &$value ) { |
| 1495 | if ( $value['ck'] === 'width' ) { |
| 1496 | $value['ck'] = 'bogus'; |
| 1497 | } |
| 1498 | } |
| 1499 | } elseif ( $format ) { |
| 1500 | if ( !$opts['size']['v']['height'] && !$opts['size']['v']['width'] ) { |
| 1501 | $defaultWidth = $env->getSiteConfig()->widthOption(); |
| 1502 | if ( isset( $opts['upright'] ) ) { |
| 1503 | if ( $opts['upright']['v'] === 'upright' ) { // Simple option |
| 1504 | $uprightFactor = 0.75; |
| 1505 | } else { |
| 1506 | $uprightFactor = (float)$opts['upright']['v']; |
| 1507 | } |
| 1508 | $defaultWidth *= $uprightFactor; |
| 1509 | // round to nearest 10 pixels |
| 1510 | $defaultWidth = 10 * round( $defaultWidth / 10 ); |
| 1511 | } |
| 1512 | $opts['size']['v']['width'] = $defaultWidth; |
| 1513 | } |
| 1514 | } |
| 1515 | |
| 1516 | // If the format is something we *recognize*, add the subtype |
| 1517 | $rdfaType = 'mw:File' . match ( $format ) { |
| 1518 | 'manualthumb', // FIXME(T305759): Does it deserve its own type? |
| 1519 | 'thumbnail' => '/Thumb', |
| 1520 | 'framed' => '/Frame', |
| 1521 | 'frameless' => '/Frameless', |
| 1522 | default => '' |
| 1523 | }; |
| 1524 | |
| 1525 | // Tell VE that it shouldn't try to edit this |
| 1526 | if ( !empty( $dataParsoid->uneditable ) ) { |
| 1527 | $rdfaType .= ' mw:Placeholder'; |
| 1528 | } else { |
| 1529 | unset( $dataParsoid->src ); |
| 1530 | } |
| 1531 | |
| 1532 | $wrapperInfo = self::getWrapperInfo( $opts ); |
| 1533 | |
| 1534 | $isInline = $wrapperInfo['isInline']; |
| 1535 | $containerName = $isInline ? 'span' : 'figure'; |
| 1536 | |
| 1537 | $classes = $wrapperInfo['classes']; |
| 1538 | if ( !empty( $opts['class'] ) ) { |
| 1539 | PHPUtils::pushArray( $classes, explode( ' ', $opts['class']['v'] ) ); |
| 1540 | } |
| 1541 | |
| 1542 | $attribs = [ new KV( 'typeof', $rdfaType ) ]; |
| 1543 | if ( count( $classes ) > 0 ) { |
| 1544 | array_unshift( $attribs, new KV( 'class', implode( ' ', $classes ) ) ); |
| 1545 | } |
| 1546 | |
| 1547 | $container = new TagTk( $containerName, $attribs, $dataParsoid ); |
| 1548 | $containerClose = new EndTagTk( $containerName ); |
| 1549 | |
| 1550 | if ( $hasExpandableOpt ) { |
| 1551 | $container->addAttribute( 'about', $env->newAboutId() ); |
| 1552 | $container->addSpaceSeparatedAttribute( 'typeof', 'mw:ExpandedAttrs' ); |
| 1553 | } elseif ( preg_match( '/\bmw:ExpandedAttrs\b/', $token->getAttributeV( 'typeof' ) ?? '' ) ) { |
| 1554 | $container->addSpaceSeparatedAttribute( 'typeof', 'mw:ExpandedAttrs' ); |
| 1555 | } |
| 1556 | |
| 1557 | // Start off as broken media since we don't know if the file exists. |
| 1558 | // In the AddMediaInfo pass, we'll replace the node after calling getFileInfo |
| 1559 | $span = new TagTk( 'span', [ new KV( 'class', 'mw-file-element mw-broken-media' ) ] ); |
| 1560 | |
| 1561 | // "resource" and "lang" are allowed attributes on spans |
| 1562 | $span->addNormalizedAttribute( 'resource', $opts['title']['v'], $opts['title']['src'] ); |
| 1563 | if ( isset( $opts['lang'] ) ) { |
| 1564 | $span->addNormalizedAttribute( 'lang', $opts['lang']['v'], $opts['lang']['src'] ); |
| 1565 | } |
| 1566 | |
| 1567 | // Token's KV attributes only accept strings, Tokens or arrays of those. |
| 1568 | $size = $opts['size']['v']; |
| 1569 | if ( !empty( $size['width'] ) ) { |
| 1570 | $span->addAttribute( 'data-width', (string)$size['width'] ); |
| 1571 | } |
| 1572 | if ( !empty( $size['height'] ) ) { |
| 1573 | $span->addAttribute( 'data-height', (string)$size['height'] ); |
| 1574 | } |
| 1575 | if ( $uprightFactor !== null ) { |
| 1576 | $span->addAttribute( 'data-upright', (string)$uprightFactor ); |
| 1577 | } |
| 1578 | |
| 1579 | $anchor = new TagTk( 'a' ); |
| 1580 | $anchor->setAttribute( 'href', $this->specialFilePath( $target->title ) ); |
| 1581 | $anchor->setAttribute( 'class', 'new' ); |
| 1582 | $anchor->setAttribute( 'title', $target->title->getPrefixedText() ); |
| 1583 | |
| 1584 | $tokens = [ |
| 1585 | $container, |
| 1586 | $anchor, |
| 1587 | $span, |
| 1588 | $target->title->getPrefixedText(), |
| 1589 | new EndTagTk( 'span' ), |
| 1590 | new EndTagTk( 'a' ) |
| 1591 | ]; |
| 1592 | |
| 1593 | $optsCaption = $opts['caption'] ?? null; |
| 1594 | if ( $isInline ) { |
| 1595 | if ( $optsCaption ) { |
| 1596 | if ( !is_array( $optsCaption['v'] ) ) { |
| 1597 | $opts['caption']['v'] = $optsCaption['v'] = [ $optsCaption['v'] ]; |
| 1598 | } |
| 1599 | // Parse the caption |
| 1600 | $captionDOM = PipelineUtils::processContentInPipeline( |
| 1601 | $this->env, |
| 1602 | $this->manager->getFrame(), |
| 1603 | array_merge( $optsCaption['v'], [ new EOFTk() ] ), |
| 1604 | [ |
| 1605 | 'pipelineType' => 'expanded-tokens-to-fragment', |
| 1606 | 'pipelineOpts' => [ |
| 1607 | 'inlineContext' => true, |
| 1608 | 'expandTemplates' => $this->options['expandTemplates'], |
| 1609 | 'inTemplate' => $this->options['inTemplate'] |
| 1610 | ], |
| 1611 | 'srcOffsets' => $optsCaption['srcOffsets'] ?? null, |
| 1612 | 'sol' => true |
| 1613 | ] |
| 1614 | ); |
| 1615 | |
| 1616 | // Use parsed DOM given in `captionDOM` |
| 1617 | // FIXME: Does this belong in `dataMw.attribs`? |
| 1618 | $dataMw->caption = $captionDOM; |
| 1619 | } |
| 1620 | } else { |
| 1621 | // We always add a figcaption for blocks |
| 1622 | $tsr = $optsCaption['srcOffsets'] ?? null; |
| 1623 | $dp = new DataParsoid; |
| 1624 | $dp->tsr = $tsr; |
| 1625 | $tokens[] = new TagTk( 'figcaption', [], $dp ); |
| 1626 | if ( $optsCaption ) { |
| 1627 | if ( is_string( $optsCaption['v'] ) ) { |
| 1628 | $tokens[] = $optsCaption['v']; |
| 1629 | } else { |
| 1630 | $tokens[] = PipelineUtils::getDOMFragmentToken( |
| 1631 | $optsCaption['v'], |
| 1632 | $tsr, |
| 1633 | [ 'inlineContext' => true, 'token' => $token ] |
| 1634 | ); |
| 1635 | } |
| 1636 | } |
| 1637 | $tokens[] = new EndTagTk( 'figcaption' ); |
| 1638 | } |
| 1639 | |
| 1640 | if ( !$dataMw->isEmpty() ) { |
| 1641 | $container->dataMw = $dataMw; |
| 1642 | } |
| 1643 | |
| 1644 | $tokens[] = $containerClose; |
| 1645 | return $tokens; |
| 1646 | } |
| 1647 | |
| 1648 | private function specialFilePath( Title $title ): string { |
| 1649 | $filePath = Sanitizer::sanitizeTitleURI( $title->getDBkey(), false ); |
| 1650 | return "./Special:FilePath/{$filePath}"; |
| 1651 | } |
| 1652 | |
| 1653 | /** |
| 1654 | * @param Token $token |
| 1655 | * @param stdClass $target |
| 1656 | * @param list<DataMwError> $errs |
| 1657 | * @param ?array{url?:string} $info |
| 1658 | * @return array<string|Token> |
| 1659 | */ |
| 1660 | private function linkToMedia( Token $token, stdClass $target, array $errs, ?array $info ): array { |
| 1661 | // Only pass in the url, since media links should not link to the thumburl |
| 1662 | $imgHref = $info['url'] ?? $this->specialFilePath( $target->title ); // Copied from getPath |
| 1663 | $imgHrefFileName = preg_replace( '#.*/#', '', $imgHref, 1 ); |
| 1664 | |
| 1665 | $link = new TagTk( 'a' ); |
| 1666 | |
| 1667 | try { |
| 1668 | $content = $this->addLinkAttributesAndGetContent( $link, $token, $target ); |
| 1669 | } catch ( InternalException ) { |
| 1670 | return self::bailTokens( $this->manager, $token ); |
| 1671 | } |
| 1672 | |
| 1673 | // Change the rel to be mw:MediaLink |
| 1674 | $link->getAttributeKV( 'rel' )->v = 'mw:MediaLink'; |
| 1675 | |
| 1676 | $link->addNormalizedAttribute( 'href', $imgHref, $target->hrefSrc ); |
| 1677 | |
| 1678 | // html2wt will use the resource rather than try to parse the href. |
| 1679 | $link->addNormalizedAttribute( |
| 1680 | 'resource', |
| 1681 | $this->env->makeLink( $target->title ), |
| 1682 | $target->hrefSrc |
| 1683 | ); |
| 1684 | |
| 1685 | // Normalize title according to how PHP parser does it currently |
| 1686 | $link->setAttribute( 'title', str_replace( '_', ' ', $imgHrefFileName ) ); |
| 1687 | |
| 1688 | if ( count( $errs ) > 0 ) { |
| 1689 | // Set RDFa type to mw:Error so VE and other clients |
| 1690 | // can use this to do client-specific action on these. |
| 1691 | if ( !TokenUtils::hasTypeOf( $link, 'mw:Error' ) ) { |
| 1692 | $link->addSpaceSeparatedAttribute( 'typeof', 'mw:Error' ); |
| 1693 | } |
| 1694 | |
| 1695 | // Update data-mw |
| 1696 | $dataMw = $token->dataMw ?? new DataMw; |
| 1697 | if ( is_array( $dataMw->errors ?? null ) ) { |
| 1698 | array_push( $dataMw->errors, ...$errs ); |
| 1699 | } else { |
| 1700 | $dataMw->errors = $errs; |
| 1701 | } |
| 1702 | $link->dataMw = $dataMw; |
| 1703 | } |
| 1704 | |
| 1705 | return array_merge( [ $link ], $content, [ new EndTagTk( 'a' ) ] ); |
| 1706 | } |
| 1707 | |
| 1708 | /** |
| 1709 | * FIXME: The media request here is only used to determine if this is a |
| 1710 | * redlink and deserves to be handling in the redlink post-processing pass. |
| 1711 | * @return array<string|Token> |
| 1712 | */ |
| 1713 | private function renderMedia( Token $token, stdClass $target ): array { |
| 1714 | $env = $this->env; |
| 1715 | $title = $target->title; |
| 1716 | $errs = []; |
| 1717 | $info = $env->getDataAccess()->getFileInfo( |
| 1718 | $env->getPageConfig(), |
| 1719 | [ [ $title->getDBkey(), [ 'height' => null, 'width' => null ] ] ] |
| 1720 | )[0]; |
| 1721 | if ( !$info ) { |
| 1722 | $errs[] = new DataMwError( 'apierror-filedoesnotexist', [], 'This image does not exist.' ); |
| 1723 | } elseif ( isset( $info['thumberror'] ) ) { |
| 1724 | $errs[] = new DataMwError( 'apierror-unknownerror', [], $info['thumberror'] ); |
| 1725 | } |
| 1726 | return $this->linkToMedia( $token, $target, $errs, $info ); |
| 1727 | } |
| 1728 | |
| 1729 | /** @inheritDoc */ |
| 1730 | public function onTag( XMLTagTk $token ): ?array { |
| 1731 | return match ( $token->getName() ) { |
| 1732 | 'wikilink' => $this->onWikiLink( $token ), |
| 1733 | 'mw:redirect' => $this->onRedirect( $token ), |
| 1734 | default => null |
| 1735 | }; |
| 1736 | } |
| 1737 | } |