Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
17.93% |
45 / 251 |
|
14.52% |
9 / 62 |
CRAP | |
0.00% |
0 / 1 |
| WTUtils | |
17.93% |
45 / 251 |
|
14.52% |
9 / 62 |
14850.76 | |
0.00% |
0 / 1 |
| hasLiteralHTMLMarker | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| isLiteralHTMLNode | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isZeroWidthWikitextElt | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isBlockNodeWithVisibleWT | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isATagFromWikiLinkSyntax | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
30 | |||
| isATagFromExtLinkSyntax | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
30 | |||
| isATagFromURLLinkSyntax | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| isATagFromMagicLinkSyntax | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| matchTplType | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| hasExpandedAttrsType | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isTplMarkerMeta | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isTplStartMarkerMeta | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isTplEndMarkerMeta | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| findFirstEncapsulationWrapperNode | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
20 | |||
| isNewElt | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| isIndentPre | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| isInlineMedia | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isGeneratedFigure | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| indentPreDSRCorrection | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
20 | |||
| isEncapsulatedDOMForestRoot | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| isRedirectLink | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| isCategoryLink | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| isSolTransparentLink | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| emitsSolTransparentSingleLineWT | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| isFallbackIdSpan | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isRenderingTransparentNode | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
56 | |||
| inHTMLTableTag | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
20 | |||
| isFirstEncapsulationWrapperNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isFirstExtensionWrapperNode | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isExtensionOutputtingCoreMwDomSpec | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| isEncapsulationWrapper | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| isDOMFragmentWrapper | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isSealedFragmentOfType | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isParsoidSectionTag | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| fromExtensionContent | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
5 | |||
| fromEncapsulatedContent | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| getWTSource | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getAboutSiblings | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
90 | |||
| skipOverEncapsulatedContent | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| encodeComment | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| decodeComment | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| decodedCommentLength | |
58.33% |
7 / 12 |
|
0.00% |
0 / 1 |
8.60 | |||
| getExtTagName | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getNativeExt | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isIncludeTag | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
12 | |||
| isAnnOrExtTag | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
30 | |||
| createEmptyLocalizationFragment | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| createPageContentI18nFragment | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| createInterfaceI18nFragment | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| createLangI18nFragment | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| addPageContentI18nAttribute | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| addInterfaceI18nAttribute | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| addLangI18nAttribute | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| matchAnnotationMeta | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| extractAnnotationType | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| isAnnotationStartMarkerMeta | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| isAnnotationEndMarkerMeta | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| isMovedMetaTag | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| isMarkerAnnotation | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| getMediaFormat | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| hasVisibleCaption | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| textContentFromCaption | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
42 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | // Suppress UnusedPluginSuppression because |
| 4 | // Phan on PHP 7.4 and PHP 8.1 need different suppressions |
| 5 | // @phan-file-suppress UnusedPluginSuppression,UnusedPluginFileSuppression |
| 6 | |
| 7 | namespace Wikimedia\Parsoid\Utils; |
| 8 | |
| 9 | use DOMException; |
| 10 | use Wikimedia\Assert\UnreachableException; |
| 11 | use Wikimedia\Bcp47Code\Bcp47Code; |
| 12 | use Wikimedia\Parsoid\Config\Env; |
| 13 | use Wikimedia\Parsoid\DOM\Comment; |
| 14 | use Wikimedia\Parsoid\DOM\Document; |
| 15 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 16 | use Wikimedia\Parsoid\DOM\Element; |
| 17 | use Wikimedia\Parsoid\DOM\Node; |
| 18 | use Wikimedia\Parsoid\DOM\Text; |
| 19 | use Wikimedia\Parsoid\Ext\ExtensionTagHandler; |
| 20 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 21 | use Wikimedia\Parsoid\NodeData\I18nInfo; |
| 22 | use Wikimedia\Parsoid\NodeData\TempData; |
| 23 | use Wikimedia\Parsoid\Tokens\CommentTk; |
| 24 | use Wikimedia\Parsoid\Wikitext\Consts; |
| 25 | use Wikimedia\Parsoid\Wt2Html\Frame; |
| 26 | |
| 27 | /** |
| 28 | * These utilites pertain to querying / extracting / modifying wikitext information from the DOM. |
| 29 | */ |
| 30 | class WTUtils { |
| 31 | private const FIRST_ENCAP_REGEXP = |
| 32 | '#(?:^|\s)(mw:(?:Transclusion|Param|LanguageVariant|Extension(/\S+)))(?=$|\s)#D'; |
| 33 | |
| 34 | /** |
| 35 | * Regex corresponding to FIRST_ENCAP_REGEXP, but excluding extensions. If FIRST_ENCAP_REGEXP is |
| 36 | * updated, this one should be as well. |
| 37 | */ |
| 38 | private const NON_EXTENSION_ENCAP_REGEXP = |
| 39 | '#(?:^|\s)(mw:(?:Transclusion|Param|LanguageVariant))(?=$|\s)#D'; |
| 40 | |
| 41 | /** |
| 42 | * Regexp for checking marker metas typeofs representing |
| 43 | * transclusion markup or template param markup. |
| 44 | */ |
| 45 | private const TPL_META_TYPE_REGEXP = '#^mw:(?:Transclusion|Param)(?:/End)?$#D'; |
| 46 | |
| 47 | /** |
| 48 | * Regexp for checking marker metas typeofs representing |
| 49 | * annotation markup |
| 50 | */ |
| 51 | public const ANNOTATION_META_TYPE_REGEXP = '#^mw:(?:Annotation/([\w\d]+))(?:/End)?$#uD'; |
| 52 | |
| 53 | /** |
| 54 | * Check whether a node's data-parsoid object includes |
| 55 | * an indicator that the original wikitext was a literal |
| 56 | * HTML element (like table or p) |
| 57 | * |
| 58 | * @param DataParsoid $dp |
| 59 | * @return bool |
| 60 | */ |
| 61 | public static function hasLiteralHTMLMarker( DataParsoid $dp ): bool { |
| 62 | return isset( $dp->stx ) && $dp->stx === 'html'; |
| 63 | } |
| 64 | |
| 65 | /** |
| 66 | * Run a node through {@link #hasLiteralHTMLMarker}. |
| 67 | * @param ?Node $node |
| 68 | * @return bool |
| 69 | */ |
| 70 | public static function isLiteralHTMLNode( ?Node $node ): bool { |
| 71 | return $node instanceof Element && |
| 72 | self::hasLiteralHTMLMarker( DOMDataUtils::getDataParsoid( $node ) ); |
| 73 | } |
| 74 | |
| 75 | /** |
| 76 | * @param Node $node |
| 77 | * @return bool |
| 78 | */ |
| 79 | public static function isZeroWidthWikitextElt( Node $node ): bool { |
| 80 | return isset( Consts::$ZeroWidthWikitextTags[DOMCompat::nodeName( $node )] ) && |
| 81 | !self::isLiteralHTMLNode( $node ); |
| 82 | } |
| 83 | |
| 84 | /** |
| 85 | * Is `$node` a block node that is also visible in wikitext? |
| 86 | * An example of an invisible block node is a `<p>`-tag that |
| 87 | * Parsoid generated, or a `<ul>`, `<ol>` tag. |
| 88 | * |
| 89 | * @param Node $node |
| 90 | * @return bool |
| 91 | */ |
| 92 | public static function isBlockNodeWithVisibleWT( Node $node ): bool { |
| 93 | return DOMUtils::isWikitextBlockNode( $node ) && |
| 94 | !self::isZeroWidthWikitextElt( $node ); |
| 95 | } |
| 96 | |
| 97 | /** |
| 98 | * Helper functions to detect when an A-$node uses [[..]]/[..]/... style |
| 99 | * syntax (for wikilinks, ext links, url links). rel-type is not sufficient |
| 100 | * anymore since mw:ExtLink is used for all the three link syntaxes. |
| 101 | * |
| 102 | * @param Element $node |
| 103 | * @return bool |
| 104 | */ |
| 105 | public static function isATagFromWikiLinkSyntax( Element $node ): bool { |
| 106 | if ( DOMCompat::nodeName( $node ) !== 'a' ) { |
| 107 | return false; |
| 108 | } |
| 109 | |
| 110 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 111 | return DOMUtils::hasRel( $node, 'mw:WikiLink' ) || |
| 112 | ( isset( $dp->stx ) && $dp->stx !== "url" && $dp->stx !== "magiclink" ); |
| 113 | } |
| 114 | |
| 115 | /** |
| 116 | * Helper function to detect when an A-node uses ext-link syntax. |
| 117 | * rel attribute is not sufficient anymore since mw:ExtLink is used for |
| 118 | * multiple link types |
| 119 | * |
| 120 | * @param Element $node |
| 121 | * @return bool |
| 122 | */ |
| 123 | public static function isATagFromExtLinkSyntax( Element $node ): bool { |
| 124 | if ( DOMCompat::nodeName( $node ) !== 'a' ) { |
| 125 | return false; |
| 126 | } |
| 127 | |
| 128 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 129 | return DOMUtils::hasRel( $node, 'mw:ExtLink' ) && |
| 130 | ( !isset( $dp->stx ) || ( $dp->stx !== "url" && $dp->stx !== "magiclink" ) ); |
| 131 | } |
| 132 | |
| 133 | /** |
| 134 | * Helper function to detect when an A-node uses url-link syntax. |
| 135 | * rel attribute is not sufficient anymore since mw:ExtLink is used for |
| 136 | * multiple link types |
| 137 | * |
| 138 | * @param Element $node |
| 139 | * @return bool |
| 140 | */ |
| 141 | public static function isATagFromURLLinkSyntax( Element $node ): bool { |
| 142 | if ( DOMCompat::nodeName( $node ) !== 'a' ) { |
| 143 | return false; |
| 144 | } |
| 145 | |
| 146 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 147 | return DOMUtils::hasRel( $node, 'mw:ExtLink' ) && |
| 148 | isset( $dp->stx ) && $dp->stx === "url"; |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * Helper function to detect when an A-node uses magic-link syntax. |
| 153 | * rel attribute is not sufficient anymore since mw:ExtLink is used for |
| 154 | * multiple link types |
| 155 | * |
| 156 | * @param Element $node |
| 157 | * @return bool |
| 158 | */ |
| 159 | public static function isATagFromMagicLinkSyntax( Element $node ): bool { |
| 160 | if ( DOMCompat::nodeName( $node ) !== 'a' ) { |
| 161 | return false; |
| 162 | } |
| 163 | |
| 164 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 165 | return DOMUtils::hasRel( $node, 'mw:ExtLink' ) && |
| 166 | isset( $dp->stx ) && $dp->stx === 'magiclink'; |
| 167 | } |
| 168 | |
| 169 | /** |
| 170 | * Check whether a node's typeof indicates that it is a template expansion. |
| 171 | * |
| 172 | * @param Element $node |
| 173 | * @return ?string The matched type, or null if no match. |
| 174 | */ |
| 175 | public static function matchTplType( Element $node ): ?string { |
| 176 | return DOMUtils::matchTypeOf( $node, self::TPL_META_TYPE_REGEXP ); |
| 177 | } |
| 178 | |
| 179 | /** |
| 180 | * Check whether a typeof indicates that it signifies an |
| 181 | * expanded attribute. |
| 182 | * |
| 183 | * @param Element $node |
| 184 | * @return bool |
| 185 | */ |
| 186 | public static function hasExpandedAttrsType( Element $node ): bool { |
| 187 | return DOMUtils::matchTypeOf( $node, '/^mw:ExpandedAttrs(\/\S+)*$/' ) !== null; |
| 188 | } |
| 189 | |
| 190 | /** |
| 191 | * Check whether a node is a meta tag that signifies a template expansion. |
| 192 | * |
| 193 | * @param Node $node |
| 194 | * @return bool |
| 195 | */ |
| 196 | public static function isTplMarkerMeta( Node $node ): bool { |
| 197 | return DOMUtils::matchNameAndTypeOf( $node, 'meta', self::TPL_META_TYPE_REGEXP ) !== null; |
| 198 | } |
| 199 | |
| 200 | /** |
| 201 | * Check whether a node is a meta signifying the start of a template expansion. |
| 202 | * |
| 203 | * @param Node $node |
| 204 | * @return bool |
| 205 | */ |
| 206 | public static function isTplStartMarkerMeta( Node $node ): bool { |
| 207 | $t = DOMUtils::matchNameAndTypeOf( $node, 'meta', self::TPL_META_TYPE_REGEXP ); |
| 208 | return $t !== null && !str_ends_with( $t, '/End' ); |
| 209 | } |
| 210 | |
| 211 | /** |
| 212 | * Check whether a node is a meta signifying the end of a template expansion. |
| 213 | * |
| 214 | * @param Node $node |
| 215 | * @return bool |
| 216 | */ |
| 217 | public static function isTplEndMarkerMeta( Node $node ): bool { |
| 218 | $t = DOMUtils::matchNameAndTypeOf( $node, 'meta', self::TPL_META_TYPE_REGEXP ); |
| 219 | return $t !== null && str_ends_with( $t, '/End' ); |
| 220 | } |
| 221 | |
| 222 | /** |
| 223 | * Find the first wrapper element of encapsulated content. |
| 224 | * @param Node $node |
| 225 | * @return Element|null |
| 226 | */ |
| 227 | public static function findFirstEncapsulationWrapperNode( Node $node ): ?Element { |
| 228 | if ( !self::isEncapsulatedDOMForestRoot( $node ) ) { |
| 229 | return null; |
| 230 | } |
| 231 | /** @var Element $node */ |
| 232 | DOMUtils::assertElt( $node ); |
| 233 | |
| 234 | $about = DOMCompat::getAttribute( $node, 'about' ); |
| 235 | $prev = $node; |
| 236 | do { |
| 237 | $node = $prev; |
| 238 | $prev = DiffDOMUtils::previousNonDeletedSibling( $node ); |
| 239 | } while ( |
| 240 | $prev instanceof Element && |
| 241 | DOMCompat::getAttribute( $prev, 'about' ) === $about |
| 242 | ); |
| 243 | // NOTE: findFirstEncapsulationWrapperNode can be called by code |
| 244 | // even before templates have been fully encapsulated everywhere. |
| 245 | // ProcessTreeBuilderFixups::removeAutoInsertedEmptyTags is the main |
| 246 | // culprit here and it makes the contract for this helper murky |
| 247 | // by hiding potential brokenness since this should never return null |
| 248 | // once all templates have been encapsulated! |
| 249 | $elt = self::isFirstEncapsulationWrapperNode( $node ) ? $node : null; |
| 250 | '@phan-var ?Element $elt'; // @var ?Element $elt |
| 251 | return $elt; |
| 252 | } |
| 253 | |
| 254 | /** |
| 255 | * This tests whether a DOM node is a new node added during an edit session |
| 256 | * or an existing node from parsed wikitext. |
| 257 | * |
| 258 | * As written, this function can only be used on non-template/extension content |
| 259 | * or on the top-level nodes of template/extension content. This test will |
| 260 | * return the wrong results on non-top-level $nodes of template/extension content. |
| 261 | * |
| 262 | * @param Node $node |
| 263 | * @return bool |
| 264 | */ |
| 265 | public static function isNewElt( Node $node ): bool { |
| 266 | // We cannot determine newness on text/comment $nodes. |
| 267 | if ( !( $node instanceof Element ) ) { |
| 268 | return false; |
| 269 | } |
| 270 | |
| 271 | // For template/extension content, newness should be |
| 272 | // checked on the encapsulation wrapper $node. |
| 273 | $node = self::findFirstEncapsulationWrapperNode( $node ) ?? $node; |
| 274 | return DOMDataUtils::getDataParsoid( $node )->getTempFlag( TempData::IS_NEW ); |
| 275 | } |
| 276 | |
| 277 | /** |
| 278 | * Check whether a pre is caused by indentation in the original wikitext. |
| 279 | * @param Node $node |
| 280 | * @return bool |
| 281 | */ |
| 282 | public static function isIndentPre( Node $node ): bool { |
| 283 | return DOMCompat::nodeName( $node ) === "pre" && !self::isLiteralHTMLNode( $node ); |
| 284 | } |
| 285 | |
| 286 | /** |
| 287 | * @param Node $node |
| 288 | * @return bool |
| 289 | */ |
| 290 | public static function isInlineMedia( Node $node ): bool { |
| 291 | return self::isGeneratedFigure( $node ) && |
| 292 | DOMCompat::nodeName( $node ) !== 'figure'; // span, figure-inline |
| 293 | } |
| 294 | |
| 295 | /** |
| 296 | * @param Node $node |
| 297 | * @return bool |
| 298 | */ |
| 299 | public static function isGeneratedFigure( Node $node ): bool { |
| 300 | // TODO: Remove "Image|Video|Audio" when version 2.4.0 of the content |
| 301 | // is no longer supported |
| 302 | return DOMUtils::matchTypeOf( $node, '#^mw:(File|Image|Video|Audio)($|/)#D' ) !== null; |
| 303 | } |
| 304 | |
| 305 | /** |
| 306 | * Find how much offset is necessary for the DSR of an |
| 307 | * indent-originated pre tag. |
| 308 | * |
| 309 | * @param Node $textNode |
| 310 | * @return int |
| 311 | */ |
| 312 | public static function indentPreDSRCorrection( Node $textNode ): int { |
| 313 | // NOTE: This assumes a text-node and doesn't check that it is one. |
| 314 | // |
| 315 | // FIXME: Doesn't handle text nodes that are not direct children of the pre |
| 316 | if ( self::isIndentPre( $textNode->parentNode ) ) { |
| 317 | $numNLs = substr_count( $textNode->nodeValue, "\n" ); |
| 318 | if ( $textNode->parentNode->lastChild === $textNode ) { |
| 319 | // We dont want the trailing newline of the last child of the pre |
| 320 | // to contribute a pre-correction since it doesn't add new content |
| 321 | // in the pre-node after the text |
| 322 | if ( str_ends_with( $textNode->nodeValue, "\n" ) ) { |
| 323 | $numNLs--; |
| 324 | } |
| 325 | } |
| 326 | return $numNLs; |
| 327 | } else { |
| 328 | return 0; |
| 329 | } |
| 330 | } |
| 331 | |
| 332 | /** |
| 333 | * Check if $node is a root in an encapsulated DOM forest. |
| 334 | * |
| 335 | * @param Node $node |
| 336 | * @return bool |
| 337 | */ |
| 338 | public static function isEncapsulatedDOMForestRoot( Node $node ): bool { |
| 339 | $about = $node instanceof Element ? |
| 340 | DOMCompat::getAttribute( $node, 'about' ) : null; |
| 341 | if ( $about !== null ) { |
| 342 | // FIXME: Ensure that our DOM spec clarifies this expectation |
| 343 | return Utils::isParsoidObjectId( $about ); |
| 344 | } else { |
| 345 | return false; |
| 346 | } |
| 347 | } |
| 348 | |
| 349 | /** |
| 350 | * Does $node represent a redirect link? |
| 351 | * |
| 352 | * @param Node $node |
| 353 | * @return bool |
| 354 | */ |
| 355 | public static function isRedirectLink( Node $node ): bool { |
| 356 | return $node instanceof Element && |
| 357 | DOMCompat::nodeName( $node ) === 'link' && |
| 358 | DOMUtils::matchRel( $node, '#\bmw:PageProp/redirect\b#' ) !== null; |
| 359 | } |
| 360 | |
| 361 | /** |
| 362 | * Does $node represent a category link? |
| 363 | * |
| 364 | * @param ?Node $node |
| 365 | * @return bool |
| 366 | */ |
| 367 | public static function isCategoryLink( ?Node $node ): bool { |
| 368 | return $node instanceof Element && |
| 369 | DOMCompat::nodeName( $node ) === 'link' && |
| 370 | DOMUtils::matchRel( $node, '#\bmw:PageProp/Category\b#' ) !== null; |
| 371 | } |
| 372 | |
| 373 | /** |
| 374 | * Does $node represent a link that is sol-transparent? |
| 375 | * |
| 376 | * @param Node $node |
| 377 | * @return bool |
| 378 | */ |
| 379 | public static function isSolTransparentLink( Node $node ): bool { |
| 380 | return $node instanceof Element && |
| 381 | DOMCompat::nodeName( $node ) === 'link' && |
| 382 | DOMUtils::matchRel( $node, TokenUtils::SOL_TRANSPARENT_LINK_REGEX ) !== null; |
| 383 | } |
| 384 | |
| 385 | /** |
| 386 | * Check if '$node' emits wikitext that is sol-transparent in wikitext form. |
| 387 | * This is a test for wikitext that doesn't introduce line breaks. |
| 388 | * |
| 389 | * Comment, whitespace text $nodes, category links, redirect links, behavior |
| 390 | * switches, and include directives currently satisfy this definition. |
| 391 | * |
| 392 | * This should come close to matching TokenUtils.isSolTransparent() |
| 393 | * |
| 394 | * @param Node $node |
| 395 | * @return bool |
| 396 | */ |
| 397 | public static function emitsSolTransparentSingleLineWT( Node $node ): bool { |
| 398 | if ( $node instanceof Text ) { |
| 399 | // NB: We differ here to meet the nl condition. |
| 400 | return (bool)preg_match( '/^[ \t]*$/D', $node->nodeValue ); |
| 401 | } elseif ( self::isRenderingTransparentNode( $node ) ) { |
| 402 | // NB: The only metas in a DOM should be for behavior switches and |
| 403 | // include directives, other than explicit HTML meta tags. This |
| 404 | // differs from our counterpart in Util where ref meta tokens |
| 405 | // haven't been expanded to spans yet. |
| 406 | return true; |
| 407 | } else { |
| 408 | return false; |
| 409 | } |
| 410 | } |
| 411 | |
| 412 | /** |
| 413 | * This is the span added to headings to add fallback ids for when legacy |
| 414 | * and HTML5 ids don't match up. This prevents broken links to legacy ids. |
| 415 | * |
| 416 | * @param Node $node |
| 417 | * @return bool |
| 418 | */ |
| 419 | public static function isFallbackIdSpan( Node $node ): bool { |
| 420 | return DOMUtils::hasNameAndTypeOf( $node, 'span', 'mw:FallbackId' ); |
| 421 | } |
| 422 | |
| 423 | /** |
| 424 | * These are primarily 'metadata'-like $nodes that don't show up in output rendering. |
| 425 | * - In Parsoid output, they are represented by link/meta tags. |
| 426 | * - In the PHP parser, they are completely stripped from the input early on. |
| 427 | * Because of this property, these rendering-transparent $nodes are also |
| 428 | * SOL-transparent for the purposes of parsing behavior. |
| 429 | * |
| 430 | * @param Node $node |
| 431 | * @return bool |
| 432 | */ |
| 433 | public static function isRenderingTransparentNode( Node $node ): bool { |
| 434 | // FIXME: Can we change this entire thing to |
| 435 | // $node instanceof Comment || |
| 436 | // DOMUtils::getDataParsoid($node).stx !== 'html' && |
| 437 | // (DOMCompat::nodeName($node) === 'meta' || DOMCompat::nodeName($node) === 'link') |
| 438 | // |
| 439 | return $node instanceof Comment || |
| 440 | self::isSolTransparentLink( $node ) || ( |
| 441 | // Catch-all for everything else. |
| 442 | $node instanceof Element && |
| 443 | DOMCompat::nodeName( $node ) === 'meta' && |
| 444 | !self::isMarkerAnnotation( $node ) && |
| 445 | ( DOMDataUtils::getDataParsoid( $node )->stx ?? '' ) !== 'html' |
| 446 | ) || self::isFallbackIdSpan( $node ); |
| 447 | } |
| 448 | |
| 449 | /** |
| 450 | * Is $node nested inside a table tag that uses HTML instead of native |
| 451 | * wikitext? |
| 452 | * |
| 453 | * @param Node $node |
| 454 | * @return bool |
| 455 | */ |
| 456 | public static function inHTMLTableTag( Node $node ): bool { |
| 457 | $p = $node->parentNode; |
| 458 | while ( DOMUtils::isTableTag( $p ) ) { |
| 459 | if ( self::isLiteralHTMLNode( $p ) ) { |
| 460 | return true; |
| 461 | } elseif ( DOMCompat::nodeName( $p ) === 'table' ) { |
| 462 | // Don't cross <table> boundaries |
| 463 | return false; |
| 464 | } |
| 465 | $p = $p->parentNode; |
| 466 | } |
| 467 | |
| 468 | return false; |
| 469 | } |
| 470 | |
| 471 | /** |
| 472 | * Is $node the first wrapper element of encapsulated content? |
| 473 | * |
| 474 | * @param Node $node |
| 475 | * @return bool |
| 476 | */ |
| 477 | public static function isFirstEncapsulationWrapperNode( Node $node ): bool { |
| 478 | return DOMUtils::matchTypeOf( $node, self::FIRST_ENCAP_REGEXP ) !== null; |
| 479 | } |
| 480 | |
| 481 | /** |
| 482 | * Is $node the first wrapper element of extension content? |
| 483 | * |
| 484 | * @param Node $node |
| 485 | * @return bool |
| 486 | */ |
| 487 | public static function isFirstExtensionWrapperNode( Node $node ): bool { |
| 488 | return DOMUtils::matchTypeOf( $node, "#mw:Extension/#" ) !== null; |
| 489 | } |
| 490 | |
| 491 | /** |
| 492 | * Checks whether a first encapsulation wrapper node is encapsulating an extension |
| 493 | * that outputs MediaWiki Core DOM Spec HTML (https://www.mediawiki.org/wiki/Specs/HTML) |
| 494 | * @param Node $node |
| 495 | * @param Env $env |
| 496 | * @return bool |
| 497 | */ |
| 498 | public static function isExtensionOutputtingCoreMwDomSpec( Node $node, Env $env ): bool { |
| 499 | if ( DOMUtils::matchTypeOf( $node, self::NON_EXTENSION_ENCAP_REGEXP ) !== null ) { |
| 500 | return false; |
| 501 | } |
| 502 | $extTagName = self::getExtTagName( $node ); |
| 503 | $extConfig = $env->getSiteConfig()->getExtTagConfig( $extTagName ); |
| 504 | $htmlType = $extConfig['options']['outputHasCoreMwDomSpecMarkup'] ?? null; |
| 505 | return $htmlType === true; |
| 506 | } |
| 507 | |
| 508 | /** |
| 509 | * Is $node an encapsulation wrapper elt? |
| 510 | * |
| 511 | * All root-level $nodes of generated content are considered |
| 512 | * encapsulation wrappers and share an about-id. |
| 513 | * |
| 514 | * @param Node $node |
| 515 | * @return bool |
| 516 | */ |
| 517 | public static function isEncapsulationWrapper( Node $node ): bool { |
| 518 | // True if it has an encapsulation type or while walking backwards |
| 519 | // over elts with identical about ids, we run into a $node with an |
| 520 | // encapsulation type. |
| 521 | if ( !( $node instanceof Element ) ) { |
| 522 | return false; |
| 523 | } |
| 524 | return self::findFirstEncapsulationWrapperNode( $node ) !== null; |
| 525 | } |
| 526 | |
| 527 | /** |
| 528 | * Is $node a DOMFragment wrapper? |
| 529 | * |
| 530 | * @param Node $node |
| 531 | * @return bool |
| 532 | */ |
| 533 | public static function isDOMFragmentWrapper( Node $node ): bool { |
| 534 | // See TokenUtils::hasDOMFragmentType |
| 535 | return DOMUtils::matchTypeOf( $node, '#^mw:DOMFragment(/sealed/\w+)?$#D' ) !== null; |
| 536 | } |
| 537 | |
| 538 | /** |
| 539 | * Is $node a sealed DOMFragment of a specific type? |
| 540 | * |
| 541 | * @param Node $node |
| 542 | * @param string $type |
| 543 | * @return bool |
| 544 | */ |
| 545 | public static function isSealedFragmentOfType( Node $node, string $type ): bool { |
| 546 | return DOMUtils::hasTypeOf( $node, "mw:DOMFragment/sealed/$type" ); |
| 547 | } |
| 548 | |
| 549 | /** |
| 550 | * Is $node a Parsoid-generated <section> tag? |
| 551 | * |
| 552 | * @param Node $node |
| 553 | * @return bool |
| 554 | */ |
| 555 | public static function isParsoidSectionTag( Node $node ): bool { |
| 556 | return $node instanceof Element && |
| 557 | DOMCompat::nodeName( $node ) === 'section' && |
| 558 | $node->hasAttribute( 'data-mw-section-id' ); |
| 559 | } |
| 560 | |
| 561 | /** |
| 562 | * Is the $node from extension content? |
| 563 | * @param Node $node |
| 564 | * @param ?string $extType If non-null, checks for that specific extension |
| 565 | * @return bool |
| 566 | */ |
| 567 | public static function fromExtensionContent( Node $node, ?string $extType = null ): bool { |
| 568 | $re = $extType ? "#mw:Extension/$extType#" : "#mw:Extension/\w+#"; |
| 569 | while ( $node && !DOMUtils::atTheTop( $node ) ) { |
| 570 | if ( DOMUtils::matchTypeOf( $node, $re ) ) { |
| 571 | return true; |
| 572 | } |
| 573 | $node = $node->parentNode; |
| 574 | } |
| 575 | return false; |
| 576 | } |
| 577 | |
| 578 | /** |
| 579 | * Is $node from encapsulated (template, extension, etc.) content? |
| 580 | * @param Node $node |
| 581 | * @return bool |
| 582 | */ |
| 583 | public static function fromEncapsulatedContent( Node $node ): bool { |
| 584 | while ( $node && !DOMUtils::atTheTop( $node ) ) { |
| 585 | if ( self::findFirstEncapsulationWrapperNode( $node ) !== null ) { |
| 586 | return true; |
| 587 | } |
| 588 | $node = $node->parentNode; |
| 589 | } |
| 590 | return false; |
| 591 | } |
| 592 | |
| 593 | /** |
| 594 | * Compute, when possible, the wikitext source for a $node in |
| 595 | * an environment env. Returns null if the source cannot be |
| 596 | * extracted. |
| 597 | * @param Frame $frame |
| 598 | * @param Element $node |
| 599 | * @return string|null |
| 600 | */ |
| 601 | public static function getWTSource( Frame $frame, Element $node ): ?string { |
| 602 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 603 | $dsr = $dp->dsr ?? null; |
| 604 | // FIXME: We could probably change the null return to '' |
| 605 | // Just need to verify that code that uses this won't break |
| 606 | return Utils::isValidDSR( $dsr ) ? |
| 607 | $dsr->substr( $frame->getSrcText() ) : null; |
| 608 | } |
| 609 | |
| 610 | /** |
| 611 | * Gets all siblings that follow '$node' that have an 'about' as |
| 612 | * their about id. |
| 613 | * |
| 614 | * This is used to fetch transclusion/extension content by using |
| 615 | * the about-id as the key. This works because |
| 616 | * transclusion/extension content is a forest of dom-trees formed |
| 617 | * by adjacent dom-nodes. This is the contract that template |
| 618 | * encapsulation, dom-reuse, and VE code all have to abide by. |
| 619 | * |
| 620 | * The only exception to this adjacency rule is IEW nodes in |
| 621 | * fosterable positions (in tables) which are not span-wrapped to |
| 622 | * prevent them from getting fostered out. |
| 623 | * |
| 624 | * @param Node $node |
| 625 | * @param ?string $about |
| 626 | * @return Node[] |
| 627 | */ |
| 628 | public static function getAboutSiblings( Node $node, ?string $about ): array { |
| 629 | $nodes = [ $node ]; |
| 630 | |
| 631 | if ( $about === null ) { |
| 632 | return $nodes; |
| 633 | } |
| 634 | |
| 635 | $node = $node->nextSibling; |
| 636 | while ( $node && ( |
| 637 | ( $node instanceof Element && DOMCompat::getAttribute( $node, 'about' ) === $about ) || |
| 638 | ( DOMUtils::isFosterablePosition( $node ) && DOMUtils::isIEW( $node ) ) |
| 639 | ) ) { |
| 640 | $nodes[] = $node; |
| 641 | $node = $node->nextSibling; |
| 642 | } |
| 643 | |
| 644 | // Remove already consumed trailing IEW, if any |
| 645 | while ( count( $nodes ) > 0 && DOMUtils::isIEW( $nodes[count( $nodes ) - 1] ) ) { |
| 646 | array_pop( $nodes ); |
| 647 | } |
| 648 | |
| 649 | return $nodes; |
| 650 | } |
| 651 | |
| 652 | /** |
| 653 | * This function is only intended to be used on encapsulated $nodes |
| 654 | * (Template/Extension/Param content). |
| 655 | * |
| 656 | * Given a '$node' that has an about-id, it is assumed that it is generated |
| 657 | * by templates or extensions. This function skips over all |
| 658 | * following content nodes and returns the first non-template node |
| 659 | * that follows it. |
| 660 | * |
| 661 | * @param Node $node |
| 662 | * @return Node|null |
| 663 | */ |
| 664 | public static function skipOverEncapsulatedContent( Node $node ): ?Node { |
| 665 | $about = $node instanceof Element ? |
| 666 | DOMCompat::getAttribute( $node, 'about' ) : null; |
| 667 | if ( $about !== null ) { |
| 668 | // Guaranteed not to be empty. It will at least include $node. |
| 669 | $aboutSiblings = self::getAboutSiblings( $node, $about ); |
| 670 | return end( $aboutSiblings )->nextSibling; |
| 671 | } else { |
| 672 | return $node->nextSibling; |
| 673 | } |
| 674 | } |
| 675 | |
| 676 | /** |
| 677 | * Comment encoding/decoding. |
| 678 | * |
| 679 | * * Some relevant phab tickets: T94055, T70146, T60184, T95039 |
| 680 | * |
| 681 | * The wikitext comment rule is very simple: <!-- starts a comment, |
| 682 | * and --> ends a comment. This means we can have almost anything as the |
| 683 | * contents of a comment (except the string "-->", but see below), including |
| 684 | * several things that are not valid in HTML5 comments: |
| 685 | * |
| 686 | * * For one, the html5 comment parsing algorithm [0] leniently accepts |
| 687 | * --!> as a closing comment tag, which differs from the php+tidy combo. |
| 688 | * |
| 689 | * * If the comment's data matches /^-?>/, html5 will end the comment. |
| 690 | * For example, <!-->stuff<--> breaks up as |
| 691 | * <!--> (the comment) followed by, stuff<--> (as text). |
| 692 | * |
| 693 | * * Finally, comment data shouldn't contain two consecutive hyphen-minus |
| 694 | * characters (--), nor end in a hyphen-minus character (/-$/) as defined |
| 695 | * in the spec [1]. |
| 696 | * |
| 697 | * We work around all these problems by using HTML entity encoding inside |
| 698 | * the comment body. The characters -, >, and & must be encoded in order |
| 699 | * to prevent premature termination of the comment by one of the cases |
| 700 | * above. Encoding other characters is optional; all entities will be |
| 701 | * decoded during wikitext serialization. |
| 702 | * |
| 703 | * In order to allow *arbitrary* content inside a wikitext comment, |
| 704 | * including the forbidden string "-->" we also do some minimal entity |
| 705 | * decoding on the wikitext. We are also limited by our inability |
| 706 | * to encode DSR attributes on the comment $node, so our wikitext entity |
| 707 | * decoding must be 1-to-1: that is, there must be a unique "decoded" |
| 708 | * string for every wikitext sequence, and for every decoded string there |
| 709 | * must be a unique wikitext which creates it. |
| 710 | * |
| 711 | * The basic idea here is to replace every string ab*c with the string with |
| 712 | * one more b in it. This creates a string with no instance of "ac", |
| 713 | * so you can use 'ac' to encode one more code point. In this case |
| 714 | * a is "--&", "b" is "amp;", and "c" is "gt;" and we use ac to |
| 715 | * encode "-->" (which is otherwise unspeakable in wikitext). |
| 716 | * |
| 717 | * Note that any user content which does not match the regular |
| 718 | * expression /--(>|&(amp;)*gt;)/ is unchanged in its wikitext |
| 719 | * representation, as shown in the first two examples below. |
| 720 | * |
| 721 | * User-authored comment text Wikitext HTML5 DOM |
| 722 | * -------------------------- ------------- ---------------------- |
| 723 | * & - > & - > & + > |
| 724 | * Use > here Use > here Use &gt; here |
| 725 | * --> --> ++> |
| 726 | * --> --&gt; ++&gt; |
| 727 | * --&gt; --&amp;gt; ++&amp;gt; |
| 728 | * |
| 729 | * [0] http://www.w3.org/TR/html5/syntax.html#comment-start-state |
| 730 | * [1] http://www.w3.org/TR/html5/syntax.html#comments |
| 731 | * |
| 732 | * Map a wikitext-escaped comment to an HTML DOM-escaped comment. |
| 733 | * |
| 734 | * @param string $comment Wikitext-escaped comment. |
| 735 | * @return string DOM-escaped comment. |
| 736 | */ |
| 737 | public static function encodeComment( string $comment ): string { |
| 738 | // Undo wikitext escaping to obtain "true value" of comment. |
| 739 | $trueValue = preg_replace_callback( '/--&(amp;)*gt;/', static function ( $m ) { |
| 740 | return Utils::decodeWtEntities( $m[0] ); |
| 741 | }, $comment ); |
| 742 | |
| 743 | // Now encode '-', '>' and '&' in the "true value" as HTML entities, |
| 744 | // so that they can be safely embedded in an HTML comment. |
| 745 | // This part doesn't have to map strings 1-to-1. |
| 746 | return preg_replace_callback( '/[->&]/', static function ( $m ) { |
| 747 | return Utils::entityEncodeAll( $m[0] ); |
| 748 | }, $trueValue ); |
| 749 | } |
| 750 | |
| 751 | /** |
| 752 | * Map an HTML DOM-escaped comment to a wikitext-escaped comment. |
| 753 | * @param string $comment DOM-escaped comment. |
| 754 | * @return string Wikitext-escaped comment. |
| 755 | */ |
| 756 | public static function decodeComment( string $comment ): string { |
| 757 | // Undo HTML entity escaping to obtain "true value" of comment. |
| 758 | $trueValue = Utils::decodeWtEntities( $comment ); |
| 759 | |
| 760 | // ok, now encode this "true value" of the comment in such a way |
| 761 | // that the string "-->" never shows up. (See above.) |
| 762 | return preg_replace_callback( '/--(&(amp;)*gt;|>)/', static function ( $m ) { |
| 763 | $s = $m[0]; |
| 764 | return $s === '-->' ? '-->' : '--&' . substr( $s, 3 ); |
| 765 | }, $trueValue ); |
| 766 | } |
| 767 | |
| 768 | /** |
| 769 | * Utility function: we often need to know the wikitext DSR length for |
| 770 | * an HTML DOM comment value. |
| 771 | * |
| 772 | * @param Comment|CommentTk $node A comment node containing a DOM-escaped comment. |
| 773 | * @return int The wikitext length in UTF-8 bytes necessary to encode this |
| 774 | * comment, including 7 characters for the `<!--` and `-->` delimiters. |
| 775 | */ |
| 776 | public static function decodedCommentLength( $node ): int { |
| 777 | // Add 7 for the "<!--" and "-->" delimiters in wikitext. |
| 778 | $syntaxLen = 7; |
| 779 | if ( $node instanceof Comment ) { |
| 780 | $value = $node->nodeValue; |
| 781 | if ( $node->previousSibling && |
| 782 | DOMUtils::hasTypeOf( $node->previousSibling, "mw:Placeholder/UnclosedComment" ) |
| 783 | ) { |
| 784 | $syntaxLen = 4; |
| 785 | } |
| 786 | } elseif ( $node instanceof CommentTk ) { |
| 787 | // @phan-suppress-next-line PhanUndeclaredProperty dynamic property |
| 788 | if ( isset( $node->dataParsoid->unclosedComment ) ) { |
| 789 | $syntaxLen = 4; |
| 790 | } |
| 791 | $value = $node->value; |
| 792 | } else { |
| 793 | throw new UnreachableException( 'Should not be here!' ); |
| 794 | } |
| 795 | return strlen( self::decodeComment( $value ) ) + $syntaxLen; |
| 796 | } |
| 797 | |
| 798 | /** |
| 799 | * @param Node $node |
| 800 | * @return ?string |
| 801 | */ |
| 802 | public static function getExtTagName( Node $node ): ?string { |
| 803 | $match = DOMUtils::matchTypeOf( $node, '#^mw:Extension/(.+?)$#D' ); |
| 804 | return $match ? mb_strtolower( substr( $match, strlen( 'mw:Extension/' ) ) ) : null; |
| 805 | } |
| 806 | |
| 807 | /** |
| 808 | * @param Env $env |
| 809 | * @param Node $node |
| 810 | * @return ?ExtensionTagHandler |
| 811 | */ |
| 812 | public static function getNativeExt( Env $env, Node $node ): ?ExtensionTagHandler { |
| 813 | $extTagName = self::getExtTagName( $node ); |
| 814 | return $extTagName ? $env->getSiteConfig()->getExtTagImpl( $extTagName ) : null; |
| 815 | } |
| 816 | |
| 817 | /** |
| 818 | * Is this an include directive? |
| 819 | * @param string $name |
| 820 | * @return bool |
| 821 | */ |
| 822 | public static function isIncludeTag( string $name ): bool { |
| 823 | return $name === 'includeonly' || $name === 'noinclude' || $name === 'onlyinclude'; |
| 824 | } |
| 825 | |
| 826 | /** |
| 827 | * Check if tag is annotation or extension directive |
| 828 | * Adapted from similar grammar function |
| 829 | * |
| 830 | * @param Env $env |
| 831 | * @param string $name |
| 832 | * @return bool |
| 833 | */ |
| 834 | public static function isAnnOrExtTag( Env $env, string $name ): bool { |
| 835 | $tagName = mb_strtolower( $name ); |
| 836 | $siteConfig = $env->getSiteConfig(); |
| 837 | $extTags = $siteConfig->getExtensionTagNameMap(); |
| 838 | $isInstalledExt = isset( $extTags[$tagName] ); |
| 839 | $isIncludeTag = self::isIncludeTag( $tagName ); |
| 840 | $isAnnotationTag = $siteConfig->isAnnotationTag( $tagName ); |
| 841 | |
| 842 | if ( !$isAnnotationTag ) { |
| 843 | // avoid crashing on <tvar|name> even if we don't support that syntax explicitly |
| 844 | $pipepos = strpos( $tagName, '|' ); |
| 845 | if ( $pipepos ) { |
| 846 | $strBeforePipe = substr( $tagName, 0, $pipepos ); |
| 847 | $isAnnotationTag = $siteConfig->isAnnotationTag( $strBeforePipe ); |
| 848 | } |
| 849 | } |
| 850 | return $isInstalledExt || $isIncludeTag || $isAnnotationTag; |
| 851 | } |
| 852 | |
| 853 | /** |
| 854 | * Creates a DocumentFragment containing a single span with type "mw:I18n". The created span |
| 855 | * should be filled in with setDataNodeI18n to be valid. |
| 856 | * @param Document $doc |
| 857 | * @return DocumentFragment |
| 858 | * @throws DOMException |
| 859 | */ |
| 860 | public static function createEmptyLocalizationFragment( Document $doc ): DocumentFragment { |
| 861 | $frag = $doc->createDocumentFragment(); |
| 862 | $span = $doc->createElement( 'span' ); |
| 863 | DOMUtils::addTypeOf( $span, 'mw:I18n' ); |
| 864 | $frag->appendChild( $span ); |
| 865 | return $frag; |
| 866 | } |
| 867 | |
| 868 | /** |
| 869 | * Creates an internationalization (i18n) message that will be localized into the page content |
| 870 | * language. The returned DocumentFragment contains, as a single child, a span |
| 871 | * element with the appropriate information for later localization. |
| 872 | * @param Document $doc |
| 873 | * @param string $key message key for the message to be localized |
| 874 | * @param ?array $params parameters for localization |
| 875 | * @return DocumentFragment |
| 876 | * @throws DOMException |
| 877 | */ |
| 878 | public static function createPageContentI18nFragment( |
| 879 | Document $doc, string $key, ?array $params = null |
| 880 | ): DocumentFragment { |
| 881 | $frag = self::createEmptyLocalizationFragment( $doc ); |
| 882 | $i18n = I18nInfo::createPageContentI18n( $key, $params ); |
| 883 | DOMDataUtils::setDataNodeI18n( $frag->firstChild, $i18n ); |
| 884 | return $frag; |
| 885 | } |
| 886 | |
| 887 | /** |
| 888 | * Creates an internationalization (i18n) message that will be localized into the user |
| 889 | * interface language. The returned DocumentFragment contains, as a single child, a span |
| 890 | * element with the appropriate information for later localization. |
| 891 | * @param Document $doc |
| 892 | * @param string $key message key for the message to be localized |
| 893 | * @param ?array $params parameters for localization |
| 894 | * @return DocumentFragment |
| 895 | * @throws DOMException |
| 896 | */ |
| 897 | public static function createInterfaceI18nFragment( |
| 898 | Document $doc, string $key, ?array $params = null |
| 899 | ): DocumentFragment { |
| 900 | $frag = self::createEmptyLocalizationFragment( $doc ); |
| 901 | $i18n = I18nInfo::createInterfaceI18n( $key, $params ); |
| 902 | DOMDataUtils::setDataNodeI18n( $frag->firstChild, $i18n ); |
| 903 | return $frag; |
| 904 | } |
| 905 | |
| 906 | /** |
| 907 | * Creates an internationalization (i18n) message that will be localized into an arbitrary |
| 908 | * language. The returned DocumentFragment contains, as a single child, a span |
| 909 | * element with the appropriate information for later localization. |
| 910 | * The use of this method is discouraged; use ::createPageContentI18nFragment(...) and |
| 911 | * ::createInterfaceI18nFragment(...) where possible rather than, respectively, |
| 912 | * ::createLangI18nFragment(..., $wgContLang, ...) and |
| 913 | * ::createLangI18nFragment(..., $wgLang,...). |
| 914 | * @param Document $doc |
| 915 | * @param Bcp47Code $lang language for the localization |
| 916 | * @param string $key message key for the message to be localized |
| 917 | * @param ?array $params parameters for localization |
| 918 | * @return DocumentFragment |
| 919 | * @throws DOMException |
| 920 | */ |
| 921 | public static function createLangI18nFragment( |
| 922 | Document $doc, Bcp47Code $lang, string $key, ?array $params = null |
| 923 | ): DocumentFragment { |
| 924 | $frag = self::createEmptyLocalizationFragment( $doc ); |
| 925 | $i18n = I18nInfo::createLangI18n( $lang, $key, $params ); |
| 926 | DOMDataUtils::setDataNodeI18n( $frag->firstChild, $i18n ); |
| 927 | return $frag; |
| 928 | } |
| 929 | |
| 930 | /** |
| 931 | * Adds to $element the internationalization information needed for the attribute $name to be |
| 932 | * localized in a later pass into the page content language. |
| 933 | * @param Element $element element on which to add internationalization information |
| 934 | * @param string $name name of the attribute whose value will be localized |
| 935 | * @param string $key message key used for the attribute value localization |
| 936 | * @param ?array $params parameters for localization |
| 937 | */ |
| 938 | public static function addPageContentI18nAttribute( |
| 939 | Element $element, string $name, string $key, ?array $params = null |
| 940 | ): void { |
| 941 | $i18n = I18nInfo::createPageContentI18n( $key, $params ); |
| 942 | DOMUtils::addTypeOf( $element, 'mw:LocalizedAttrs' ); |
| 943 | DOMDataUtils::setDataAttrI18n( $element, $name, $i18n ); |
| 944 | } |
| 945 | |
| 946 | /** Adds to $element the internationalization information needed for the attribute $name to be |
| 947 | * localized in a later pass into the user interface language. |
| 948 | * @param Element $element element on which to add internationalization information |
| 949 | * @param string $name name of the attribute whose value will be localized |
| 950 | * @param string $key message key used for the attribute value localization |
| 951 | * @param ?array $params parameters for localization |
| 952 | */ |
| 953 | public static function addInterfaceI18nAttribute( |
| 954 | Element $element, string $name, string $key, ?array $params = null |
| 955 | ): void { |
| 956 | $i18n = I18nInfo::createInterfaceI18n( $key, $params ); |
| 957 | DOMUtils::addTypeOf( $element, 'mw:LocalizedAttrs' ); |
| 958 | DOMDataUtils::setDataAttrI18n( $element, $name, $i18n ); |
| 959 | } |
| 960 | |
| 961 | /** |
| 962 | * Adds to $element the internationalization information needed for the attribute $name to be |
| 963 | * localized in a later pass into the provided language. |
| 964 | * The use of this method is discouraged; ; use ::addPageContentI18nAttribute(...) and |
| 965 | * ::addInterfaceI18nAttribute(...) where possible rather than, respectively, |
| 966 | * ::addLangI18nAttribute(..., $wgContLang, ...) and ::addLangI18nAttribute(..., $wgLang, ...). |
| 967 | * @param Element $element element on which to add internationalization information |
| 968 | * @param Bcp47Code $lang language in which the message will be localized |
| 969 | * @param string $name name of the attribute whose value will be localized |
| 970 | * @param string $key message key used for the attribute value localization |
| 971 | * @param ?array $params parameters for localization |
| 972 | */ |
| 973 | public static function addLangI18nAttribute( |
| 974 | Element $element, Bcp47Code $lang, string $name, string $key, ?array $params = null |
| 975 | ): void { |
| 976 | $i18n = I18nInfo::createLangI18n( $lang, $key, $params ); |
| 977 | DOMUtils::addTypeOf( $element, 'mw:LocalizedAttrs' ); |
| 978 | DOMDataUtils::setDataAttrI18n( $element, $name, $i18n ); |
| 979 | } |
| 980 | |
| 981 | /** Check whether a node is an annotation meta; if yes, returns its type |
| 982 | * @param Node $node |
| 983 | * @return ?string |
| 984 | */ |
| 985 | public static function matchAnnotationMeta( Node $node ): ?string { |
| 986 | return DOMUtils::matchNameAndTypeOf( $node, 'meta', self::ANNOTATION_META_TYPE_REGEXP ); |
| 987 | } |
| 988 | |
| 989 | /** |
| 990 | * Extract the annotation type, excluding potential "/End" suffix; returns null if not a valid |
| 991 | * annotation meta. &$isStart is set to true if the annotation is a start tag, false otherwise. |
| 992 | * |
| 993 | * @param Node $node |
| 994 | * @param bool &$isStart |
| 995 | * @return ?string The matched type, or null if no match. |
| 996 | */ |
| 997 | public static function extractAnnotationType( Node $node, bool &$isStart = false ): ?string { |
| 998 | $t = DOMUtils::matchTypeOf( $node, self::ANNOTATION_META_TYPE_REGEXP ); |
| 999 | if ( $t !== null && preg_match( self::ANNOTATION_META_TYPE_REGEXP, $t, $matches ) ) { |
| 1000 | $isStart = !str_ends_with( $t, '/End' ); |
| 1001 | return $matches[1]; |
| 1002 | } |
| 1003 | return null; |
| 1004 | } |
| 1005 | |
| 1006 | /** |
| 1007 | * Check whether a node is a meta signifying the start of an annotated part of the DOM |
| 1008 | * |
| 1009 | * @param Node $node |
| 1010 | * @return bool |
| 1011 | */ |
| 1012 | public static function isAnnotationStartMarkerMeta( Node $node ): bool { |
| 1013 | if ( !$node instanceof Element || DOMCompat::nodeName( $node ) !== 'meta' ) { |
| 1014 | return false; |
| 1015 | } |
| 1016 | $isStart = false; |
| 1017 | $t = self::extractAnnotationType( $node, $isStart ); |
| 1018 | return $t !== null && $isStart; |
| 1019 | } |
| 1020 | |
| 1021 | /** |
| 1022 | * Check whether a node is a meta signifying the end of an annotated part of the DOM |
| 1023 | * |
| 1024 | * @param Node $node |
| 1025 | * @return bool |
| 1026 | */ |
| 1027 | public static function isAnnotationEndMarkerMeta( Node $node ): bool { |
| 1028 | if ( !$node instanceof Element || DOMCompat::nodeName( $node ) !== 'meta' ) { |
| 1029 | return false; |
| 1030 | } |
| 1031 | $isStart = false; |
| 1032 | $t = self::extractAnnotationType( $node, $isStart ); |
| 1033 | return $t !== null && !$isStart; |
| 1034 | } |
| 1035 | |
| 1036 | /** |
| 1037 | * Check whether the meta tag was moved from its initial position |
| 1038 | * @param Node $node |
| 1039 | * @return bool |
| 1040 | */ |
| 1041 | public static function isMovedMetaTag( Node $node ): bool { |
| 1042 | if ( $node instanceof Element && self::matchAnnotationMeta( $node ) !== null ) { |
| 1043 | $parsoidData = DOMDataUtils::getDataParsoid( $node ); |
| 1044 | if ( isset( $parsoidData->wasMoved ) ) { |
| 1045 | return $parsoidData->wasMoved; |
| 1046 | } |
| 1047 | } |
| 1048 | return false; |
| 1049 | } |
| 1050 | |
| 1051 | /** Returns true if a node is a (start or end) annotation meta tag |
| 1052 | * @param ?Node $n |
| 1053 | * @return bool |
| 1054 | */ |
| 1055 | public static function isMarkerAnnotation( ?Node $n ): bool { |
| 1056 | return $n !== null && self::matchAnnotationMeta( $n ) !== null; |
| 1057 | } |
| 1058 | |
| 1059 | /** |
| 1060 | * Extracts the media format from the attribute string |
| 1061 | * |
| 1062 | * @param Element $node |
| 1063 | * @return string |
| 1064 | */ |
| 1065 | public static function getMediaFormat( Element $node ): string { |
| 1066 | // TODO: Remove "Image|Video|Audio" when version 2.4.0 of the content |
| 1067 | // is no longer supported |
| 1068 | $mediaType = DOMUtils::matchTypeOf( $node, '#^mw:(File|Image|Video|Audio)(/|$)#' ); |
| 1069 | $parts = explode( '/', $mediaType ?? '' ); |
| 1070 | return $parts[1] ?? ''; |
| 1071 | } |
| 1072 | |
| 1073 | /** |
| 1074 | * @param Element $node |
| 1075 | * @return bool |
| 1076 | */ |
| 1077 | public static function hasVisibleCaption( Element $node ): bool { |
| 1078 | $format = self::getMediaFormat( $node ); |
| 1079 | return in_array( |
| 1080 | $format, [ 'Thumb', /* 'Manualthumb', FIXME(T305759) */ 'Frame' ], true |
| 1081 | ); |
| 1082 | } |
| 1083 | |
| 1084 | /** |
| 1085 | * Ref dom post-processing happens after adding media info, so the |
| 1086 | * linkbacks aren't available in the textContent added to the alt. |
| 1087 | * However, when serializing, they are in the caption elements. So, this |
| 1088 | * special handler drops the linkbacks for the purpose of comparison. |
| 1089 | * |
| 1090 | * @param Node $node |
| 1091 | * @return string |
| 1092 | */ |
| 1093 | public static function textContentFromCaption( Node $node ): string { |
| 1094 | $content = ''; |
| 1095 | $c = $node->firstChild; |
| 1096 | while ( $c ) { |
| 1097 | if ( $c instanceof Text ) { |
| 1098 | $content .= $c->nodeValue; |
| 1099 | } elseif ( |
| 1100 | $c instanceof Element && |
| 1101 | !DOMUtils::isMetaDataTag( $c ) && |
| 1102 | !DOMUtils::hasTypeOf( $c, "mw:Extension/ref" ) |
| 1103 | ) { |
| 1104 | $content .= self::textContentFromCaption( $c ); |
| 1105 | } |
| 1106 | $c = $c->nextSibling; |
| 1107 | } |
| 1108 | return $content; |
| 1109 | } |
| 1110 | |
| 1111 | } |