Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
92.39% |
85 / 92 |
|
50.00% |
3 / 6 |
CRAP | |
0.00% |
0 / 1 |
| HandleParsoidSectionLinks | |
92.39% |
85 / 92 |
|
50.00% |
3 / 6 |
31.42 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| shouldRun | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isHtmlHeading | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
5.07 | |||
| transformDOM | |
88.10% |
37 / 42 |
|
0.00% |
0 / 1 |
12.24 | |||
| transformHeading | |
100.00% |
36 / 36 |
|
100.00% |
1 / 1 |
10 | |||
| resolveSkin | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
2.06 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace MediaWiki\OutputTransform\Stages; |
| 5 | |
| 6 | use MediaWiki\Config\ServiceOptions; |
| 7 | use MediaWiki\Context\RequestContext; |
| 8 | use MediaWiki\OutputTransform\ContentDOMTransformStage; |
| 9 | use MediaWiki\Parser\ParserOptions; |
| 10 | use MediaWiki\Parser\ParserOutput; |
| 11 | use MediaWiki\Parser\ParserOutputFlags; |
| 12 | use MediaWiki\Parser\Sanitizer; |
| 13 | use MediaWiki\Skin\Skin; |
| 14 | use MediaWiki\Title\TitleFactory; |
| 15 | use Psr\Log\LoggerInterface; |
| 16 | use Wikimedia\Parsoid\Core\SectionMetadata; |
| 17 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 18 | use Wikimedia\Parsoid\DOM\Element; |
| 19 | use Wikimedia\Parsoid\DOM\Node; |
| 20 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 21 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 22 | use Wikimedia\Parsoid\Utils\DOMTraverser; |
| 23 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 24 | |
| 25 | /** |
| 26 | * Add anchors and other heading formatting, and replace the section link placeholders. |
| 27 | * @internal |
| 28 | */ |
| 29 | class HandleParsoidSectionLinks extends ContentDOMTransformStage { |
| 30 | // See below: if/when PHP implements DocumentFragment::getElementById() |
| 31 | // efficiently, set this to true. |
| 32 | private static bool $useGetElementById = false; |
| 33 | |
| 34 | private TitleFactory $titleFactory; |
| 35 | |
| 36 | public function __construct( |
| 37 | ServiceOptions $options, LoggerInterface $logger, TitleFactory $titleFactory |
| 38 | ) { |
| 39 | parent::__construct( $options, $logger ); |
| 40 | $this->titleFactory = $titleFactory; |
| 41 | } |
| 42 | |
| 43 | public function shouldRun( ParserOutput $po, ?ParserOptions $popts, array $options = [] ): bool { |
| 44 | // Only run this stage if it is parsoid content |
| 45 | return $po->getContentHolder()->isParsoidContent(); |
| 46 | } |
| 47 | |
| 48 | /** |
| 49 | * Check if the heading has attributes that can only be added using HTML syntax. |
| 50 | * |
| 51 | * In the Parsoid default future, we might prefer checking for stx=html. |
| 52 | */ |
| 53 | private static function isHtmlHeading( Element $h ): bool { |
| 54 | foreach ( $h->attributes as $attr ) { |
| 55 | // Condition matches DiscussionTool's CommentFormatter::handleHeading |
| 56 | if ( |
| 57 | !in_array( $attr->name, [ 'id', 'data-object-id', 'about', 'typeof' ], true ) && |
| 58 | !Sanitizer::isReservedDataAttribute( $attr->name ) |
| 59 | ) { |
| 60 | return true; |
| 61 | } |
| 62 | } |
| 63 | // FIXME(T100856): stx info probably shouldn't be in data-parsoid |
| 64 | // Id is ignored above since it's a special case, make use of metadata |
| 65 | // to determine if it came from wikitext |
| 66 | if ( DOMDataUtils::getDataParsoid( $h )->reusedId ?? false ) { |
| 67 | return true; |
| 68 | } |
| 69 | return false; |
| 70 | } |
| 71 | |
| 72 | public function transformDOM( |
| 73 | DocumentFragment $df, ParserOutput $po, ?ParserOptions $popts, array &$options |
| 74 | ): DocumentFragment { |
| 75 | $skin = $this->resolveSkin( $options ); |
| 76 | // Transform: |
| 77 | // <section data-mw-section-id=...> |
| 78 | // <h2 id=...><span id=... typeof="mw:FallbackId"></span> ... </h2> |
| 79 | // ...section contents.. |
| 80 | // To: |
| 81 | // <section data-mw-section-id=...> |
| 82 | // <div class="mw-heading mw-heading2"> |
| 83 | // <h2 id=...><span id=... typeof="mw:FallbackId"></span> ... </h2> |
| 84 | // <span class="mw-editsection">...section edit link...</span> |
| 85 | // </div> |
| 86 | // That is, we're wrapping a <div> around the <h2> generated by |
| 87 | // Parsoid, and then (assuming section edit links are enabled) |
| 88 | // adding a <span> with the section edit link |
| 89 | // inside that <div> |
| 90 | // |
| 91 | // If COLLAPSIBLE_SECTIONS is set, then we also wrap a <div> |
| 92 | // around the section *contents*. |
| 93 | $toc = $po->getTOCData(); |
| 94 | $sections = ( $toc !== null ) ? $toc->getSections() : []; |
| 95 | $sectionMap = []; |
| 96 | foreach ( $sections as $section ) { |
| 97 | if ( $section->anchor === '' ) { |
| 98 | // T375002 / T368722: The empty string isn't a valid id so |
| 99 | // Parsoid will have reassigned it and we'll never be able |
| 100 | // to select by it below. There's no sense in logging an |
| 101 | // error since it's a common enough occurrence at present. |
| 102 | continue; |
| 103 | } |
| 104 | $sectionMap[$section->anchor] = [ |
| 105 | 'processed' => false, |
| 106 | 'section' => $section |
| 107 | ]; |
| 108 | } |
| 109 | |
| 110 | if ( self::$useGetElementById ) { |
| 111 | // This version will be faster if we have an efficient O(1) |
| 112 | // implementation of DocumentFragment::getElementById() |
| 113 | // https://github.com/php/php-src/issues/20282 |
| 114 | foreach ( $sectionMap as $anchor => &$info ) { |
| 115 | $h = DOMCompat::getElementById( $df, $anchor ); |
| 116 | if ( $h !== null ) { |
| 117 | $this->transformHeading( $df, $h, $po, $options, $skin, $info ); |
| 118 | } |
| 119 | } |
| 120 | } else { |
| 121 | // Older PHP versions must traverse the entire DOM to find the |
| 122 | // heading nodes. |
| 123 | $traverser = new DOMTraverser( false, false ); |
| 124 | $headings = array_fill_keys( |
| 125 | [ 'h1', 'h2', 'h3', 'h4', 'h5', 'h6' ], true |
| 126 | ); |
| 127 | $traverser->addHandler( null, function ( Node $node ) use ( |
| 128 | $df, $po, $options, $skin, &$sectionMap, $headings |
| 129 | ) { |
| 130 | if ( !( $headings[DOMUtils::nodeName( $node )] ?? false ) ) { |
| 131 | return true; |
| 132 | } |
| 133 | '@phan-var Element $node'; |
| 134 | $id = DOMCompat::getAttribute( $node, 'id' ); |
| 135 | if ( $id === null ) { |
| 136 | return true; |
| 137 | } |
| 138 | if ( !isset( $sectionMap[$id] ) ) { |
| 139 | return true; |
| 140 | } |
| 141 | return $this->transformHeading( |
| 142 | $df, $node, $po, $options, $skin, $sectionMap[$id] |
| 143 | ); |
| 144 | } ); |
| 145 | $traverser->traverse( null, $df ); |
| 146 | } |
| 147 | |
| 148 | foreach ( $sectionMap as $id => $sectionInfo ) { |
| 149 | if ( !$sectionInfo['processed'] ) { |
| 150 | $this->logger->error( |
| 151 | __METHOD__ . ': Heading missing for anchor', |
| 152 | $sectionInfo['section']->toLegacy() |
| 153 | ); |
| 154 | } |
| 155 | } |
| 156 | return $df; |
| 157 | } |
| 158 | |
| 159 | /** |
| 160 | * @param DocumentFragment $df |
| 161 | * @param Element $h |
| 162 | * @param ParserOutput $po |
| 163 | * @param array $options |
| 164 | * @param Skin $skin |
| 165 | * @param array{section:SectionMetadata,processed:bool} &$sectionInfo |
| 166 | * @return Node|null|bool |
| 167 | */ |
| 168 | private function transformHeading( |
| 169 | DocumentFragment $df, Element $h, ParserOutput $po, array $options, Skin $skin, array &$sectionInfo |
| 170 | ) { |
| 171 | $sectionInfo['processed'] = true; |
| 172 | $section = $sectionInfo['section']; |
| 173 | |
| 174 | if ( self::isHtmlHeading( $h ) ) { |
| 175 | // This is a <h#> tag with attributes added using HTML syntax. |
| 176 | // Mark it with a class to make them easier to distinguish (T68637). |
| 177 | DOMCompat::getClassList( $h )->add( 'mw-html-heading' ); |
| 178 | |
| 179 | // Do not add the wrapper if the heading has attributes added using HTML syntax (T353489). |
| 180 | return true; |
| 181 | } |
| 182 | |
| 183 | $next = $h->nextSibling; |
| 184 | |
| 185 | $fromTitle = $section->fromTitle; |
| 186 | $div = $df->ownerDocument->createElement( 'div' ); |
| 187 | if ( |
| 188 | $fromTitle !== null && |
| 189 | ( $options['enableSectionEditLinks'] ?? true ) && |
| 190 | !$po->getOutputFlag( ParserOutputFlags::NO_SECTION_EDIT_LINKS ) |
| 191 | ) { |
| 192 | $editPage = $this->titleFactory->newFromTextThrow( $fromTitle ); |
| 193 | $html = $skin->doEditSectionLink( |
| 194 | $editPage, $section->index, $h->textContent, |
| 195 | $skin->getLanguage() |
| 196 | ); |
| 197 | DOMCompat::setInnerHTML( $div, $html ); |
| 198 | } |
| 199 | |
| 200 | // Reuse existing wrapper if present. |
| 201 | $maybeWrapper = $h->parentNode; |
| 202 | '@phan-var \Wikimedia\Parsoid\DOM\Element $maybeWrapper'; |
| 203 | if ( |
| 204 | DOMUtils::nodeName( $maybeWrapper ) === 'div' && |
| 205 | DOMCompat::getClassList( $maybeWrapper )->contains( 'mw-heading' ) |
| 206 | ) { |
| 207 | // Transfer section edit link children to existing wrapper |
| 208 | // All contents of the div (the section edit link) will be |
| 209 | // inserted immediately following the <h> tag |
| 210 | $ref = $h->nextSibling; |
| 211 | while ( $div->firstChild !== null ) { |
| 212 | // @phan-suppress-next-line PhanTypeMismatchArgumentNullableInternal firstChild is non-null (PHP81) |
| 213 | $maybeWrapper->insertBefore( $div->firstChild, $ref ); |
| 214 | } |
| 215 | $div = $maybeWrapper; // for use below |
| 216 | } else { |
| 217 | // Move <hX> to new wrapper: the div contents are currently |
| 218 | // the section edit link. We first replace the h with the |
| 219 | // div, then insert the <h> as the first child of the div |
| 220 | // so the section edit link is immediately following the <h>. |
| 221 | $div->setAttribute( |
| 222 | 'class', 'mw-heading mw-heading' . $section->hLevel |
| 223 | ); |
| 224 | $h->parentNode->replaceChild( $div, $h ); |
| 225 | // Work around bug in phan (https://github.com/phan/phan/pull/4837) |
| 226 | // by asserting that $div->firstChild is non-null here. Actually, |
| 227 | // ::insertBefore will work fine if $div->firstChild is null (if |
| 228 | // "doEditSectionLink" returned nothing, for instance), but |
| 229 | // phan incorrectly thinks the second argument must be non-null. |
| 230 | $divFirstChild = $div->firstChild; |
| 231 | '@phan-var \DOMNode $divFirstChild'; // asserting non-null (PHP81) |
| 232 | $div->insertBefore( $h, $divFirstChild ); |
| 233 | } |
| 234 | // Create collapsible section wrapper if requested. |
| 235 | if ( $po->getOutputFlag( ParserOutputFlags::COLLAPSIBLE_SECTIONS ) ) { |
| 236 | $contentsDiv = $df->ownerDocument->createElement( 'div' ); |
| 237 | while ( $div->nextSibling !== null ) { |
| 238 | // @phan-suppress-next-line PhanTypeMismatchArgumentNullableInternal |
| 239 | $contentsDiv->appendChild( $div->nextSibling ); |
| 240 | } |
| 241 | $div->parentNode->appendChild( $contentsDiv ); |
| 242 | } |
| 243 | |
| 244 | return $next; |
| 245 | } |
| 246 | |
| 247 | /** |
| 248 | * Extracts the skin from the $options array, with a fallback on request context skin |
| 249 | * @param array $options |
| 250 | * @return Skin |
| 251 | */ |
| 252 | private function resolveSkin( array $options ): Skin { |
| 253 | $skin = $options[ 'skin' ] ?? null; |
| 254 | if ( !$skin ) { |
| 255 | // T348853 passing $skin will be mandatory in the future |
| 256 | $skin = RequestContext::getMain()->getSkin(); |
| 257 | } |
| 258 | return $skin; |
| 259 | } |
| 260 | } |