Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 93 |
|
0.00% |
0 / 7 |
CRAP | |
0.00% |
0 / 1 |
| EncapsulatedContentHandler | |
0.00% |
0 / 93 |
|
0.00% |
0 / 7 |
1892 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| handle | |
0.00% |
0 / 36 |
|
0.00% |
0 / 1 |
90 | |||
| before | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
110 | |||
| after | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
42 | |||
| handleListPrefix | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
56 | |||
| isTplListWithoutSharedPrefix | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
42 | |||
| parentBulletsHaveBeenEmitted | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
20 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Html2Wt\DOMHandlers; |
| 5 | |
| 6 | use LogicException; |
| 7 | use Wikimedia\Assert\Assert; |
| 8 | use Wikimedia\Parsoid\Core\ClientError; |
| 9 | use Wikimedia\Parsoid\DOM\Element; |
| 10 | use Wikimedia\Parsoid\DOM\Node; |
| 11 | use Wikimedia\Parsoid\Html2Wt\SerializerState; |
| 12 | use Wikimedia\Parsoid\Utils\DiffDOMUtils; |
| 13 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 14 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 15 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 16 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 17 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 18 | |
| 19 | class EncapsulatedContentHandler extends DOMHandler { |
| 20 | |
| 21 | /** @var array[] Maps list item HTML elements to the expected parent element */ |
| 22 | private $parentMap = [ |
| 23 | 'li' => [ 'ul', 'ol' ], |
| 24 | 'dt' => [ 'dl' ], |
| 25 | 'dd' => [ 'dl' ], |
| 26 | ]; |
| 27 | |
| 28 | public function __construct() { |
| 29 | parent::__construct( false ); |
| 30 | } |
| 31 | |
| 32 | /** |
| 33 | * @inheritDoc |
| 34 | * @throws ClientError |
| 35 | */ |
| 36 | public function handle( |
| 37 | Element $node, SerializerState $state, bool $wrapperUnmodified = false |
| 38 | ): ?Node { |
| 39 | $env = $state->getEnv(); |
| 40 | $serializer = $state->serializer; |
| 41 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 42 | $dataMw = DOMDataUtils::getDataMw( $node ); |
| 43 | $src = null; |
| 44 | $transclusionType = DOMUtils::matchTypeOf( $node, '/^mw:(Transclusion|Param)$/' ); |
| 45 | $extTagName = WTUtils::getExtTagName( $node ); |
| 46 | if ( $transclusionType ) { |
| 47 | if ( is_array( $dataMw->parts ?? null ) ) { |
| 48 | $src = $serializer->serializeFromParts( $state, $node, $dataMw->parts ); |
| 49 | } elseif ( isset( $dp->src ) ) { |
| 50 | $env->log( 'error', 'data-mw.parts is not an array: ', DOMCompat::getOuterHTML( $node ), |
| 51 | PHPUtils::jsonEncode( $dataMw ) ); |
| 52 | $src = $dp->src; |
| 53 | } else { |
| 54 | throw new ClientError( |
| 55 | "Cannot serialize $transclusionType without data-mw.parts or data-parsoid.src" |
| 56 | ); |
| 57 | } |
| 58 | } elseif ( $extTagName ) { |
| 59 | // Set name since downstream code assumes it |
| 60 | if ( ( $dataMw->name ?? '' ) === '' ) { |
| 61 | $dataMw->name = $extTagName; |
| 62 | } |
| 63 | $src = false; |
| 64 | $ext = $env->getSiteConfig()->getExtTagImpl( $extTagName ); |
| 65 | if ( $ext ) { |
| 66 | $src = $ext->domToWikitext( $state->extApi, $node, $wrapperUnmodified ); |
| 67 | } |
| 68 | if ( $src === false ) { |
| 69 | $src = $serializer->defaultExtensionHandler( $node, $state ); |
| 70 | } |
| 71 | } elseif ( DOMUtils::hasTypeOf( $node, 'mw:LanguageVariant' ) ) { |
| 72 | $state->serializer->languageVariantHandler( $node ); |
| 73 | return $node->nextSibling; |
| 74 | } else { |
| 75 | throw new LogicException( 'Should never reach here' ); |
| 76 | } |
| 77 | $state->singleLineContext->disable(); |
| 78 | // FIXME: https://phabricator.wikimedia.org/T184779 |
| 79 | $src = ( $dataMw->extPrefix ?? '' ) . $src |
| 80 | . ( $dataMw->extSuffix ?? '' ); |
| 81 | $serializer->emitWikitext( $this->handleListPrefix( $node, $state ) . $src, $node ); |
| 82 | $state->singleLineContext->pop(); |
| 83 | return WTUtils::skipOverEncapsulatedContent( $node ); |
| 84 | } |
| 85 | |
| 86 | // XXX: This is questionable, as the template can expand |
| 87 | // to newlines too. Which default should we pick for new |
| 88 | // content? We don't really want to make separator |
| 89 | // newlines in HTML significant for the semantics of the |
| 90 | // template content. |
| 91 | |
| 92 | /** @inheritDoc */ |
| 93 | public function before( Element $node, Node $otherNode, SerializerState $state ): array { |
| 94 | // Handle native extension constraints. Only apply to plain extension tags. |
| 95 | $extTagName = WTUtils::getExtTagName( $node ); |
| 96 | if ( $extTagName && !DOMUtils::hasTypeOf( $node, 'mw:Transclusion' ) ) { |
| 97 | $extConfig = $state->getEnv()->getSiteConfig()->getExtTagConfig( $extTagName ); |
| 98 | if ( |
| 99 | ( $extConfig['options']['html2wt']['format'] ?? '' ) === 'block' && |
| 100 | WTUtils::isNewElt( $node ) |
| 101 | ) { |
| 102 | return [ 'min' => 1, 'max' => 2 ]; |
| 103 | } |
| 104 | } |
| 105 | |
| 106 | // If this content came from a multi-part-template-block |
| 107 | // use the first node in that block for determining |
| 108 | // newline constraints. |
| 109 | $dp = DOMDataUtils::getDataParsoid( $node ); |
| 110 | if ( isset( $dp->firstWikitextNode ) ) { |
| 111 | // Note: this should match the case returned by DOMCompat::nodeName |
| 112 | // so that this is effectively a case-insensitive comparison here. |
| 113 | // (ie, data-parsoid could have either uppercase tag names or |
| 114 | // lowercase tag names and this code should still work.) |
| 115 | $ftn = mb_strtolower( $dp->firstWikitextNode, "UTF-8" ); |
| 116 | $h = ( new DOMHandlerFactory )->newFromTagHandler( $ftn ); |
| 117 | if ( !$h && ( $dp->stx ?? null ) === 'html' && $ftn !== 'a_html' ) { |
| 118 | $h = new FallbackHTMLHandler(); |
| 119 | } |
| 120 | if ( $h ) { |
| 121 | return $h->before( $node, $otherNode, $state ); |
| 122 | } |
| 123 | } |
| 124 | |
| 125 | // default behavior |
| 126 | return []; |
| 127 | } |
| 128 | |
| 129 | /** @inheritDoc */ |
| 130 | public function after( Element $node, Node $otherNode, SerializerState $state ): array { |
| 131 | // Handle native extension constraints. Only apply to plain extension tags. |
| 132 | $extTagName = WTUtils::getExtTagName( $node ); |
| 133 | if ( $extTagName && !DOMUtils::hasTypeOf( $node, 'mw:Transclusion' ) ) { |
| 134 | $extConfig = $state->getEnv()->getSiteConfig()->getExtTagConfig( $extTagName ); |
| 135 | if ( |
| 136 | ( $extConfig['options']['html2wt']['format'] ?? '' ) === 'block' && |
| 137 | WTUtils::isNewElt( $node ) && !DOMUtils::atTheTop( $otherNode ) |
| 138 | ) { |
| 139 | return [ 'min' => 1, 'max' => 2 ]; |
| 140 | } |
| 141 | } |
| 142 | |
| 143 | // default behavior |
| 144 | return []; |
| 145 | } |
| 146 | |
| 147 | private function handleListPrefix( Element $node, SerializerState $state ): string { |
| 148 | $bullets = ''; |
| 149 | if ( DOMUtils::isListOrListItem( $node ) |
| 150 | && !$this->parentBulletsHaveBeenEmitted( $node ) |
| 151 | && !DiffDOMUtils::previousNonSepSibling( $node ) // Maybe consider parentNode. |
| 152 | && $this->isTplListWithoutSharedPrefix( $node ) |
| 153 | // Nothing to do for definition list rows, |
| 154 | // since we're emitting for the parent node. |
| 155 | && !( DOMCompat::nodeName( $node ) === 'dd' |
| 156 | && ( DOMDataUtils::getDataParsoid( $node )->stx ?? null ) === 'row' ) |
| 157 | ) { |
| 158 | // phan fails to infer that the parent of a Element is always a Element |
| 159 | $parentNode = $node->parentNode; |
| 160 | '@phan-var Element $parentNode'; |
| 161 | $bullets = $this->getListBullets( $state, $parentNode ); |
| 162 | } |
| 163 | return $bullets; |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * Normally we wait until hitting the deepest nested list element before |
| 168 | * emitting bullets. However, if one of those list elements is about-id |
| 169 | * marked, the tag handler will serialize content from data-mw parts or src. |
| 170 | * This is a problem when a list wasn't assigned the shared prefix of bullets. |
| 171 | * For example, |
| 172 | * |
| 173 | * ** a |
| 174 | * ** b |
| 175 | * |
| 176 | * Will assign bullets as, |
| 177 | * |
| 178 | * <ul><li-*> |
| 179 | * <ul> |
| 180 | * <li-*> a</li> <!-- no shared prefix --> |
| 181 | * <li-**> b</li> <!-- gets both bullets --> |
| 182 | * </ul> |
| 183 | * </li></ul> |
| 184 | * |
| 185 | * For the b-li, getListsBullets will walk up and emit the two bullets it was |
| 186 | * assigned. If it was about-id marked, the parts would contain the two bullet |
| 187 | * start tag it was assigned. However, for the a-li, only one bullet is |
| 188 | * associated. When it's about-id marked, serializing the data-mw parts or |
| 189 | * src would miss the bullet assigned to the container li. |
| 190 | * |
| 191 | * @param Element $node |
| 192 | * @return bool |
| 193 | */ |
| 194 | private function isTplListWithoutSharedPrefix( Element $node ): bool { |
| 195 | if ( !WTUtils::isEncapsulationWrapper( $node ) ) { |
| 196 | return false; |
| 197 | } |
| 198 | |
| 199 | if ( DOMUtils::hasTypeOf( $node, 'mw:Transclusion' ) ) { |
| 200 | // If the first part is a string, template ranges were expanded to |
| 201 | // include this list element. That may be trouble. Otherwise, |
| 202 | // containers aren't part of the template source and we should emit |
| 203 | // them. |
| 204 | $dataMw = DOMDataUtils::getDataMw( $node ); |
| 205 | if ( !isset( $dataMw->parts ) || !is_string( $dataMw->parts[0] ) ) { |
| 206 | return true; |
| 207 | } |
| 208 | // Less than two bullets indicates that a shared prefix was not |
| 209 | // assigned to this element. A safe indication that we should call |
| 210 | // getListsBullets on the containing list element. |
| 211 | return !preg_match( '/^[*#:;]{2,}$/D', $dataMw->parts[0] ); |
| 212 | } elseif ( DOMUtils::matchTypeOf( $node, '/^mw:(Extension|Param)/' ) ) { |
| 213 | // Containers won't ever be part of the src here, so emit them. |
| 214 | return true; |
| 215 | } else { |
| 216 | return false; |
| 217 | } |
| 218 | } |
| 219 | |
| 220 | private function parentBulletsHaveBeenEmitted( Element $node ): bool { |
| 221 | if ( WTUtils::isLiteralHTMLNode( $node ) ) { |
| 222 | return true; |
| 223 | } elseif ( DOMUtils::isList( $node ) ) { |
| 224 | return !DOMUtils::isListItem( $node->parentNode ); |
| 225 | } else { |
| 226 | Assert::invariant( DOMUtils::isListItem( $node ), |
| 227 | '$node must be a list, list item or literal html node' ); |
| 228 | $parentNode = $node->parentNode; |
| 229 | // Skip builder-inserted wrappers |
| 230 | while ( $this->isBuilderInsertedElt( $parentNode ) ) { |
| 231 | $parentNode = $parentNode->parentNode; |
| 232 | } |
| 233 | return !in_array( |
| 234 | DOMCompat::nodeName( $parentNode ), |
| 235 | $this->parentMap[DOMCompat::nodeName( $node )], |
| 236 | true |
| 237 | ); |
| 238 | } |
| 239 | } |
| 240 | } |