Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
35.58% |
195 / 548 |
|
14.93% |
10 / 67 |
CRAP | |
0.00% |
0 / 1 |
| DOMDataUtils | |
35.58% |
195 / 548 |
|
14.93% |
10 / 67 |
15250.40 | |
0.00% |
0 / 1 |
| setExtensionData | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getExtensionData | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| hasExtensionData | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| removeExtensionData | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getBag | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getCodec | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| isPrepared | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isPreparedAndLoaded | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| prepareAndLoadDoc | |
93.33% |
14 / 15 |
|
0.00% |
0 / 1 |
3.00 | |||
| storeAndUnprepareDoc | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| stashObjectInDoc | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| eagerlyLoadRichAttributes | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
30 | |||
| hasInlineRichAttributes | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
30 | |||
| dedupeNodeData | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| dedupeNodeDataVisitor | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
90 | |||
| noAttrs | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| loadFromPageBundle | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
30 | |||
| getNodeData | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
30 | |||
| setNodeData | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getDataParsoid | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setDataParsoid | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDataMwVariant | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDataMwI18n | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDataMwI18nDefault | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDataNodeI18n | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| setDataNodeI18n | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getDataAttrI18n | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| setDataAttrI18n | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getDataAttrI18nNames | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getDataParsoidDiff | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDataParsoidDiffDefault | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setDataParsoidDiff | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getDataMw | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDataMwIfExists | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setDataMw | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getJSONAttribute | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
12 | |||
| setJSONAttribute | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| setShadowInfo | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
2 | |||
| setShadowInfoIfModified | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
42 | |||
| addNormalizedAttribute | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| storeInPageBundle | |
83.33% |
15 / 18 |
|
0.00% |
0 / 1 |
6.17 | |||
| visitAndLoadDataAttribs | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| loadDataAttribs | |
75.00% |
6 / 8 |
|
0.00% |
0 / 1 |
3.14 | |||
| usedIdIndex | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
20 | |||
| visitAndStoreDataAttribs | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
3 | |||
| storeDataAttribs | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
4 | |||
| cloneNode | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| cloneElement | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| cloneDocumentFragment | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| cloneDocument | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
3 | |||
| isHtmlAttributeWithSpecialSemantics | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| nodeDataPropName | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
12 | |||
| decodeAttribute | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
6 | |||
| getAttributeObject | |
77.78% |
14 / 18 |
|
0.00% |
0 / 1 |
6.40 | |||
| getAttributeObjectDefault | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
4 | |||
| setAttributeObject | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| setAttributeObjectNodeData | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| removeAttributeObject | |
90.00% |
9 / 10 |
|
0.00% |
0 / 1 |
3.01 | |||
| removeFromExpandedAttrs | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
42 | |||
| getAttributeDom | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| getAttributeDomDefault | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| setAttributeDom | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| removeAttributeDom | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
2 | |||
| loadRichAttributes | |
0.00% |
0 / 43 |
|
0.00% |
0 / 1 |
272 | |||
| storeAttribute | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
240 | |||
| storeRichAttributes | |
83.10% |
59 / 71 |
|
0.00% |
0 / 1 |
47.72 | |||
| dumpRichAttribs | |
0.00% |
0 / 36 |
|
0.00% |
0 / 1 |
210 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Utils; |
| 5 | |
| 6 | use Composer\Semver\Semver; |
| 7 | use InvalidArgumentException; |
| 8 | use LogicException; |
| 9 | use stdClass; |
| 10 | use WeakMap; |
| 11 | use Wikimedia\Assert\Assert; |
| 12 | use Wikimedia\JsonCodec\Hint; |
| 13 | use Wikimedia\Parsoid\Config\SiteConfig; |
| 14 | use Wikimedia\Parsoid\Core\BasePageBundle; |
| 15 | use Wikimedia\Parsoid\Core\DOMCompat; |
| 16 | use Wikimedia\Parsoid\Core\DomPageBundle; |
| 17 | use Wikimedia\Parsoid\DOM\Document; |
| 18 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 19 | use Wikimedia\Parsoid\DOM\Element; |
| 20 | use Wikimedia\Parsoid\DOM\Node; |
| 21 | use Wikimedia\Parsoid\NodeData\DataBag; |
| 22 | use Wikimedia\Parsoid\NodeData\DataMw; |
| 23 | use Wikimedia\Parsoid\NodeData\DataMwAttrib; |
| 24 | use Wikimedia\Parsoid\NodeData\DataMwI18n; |
| 25 | use Wikimedia\Parsoid\NodeData\DataMwVariant; |
| 26 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 27 | use Wikimedia\Parsoid\NodeData\DataParsoidDiff; |
| 28 | use Wikimedia\Parsoid\NodeData\I18nInfo; |
| 29 | use Wikimedia\Parsoid\NodeData\NodeData; |
| 30 | use Wikimedia\Parsoid\NodeData\TempData; |
| 31 | |
| 32 | /** |
| 33 | * These helpers pertain to HTML and data attributes of a node. |
| 34 | */ |
| 35 | class DOMDataUtils { |
| 36 | public const DATA_OBJECT_ATTR_NAME = 'data-object-id'; |
| 37 | |
| 38 | /** The internal property prefix used for rich attribute data. */ |
| 39 | private const RICH_ATTR_DATA_PREFIX = 'rich-data-'; |
| 40 | |
| 41 | /** The internal property prefix used for rich attribute type hints. */ |
| 42 | private const RICH_ATTR_HINT_PREFIX = 'rich-hint-'; |
| 43 | |
| 44 | /** Backing storage for "extension data" on a Document. */ |
| 45 | private static array $docMap = []; |
| 46 | |
| 47 | /** |
| 48 | * Associate additional data with a Document, without resorting to |
| 49 | * a dynamic property. |
| 50 | * |
| 51 | * Note that our data is hung off the top-level Document because it is |
| 52 | * the only object guaranteed to be kept live during the duration of a |
| 53 | * parse. Individual Nodes and Elements will be garbage collected as |
| 54 | * soon as local variable references are dropped (the fact that backing |
| 55 | * data is referenced from the Document does not prevent this) which |
| 56 | * would erase the entry from the WeakMap used here. |
| 57 | * |
| 58 | * @param Document $doc |
| 59 | * @param string $key |
| 60 | * @param mixed $value |
| 61 | */ |
| 62 | private static function setExtensionData( Document $doc, string $key, $value ): void { |
| 63 | if ( !isset( self::$docMap[$key] ) ) { |
| 64 | self::$docMap[$key] = new WeakMap(); |
| 65 | } |
| 66 | self::$docMap[$key]->offsetSet( $doc, $value ); |
| 67 | } |
| 68 | |
| 69 | /** |
| 70 | * Retrieve additional data associated with a Document. |
| 71 | * |
| 72 | * See discussion in ::setExtensionData() above. |
| 73 | * |
| 74 | * @param Document $doc |
| 75 | * @param string $key |
| 76 | * @return mixed |
| 77 | * @throws \Error if $key is not found. |
| 78 | */ |
| 79 | private static function getExtensionData( Document $doc, string $key ) { |
| 80 | if ( !isset( self::$docMap[$key] ) ) { |
| 81 | throw new \Error( "$key not found" ); |
| 82 | } |
| 83 | return self::$docMap[$key]->offsetGet( $doc ); |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * Determine if there is extension data for $key associated with the |
| 88 | * given document. |
| 89 | */ |
| 90 | private static function hasExtensionData( Document $doc, string $key ): bool { |
| 91 | if ( !isset( self::$docMap[$key] ) ) { |
| 92 | return false; |
| 93 | } |
| 94 | return self::$docMap[$key]->offsetExists( $doc ); |
| 95 | } |
| 96 | |
| 97 | /** |
| 98 | * Remove the association of extra data with a document. |
| 99 | * @param Document $doc |
| 100 | * @param string $key |
| 101 | * @see ::setExtensionData |
| 102 | */ |
| 103 | private static function removeExtensionData( Document $doc, string $key ): void { |
| 104 | if ( !isset( self::$docMap[$key] ) ) { |
| 105 | throw new \Error( "$key not found" ); |
| 106 | } |
| 107 | self::$docMap[$key]->offsetUnset( $doc ); |
| 108 | } |
| 109 | |
| 110 | /** |
| 111 | * Return the dynamic "bag" property of a Document. |
| 112 | * @param Document $doc |
| 113 | * @return DataBag |
| 114 | */ |
| 115 | public static function getBag( Document $doc ): DataBag { |
| 116 | return self::getExtensionData( $doc, "bag" ); |
| 117 | } |
| 118 | |
| 119 | /** |
| 120 | * Return the JsonCodec used for rich attributes in a Document. |
| 121 | * @param Node $node |
| 122 | * @return DOMDataCodec |
| 123 | */ |
| 124 | public static function getCodec( Node $node ): DOMDataCodec { |
| 125 | // Owner document is set for all nodes except Document itself. |
| 126 | $doc = $node->ownerDocument ?? $node; |
| 127 | return self::getExtensionData( $doc, "codec" ); |
| 128 | } |
| 129 | |
| 130 | private static function isPrepared( Document $doc ): bool { |
| 131 | return self::hasExtensionData( $doc, "bag" ); |
| 132 | } |
| 133 | |
| 134 | public static function isPreparedAndLoaded( Document $doc ): bool { |
| 135 | return self::isPrepared( $doc ) && self::getBag( $doc )->loaded; |
| 136 | } |
| 137 | |
| 138 | public static function prepareAndLoadDoc( |
| 139 | Document $doc, SiteConfig $siteConfig, array $options = [] |
| 140 | ): void { |
| 141 | $bag = new DataBag(); |
| 142 | $codec = new DOMDataCodec( $doc, [] ); |
| 143 | self::setExtensionData( $doc, "bag", $bag ); |
| 144 | self::setExtensionData( $doc, "codec", $codec ); |
| 145 | |
| 146 | $pb = $options['loadFromPageBundle'] ?? null; |
| 147 | '@phan-var ?BasePageBundle $pb'; // @var ?BasePageBundle $pb |
| 148 | $bag->inputPageBundle = $pb; |
| 149 | $bag->serializeNewEmptyDp = $options['serializeNewEmptyDp'] ?? false; |
| 150 | |
| 151 | if ( $bag->inputPageBundle?->hasValidCounters() ) { |
| 152 | $bag->updateCountersFromPageBundle( $bag->inputPageBundle ); |
| 153 | } else { |
| 154 | self::visitAndLoadDataAttribs( DOMCompat::getBody( $doc ) ); |
| 155 | foreach ( $options['fragments'] ?? [] as $f ) { |
| 156 | self::visitAndLoadDataAttribs( $f ); |
| 157 | } |
| 158 | } |
| 159 | |
| 160 | // Mark the document as loaded so we can try to catch errors which |
| 161 | // might try to reload this again later. |
| 162 | $bag->loaded = true; |
| 163 | |
| 164 | // Cache the head and body. |
| 165 | DOMCompat::getHead( $doc ); |
| 166 | DOMCompat::getBody( $doc ); |
| 167 | } |
| 168 | |
| 169 | /** @internal */ |
| 170 | public static function storeAndUnprepareDoc( Document $doc, array $options ): void { |
| 171 | self::visitAndStoreDataAttribs( DOMCompat::getBody( $doc ), $options ); |
| 172 | foreach ( $options['fragments'] ?? [] as $f ) { |
| 173 | self::visitAndStoreDataAttribs( $f, $options ); |
| 174 | } |
| 175 | self::setExtensionData( $doc, "bag", null ); |
| 176 | self::setExtensionData( $doc, "codec", null ); |
| 177 | } |
| 178 | |
| 179 | /** |
| 180 | * Stash $obj in $doc and return an id for later retrieval |
| 181 | * @param Document $doc |
| 182 | * @param NodeData $obj |
| 183 | * @return int |
| 184 | */ |
| 185 | public static function stashObjectInDoc( Document $doc, NodeData $obj ): int { |
| 186 | return self::getBag( $doc )->stashObject( $obj ); |
| 187 | } |
| 188 | |
| 189 | /** @internal */ |
| 190 | public static function eagerlyLoadRichAttributes( Element $node ): void { |
| 191 | // This is a list of all the attributes which could have embedded |
| 192 | // HTML, which we should ensure are loaded before serialization in |
| 193 | // certain circumstances so that format conversion is performed. |
| 194 | // XXX: it would be best if these attributes were marked as |
| 195 | // proposed in [[mw:Parsoid/MediaWiki_DOM_spec/Rich_Attributes]] |
| 196 | // Phase 3 so this wasn't so ad-hoc |
| 197 | if ( $node->hasAttribute( 'data-mw' ) ) { |
| 198 | self::getDataMw( $node ); |
| 199 | } |
| 200 | if ( $node->hasAttribute( 'data-mw-variant' ) ) { |
| 201 | // DataMwVariant::$twoway,$oneway,$filter,$disabled,$name etc |
| 202 | self::getDataMwVariant( $node ); |
| 203 | } |
| 204 | if ( $node->hasAttribute( 'data-mw-i18n' ) ) { |
| 205 | // I18nInfo::$params could potentially contain embedded HTML |
| 206 | self::getDataNodeI18n( $node ); |
| 207 | } |
| 208 | if ( $node->hasAttribute( 'data-parsoid' ) ) { |
| 209 | // embedded HTML in DataParsoid::$html |
| 210 | self::getDataParsoid( $node ); |
| 211 | } |
| 212 | // data-parsoid-diff is a rich attribute, but it doesn't contain HTML |
| 213 | } |
| 214 | |
| 215 | public static function hasInlineRichAttributes( Element $node ): bool { |
| 216 | return $node->hasAttribute( 'data-parsoid' ) || |
| 217 | $node->hasAttribute( 'data-mw' ) || |
| 218 | $node->hasAttribute( 'data-mw-variant' ) || |
| 219 | $node->hasAttribute( 'data-mw-i18n' ) || |
| 220 | // data-parsoid-diff is a rich attribute, but it doesn't contain HTML |
| 221 | false; |
| 222 | } |
| 223 | |
| 224 | public static function dedupeNodeData( Node $clonedRoot ): void { |
| 225 | $bag = self::getBag( $clonedRoot->ownerDocument ); |
| 226 | $aboutMap = []; |
| 227 | self::dedupeNodeDataVisitor( $bag, $aboutMap, $clonedRoot ); |
| 228 | } |
| 229 | |
| 230 | private static function dedupeNodeDataVisitor( |
| 231 | DataBag $bag, array &$aboutMap, Node $node |
| 232 | ): void { |
| 233 | if ( $node instanceof Element ) { |
| 234 | $nd = self::getNodeData( $node, init: false ); |
| 235 | if ( $nd ) { |
| 236 | // All we are doing here is cloning the node-data object |
| 237 | // and reassigning it to the same node. So, any unloaded |
| 238 | // data continues to be available. But, we eagerly load |
| 239 | // all rich attributes to ensure that embedded HTML fragments |
| 240 | // get their (about,annotation,node) ids deduplicated correctly. |
| 241 | self::eagerlyLoadRichAttributes( $node ); |
| 242 | |
| 243 | $node->removeAttribute( self::DATA_OBJECT_ATTR_NAME ); |
| 244 | $nd = $nd->cloneNodeData(); |
| 245 | self::setNodeData( $node, $nd ); |
| 246 | |
| 247 | // Deduplicate annotation range ids |
| 248 | // These can occur multiple times in a given subtree, so we |
| 249 | // need to record the mapping for future use. |
| 250 | // (There's no DataMw unless there was a DATA_OBJECT_ATTR_NAME) |
| 251 | if ( isset( $nd->mw->rangeId ) ) { |
| 252 | $oldAbout = $nd->mw->rangeId; |
| 253 | $isStart = false; |
| 254 | $type = WTUtils::extractAnnotationType( $node, $isStart ); |
| 255 | if ( $type !== null && $isStart ) { |
| 256 | $aboutMap[$oldAbout] = $bag->newAnnotationId(); |
| 257 | } |
| 258 | $nd->mw->rangeId = $aboutMap[$oldAbout] ?? $oldAbout; |
| 259 | } |
| 260 | |
| 261 | if ( $node->hasAttribute( 'about' ) ) { |
| 262 | // Deduplicate transclusion ids |
| 263 | // As with annotation ranges, these can occur multiple times |
| 264 | // in a given subtree, so we need to record the mapping used. |
| 265 | $oldAbout = DOMCompat::getAttribute( $node, 'about' ); |
| 266 | if ( WTUtils::isFirstEncapsulationWrapperNode( $node ) ) { |
| 267 | $aboutMap[$oldAbout] = $bag->newAboutId(); |
| 268 | } |
| 269 | $node->setAttribute( 'about', $aboutMap[$oldAbout] ?? $oldAbout ); |
| 270 | } |
| 271 | } |
| 272 | } |
| 273 | foreach ( DOMUtils::childNodes( $node ) as $child ) { |
| 274 | self::dedupeNodeDataVisitor( $bag, $aboutMap, $child ); |
| 275 | } |
| 276 | } |
| 277 | |
| 278 | /** |
| 279 | * Does this node have any attributes? |
| 280 | * @param Element $node |
| 281 | * @return bool |
| 282 | */ |
| 283 | public static function noAttrs( Element $node ): bool { |
| 284 | // The 'xmlns' attribute is "invisible" T235295 |
| 285 | if ( $node->hasAttribute( 'xmlns' ) ) { |
| 286 | return false; |
| 287 | } |
| 288 | $numAttrs = count( $node->attributes ); |
| 289 | return $numAttrs === 0 || |
| 290 | ( $numAttrs === 1 && $node->hasAttribute( self::DATA_OBJECT_ATTR_NAME ) ); |
| 291 | } |
| 292 | |
| 293 | public static function loadFromPageBundle( Element $node ): ?NodeData { |
| 294 | $bag = self::getBag( $node->ownerDocument ); |
| 295 | $pb = $bag->inputPageBundle; |
| 296 | if ( $pb === null ) { |
| 297 | return null; |
| 298 | } |
| 299 | |
| 300 | $id = DOMCompat::getAttribute( $node, 'id' ); |
| 301 | if ( $id === null ) { |
| 302 | return null; |
| 303 | } |
| 304 | |
| 305 | $nodeData = new NodeData; |
| 306 | // See if there is data-parsoid or data-mw in the page bundle |
| 307 | if ( isset( $pb->parsoid['ids'][$id] ) ) { |
| 308 | $nodeData->parsoid = [ $pb->parsoid['ids'][$id] ]; // Undecoded JSON blob |
| 309 | } |
| 310 | if ( isset( $pb->mw['ids'][$id] ) ) { |
| 311 | $nodeData->mw = [ $pb->mw['ids'][$id] ]; // Undecoded JSON blob |
| 312 | } |
| 313 | self::setNodeData( $node, $nodeData ); |
| 314 | return $nodeData; |
| 315 | } |
| 316 | |
| 317 | /** |
| 318 | * Fetch node data from the data-bag OR pagebundle. |
| 319 | * Initialize if not found (unless disabled via the $init param) |
| 320 | */ |
| 321 | public static function getNodeData( Element $node, bool $init = true ): ?NodeData { |
| 322 | $nodeId = DOMCompat::getAttribute( $node, self::DATA_OBJECT_ATTR_NAME ); |
| 323 | $bag = self::getBag( $node->ownerDocument ); |
| 324 | if ( $nodeId === null ) { |
| 325 | $nodeData = self::loadFromPageBundle( $node ); |
| 326 | if ( $nodeData === null && $init ) { |
| 327 | // Initialized on first request |
| 328 | $nodeData = new NodeData; |
| 329 | self::setNodeData( $node, $nodeData ); |
| 330 | } |
| 331 | return $nodeData; |
| 332 | } |
| 333 | |
| 334 | $nodeData = $bag->getObject( (int)$nodeId ); |
| 335 | Assert::invariant( $nodeData !== null, 'Bogus nodeId given!' ); |
| 336 | if ( isset( $nodeData->storedId ) ) { |
| 337 | throw new LogicException( |
| 338 | 'Trying to fetch node data without loading! ' . |
| 339 | // If this node's data-object id is different from storedId, |
| 340 | // it will indicate that the data-parsoid object was shared |
| 341 | // between nodes without getting cloned. Useful for debugging. |
| 342 | 'Node id: ' . $nodeId . ' ' . |
| 343 | 'Stored data: ' . PHPUtils::jsonEncode( $nodeData ) |
| 344 | ); |
| 345 | } |
| 346 | return $nodeData; |
| 347 | } |
| 348 | |
| 349 | /** |
| 350 | * Set node data. |
| 351 | * |
| 352 | * @param Element $node node |
| 353 | * @param NodeData $data data |
| 354 | */ |
| 355 | public static function setNodeData( Element $node, NodeData $data ): void { |
| 356 | $nodeId = self::stashObjectInDoc( $node->ownerDocument, $data ); |
| 357 | $node->setAttribute( self::DATA_OBJECT_ATTR_NAME, (string)$nodeId ); |
| 358 | } |
| 359 | |
| 360 | /** |
| 361 | * Get data parsoid info from a node. |
| 362 | * |
| 363 | * @param Element $node node |
| 364 | * @return DataParsoid |
| 365 | */ |
| 366 | public static function getDataParsoid( Element $node ): DataParsoid { |
| 367 | return self::getNodeData( $node )->getDataParsoid( $node ); |
| 368 | } |
| 369 | |
| 370 | /** |
| 371 | * Set data parsoid info on a node. |
| 372 | * |
| 373 | * @param Element $node node |
| 374 | * @param DataParsoid $dp data-parsoid |
| 375 | */ |
| 376 | public static function setDataParsoid( Element $node, DataParsoid $dp ): void { |
| 377 | self::getNodeData( $node )->setDataParsoid( $node, $dp ); |
| 378 | } |
| 379 | |
| 380 | /** |
| 381 | * Returns the language variant information of a node. |
| 382 | * @param Element $node |
| 383 | * @return ?DataMwVariant |
| 384 | */ |
| 385 | public static function getDataMwVariant( Element $node ): ?DataMwVariant { |
| 386 | // No default value; returns null if not present. |
| 387 | return self::getAttributeObject( $node, 'data-mw-variant', DataMwVariant::hint() ); |
| 388 | } |
| 389 | |
| 390 | /** |
| 391 | * Returns the i18n information of a node. This is in private access because it shouldn't |
| 392 | * typically be used directly; instead getDataNodeI18n and getDataAttrI18n should be used. |
| 393 | * @param Element $node |
| 394 | * @return DataMwI18n|null |
| 395 | */ |
| 396 | private static function getDataMwI18n( Element $node ): ?DataMwI18n { |
| 397 | // No default value; returns null if not present. |
| 398 | return self::getAttributeObject( $node, 'data-mw-i18n', DataMwI18n::hint() ); |
| 399 | } |
| 400 | |
| 401 | /** |
| 402 | * Returns the i18n information of a node, setting it to a default |
| 403 | * value if it is missing. This should not typically be used |
| 404 | * directly; instead setDataNodeI18n and setDataAttrI18n should be |
| 405 | * used. |
| 406 | * |
| 407 | * @param Element $node |
| 408 | * @return DataMwI18n $i18n |
| 409 | */ |
| 410 | private static function getDataMwI18nDefault( Element $node ): DataMwI18n { |
| 411 | return self::getAttributeObjectDefault( $node, 'data-mw-i18n', DataMwI18n::hint() ); |
| 412 | } |
| 413 | |
| 414 | /** |
| 415 | * Retrieves internationalization (i18n) information of a node (typically for localization) |
| 416 | * @param Element $node |
| 417 | * @return ?I18nInfo |
| 418 | */ |
| 419 | public static function getDataNodeI18n( Element $node ): ?I18nInfo { |
| 420 | $i18n = self::getDataMwI18n( $node ); |
| 421 | if ( $i18n === null ) { |
| 422 | return null; |
| 423 | } |
| 424 | return $i18n->getSpanInfo(); |
| 425 | } |
| 426 | |
| 427 | /** |
| 428 | * Sets internationalization (i18n) information of a node, used for later localization |
| 429 | * @param Element $node |
| 430 | * @param I18nInfo $info |
| 431 | * @return void |
| 432 | */ |
| 433 | public static function setDataNodeI18n( Element $node, I18nInfo $info ) { |
| 434 | $i18n = self::getDataMwI18nDefault( $node ); |
| 435 | $i18n->setSpanInfo( $info ); |
| 436 | } |
| 437 | |
| 438 | /** |
| 439 | * Retrieves internationalization (i18n) information of an attribute value (typically for |
| 440 | * localization) |
| 441 | * @param Element $node |
| 442 | * @param string $name |
| 443 | * @return ?I18nInfo |
| 444 | */ |
| 445 | public static function getDataAttrI18n( Element $node, string $name ): ?I18nInfo { |
| 446 | $i18n = self::getDataMwI18n( $node ); |
| 447 | if ( $i18n === null ) { |
| 448 | return null; |
| 449 | } |
| 450 | return $i18n->getAttributeInfo( $name ); |
| 451 | } |
| 452 | |
| 453 | /** |
| 454 | * Sets internationalization (i18n) information of a attribute value, used for later |
| 455 | * localization |
| 456 | * @param Element $node |
| 457 | * @param string $name |
| 458 | * @param I18nInfo $info |
| 459 | * @return void |
| 460 | */ |
| 461 | public static function setDataAttrI18n( Element $node, string $name, I18nInfo $info ) { |
| 462 | $i18n = self::getDataMwI18nDefault( $node ); |
| 463 | $i18n->setAttributeInfo( $name, $info ); |
| 464 | } |
| 465 | |
| 466 | /** |
| 467 | * @param Element $node |
| 468 | * @return array |
| 469 | */ |
| 470 | public static function getDataAttrI18nNames( Element $node ): array { |
| 471 | $i18n = self::getDataMwI18n( $node ); |
| 472 | if ( $i18n === null ) { |
| 473 | // We won't set a default value for this property |
| 474 | return []; |
| 475 | } |
| 476 | return $i18n->getAttributeNames(); |
| 477 | } |
| 478 | |
| 479 | /** |
| 480 | * Get data diff info from a node. |
| 481 | * |
| 482 | * @param Element $node node |
| 483 | * @return ?DataParsoidDiff |
| 484 | */ |
| 485 | public static function getDataParsoidDiff( Element $node ): ?DataParsoidDiff { |
| 486 | // No default value; returns null if not present. |
| 487 | return self::getAttributeObject( $node, 'data-parsoid-diff', DataParsoidDiff::hint() ); |
| 488 | } |
| 489 | |
| 490 | /** |
| 491 | * Get data diff info from a node, setting a default value if not present. |
| 492 | * |
| 493 | * @param Element $node node |
| 494 | * @return DataParsoidDiff |
| 495 | */ |
| 496 | public static function getDataParsoidDiffDefault( Element $node ): DataParsoidDiff { |
| 497 | return self::getAttributeObjectDefault( $node, 'data-parsoid-diff', DataParsoidDiff::hint() ); |
| 498 | } |
| 499 | |
| 500 | /** |
| 501 | * Set data diff info on a node. |
| 502 | * |
| 503 | * @param Element $node node |
| 504 | * @param ?DataParsoidDiff $diffObj data-parsoid-diff object |
| 505 | */ |
| 506 | public static function setDataParsoidDiff( Element $node, ?DataParsoidDiff $diffObj ): void { |
| 507 | if ( $diffObj !== null ) { |
| 508 | self::setAttributeObject( $node, 'data-parsoid-diff', $diffObj, DataParsoidDiff::hint() ); |
| 509 | } else { |
| 510 | self::removeAttributeObject( $node, 'data-parsoid-diff' ); |
| 511 | } |
| 512 | } |
| 513 | |
| 514 | /** |
| 515 | * Get data meta wiki info from a node. |
| 516 | * |
| 517 | * @param Element $node node |
| 518 | * @return DataMw |
| 519 | */ |
| 520 | public static function getDataMw( Element $node ): DataMw { |
| 521 | return self::getNodeData( $node )->getDataMw( $node ); |
| 522 | } |
| 523 | |
| 524 | /** |
| 525 | * Get data meta wiki info, but don't create it if it doesn't exist. |
| 526 | */ |
| 527 | public static function getDataMwIfExists( Element $node ): ?DataMw { |
| 528 | return self::getAttributeObject( $node, 'data-mw', DataMw::hint() ); |
| 529 | } |
| 530 | |
| 531 | /** |
| 532 | * Set data meta wiki info from a node. |
| 533 | * |
| 534 | * @param Element $node node |
| 535 | * @param ?DataMw $dmw data-mw |
| 536 | */ |
| 537 | public static function setDataMw( Element $node, ?DataMw $dmw ): void { |
| 538 | self::getNodeData( $node )->setDataMw( $node, $dmw ); |
| 539 | } |
| 540 | |
| 541 | /** |
| 542 | * Get an object from a JSON-encoded XML attribute on a node. |
| 543 | * |
| 544 | * @param Element $node node |
| 545 | * @param string $name name |
| 546 | * @param mixed $defaultVal |
| 547 | * @return mixed |
| 548 | * @deprecated since 0.23; use getAttributeObject() |
| 549 | */ |
| 550 | public static function getJSONAttribute( Element $node, string $name, $defaultVal ) { |
| 551 | PHPUtils::deprecated( __METHOD__, '0.23' ); |
| 552 | $attVal = DOMCompat::getAttribute( $node, $name ); |
| 553 | if ( $attVal === null ) { |
| 554 | return $defaultVal; |
| 555 | } |
| 556 | $decoded = PHPUtils::jsonDecode( $attVal, false ); |
| 557 | if ( $decoded !== null ) { |
| 558 | return $decoded; |
| 559 | } else { |
| 560 | error_log( 'ERROR: Could not decode attribute-val ' . $attVal . |
| 561 | ' for ' . $name . ' on node ' . DOMUtils::nodeName( $node ) ); |
| 562 | return $defaultVal; |
| 563 | } |
| 564 | } |
| 565 | |
| 566 | /** |
| 567 | * Set a attribute on a node with a JSON-encoded object. |
| 568 | * |
| 569 | * @param Element $node node |
| 570 | * @param string $name Name of the attribute. |
| 571 | * @param mixed $obj value of the attribute to |
| 572 | * @deprecated since 0.23; use setAttributeObject() |
| 573 | */ |
| 574 | public static function setJSONAttribute( Element $node, string $name, $obj ): void { |
| 575 | PHPUtils::deprecated( __METHOD__, '0.23' ); |
| 576 | $val = $obj === [] ? '{}' : PHPUtils::jsonEncode( $obj ); |
| 577 | $node->setAttribute( $name, $val ); |
| 578 | } |
| 579 | |
| 580 | // Shadow attributes should probably be unified with rich attributes |
| 581 | // at some point. [CSA 2024-10-15] |
| 582 | |
| 583 | /** |
| 584 | * Set shadow info on a node; similar to the method on tokens. |
| 585 | * Records a key = value pair in data-parsoid['a'] property. |
| 586 | * |
| 587 | * This is effectively a call of 'setShadowInfoIfModified' except |
| 588 | * there is no original value, so by definition, $val is modified. |
| 589 | * |
| 590 | * @param Element $node node |
| 591 | * @param string $name Name of the attribute. |
| 592 | * @param mixed $val val |
| 593 | */ |
| 594 | public static function setShadowInfo( Element $node, string $name, $val ): void { |
| 595 | $dp = self::getDataParsoid( $node ); |
| 596 | $dp->a ??= []; |
| 597 | $dp->sa ??= []; |
| 598 | $dp->a[$name] = $val; |
| 599 | } |
| 600 | |
| 601 | /** |
| 602 | * Set shadow info on a node; similar to the method on tokens. |
| 603 | * |
| 604 | * If the new value ($val) for the key ($name) is different from the |
| 605 | * original value ($origVal): |
| 606 | * - the new value is recorded in data-parsoid->a and |
| 607 | * - the original value is recorded in data-parsoid->sa |
| 608 | * |
| 609 | * @param Element $node node |
| 610 | * @param string $name Name of the attribute. |
| 611 | * @param mixed $val val |
| 612 | * @param mixed $origVal original value (null is a valid value) |
| 613 | * @param bool $skipOrig |
| 614 | */ |
| 615 | public static function setShadowInfoIfModified( |
| 616 | Element $node, string $name, $val, $origVal, bool $skipOrig = false |
| 617 | ): void { |
| 618 | if ( !$skipOrig && ( $val === $origVal || $origVal === null ) ) { |
| 619 | return; |
| 620 | } |
| 621 | $dp = self::getDataParsoid( $node ); |
| 622 | $dp->a ??= []; |
| 623 | $dp->sa ??= []; |
| 624 | // FIXME: This is a hack to not overwrite already shadowed info. |
| 625 | // We should either fix the call site that depends on this |
| 626 | // behaviour to do an explicit check, or double down on this |
| 627 | // by porting it to the token method as well. |
| 628 | if ( !$skipOrig && !array_key_exists( $name, $dp->a ) ) { |
| 629 | $dp->sa[$name] = $origVal; |
| 630 | } |
| 631 | $dp->a[$name] = $val; |
| 632 | } |
| 633 | |
| 634 | /** |
| 635 | * Set an attribute and shadow info to a node. |
| 636 | * Similar to the method on tokens |
| 637 | * |
| 638 | * @param Element $node node |
| 639 | * @param string $name Name of the attribute. |
| 640 | * @param mixed $val value |
| 641 | * @param mixed $origVal original value |
| 642 | * @param bool $skipOrig |
| 643 | */ |
| 644 | public static function addNormalizedAttribute( |
| 645 | Element $node, string $name, $val, $origVal, bool $skipOrig = false |
| 646 | ): void { |
| 647 | if ( $name === 'id' ) { |
| 648 | DOMCompat::setIdAttribute( $node, $val ); |
| 649 | } else { |
| 650 | $node->setAttribute( $name, $val ); |
| 651 | } |
| 652 | self::setShadowInfoIfModified( $node, $name, $val, $origVal, $skipOrig ); |
| 653 | } |
| 654 | |
| 655 | /** |
| 656 | * Removes the `data-*` attribute from a node, and migrates the data to the |
| 657 | * given DomPageBundle. Generates a unique id but attempts to keep user defined ids. |
| 658 | * |
| 659 | * TODO: Note that $data is effective a partial PageBundle containing |
| 660 | * only the 'parsoid' and 'mw' properties. |
| 661 | * |
| 662 | * @param DomPageBundle $pb |
| 663 | * @param Element $node node |
| 664 | * @param stdClass $data data |
| 665 | * @param array $idIndex Index of used id attributes in the DOM |
| 666 | */ |
| 667 | private static function storeInPageBundle( |
| 668 | DomPageBundle $pb, Element $node, stdClass $data, array $idIndex |
| 669 | ): void { |
| 670 | $bag = self::getBag( $node->ownerDocument ); |
| 671 | $bag->updateCountersInPageBundle( $pb ); |
| 672 | |
| 673 | $uid = DOMCompat::getAttribute( $node, 'id' ); |
| 674 | $codec = self::getCodec( $node ); |
| 675 | $pbCounters = &$pb->counters; |
| 676 | if ( $uid !== null && array_key_exists( $uid, $pb->parsoid['ids'] ) ) { |
| 677 | // Forcibly reset the ID if there's a conflict |
| 678 | // T415477: Preserve the original id as in TokenizerUtils::protectAttrs() |
| 679 | // so that it's still present in the HTML |
| 680 | $node->setAttribute( 'data-x-id', $uid ); |
| 681 | $uid = null; |
| 682 | } |
| 683 | if ( $uid === '' ) { |
| 684 | // Forcibly reset the ID if it is invalid |
| 685 | // Don't bother with a data-x-id here, the sanitizer will have |
| 686 | // already stripped and shadowed an empty id coming from source |
| 687 | $uid = null; |
| 688 | } |
| 689 | if ( $uid === null ) { |
| 690 | do { |
| 691 | $pbCounters['nodedata'] += 1; |
| 692 | // The idIndex maps all *existing* ids from the original |
| 693 | // document, so that we can ensure than any *newly assigned* |
| 694 | // UIDs don't happen to step on them. We don't need to update |
| 695 | // the idIndex here because (a) we only add a new UID if it |
| 696 | // doesn't conflict with an existing ID, and (b) by |
| 697 | // construction, none of our new UIDs will conflict with each |
| 698 | // other. |
| 699 | $uid = CounterType::NODE_DATA_ID->counterToId( $pbCounters['nodedata'] ); |
| 700 | } while ( isset( $idIndex[$uid] ) ); |
| 701 | DOMCompat::setIdAttribute( $node, $uid ); |
| 702 | } |
| 703 | $pb->parsoid['ids'][$uid] = $data->parsoid; |
| 704 | if ( isset( $data->mw ) ) { |
| 705 | $pb->mw['ids'][$uid] = $data->mw; |
| 706 | } |
| 707 | } |
| 708 | |
| 709 | /** |
| 710 | * Walk DOM from $node downward calling loadDataAttribs. |
| 711 | * This forces full eager loading of this subtree. |
| 712 | */ |
| 713 | public static function visitAndLoadDataAttribs( Node $node ): void { |
| 714 | $doc = $node->ownerDocument; |
| 715 | Assert::invariant( self::isPrepared( $doc ), "document should be prepared" ); |
| 716 | |
| 717 | DOMUtils::visitDOM( $node, function ( Node $node ) { |
| 718 | if ( $node instanceof Element ) { |
| 719 | self::loadDataAttribs( $node ); |
| 720 | } |
| 721 | } ); |
| 722 | } |
| 723 | |
| 724 | /** |
| 725 | * These are intended be used on a document after post-processing, so that |
| 726 | * the underlying .dataobject is transparently applied (in the store case) |
| 727 | * and reloaded (in the load case), rather than worrying about keeping |
| 728 | * the attributes up-to-date throughout that phase. For the most part, |
| 729 | * using this.ppTo* should be sufficient and using these directly should be |
| 730 | * avoided. |
| 731 | */ |
| 732 | private static function loadDataAttribs( Element $node ): void { |
| 733 | $bag = self::getBag( $node->ownerDocument ?? $node ); |
| 734 | $about = DOMCompat::getAttribute( $node, 'about' ); |
| 735 | if ( $about !== null ) { |
| 736 | $bag->seenAboutId( $about ); |
| 737 | } |
| 738 | $nodeData = self::getNodeData( $node ); |
| 739 | |
| 740 | // Force load of data-mw to lookup annotation range id. |
| 741 | $nodeData->getDataMwIfExists( $node ); |
| 742 | if ( isset( $nodeData->mw->rangeId ) ) { |
| 743 | $bag->seenAnnotationId( $nodeData->mw->rangeId ); |
| 744 | } |
| 745 | |
| 746 | // We don't load rich attributes or data-parsoid: that will be done |
| 747 | // lazily on demand when getDataParsoid or getAttributeObject()/etc |
| 748 | // methods are called. For OutputTransformPipeline stages, this is a |
| 749 | // performance optimization. For non-data-parsoid, in addition, we also |
| 750 | // don't know the true types of the rich values yet. |
| 751 | } |
| 752 | |
| 753 | /** |
| 754 | * Builds an index of id attributes seen in the DOM |
| 755 | * |
| 756 | * @param SiteConfig $siteConfig A SiteConfig is required to properly |
| 757 | * traverse document fragments embedded in extension DOM. |
| 758 | * @param Document $doc |
| 759 | * @param array<string,DocumentFragment> $fragments |
| 760 | * @return array<string, true> |
| 761 | */ |
| 762 | public static function usedIdIndex( SiteConfig $siteConfig, Document $doc, array $fragments = [] ): array { |
| 763 | $index = []; |
| 764 | $t = new DOMTraverser( false, true ); |
| 765 | $t->addHandler( null, static function ( $n, $state ) use ( &$index ) { |
| 766 | if ( $n instanceof Element ) { |
| 767 | $id = DOMCompat::getAttribute( $n, 'id' ); |
| 768 | if ( $id !== null ) { |
| 769 | $index[$id] = true; |
| 770 | } |
| 771 | } |
| 772 | return true; |
| 773 | } ); |
| 774 | $t->traverse( $siteConfig, DOMCompat::getBody( $doc ) ); |
| 775 | foreach ( $fragments as $name => $f ) { |
| 776 | $t->traverse( $siteConfig, $f ); |
| 777 | } |
| 778 | return $index; |
| 779 | } |
| 780 | |
| 781 | /** |
| 782 | * Walk DOM from node downward calling storeDataAttribs |
| 783 | * |
| 784 | * @param Node $node node |
| 785 | * @param array $options options |
| 786 | */ |
| 787 | public static function visitAndStoreDataAttribs( Node $node, array $options = [] ): void { |
| 788 | $bag = self::getBag( $node->ownerDocument ?? $node ); |
| 789 | Assert::invariant( $bag->loaded, "store without load" ); |
| 790 | // PORT-FIXME: storeDataAttribs calls storeInPageBundle which calls getElementById. |
| 791 | // PHP's `getElementById` implementation is broken, and we work around that by |
| 792 | // using Zest which uses XPath. So, getElementById call can be O(n) and calling it |
| 793 | // on on every element of the DOM via vistDOM here makes it O(n^2) instead of O(n). |
| 794 | // So, we work around that by building an index and avoiding getElementById entirely |
| 795 | // in storeInPageBundle. |
| 796 | if ( !empty( $options['storeInPageBundle'] ) ) { |
| 797 | Assert::invariant( isset( $options['idIndex'] ), |
| 798 | "Page bundle requires idIndex to avoid conflicts" ); |
| 799 | } |
| 800 | |
| 801 | Assert::invariant( empty( $options['discardDataParsoid'] ) || empty( $options['keepTmp'] ), |
| 802 | 'Conflicting options: discardDataParsoid and keepTmp are both enabled.' ); |
| 803 | |
| 804 | $options['serializeNewEmptyDp'] = $bag->serializeNewEmptyDp; |
| 805 | |
| 806 | // Set the "storage options" |
| 807 | $codec = self::getCodec( $node ); |
| 808 | $codec->setOptions( $options ); |
| 809 | DOMUtils::visitDOM( $node, function ( Node $node, array $options ) { |
| 810 | self::storeDataAttribs( $node, $options ); |
| 811 | }, $options ); |
| 812 | } |
| 813 | |
| 814 | /** |
| 815 | * Copy data attributes from the bag to either JSON-encoded attributes on |
| 816 | * each node, or the page bundle, erasing the data-object-id attributes. |
| 817 | * |
| 818 | * @param Node $node node |
| 819 | * @param array $options options |
| 820 | * - discardDataParsoid: Discard DataParsoid objects instead of storing them |
| 821 | * - keepTmp: Preserve DataParsoid::$tmp |
| 822 | * - storeInPageBundle: If set to a DomPageBundle, data will be stored |
| 823 | * in the given page bundle instead of data-parsoid and data-mw. |
| 824 | * - outputContentVersion: Version of output we're storing. The page bundle |
| 825 | * didn't have data-mw before 999.x |
| 826 | * - idIndex: Array of used ID attributes |
| 827 | */ |
| 828 | private static function storeDataAttribs( Node $node, array $options ): void { |
| 829 | if ( !( $node instanceof Element ) ) { |
| 830 | return; |
| 831 | } |
| 832 | |
| 833 | $pbData = self::storeRichAttributes( $node, $options ); |
| 834 | |
| 835 | // Indicate that this node's data has been stored so that if we try |
| 836 | // to access it after the fact we're aware and remove the attribute |
| 837 | // since it's no longer needed. |
| 838 | $nd = self::getNodeData( $node ); |
| 839 | $id = DOMCompat::getAttribute( $node, self::DATA_OBJECT_ATTR_NAME ); |
| 840 | $nd->storedId = $id !== null ? intval( $id ) : null; |
| 841 | |
| 842 | // Store pagebundle |
| 843 | if ( $pbData !== null ) { |
| 844 | self::storeInPageBundle( $options['storeInPageBundle'], $node, $pbData, $options['idIndex'] ); |
| 845 | } |
| 846 | |
| 847 | $node->removeAttribute( self::DATA_OBJECT_ATTR_NAME ); |
| 848 | } |
| 849 | |
| 850 | /** |
| 851 | * Clones a node and its data bag. |
| 852 | */ |
| 853 | public static function cloneNode( Node $elt, bool $deep ): Node { |
| 854 | if ( self::hasExtensionData( $elt->ownerDocument, "cloneTarget" ) ) { |
| 855 | // Special case for "clone into a brand new document", which |
| 856 | // is the exception rather than the rule since Parsoid tries |
| 857 | // to maintain a single ownerDocument context. |
| 858 | $cloneTarget = self::getExtensionData( $elt->ownerDocument, "cloneTarget" ); |
| 859 | return $cloneTarget->importNode( $elt, $deep ); |
| 860 | } |
| 861 | $clone = $elt->cloneNode( $deep ); |
| 862 | '@phan-var Element $clone'; // @var Element $clone |
| 863 | // We do not need to worry about $deep because a shallow clone does not have child nodes, |
| 864 | // so it's always cloning data on the cloned tree (which may be empty). |
| 865 | self::dedupeNodeData( $clone ); |
| 866 | return $clone; |
| 867 | } |
| 868 | |
| 869 | // Two specific helper methods to work around the lack of constrainted |
| 870 | // templated types in phan. |
| 871 | |
| 872 | /** |
| 873 | * Clones an element and its data bag(s) |
| 874 | */ |
| 875 | public static function cloneElement( Element $elt, bool $deep ): Element { |
| 876 | $clone = self::cloneNode( $elt, $deep ); |
| 877 | '@phan-var Element $clone'; // @var Element $clone |
| 878 | return $clone; |
| 879 | } |
| 880 | |
| 881 | /** |
| 882 | * Deep clone a DocumentFragment and its associated data bags |
| 883 | */ |
| 884 | public static function cloneDocumentFragment( DocumentFragment $df ): DocumentFragment { |
| 885 | $clone = self::cloneNode( $df, true ); |
| 886 | '@phan-var DocumentFragment $clone'; // @var DocumentFragment $clone |
| 887 | return $clone; |
| 888 | } |
| 889 | |
| 890 | /** |
| 891 | * Deep clone a Document and its associated data bags |
| 892 | */ |
| 893 | public static function cloneDocument( Document $doc ): Document { |
| 894 | // Standard PHP clone works fine for the Document |
| 895 | $clone = clone $doc; |
| 896 | if ( !DOMCompat::isStandardsMode( $doc ) ) { |
| 897 | // Uncache head & body since they point to the old doc. |
| 898 | $clone->head = null; |
| 899 | $clone->body = null; |
| 900 | } |
| 901 | // But now we need to duplicate the extension data. |
| 902 | if ( self::isPrepared( $doc ) ) { |
| 903 | $codec = new DOMDataCodec( $clone, [ |
| 904 | // No loading options needed; any info needed for |
| 905 | // lazy-loading will come from the (cloned) DataBag |
| 906 | ] ); |
| 907 | self::setExtensionData( $clone, "codec", $codec ); |
| 908 | |
| 909 | // Overwrite the empty Bag with a clone, after setting up |
| 910 | // to importNode rich data |
| 911 | self::setExtensionData( $doc, "cloneTarget", $clone ); |
| 912 | self::setExtensionData( $clone, "bag", clone self::getBag( $doc ) ); |
| 913 | self::removeExtensionData( $doc, "cloneTarget" ); |
| 914 | } |
| 915 | return $clone; |
| 916 | } |
| 917 | |
| 918 | // This is a generic (and somewhat optimistic) interface for |
| 919 | // complex-valued attributes in a DOM tree. The object and DOM |
| 920 | // values are "live"; that is, they are passed by-reference and |
| 921 | // mutations to the object and DOM persist in the document. |
| 922 | // These values are only "frozen" into a standards-compliant |
| 923 | // HTML5 attribute representation when the document is serialized. |
| 924 | // (A corresponding 'parse' stage needs to occur on a new document |
| 925 | // to "thaw out" the HTML5 attribute representations.) |
| 926 | |
| 927 | // Note that although we are expanding the possible attribute *values* |
| 928 | // we are still deliberately keeping attribute *names* restricted. |
| 929 | // This is a deliberate design choice. Dynamically-generated |
| 930 | // attribute names are best handled by the "key value pair" |
| 931 | // fragment datatype, which is one of the fragment types from which |
| 932 | // the output document can be composed -- but that composition |
| 933 | // mechanism and the way the fragment composition is reflected in |
| 934 | // the DOM is out-of-scope for this API. This just provides a |
| 935 | // richer way to embed complex information of that sort into a |
| 936 | // DOM document. |
| 937 | |
| 938 | // An important design decision here was not to embed type information |
| 939 | // for attributes into the representation, which is done to avoid |
| 940 | // HTML bloat. This leads directly to a "lazy load" implementation, |
| 941 | // as we can't actually load an attribute value until we know what |
| 942 | // its class type is, and that's only provided when the call to |
| 943 | // ::getAttributeObject() is made. In order to implement an "eager |
| 944 | // load" implementation, we would need a schema for the document |
| 945 | // which maps every named attribute to an appropriate type. This |
| 946 | // is possible if eager loading is desired in the future, or because |
| 947 | // you like the added structural documentation provided by a schema. |
| 948 | |
| 949 | // Certain attributes have semantics given by HTML. For example, |
| 950 | // the `class` and `alt` attributes shouldn't be serialized as a |
| 951 | // JSON blob, even if you want to store a rich value. For these |
| 952 | // "HTML attributes with special semantics" (everything not |
| 953 | // starting with data-* at the moment) we tolerate a bit of bloat |
| 954 | // and store a flattened string representation of the rich value |
| 955 | // in the direct HTML attribute, and store the serialized rich |
| 956 | // value elsewhere. This value is used to provide the appropriate |
| 957 | // HTML semantics (ie, the browser will apply CSS styling to the |
| 958 | // flattened `class`, use the flattened `href` to navigate) but |
| 959 | // should not be used by clients /of the MediaWiki DOM spec/ |
| 960 | // (including Parsoid), which should ignore the flattened value |
| 961 | // and consistently use the rich value in order to avoid |
| 962 | // losing/overwriting data. |
| 963 | |
| 964 | // The JSON representation of a rich valued attribute can be |
| 965 | // customized using the mechanisms provided by the wikimedia/json-codec |
| 966 | // library; in particular you will want to use the "implicit typing" |
| 967 | // mechanism provided by the library to avoid bloating the output |
| 968 | // with explicit references to the PHP implementation classes. |
| 969 | |
| 970 | // See |
| 971 | // https://www.mediawiki.org/wiki/Parsoid/MediaWiki_DOM_spec/Rich_Attributes |
| 972 | // for a more detailed discussion of this design. The present |
| 973 | // implementation corresponds to "proposal 1a", the first step in |
| 974 | // the full proposal. |
| 975 | |
| 976 | /** |
| 977 | * Determine whether the given attribute name has "special" HTML |
| 978 | * semantics. For these attributes, a "stringified" flattened |
| 979 | * version of the attribute is stored in the attribute, for |
| 980 | * semantic compatibility with browsers etc, and the "rich" form |
| 981 | * of the attribute is stored in a separate attribute. |
| 982 | * |
| 983 | * Although in theory we could minimize this by looking at the |
| 984 | * names of attributes explicitly reserved for each tag name in |
| 985 | * the HTML spec, at this time we're going to be conservative and |
| 986 | * assume every attribute has "special" semantics that we should |
| 987 | * preserve except for those attributes whose names begin with |
| 988 | * `data-*`. |
| 989 | * |
| 990 | * In the future we might tweak the set of attributes with special |
| 991 | * semantics in order to reduce unnecessary bloat (ie storing |
| 992 | * flattened versions of attributes where the flattened value will |
| 993 | * never be used) and/or to include flattened values for certain |
| 994 | * data-* attributes (for example, if a gadget were to rely on a |
| 995 | * flattened value in `data-time`). |
| 996 | * |
| 997 | * @param string $tagName The tag name of the Element containing the |
| 998 | * attribute |
| 999 | * @param string $attrName The name of the attribute |
| 1000 | * @return bool True if the named attribute has special HTML semantics |
| 1001 | */ |
| 1002 | private static function isHtmlAttributeWithSpecialSemantics( string $tagName, string $attrName ): bool { |
| 1003 | return !(bool)preg_match( '/^data-/i', $attrName ); |
| 1004 | } |
| 1005 | |
| 1006 | private static function nodeDataPropName( string $name ): string { |
| 1007 | return ( $name === 'data-mw' || $name === 'data-parsoid' ) ? |
| 1008 | preg_replace( '/data-/', '', $name ) : self::RICH_ATTR_DATA_PREFIX . $name; |
| 1009 | } |
| 1010 | |
| 1011 | /** |
| 1012 | * @param DOMDataCodec $codec |
| 1013 | * @param object|array $value |
| 1014 | * @param Hint $classHint |
| 1015 | * @return object |
| 1016 | */ |
| 1017 | private static function decodeAttribute( DOMDataCodec $codec, $value, $classHint ) { |
| 1018 | // This value should be decoded |
| 1019 | $value = $codec->newFromJsonArray( $value, $classHint ); |
| 1020 | if ( is_array( $value ) ) { |
| 1021 | // JsonCodec allows class hints to indicate that the value |
| 1022 | // is an array of some object type, but for our purposes |
| 1023 | // the result must always be an object so that it is live. |
| 1024 | $value = (object)$value; |
| 1025 | } |
| 1026 | // To signal that it's been decoded already we need $value |
| 1027 | // not to be an array |
| 1028 | Assert::invariant( |
| 1029 | !is_array( $value ), "rich attribute can't be array" |
| 1030 | ); |
| 1031 | |
| 1032 | return $value; |
| 1033 | } |
| 1034 | |
| 1035 | /** |
| 1036 | * Return the value of a rich attribute as a live (by-reference) object. |
| 1037 | * This also serves as an assertion that there are not conflicting types. |
| 1038 | * |
| 1039 | * @template T |
| 1040 | * @param Element $node The node on which the attribute is to be found. |
| 1041 | * @param string $name The name of the attribute. |
| 1042 | * @param class-string<T>|Hint<T> $classHint |
| 1043 | * @return T|null The attribute value, or null if not present. |
| 1044 | */ |
| 1045 | public static function getAttributeObject( |
| 1046 | Element $node, string $name, $classHint |
| 1047 | ): ?object { |
| 1048 | self::loadRichAttributes( $node, $name ); // lazy load |
| 1049 | if ( !$node->hasAttribute( self::DATA_OBJECT_ATTR_NAME ) ) { |
| 1050 | // Don't create an empty node data object if we don't need to. |
| 1051 | $value = null; |
| 1052 | } else { |
| 1053 | $nodeData = self::getNodeData( $node ); |
| 1054 | $propName = self::nodeDataPropName( $name ); |
| 1055 | $value = $nodeData->$propName ?? null; |
| 1056 | // We lazily decode rich values, because we need to know the $classHint |
| 1057 | // before we decode. Undecoded values are wrapped with an array so |
| 1058 | // we can tell whether the value has been decoded already or not. |
| 1059 | if ( is_array( $value ) ) { |
| 1060 | $value = self::decodeAttribute( self::getCodec( $node ), $value[0], $classHint ); |
| 1061 | $nodeData->$propName = $value; |
| 1062 | $hintName = self::RICH_ATTR_HINT_PREFIX . $name; |
| 1063 | $nodeData->$hintName = $classHint; |
| 1064 | } |
| 1065 | } |
| 1066 | // Special case handling for DocumentFragments with plain-text |
| 1067 | // values. |
| 1068 | if ( |
| 1069 | $value === null && |
| 1070 | $classHint === DocumentFragment::class && |
| 1071 | $node->hasAttribute( $name ) |
| 1072 | ) { |
| 1073 | $value = $node->ownerDocument->createDocumentFragment(); |
| 1074 | DOMCompat::append( $value, DOMCompat::getAttribute( $node, $name ) ); |
| 1075 | self::setAttributeDom( $node, $name, $value ); |
| 1076 | } |
| 1077 | return $value; |
| 1078 | } |
| 1079 | |
| 1080 | /** |
| 1081 | * Return the value of a rich attribute as a live (by-reference) |
| 1082 | * object. This also serves as an assertion that there are not |
| 1083 | * conflicting types. If the value is not present, a default value |
| 1084 | * will be created using `$codec->defaultValue()` falling back to |
| 1085 | * `$className::defaultValue()` and stored as the value of the |
| 1086 | * attribute. |
| 1087 | * |
| 1088 | * @note The $className should have be JsonCodecable (either directly |
| 1089 | * or via a custom JsonClassCodec). |
| 1090 | * |
| 1091 | * @template T |
| 1092 | * @param Element $node The node on which the attribute is to be found. |
| 1093 | * @param string $name The name of the attribute. |
| 1094 | * @param class-string<T>|Hint<T> $classHint |
| 1095 | * @return T|null The attribute value, or null if not present. |
| 1096 | */ |
| 1097 | public static function getAttributeObjectDefault( |
| 1098 | Element $node, string $name, $classHint |
| 1099 | ): ?object { |
| 1100 | $value = self::getAttributeObject( $node, $name, $classHint ); |
| 1101 | if ( $value === null ) { |
| 1102 | $className = $classHint; |
| 1103 | while ( $className instanceof Hint ) { |
| 1104 | Assert::invariant( |
| 1105 | $className->modifier !== Hint::LIST && |
| 1106 | $className->modifier !== Hint::STDCLASS, |
| 1107 | "Can't create default value for list or object" |
| 1108 | ); |
| 1109 | $className = $className->parent; |
| 1110 | } |
| 1111 | '@phan-var string $className'; |
| 1112 | $codec = self::getCodec( $node ); |
| 1113 | $value = $codec->defaultValue( $className ); |
| 1114 | $value ??= new $className; |
| 1115 | self::setAttributeObject( $node, $name, $value, $classHint ); |
| 1116 | } |
| 1117 | return $value; |
| 1118 | } |
| 1119 | |
| 1120 | /** |
| 1121 | * Set the value of a rich attribute, overwriting any previous |
| 1122 | * value. Generally mutating the result returned by the |
| 1123 | * `::getAttribute*Default()` methods should be done instead of |
| 1124 | * using this method, since the objects returned are live. |
| 1125 | * |
| 1126 | * @note For attribute names where |
| 1127 | * `::isHtmlAttributeWithSpecialSemantics()` returns `true` you |
| 1128 | * can customize the "flattened" representation used for HTML |
| 1129 | * semantics via `$codec->flatten()` which falls back to |
| 1130 | * `$className::flatten()`. |
| 1131 | * |
| 1132 | * @param Element $node The node on which the attribute is to be found. |
| 1133 | * @param string $name The name of the attribute. |
| 1134 | * @param object $value The new (object) value for the attribute |
| 1135 | * @param class-string|Hint|null $classHint Optional serialization hint |
| 1136 | */ |
| 1137 | public static function setAttributeObject( |
| 1138 | Element $node, string $name, object $value, $classHint = null |
| 1139 | ): void { |
| 1140 | // Remove attribute from DOM; will be rewritten from node data during |
| 1141 | // serialization. |
| 1142 | self::removeAttributeObject( $node, $name ); |
| 1143 | $nodeData = self::getNodeData( $node ); |
| 1144 | self::setAttributeObjectNodeData( $nodeData, $name, $value, $classHint ); |
| 1145 | } |
| 1146 | |
| 1147 | /** |
| 1148 | * @internal For use by TreeBuilderStage only |
| 1149 | * @param NodeData $nodeData |
| 1150 | * @param string $name The name of the attribute. |
| 1151 | * @param object $value The new (object) value for the attribute |
| 1152 | * @param class-string|Hint|null $classHint Optional serialization hint |
| 1153 | */ |
| 1154 | public static function setAttributeObjectNodeData( |
| 1155 | NodeData $nodeData, string $name, object $value, $classHint = null |
| 1156 | ): void { |
| 1157 | $propName = self::nodeDataPropName( $name ); |
| 1158 | $nodeData->$propName = $value; |
| 1159 | if ( $classHint === null && is_a( $value, RichCodecable::class ) ) { |
| 1160 | $className = get_class( $value ); |
| 1161 | $classHint = $className::hint(); |
| 1162 | } |
| 1163 | $hintName = self::RICH_ATTR_HINT_PREFIX . $name; |
| 1164 | $nodeData->$hintName = $classHint; |
| 1165 | } |
| 1166 | |
| 1167 | /** |
| 1168 | * Remove a rich attribute. |
| 1169 | * |
| 1170 | * @param Element $node The node on which the attribute is to be found. |
| 1171 | * @param string $name The name of the attribute. |
| 1172 | */ |
| 1173 | public static function removeAttributeObject( |
| 1174 | Element $node, string $name |
| 1175 | ): void { |
| 1176 | $node->removeAttribute( $name ); |
| 1177 | self::removeFromExpandedAttrs( $node, $name ); |
| 1178 | if ( $node->hasAttribute( self::DATA_OBJECT_ATTR_NAME ) ) { |
| 1179 | $nodeData = self::getNodeData( $node ); |
| 1180 | $propName = self::nodeDataPropName( $name ); |
| 1181 | if ( in_array( $propName, NodeData::PERSISTENT_ATTR_NAMES, true ) ) { |
| 1182 | $nodeData->$propName = null; |
| 1183 | } else { |
| 1184 | unset( $nodeData->$propName ); |
| 1185 | } |
| 1186 | $hintName = self::RICH_ATTR_HINT_PREFIX . $name; |
| 1187 | unset( $nodeData->$hintName ); |
| 1188 | } |
| 1189 | } |
| 1190 | |
| 1191 | /** |
| 1192 | * Helper function to remove any entries from data-mw.attribs which match |
| 1193 | * this attribute name. They will be rewritten during rich attribute |
| 1194 | * serialization if necessary. |
| 1195 | * @param Element $node |
| 1196 | * @param string $name |
| 1197 | */ |
| 1198 | private static function removeFromExpandedAttrs( |
| 1199 | Element $node, string $name |
| 1200 | ): void { |
| 1201 | if ( !self::isHtmlAttributeWithSpecialSemantics( $node->tagName, $name ) ) { |
| 1202 | return; |
| 1203 | } |
| 1204 | // Don't create a new data-mw yet if we don't need one. |
| 1205 | $dataMw = self::getDataMwIfExists( $node ); |
| 1206 | if ( $dataMw === null ) { |
| 1207 | return; |
| 1208 | } |
| 1209 | // If there was a data-mw.attribs for this attribute, remove it |
| 1210 | // (it will be rewritten during serialization later) |
| 1211 | $dataMw->attribs = array_values( array_filter( |
| 1212 | $dataMw->attribs ?? [], |
| 1213 | static function ( $a ) use ( $name ) { |
| 1214 | if ( !( $a instanceof DataMwAttrib ) ) { |
| 1215 | return true; |
| 1216 | } |
| 1217 | if ( $a->getKeyString() === $name ) { |
| 1218 | return false; // Remove this entry |
| 1219 | } |
| 1220 | return true; |
| 1221 | } |
| 1222 | ) ); |
| 1223 | if ( count( $dataMw->attribs ) === 0 ) { |
| 1224 | unset( $dataMw->attribs ); |
| 1225 | DOMUtils::removeTypeOf( $node, 'mw:ExpandedAttrs' ); |
| 1226 | } |
| 1227 | } |
| 1228 | |
| 1229 | /** |
| 1230 | * Return the value of a rich attribute as a live `DocumentFragment`. |
| 1231 | * This also serves as an assertion that there are not conflicting types. |
| 1232 | * |
| 1233 | * @note A string-valued attribute will be returned as a DocumentFragment |
| 1234 | * with a single Text node. This supports the efficient serialization |
| 1235 | * of 'simple' DocumentFragments as simple strings. |
| 1236 | * |
| 1237 | * @param Element $node The node on which the attribute is to be found. |
| 1238 | * @param string $name The name of the attribute. |
| 1239 | * @return ?DocumentFragment The attribute value, or null if not present. |
| 1240 | */ |
| 1241 | public static function getAttributeDom( |
| 1242 | Element $node, string $name |
| 1243 | ): ?DocumentFragment { |
| 1244 | // As it turns out, the implementation for a DocumentFragment is |
| 1245 | // the same; all the implementation differences are in the codec |
| 1246 | return self::getAttributeObject( |
| 1247 | $node, $name, DocumentFragment::class |
| 1248 | ); |
| 1249 | } |
| 1250 | |
| 1251 | /** |
| 1252 | * Return the value of a rich attribute as a `DocumentFragment`, |
| 1253 | * creating a new document fragment and setting the attribute if the |
| 1254 | * attribute was not previously present. |
| 1255 | * |
| 1256 | * @param Element $node The node on which the attribute is to be found. |
| 1257 | * @param string $name The name of the attribute. |
| 1258 | * @return DocumentFragment The attribute value. |
| 1259 | */ |
| 1260 | public static function getAttributeDomDefault( |
| 1261 | Element $node, string $name |
| 1262 | ): DocumentFragment { |
| 1263 | $value = self::getAttributeDom( $node, $name ); |
| 1264 | if ( $value === null ) { |
| 1265 | $value = $node->ownerDocument->createDocumentFragment(); |
| 1266 | self::setAttributeDom( $node, $name, $value ); |
| 1267 | } |
| 1268 | return $value; |
| 1269 | } |
| 1270 | |
| 1271 | /** |
| 1272 | * Set the value of a rich attribute, overwriting any previous |
| 1273 | * value. Generally mutating the result returned by the |
| 1274 | * `::getAttribute*Default()` methods should be done instead of |
| 1275 | * using this method, since the objects returned are live. |
| 1276 | * |
| 1277 | * @param Element $node The node on which the attribute is to be found. |
| 1278 | * @param string $name The name of the attribute. |
| 1279 | * @param DocumentFragment $value |
| 1280 | */ |
| 1281 | public static function setAttributeDom( |
| 1282 | Element $node, string $name, DocumentFragment $value |
| 1283 | ): void { |
| 1284 | // Remove attribute from DOM; will be rewritten from node data during |
| 1285 | // serialization. |
| 1286 | self::removeAttributeDom( $node, $name ); |
| 1287 | $nodeData = self::getNodeData( $node ); |
| 1288 | $propName = self::nodeDataPropName( $name ); |
| 1289 | $nodeData->$propName = $value; |
| 1290 | $hintName = self::RICH_ATTR_HINT_PREFIX . $name; |
| 1291 | $nodeData->$hintName = DocumentFragment::class; |
| 1292 | } |
| 1293 | |
| 1294 | /** |
| 1295 | * Remove a rich attribute. |
| 1296 | * |
| 1297 | * @param Element $node The node on which the attribute is to be found. |
| 1298 | * @param string $name The name of the attribute. |
| 1299 | */ |
| 1300 | public static function removeAttributeDom( |
| 1301 | Element $node, string $name |
| 1302 | ): void { |
| 1303 | // Remove attribute from DOM; will be rewritten from node data during |
| 1304 | // serialization. |
| 1305 | $node->removeAttribute( $name ); |
| 1306 | self::removeFromExpandedAttrs( $node, $name ); |
| 1307 | if ( $node->hasAttribute( self::DATA_OBJECT_ATTR_NAME ) ) { |
| 1308 | $nodeData = self::getNodeData( $node ); |
| 1309 | $propName = self::nodeDataPropName( $name ); |
| 1310 | unset( $nodeData->$propName ); |
| 1311 | $hintName = self::RICH_ATTR_HINT_PREFIX . $name; |
| 1312 | unset( $nodeData->$hintName ); |
| 1313 | } |
| 1314 | } |
| 1315 | |
| 1316 | // Serialization/deserialization support for rich attributes. |
| 1317 | |
| 1318 | // There are many possible serializations which could be used. |
| 1319 | // For the moment we've chosen the simplest possible one, which |
| 1320 | // embeds big JSON blobs in attribute values. For "attributes |
| 1321 | // with special HTML semantics" the JSON blobs are stored in |
| 1322 | // data-mw.attribs and the straight HTML attribute value is a |
| 1323 | // flattened form of the true value. |
| 1324 | |
| 1325 | /** |
| 1326 | * Internal function to lazy-load rich attribute data from the HTML |
| 1327 | * DOM representation. |
| 1328 | * @param Element $node The node possibly containing the rich attribute |
| 1329 | * @param string $name The attribute name we are going to load values for |
| 1330 | */ |
| 1331 | private static function loadRichAttributes( |
| 1332 | Element $node, string $name |
| 1333 | ): void { |
| 1334 | // Because we don't have a complete schema for the document which |
| 1335 | // identifies which attributes are 'rich' and which are not, we |
| 1336 | // lazily-load attributes one-by-one once we know their names and types |
| 1337 | // instead of trying to preload them in bulk. |
| 1338 | |
| 1339 | // *However* in order to avoid O(N^2) manipulation of the |
| 1340 | // data-mw.attribs list, we do move all the values from data-mw.attribs |
| 1341 | // into NodeData, even those not matching our given name. We can't |
| 1342 | // decode those yet: they will be decoded once getAttributeObject() |
| 1343 | // is called on them to provide the proper type hint (or else they |
| 1344 | // will eventually be reserialized in their undecoded form). |
| 1345 | |
| 1346 | $flatValue = DOMCompat::getAttribute( $node, $name ); |
| 1347 | if ( $flatValue === null ) { |
| 1348 | // Use the presence of the attribute in the DOM to indicate |
| 1349 | // whether this attribute has been loaded; this avoids (for |
| 1350 | // example) traversing AttributeExpander entries in |
| 1351 | // data-mw.attribs multiple times looking for the name of a |
| 1352 | // rich attribute. If the attribute is not in the DOM either |
| 1353 | // there is no attribute of this name or it has already been |
| 1354 | // loaded. |
| 1355 | |
| 1356 | // Since we are lazy-loading node data, if we are missing a flat-value |
| 1357 | // look in the pagebundle and load from there, if necessary. |
| 1358 | if ( ( $name === 'data-parsoid' || $name === 'data-mw' ) && |
| 1359 | !$node->hasAttribute( self::DATA_OBJECT_ATTR_NAME ) |
| 1360 | ) { |
| 1361 | self::loadFromPageBundle( $node ); |
| 1362 | } |
| 1363 | |
| 1364 | return; |
| 1365 | } |
| 1366 | |
| 1367 | if ( self::isHtmlAttributeWithSpecialSemantics( $node->tagName, $name ) ) { |
| 1368 | // Look aside at data-mw for attributes with special semantics |
| 1369 | $dataMw = self::getDataMwIfExists( $node ); |
| 1370 | if ( $dataMw === null ) { |
| 1371 | // No data-mw, so no rich value for this attribute |
| 1372 | return; |
| 1373 | } |
| 1374 | // Load all attribute values from $dataMw->attribs to avoid O(N^2) |
| 1375 | // loading of list |
| 1376 | if ( $dataMw->attribs ?? false ) { |
| 1377 | $unused = []; |
| 1378 | $nodeData = self::getNodeData( $node ); |
| 1379 | foreach ( $dataMw->attribs as $a ) { |
| 1380 | if ( !( $a instanceof DataMwAttrib ) ) { |
| 1381 | // This shouldn't happen! |
| 1382 | $unused[] = $a; |
| 1383 | continue; |
| 1384 | } |
| 1385 | $key = $a->key; |
| 1386 | $value = $a->value; |
| 1387 | if ( is_array( $key ) ) { |
| 1388 | if ( count( $key ) === 1 && isset( $key['txt'] ) ) { |
| 1389 | $key = $key['txt']; |
| 1390 | } else { |
| 1391 | // Attribute expander may use array values for |
| 1392 | // key, since it supports rich key values. |
| 1393 | // Ignore any entries created this way, since |
| 1394 | // we can't preserve their values: they will be |
| 1395 | // added to $unused and replaced. |
| 1396 | $unused[] = $a; |
| 1397 | continue; |
| 1398 | } |
| 1399 | } |
| 1400 | $propName = self::RICH_ATTR_DATA_PREFIX . $key; |
| 1401 | $hintName = self::RICH_ATTR_HINT_PREFIX . $key; |
| 1402 | if ( isset( $value['html'] ) ) { |
| 1403 | // DocumentFragment has already been decoded. |
| 1404 | $nodeData->$propName = $value['html']; |
| 1405 | $nodeData->$hintName = DocumentFragment::class; |
| 1406 | } elseif ( isset( $value['rich'] ) ) { |
| 1407 | // wrap $value with an array to indicate that |
| 1408 | // is it not yet decoded. Preserve the flattened |
| 1409 | // value as well in case we round-trip without |
| 1410 | // modifying this value. |
| 1411 | $nodeData->$propName = [ $value['rich'], $flatValue ]; |
| 1412 | // Don't know the decode hint yet. |
| 1413 | } else { |
| 1414 | $unused[] = $a; |
| 1415 | continue; |
| 1416 | } |
| 1417 | // Signal that the value has been moved to NodeData |
| 1418 | // (this will also short cut this iteration over |
| 1419 | // data-mw.attribs in future calls) |
| 1420 | $node->removeAttribute( $key ); |
| 1421 | } |
| 1422 | if ( count( $unused ) === 0 ) { |
| 1423 | unset( $dataMw->attribs ); |
| 1424 | } else { |
| 1425 | $dataMw->attribs = $unused; |
| 1426 | } |
| 1427 | } |
| 1428 | return; |
| 1429 | } |
| 1430 | // The attribute does not have "special HTML semantics" |
| 1431 | $decoded = json_decode( $flatValue, true ); |
| 1432 | // $decoded is the 'non-string' form of the value; we can't finish |
| 1433 | // deserializing it into an object until we know the appropriate type hint. |
| 1434 | self::removeAttributeObject( $node, $name ); |
| 1435 | $nodeData = self::getNodeData( $node ); |
| 1436 | $propName = self::nodeDataPropName( $name ); |
| 1437 | // Mark this as undecoded by wrapping it as an array, |
| 1438 | // since decoded values will always be objects. |
| 1439 | // (Attribute values without "special HTML semantics" do not |
| 1440 | // have flattened versions, so 2nd element to this array isn't needed.) |
| 1441 | $nodeData->$propName = [ $decoded ]; |
| 1442 | } |
| 1443 | |
| 1444 | /** |
| 1445 | * @param DOMDataCodec $codec |
| 1446 | * @param Element $node |
| 1447 | * @param bool $hasSpecialSemantics |
| 1448 | * @param NodeData $nodeData |
| 1449 | * @param string $attrName |
| 1450 | * @param mixed $v |
| 1451 | */ |
| 1452 | private static function storeAttribute( |
| 1453 | DOMDataCodec $codec, Element $node, bool $hasSpecialSemantics, NodeData $nodeData, string $attrName, $v |
| 1454 | ): void { |
| 1455 | $df = null; |
| 1456 | if ( is_array( $v ) ) { |
| 1457 | // If $v is an array, it was never decoded. |
| 1458 | $json = $v[0]; |
| 1459 | $flat = $v[1] ?? null; |
| 1460 | } else { |
| 1461 | $hintName = self::RICH_ATTR_HINT_PREFIX . $attrName; |
| 1462 | $classHint = $nodeData->$hintName ?? null; |
| 1463 | if ( is_a( $v, RichCodecable::class ) ) { |
| 1464 | $classHint ??= $v::hint(); |
| 1465 | } |
| 1466 | $classHint ??= get_class( $v ); |
| 1467 | try { |
| 1468 | // NOTE: call 'flatten()' before 'toJsonArray()' since |
| 1469 | // the latter may have side effects on $v. |
| 1470 | $flat = $codec->flatten( $v ); |
| 1471 | if ( $hasSpecialSemantics && $classHint === DocumentFragment::class ) { |
| 1472 | // Special case for attributes stored in |
| 1473 | // DataMwAttribs::$html |
| 1474 | $json = null; |
| 1475 | $df = $v; |
| 1476 | } else { |
| 1477 | $json = $codec->toJsonArray( $v, $classHint ); |
| 1478 | } |
| 1479 | } catch ( InvalidArgumentException $e ) { |
| 1480 | // For better debuggability, include the attribute name |
| 1481 | throw new InvalidArgumentException( "$attrName: " . $e->getMessage() ); |
| 1482 | } |
| 1483 | } |
| 1484 | if ( !$hasSpecialSemantics ) { |
| 1485 | $encoded = PHPUtils::jsonEncode( $json ); |
| 1486 | $node->setAttribute( $attrName, $encoded ); |
| 1487 | } else { |
| 1488 | // For compatibility, store the rich value in data-mw.attribs |
| 1489 | // and store a flattened version in the $attrName. |
| 1490 | $complex = true; |
| 1491 | '@phan-var DocumentFragment $df'; |
| 1492 | if ( $flat !== null ) { |
| 1493 | $node->setAttribute( $attrName, $flat ); |
| 1494 | // Identify special case where the html is trivial to |
| 1495 | // avoid bloating the HTML. |
| 1496 | if ( |
| 1497 | $df && $df->firstChild && |
| 1498 | $df->firstChild->nextSibling === null && |
| 1499 | $df->firstChild->nodeType === XML_TEXT_NODE && |
| 1500 | $df->firstChild->nodeValue === $flat |
| 1501 | ) { |
| 1502 | $complex = false; |
| 1503 | } |
| 1504 | } else { |
| 1505 | $node->removeAttribute( $attrName ); |
| 1506 | } |
| 1507 | // Only store rich content if it is different from the |
| 1508 | // flattened attribute value |
| 1509 | if ( $complex ) { |
| 1510 | $dataMw = self::getDataMw( $node ); |
| 1511 | $value = $df ? [ 'html' => $df ] : [ 'rich' => $json ]; |
| 1512 | $dataMw->attribs[] = new DataMwAttrib( $attrName, $value ); |
| 1513 | DOMUtils::addTypeOf( $node, 'mw:ExpandedAttrs' ); |
| 1514 | } |
| 1515 | } |
| 1516 | } |
| 1517 | |
| 1518 | /** |
| 1519 | * Internal function to encode rich attribute data into an HTML |
| 1520 | * DOM representation. |
| 1521 | * @param Element $node The node possibly containing the rich attribute |
| 1522 | * @param array $options The options provided to ::storeDataAttribs() |
| 1523 | */ |
| 1524 | private static function storeRichAttributes( Element $node, array $options ): ?stdClass { |
| 1525 | $bag = self::getBag( $node->ownerDocument ); // FIXME: Pass this in as arg? |
| 1526 | $storeInPageBundle = !empty( $options['storeInPageBundle'] ); |
| 1527 | $nodeId = DOMCompat::getAttribute( $node, self::DATA_OBJECT_ATTR_NAME ); |
| 1528 | if ( $nodeId === null ) { |
| 1529 | // The node data hasn't been loaded. |
| 1530 | // Handle four possible (PB, Inline) --> (PB, Inline) scenarios |
| 1531 | $pb = $bag->inputPageBundle; |
| 1532 | if ( $pb === null ) { |
| 1533 | if ( !$storeInPageBundle ) { |
| 1534 | // Inline --> Inline. Nothing else to do. |
| 1535 | return null; |
| 1536 | } |
| 1537 | } else { |
| 1538 | $id = DOMCompat::getAttribute( $node, 'id' ); |
| 1539 | if ( $id === null && !$storeInPageBundle ) { |
| 1540 | // PB --> Inline. |
| 1541 | // No id => no source pb-data => leave any |
| 1542 | // existing inline attribute as is and return. |
| 1543 | return null; |
| 1544 | } |
| 1545 | |
| 1546 | if ( $storeInPageBundle && !self::hasInlineRichAttributes( $node ) ) { |
| 1547 | // PB -> PB but no local data, leave page bundle as-is and return. |
| 1548 | |
| 1549 | // FIXME: we *should* only need to copy entries from the source |
| 1550 | // page bundle to the destination page bundle, but we don't know |
| 1551 | // which ones they are. If we initialized the destination page |
| 1552 | // bundle from the source page bundle this wouldn't be an issue |
| 1553 | // and we could `return null` here. |
| 1554 | // |
| 1555 | // return null; |
| 1556 | } |
| 1557 | } |
| 1558 | |
| 1559 | // Inline -> PB or PB -> PB/Inline |
| 1560 | // In all cases, load data-parsoid via loadRichAttributes |
| 1561 | // which has precedence logic there between inline & pagebundle |
| 1562 | // attributes. But, eager load data-mw and everything else |
| 1563 | // to ensure all embedded attributes are properly handled. |
| 1564 | } |
| 1565 | |
| 1566 | // loadRichAttributes only "loads" the attributes but leaves them |
| 1567 | // in decoded array form. But, eagerlyLoadRichAttributes loads *and* |
| 1568 | // decodes the attributes. Especially for data-parsoid, that is wasted |
| 1569 | // work most of the time in the OutputTransformPipeline passes in core. |
| 1570 | // So, we call loadRichAttributes first to prevent eagerlyLoadRichAttributes |
| 1571 | // from decoding the attribute. Ideally, we would do this or all the rich |
| 1572 | // attributes, but the code below doesn't handle undecoded rich attribute |
| 1573 | // correctly except for data-parsoid & data-mw. That can be a future perf opt |
| 1574 | // if deemed useful / necessary. |
| 1575 | self::loadRichAttributes( $node, "data-parsoid" ); |
| 1576 | self::eagerlyLoadRichAttributes( $node ); |
| 1577 | $nodeData = self::getNodeData( $node ); |
| 1578 | |
| 1579 | $codec = self::getCodec( $node ); |
| 1580 | |
| 1581 | // At present, rich attributes may be serialized into the data-mw attributes |
| 1582 | // which are serialized in the pagebundle; thus we need to serialize all |
| 1583 | // the "attributes with special html semantics" (which will get added to data-mw) |
| 1584 | // *before* we handle the other attributes and the page bundle. |
| 1585 | |
| 1586 | // Special attributes first |
| 1587 | foreach ( get_object_vars( $nodeData ) as $k => $v ) { |
| 1588 | // Look for dynamic properties with names w/ the proper prefix |
| 1589 | if ( str_starts_with( $k, self::RICH_ATTR_DATA_PREFIX ) ) { |
| 1590 | $attrName = substr( $k, strlen( self::RICH_ATTR_DATA_PREFIX ) ); |
| 1591 | if ( self::isHtmlAttributeWithSpecialSemantics( $node->tagName, $attrName ) ) { |
| 1592 | self::storeAttribute( $codec, $node, true, $nodeData, $attrName, $v ); |
| 1593 | unset( $nodeData->$k ); |
| 1594 | } |
| 1595 | } |
| 1596 | } |
| 1597 | |
| 1598 | // Deal with data-parsoid (and data-mw). |
| 1599 | // |
| 1600 | // Because of lazy loading, data-parsoid can be in one of three states: |
| 1601 | // * Fully loaded as DataParsoid in $nodeData->parsoid. |
| 1602 | // * Partially loaded as a decoded JSON blob in $nodeData->parsoid. |
| 1603 | // * Not loaded at all and available via the HTML data-parsoid attribute (string). |
| 1604 | $dp = $nodeData->parsoid; |
| 1605 | $discardDataParsoid = !empty( $options['discardDataParsoid'] ) || |
| 1606 | ( $dp instanceof DataParsoid && $dp->getTempFlag( TempData::DISCARDABLE_DP ) ) || |
| 1607 | ( $dp instanceof DataParsoid && !$options['serializeNewEmptyDp'] && |
| 1608 | $dp->getTempFlag( TempData::IS_NEW ) && $dp->isEmpty() ); |
| 1609 | |
| 1610 | $pbData = null; |
| 1611 | if ( !$discardDataParsoid ) { |
| 1612 | // Force a load of data-parsoid if: |
| 1613 | // * we have to migrate it from an inline-attribute to the pagebundle. |
| 1614 | // * we have to assign empty data-parsoid objects to nodes that don't have them |
| 1615 | // |
| 1616 | // In all other cases, if an inline attribute is not loaded, |
| 1617 | // we avoid the overhead of loading it from the inline attribute |
| 1618 | // just to store it back into the inline attribute unchanged. |
| 1619 | if ( $dp === null && ( $storeInPageBundle || $options['serializeNewEmptyDp'] ) ) { |
| 1620 | self::loadRichAttributes( $node, "data-parsoid" ); |
| 1621 | $dp = $nodeData->parsoid[0] ?? null; // undecoded json blob |
| 1622 | if ( $dp === [] ) { |
| 1623 | // This $dp was lazily instantiated above and so is not new. |
| 1624 | // But since it is empty, create a new object to prevent |
| 1625 | // serialization to []. |
| 1626 | $dp = new DataParsoid; |
| 1627 | } elseif ( $dp === null && $options['serializeNewEmptyDp'] ) { |
| 1628 | // NOTE: We always create an empty data-parsoid for all nodes in the DOM |
| 1629 | // to distinguish "original HTML" from "newly added nodes in edited HTML" |
| 1630 | // to aid selser. But, if we were marking new nodes, we would have |
| 1631 | // discarded dp (as above) in an eagerly-load-data-parsoid set up. |
| 1632 | // Hence also, the check if we were tracking new nodes. |
| 1633 | $dp = new DataParsoid; |
| 1634 | } |
| 1635 | } elseif ( is_array( $dp ) ) { |
| 1636 | // Unwrap the array wrapper added by getNodeData |
| 1637 | $dp = $dp[0]; |
| 1638 | if ( $dp === [] ) { |
| 1639 | // Given that this is a json blob, this was previously loaded and |
| 1640 | // as such is not new - it is only being instantiated lazily here. |
| 1641 | // But since it is empty, create a new object to prevent |
| 1642 | // serialization to []. |
| 1643 | $dp = new DataParsoid; |
| 1644 | } |
| 1645 | } |
| 1646 | |
| 1647 | if ( $dp !== null ) { |
| 1648 | if ( $dp instanceof DataParsoid ) { |
| 1649 | if ( empty( $options['keepTmp'] ) ) { |
| 1650 | // FIXME: $dp->toJsonArray drops tmp so it's discarded regardless |
| 1651 | // of this flag |
| 1652 | // @phan-suppress-next-line PhanTypeObjectUnsetDeclaredProperty |
| 1653 | unset( $dp->tmp ); |
| 1654 | } |
| 1655 | $dp = $codec->toJsonArray( $dp, DataParsoid::hint() ); |
| 1656 | } |
| 1657 | |
| 1658 | if ( $storeInPageBundle ) { |
| 1659 | $pbData ??= new stdClass; |
| 1660 | $pbData->parsoid = $dp; |
| 1661 | } else { |
| 1662 | $node->setAttribute( 'data-parsoid', PHPUtils::jsonEncode( $dp ) ); |
| 1663 | } |
| 1664 | } |
| 1665 | } |
| 1666 | |
| 1667 | // Special handling for data-mw. |
| 1668 | // (a) now that DataMw is a class type, we should never actually |
| 1669 | // have "invalid" data mw objects in practice; |
| 1670 | // (b) eventually we can remove support for output content version |
| 1671 | // older than 999.x. |
| 1672 | |
| 1673 | $storeDmwInPb = $storeInPageBundle && |
| 1674 | // The pagebundle didn't have data-mw before 999.x |
| 1675 | Semver::satisfies( $options['outputContentVersion'] ?? '0.0.0', '^999.0.0' ); |
| 1676 | |
| 1677 | $dmw = $nodeData->mw; |
| 1678 | if ( $dmw !== null ) { |
| 1679 | if ( $dmw instanceof DataMw ) { |
| 1680 | // Strip empty data-mw attributes |
| 1681 | $dmw = $dmw->isEmpty() ? null : $codec->toJsonArray( $dmw, DataMw::hint() ); |
| 1682 | } elseif ( is_array( $dmw ) ) { |
| 1683 | // Unwrap the array wrapper added by getNodeData |
| 1684 | $dmw = $dmw[0]; |
| 1685 | } |
| 1686 | } elseif ( $storeDmwInPb ) { |
| 1687 | // Force a load of data-mw to migrate it from |
| 1688 | // an inline-attribute to the pagebundle. |
| 1689 | self::loadRichAttributes( $node, "data-mw" ); |
| 1690 | $dmw = $nodeData->mw[0] ?? null; // undecoded json blob OR null |
| 1691 | } |
| 1692 | |
| 1693 | if ( $dmw !== null ) { |
| 1694 | if ( $storeDmwInPb ) { |
| 1695 | $pbData ??= new stdClass; |
| 1696 | $pbData->mw = $dmw; |
| 1697 | } else { |
| 1698 | $node->setAttribute( 'data-mw', PHPUtils::jsonEncode( $dmw ) ); |
| 1699 | } |
| 1700 | } |
| 1701 | |
| 1702 | // Non-special attributes next |
| 1703 | // (data-parsoid and data-mw could be done in this loop with the |
| 1704 | // rest of the "non-special" attributes, but at present they have |
| 1705 | // a lot of special handling. In principle they are just the same |
| 1706 | // as the rest of the lazy-loaded rich attributes, though.) |
| 1707 | foreach ( get_object_vars( $nodeData ) as $k => $v ) { |
| 1708 | // Look for dynamic properties with names w/ the proper prefix |
| 1709 | if ( str_starts_with( $k, self::RICH_ATTR_DATA_PREFIX ) ) { |
| 1710 | $attrName = substr( $k, strlen( self::RICH_ATTR_DATA_PREFIX ) ); |
| 1711 | self::storeAttribute( $codec, $node, false, $nodeData, $attrName, $v ); |
| 1712 | unset( $nodeData->$k ); |
| 1713 | } |
| 1714 | } |
| 1715 | |
| 1716 | return $pbData; |
| 1717 | } |
| 1718 | |
| 1719 | /** |
| 1720 | * Modify the attribute array, replacing data-object-id with JSON |
| 1721 | * encoded data. This is just a debugging hack, not to be confused with |
| 1722 | * DOMDataUtils::storeDataAttribs(), and does not store flattened |
| 1723 | * versions of attributes. |
| 1724 | * |
| 1725 | * @param Element $node |
| 1726 | * @param array &$attrs |
| 1727 | * @param bool $keepTmp |
| 1728 | * @param bool $storeDiffMark |
| 1729 | */ |
| 1730 | public static function dumpRichAttribs( Element $node, array &$attrs, bool $keepTmp, bool $storeDiffMark ): void { |
| 1731 | $nodeData = self::getNodeData( $node, init: false ); |
| 1732 | if ( $nodeData === null ) { |
| 1733 | return; // No rich attributes here |
| 1734 | } |
| 1735 | |
| 1736 | $codec = self::getCodec( $node ); |
| 1737 | // Reset to a default set of codec options |
| 1738 | // (in particular, make sure 'useFragmentBank' is not set) |
| 1739 | $oldOptions = $codec->setOptions( [ 'noSideEffects' => true, ] ); |
| 1740 | foreach ( get_object_vars( $nodeData ) as $k => $v ) { |
| 1741 | // Look for dynamic properties with names w/ the proper prefix |
| 1742 | $isRichAttr = false; |
| 1743 | if ( str_starts_with( $k, self::RICH_ATTR_DATA_PREFIX ) ) { |
| 1744 | $isRichAttr = true; |
| 1745 | $attrName = substr( $k, strlen( self::RICH_ATTR_DATA_PREFIX ) ); |
| 1746 | } elseif ( $k === 'parsoid' || $k === 'mw' ) { |
| 1747 | $isRichAttr = true; |
| 1748 | $attrName = "data-$k"; |
| 1749 | } |
| 1750 | if ( $isRichAttr ) { |
| 1751 | '@phan-var string $attrName'; |
| 1752 | if ( $v === null ) { |
| 1753 | $v = DOMCompat::getAttribute( $node, $attrName ); |
| 1754 | if ( $v ) { |
| 1755 | $attrs[$attrName] = $v; |
| 1756 | } |
| 1757 | } elseif ( is_array( $v ) ) { |
| 1758 | // If $v is an array, it was never decoded. |
| 1759 | $json = $v[0]; |
| 1760 | $attrs[$attrName] = PHPUtils::jsonEncode( $json ); |
| 1761 | } else { |
| 1762 | if ( $k === 'parsoid' ) { |
| 1763 | if ( !$keepTmp ) { |
| 1764 | $v = clone $v; |
| 1765 | unset( $v->tmp ); |
| 1766 | } |
| 1767 | } |
| 1768 | $hintName = self::RICH_ATTR_HINT_PREFIX . $attrName; |
| 1769 | $classHint = $nodeData->$hintName ?? null; |
| 1770 | if ( is_a( $v, RichCodecable::class ) ) { |
| 1771 | $classHint ??= $v::hint(); |
| 1772 | } |
| 1773 | $classHint ??= get_class( $v ); |
| 1774 | $json = $codec->toJsonArray( $v, $classHint ); |
| 1775 | $attrs[$attrName] = PHPUtils::jsonEncode( $json ); |
| 1776 | } |
| 1777 | } |
| 1778 | } |
| 1779 | |
| 1780 | if ( !$storeDiffMark ) { |
| 1781 | unset( $attrs['data-parsoid-diff'] ); |
| 1782 | } |
| 1783 | unset( $attrs[self::DATA_OBJECT_ATTR_NAME] ); |
| 1784 | // Restore codec options |
| 1785 | $codec->setOptions( $oldOptions ); |
| 1786 | } |
| 1787 | } |