} */ public $parsoid; /** * A map from ID to the array serialization of DataMw for the Node * with that ID. * * @var null|array{ids:array} */ public $mw; /** @var ?string */ public $version; /** * A map of HTTP headers: both name and value should be strings. * @var array|null */ public $headers; /** @var string|null */ public $contentmodel; public function __construct( string $html, ?array $parsoid = null, ?array $mw = null, ?string $version = null, ?array $headers = null, ?string $contentmodel = null ) { $this->html = $html; $this->parsoid = $parsoid; $this->mw = $mw; $this->version = $version; $this->headers = $headers; $this->contentmodel = $contentmodel; } public function toDom(): Document { $doc = DOMUtils::parseHTML( $this->html ); self::apply( $doc, $this ); return $doc; } public function toHtml(): string { return ContentUtils::toXML( $this->toDom() ); } /** * Check if this pagebundle is valid. * @param string $contentVersion Document content version to validate against. * @param ?string &$errorMessage Error message will be returned here. * @return bool */ public function validate( string $contentVersion, ?string &$errorMessage = null ) { if ( !$this->parsoid || !isset( $this->parsoid['ids'] ) ) { $errorMessage = 'Invalid data-parsoid was provided.'; return false; } elseif ( Semver::satisfies( $contentVersion, '^999.0.0' ) && ( !$this->mw || !isset( $this->mw['ids'] ) ) ) { $errorMessage = 'Invalid data-mw was provided.'; return false; } return true; } /** * @return array */ public function responseData() { $responseData = [ 'contentmodel' => $this->contentmodel ?? '', 'html' => [ 'headers' => array_merge( [ 'content-type' => 'text/html; charset=utf-8; ' . 'profile="https://www.mediawiki.org/wiki/Specs/HTML/' . $this->version . '"', ], $this->headers ?? [] ), 'body' => $this->html, ], 'data-parsoid' => [ 'headers' => [ 'content-type' => 'application/json; charset=utf-8; ' . 'profile="https://www.mediawiki.org/wiki/Specs/data-parsoid/' . $this->version . '"', ], 'body' => $this->parsoid, ], ]; if ( Semver::satisfies( $this->version, '^999.0.0' ) ) { $responseData['data-mw'] = [ 'headers' => [ 'content-type' => 'application/json; charset=utf-8; ' . 'profile="https://www.mediawiki.org/wiki/Specs/data-mw/' . $this->version . '"', ], 'body' => $this->mw, ]; } return $responseData; } /** * Applies the `data-*` attributes JSON structure to the document. * Leaves `id` attributes behind -- they are used by citation code to * extract `` body from the DOM. * * @param Document $doc doc * @param PageBundle $pb page bundle */ public static function apply( Document $doc, PageBundle $pb ): void { DOMUtils::visitDOM( DOMCompat::getBody( $doc ), static function ( Node $node ) use ( $pb ): void { if ( $node instanceof Element ) { $id = DOMCompat::getAttribute( $node, 'id' ); if ( $id === null ) { return; } if ( isset( $pb->parsoid['ids'][$id] ) ) { DOMDataUtils::setJSONAttribute( $node, 'data-parsoid', $pb->parsoid['ids'][$id] ); } if ( isset( $pb->mw['ids'][$id] ) ) { // Only apply if it isn't already set. This means // earlier applications of the pagebundle have higher // precedence, inline data being the highest. if ( !$node->hasAttribute( 'data-mw' ) ) { DOMDataUtils::setJSONAttribute( $node, 'data-mw', $pb->mw['ids'][$id] ); } } } } ); } /** * Encode some of these properties for emitting in the element of a doc * @return string */ public function encodeForHeadElement(): string { // Note that $this->parsoid and $this->mw are already serialized arrays // so a naive jsonEncode is sufficient. We don't need a codec. return PHPUtils::jsonEncode( [ 'parsoid' => $this->parsoid ?? [], 'mw' => $this->mw ?? [] ] ); } }