ContentDOMTransformStage.php

Go to the documentation of this file.

<?php
 
namespace MediaWiki\OutputTransform;
 
use MediaWiki\Parser\ParserOutput;
use MediaWiki\Parser\Parsoid\PageBundleParserOutputConverter;
use ParserOptions;
use stdClass;
use Wikimedia\Parsoid\Core\PageBundle;
use Wikimedia\Parsoid\DOM\Document;
use Wikimedia\Parsoid\Mocks\MockEnv;
use Wikimedia\Parsoid\Utils\ContentUtils;
use Wikimedia\Parsoid\Utils\DOMCompat;
use Wikimedia\Parsoid\Utils\DOMDataUtils;
use Wikimedia\Parsoid\Utils\DOMUtils;
 
abstract class ContentDOMTransformStage implements OutputTransformStage {
 
    public function transform( ParserOutput $po, ?ParserOptions $popts, array &$options ): ParserOutput {
        // TODO will use HTMLHolder in the future
        $doc = null;
        $hasPageBundle = PageBundleParserOutputConverter::hasPageBundle( $po );
        if ( $hasPageBundle ) {
            $pb = PageBundleParserOutputConverter::pageBundleFromParserOutput( $po );
            $doc = DOMUtils::parseHTML( $po->getContentHolderText() );
            PageBundle::apply( $doc, $pb );
            DOMDataUtils::prepareDoc( $doc );
            DOMDataUtils::visitAndLoadDataAttribs(
                DOMCompat::getBody( $doc )
            );
        } else {
            $doc = ContentUtils::createAndLoadDocument(
                $po->getContentHolderText(),
            );
        }
 
        $doc = $this->transformDOM( $doc, $po, $popts, $options );
 
        // TODO will use HTMLHolder in the future
        if ( $hasPageBundle ) {
            DOMDataUtils::visitAndStoreDataAttribs(
                DOMCompat::getBody( $doc ),
                [
                    'storeInPageBundle' => true,
                    'env' => new MockEnv( [] ),
                ]
            );
            $pb = DOMDataUtils::getPageBundle( $doc );
            $pb = self::workaroundT365036( $doc, $pb );
 
            PageBundleParserOutputConverter::applyPageBundleDataToParserOutput( $pb, $po );
            $text = ContentUtils::toXML( DOMCompat::getBody( $doc ), [
                'innerXML' => true,
            ] );
        } else {
            $text = ContentUtils::ppToXML( DOMCompat::getBody( $doc ), [
                'innerXML' => true,
            ] );
        }
        $po->setContentHolderText( $text );
        return $po;
    }
 
    private function workaroundT365036( Document $doc, PageBundle $pb ): stdClass {
        DOMDataUtils::injectPageBundle( $doc, $pb );
        $convertedPageBundle = DOMDataUtils::extractPageBundle( $doc );
        // Tell phan that $convertedPageBundle is non-null since $pb was non-null
        '@phan-var stdClass $convertedPageBundle';
        return $convertedPageBundle;
    }
 
    abstract public function transformDOM(
        Document $dom, ParserOutput $po, ?ParserOptions $popts, array &$options
    ): Document;
 
}