includes/tidy/RemexDriver.php

   1 <?php
   2
   3 namespace MediaWiki\Tidy;
   4
   5 use MediaWiki\Config\ServiceOptions;
   6 use MediaWiki\MainConfigNames;
   7 use Wikimedia\RemexHtml\HTMLData;
   8 use Wikimedia\RemexHtml\Serializer\Serializer;
   9 use Wikimedia\RemexHtml\Serializer\SerializerWithTracer;
  10 use Wikimedia\RemexHtml\Tokenizer\Tokenizer;
  11 use Wikimedia\RemexHtml\TreeBuilder\Dispatcher;
  12 use Wikimedia\RemexHtml\TreeBuilder\TreeBuilder;
  13 use Wikimedia\RemexHtml\TreeBuilder\TreeMutationTracer;
  14
  15 class RemexDriver extends TidyDriverBase {
  16         /** @var bool */
  17         private $treeMutationTrace;
  18         /** @var bool */
  19         private $serializerTrace;
  20         /** @var bool */
  21         private $mungerTrace;
  22         /** @var bool */
  23         private $pwrap;
  24         /** @var bool */
  25         private $enableLegacyMediaDOM;
  26
  27         /** @internal */
  28         public const CONSTRUCTOR_OPTIONS = [
  29                 MainConfigNames::TidyConfig,
  30                 MainConfigNames::ParserEnableLegacyMediaDOM,
  31         ];
  32
  33         /**
  34          * @param ServiceOptions $options
  35          */
  36         public function __construct( ServiceOptions $options ) {
  37                 $options->assertRequiredOptions( self::CONSTRUCTOR_OPTIONS );
  38                 $config = $options->get( MainConfigNames::TidyConfig );
  39                 $this->enableLegacyMediaDOM = $options->get( MainConfigNames::ParserEnableLegacyMediaDOM );
  40                 $config += [
  41                         'treeMutationTrace' => false,
  42                         'serializerTrace' => false,
  43                         'mungerTrace' => false,
  44                         'pwrap' => true
  45                 ];
  46                 $this->treeMutationTrace = $config['treeMutationTrace'];
  47                 $this->serializerTrace = $config['serializerTrace'];
  48                 $this->mungerTrace = $config['mungerTrace'];
  49                 $this->pwrap = $config['pwrap'];
  50                 parent::__construct( $config );
  51         }
  52
  53         /** @inheritDoc */
  54         public function tidy( $text, ?callable $textProcessor = null ) {
  55                 $traceCallback = static function ( $msg ) {
  56                         wfDebug( "RemexHtml: $msg" );
  57                 };
  58                 $formatter = new RemexCompatFormatter( [ 'textProcessor' => $textProcessor ] );
  59                 if ( $this->serializerTrace ) {
  60                         $serializer = new SerializerWithTracer( $formatter, null, $traceCallback );
  61                 } else {
  62                         $serializer = new Serializer( $formatter );
  63                 }
  64                 if ( $this->pwrap ) {
  65                         $munger = new RemexCompatMunger( $serializer, $this->mungerTrace );
  66                 } else {
  67                         $munger = $serializer;
  68                 }
  69                 if ( $this->treeMutationTrace ) {
  70                         $tracer = new TreeMutationTracer( $munger, $traceCallback );
  71                 } else {
  72                         $tracer = $munger;
  73                 }
  74                 $treeBuilderClass = $this->enableLegacyMediaDOM ? TreeBuilder::class : RemexCompatBuilder::class;
  75                 $treeBuilder = new $treeBuilderClass( $tracer, [
  76                         'ignoreErrors' => true,
  77                         'ignoreNulls' => true,
  78                 ] );
  79                 $dispatcher = new Dispatcher( $treeBuilder );
  80                 $tokenizer = new Tokenizer( $dispatcher, $text, [
  81                         'ignoreErrors' => true,
  82                         'ignoreCharRefs' => true,
  83                         'ignoreNulls' => true,
  84                         'skipPreprocess' => true,
  85                 ] );
  86
  87                 $tokenizer->execute( [
  88                         'fragmentNamespace' => HTMLData::NS_HTML,
  89                         'fragmentName' => 'body'
  90                 ] );
  91                 return $serializer->getResult();
  92         }
  93 }