includes/tidy/RemexDriver.php

   1 <?php
   2
   3 namespace MediaWiki\Tidy;
   4
   5 use RemexHtml\Serializer\Serializer;
   6 use RemexHtml\Serializer\SerializerWithTracer;
   7 use RemexHtml\Tokenizer\Tokenizer;
   8 use RemexHtml\TreeBuilder\Dispatcher;
   9 use RemexHtml\TreeBuilder\TreeBuilder;
  10 use RemexHtml\TreeBuilder\TreeMutationTracer;
  11
  12 class RemexDriver extends TidyDriverBase {
  13         private $treeMutationTrace;
  14         private $serializerTrace;
  15         private $mungerTrace;
  16         private $pwrap;
  17
  18         public function __construct( array $config ) {
  19                 $config += [
  20                         'treeMutationTrace' => false,
  21                         'serializerTrace' => false,
  22                         'mungerTrace' => false,
  23                         'pwrap' => true
  24                 ];
  25                 $this->treeMutationTrace = $config['treeMutationTrace'];
  26                 $this->serializerTrace = $config['serializerTrace'];
  27                 $this->mungerTrace = $config['mungerTrace'];
  28                 $this->pwrap = $config['pwrap'];
  29                 parent::__construct( $config );
  30         }
  31
  32         public function tidy( $text ) {
  33                 $traceCallback = function ( $msg ) {
  34                         wfDebug( "RemexHtml: $msg" );
  35                 };
  36
  37                 $formatter = new RemexCompatFormatter;
  38                 if ( $this->serializerTrace ) {
  39                         $serializer = new SerializerWithTracer( $formatter, null, $traceCallback );
  40                 } else {
  41                         $serializer = new Serializer( $formatter );
  42                 }
  43                 if ( $this->pwrap ) {
  44                         $munger = new RemexCompatMunger( $serializer, $this->mungerTrace );
  45                 } else {
  46                         $munger = $serializer;
  47                 }
  48                 if ( $this->treeMutationTrace ) {
  49                         $tracer = new TreeMutationTracer( $munger, $traceCallback );
  50                 } else {
  51                         $tracer = $munger;
  52                 }
  53                 $treeBuilder = new TreeBuilder( $tracer, [
  54                         'ignoreErrors' => true,
  55                         'ignoreNulls' => true,
  56                 ] );
  57                 $dispatcher = new Dispatcher( $treeBuilder );
  58                 $tokenizer = new Tokenizer( $dispatcher, $text, [
  59                         'ignoreErrors' => true,
  60                         'ignoreCharRefs' => true,
  61                         'ignoreNulls' => true,
  62                         'skipPreprocess' => true,
  63                 ] );
  64
  65                 $tokenizer->execute( [
  66                         'fragmentNamespace' => \RemexHtml\HTMLData::NS_HTML,
  67                         'fragmentName' => 'body'
  68                 ] );
  69                 return $serializer->getResult();
  70         }
  71 }