includes/tidy/RemexDriver.php

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