X-Git-Url: https://scripts.mit.edu/gitweb/autoinstallsdev/mediawiki.git/blobdiff_plain/19e297c21b10b1b8a3acad5e73fc71dcb35db44a..6932310fd58ebef145fa01eb76edf7150284d8ea:/includes/tidy/RemexDriver.php

diff --git a/includes/tidy/RemexDriver.php b/includes/tidy/RemexDriver.php
new file mode 100644
index 00000000..e02af88f
--- /dev/null
+++ b/includes/tidy/RemexDriver.php
@@ -0,0 +1,57 @@
+<?php
+
+namespace MediaWiki\Tidy;
+
+use RemexHtml\Serializer\Serializer;
+use RemexHtml\Tokenizer\Tokenizer;
+use RemexHtml\TreeBuilder\Dispatcher;
+use RemexHtml\TreeBuilder\TreeBuilder;
+use RemexHtml\TreeBuilder\TreeMutationTracer;
+
+class RemexDriver extends TidyDriverBase {
+	private $trace;
+	private $pwrap;
+
+	public function __construct( array $config ) {
+		$config += [
+			'treeMutationTrace' => false,
+			'pwrap' => true
+		];
+		$this->trace = $config['treeMutationTrace'];
+		$this->pwrap = $config['pwrap'];
+		parent::__construct( $config );
+	}
+
+	public function tidy( $text ) {
+		$formatter = new RemexCompatFormatter;
+		$serializer = new Serializer( $formatter );
+		if ( $this->pwrap ) {
+			$munger = new RemexCompatMunger( $serializer );
+		} else {
+			$munger = $serializer;
+		}
+		if ( $this->trace ) {
+			$tracer = new TreeMutationTracer( $munger, function ( $msg ) {
+				wfDebug( "RemexHtml: $msg" );
+			} );
+		} else {
+			$tracer = $munger;
+		}
+		$treeBuilder = new TreeBuilder( $tracer, [
+			'ignoreErrors' => true,
+			'ignoreNulls' => true,
+		] );
+		$dispatcher = new Dispatcher( $treeBuilder );
+		$tokenizer = new Tokenizer( $dispatcher, $text, [
+			'ignoreErrors' => true,
+			'ignoreCharRefs' => true,
+			'ignoreNulls' => true,
+			'skipPreprocess' => true,
+		] );
+		$tokenizer->execute( [
+			'fragmentNamespace' => \RemexHtml\HTMLData::NS_HTML,
+			'fragmentName' => 'body'
+		] );
+		return $serializer->getResult();
+	}
+}