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

diff --git a/includes/tidy/RemexCompatFormatter.php b/includes/tidy/RemexCompatFormatter.php
new file mode 100644
index 00000000..c8a715b1
--- /dev/null
+++ b/includes/tidy/RemexCompatFormatter.php
@@ -0,0 +1,70 @@
+<?php
+
+namespace MediaWiki\Tidy;
+
+use RemexHtml\HTMLData;
+use RemexHtml\Serializer\HtmlFormatter;
+use RemexHtml\Serializer\SerializerNode;
+
+/**
+ * @internal
+ */
+class RemexCompatFormatter extends HtmlFormatter {
+	private static $markedEmptyElements = [
+		'li' => true,
+		'p' => true,
+		'tr' => true,
+	];
+
+	public function __construct( $options = [] ) {
+		parent::__construct( $options );
+		$this->attributeEscapes["\xc2\xa0"] = '&#160;';
+		unset( $this->attributeEscapes["&"] );
+		$this->textEscapes["\xc2\xa0"] = '&#160;';
+		unset( $this->textEscapes["&"] );
+	}
+
+	public function startDocument( $fragmentNamespace, $fragmentName ) {
+		return '';
+	}
+
+	public function element( SerializerNode $parent, SerializerNode $node, $contents ) {
+		$data = $node->snData;
+		if ( $data && $data->isPWrapper ) {
+			if ( $data->nonblankNodeCount ) {
+				return "<p>$contents</p>";
+			} else {
+				return $contents;
+			}
+		}
+
+		$name = $node->name;
+		$attrs = $node->attrs;
+		if ( isset( self::$markedEmptyElements[$name] ) && $attrs->count() === 0 ) {
+			if ( strspn( $contents, "\t\n\f\r " ) === strlen( $contents ) ) {
+				return "<{$name} class=\"mw-empty-elt\">$contents</{$name}>";
+			}
+		}
+
+		$s = "<$name";
+		foreach ( $attrs->getValues() as $attrName => $attrValue ) {
+			$encValue = strtr( $attrValue, $this->attributeEscapes );
+			$s .= " $attrName=\"$encValue\"";
+		}
+		if ( $node->namespace === HTMLData::NS_HTML && isset( $this->voidElements[$name] ) ) {
+			$s .= ' />';
+			return $s;
+		}
+
+		$s .= '>';
+		if ( $node->namespace === HTMLData::NS_HTML
+			&& isset( $contents[0] ) && $contents[0] === "\n"
+			&& isset( $this->prefixLfElements[$name] )
+		) {
+			$s .= "\n$contents</$name>";
+		} else {
+			$s .= "$contents</$name>";
+		}
+		return $s;
+	}
+}