includes/tidy/RemexCompatFormatter.php

   1 <?php
   2
   3 namespace MediaWiki\Tidy;
   4
   5 use RemexHtml\HTMLData;
   6 use RemexHtml\Serializer\HtmlFormatter;
   7 use RemexHtml\Serializer\SerializerNode;
   8 use RemexHtml\Tokenizer\PlainAttributes;
   9
  10 /**
  11  * @internal
  12  */
  13 class RemexCompatFormatter extends HtmlFormatter {
  14         private static $markedEmptyElements = [
  15                 'li' => true,
  16                 'p' => true,
  17                 'tr' => true,
  18         ];
  19
  20         public function __construct( $options = [] ) {
  21                 parent::__construct( $options );
  22                 $this->attributeEscapes["\xc2\xa0"] = '&#160;';
  23                 unset( $this->attributeEscapes["&"] );
  24                 $this->textEscapes["\xc2\xa0"] = '&#160;';
  25                 unset( $this->textEscapes["&"] );
  26         }
  27
  28         public function startDocument( $fragmentNamespace, $fragmentName ) {
  29                 return '';
  30         }
  31
  32         public function element( SerializerNode $parent, SerializerNode $node, $contents ) {
  33                 $data = $node->snData;
  34                 if ( $data && $data->isPWrapper ) {
  35                         if ( $data->nonblankNodeCount ) {
  36                                 return "<p>$contents</p>";
  37                         } else {
  38                                 return $contents;
  39                         }
  40                 }
  41
  42                 $name = $node->name;
  43                 $attrs = $node->attrs;
  44                 if ( isset( self::$markedEmptyElements[$name] ) && $attrs->count() === 0 ) {
  45                         if ( strspn( $contents, "\t\n\f\r " ) === strlen( $contents ) ) {
  46                                 return "<{$name} class=\"mw-empty-elt\">$contents</{$name}>";
  47                         }
  48                 }
  49
  50                 $s = "<$name";
  51                 foreach ( $attrs->getValues() as $attrName => $attrValue ) {
  52                         $encValue = strtr( $attrValue, $this->attributeEscapes );
  53                         $s .= " $attrName=\"$encValue\"";
  54                 }
  55                 if ( $node->namespace === HTMLData::NS_HTML && isset( $this->voidElements[$name] ) ) {
  56                         $s .= ' />';
  57                         return $s;
  58                 }
  59
  60                 $s .= '>';
  61                 if ( $node->namespace === HTMLData::NS_HTML
  62                         && isset( $contents[0] ) && $contents[0] === "\n"
  63                         && isset( $this->prefixLfElements[$name] )
  64                 ) {
  65                         $s .= "\n$contents</$name>";
  66                 } else {
  67                         $s .= "$contents</$name>";
  68                 }
  69                 return $s;
  70         }
  71 }