includes/parser/RemexStripTagHandler.php

   1 <?php
   2
   3 use RemexHtml\Tokenizer\Attributes;
   4 use RemexHtml\Tokenizer\TokenHandler;
   5 use RemexHtml\Tokenizer\Tokenizer;
   6
   7 /**
   8  * @internal
   9  */
  10 class RemexStripTagHandler implements TokenHandler {
  11         private $text = '';
  12
  13         public function getResult() {
  14                 return $this->text;
  15         }
  16
  17         function startDocument( Tokenizer $t, $fns, $fn ) {
  18                 // Do nothing.
  19         }
  20
  21         function endDocument( $pos ) {
  22                 // Do nothing.
  23         }
  24
  25         function error( $text, $pos ) {
  26                 // Do nothing.
  27         }
  28
  29         function characters( $text, $start, $length, $sourceStart, $sourceLength ) {
  30                 $this->text .= substr( $text, $start, $length );
  31         }
  32
  33         function startTag( $name, Attributes $attrs, $selfClose, $sourceStart, $sourceLength ) {
  34                 // Inject whitespace for typical block-level tags to
  35                 // prevent merging unrelated<br>words.
  36                 if ( $this->isBlockLevelTag( $name ) ) {
  37                         $this->text .= ' ';
  38                 }
  39         }
  40
  41         function endTag( $name, $sourceStart, $sourceLength ) {
  42                 // Inject whitespace for typical block-level tags to
  43                 // prevent merging unrelated<br>words.
  44                 if ( $this->isBlockLevelTag( $name ) ) {
  45                         $this->text .= ' ';
  46                 }
  47         }
  48
  49         function doctype( $name, $public, $system, $quirks, $sourceStart, $sourceLength ) {
  50                 // Do nothing.
  51         }
  52
  53         function comment( $text, $sourceStart, $sourceLength ) {
  54                 // Do nothing.
  55         }
  56
  57         // Per https://developer.mozilla.org/en-US/docs/Web/HTML/Block-level_elements
  58         // retrieved on sept 12, 2018. <br> is not block level but was added anyways.
  59         // The following is a complete list of all HTML block level elements
  60         // (although "block-level" is not technically defined for elements that are
  61         // new in HTML5).
  62         // Structured as tag => true to allow O(1) membership test.
  63         private static $BLOCK_LEVEL_TAGS = [
  64                 'address' => true,
  65                 'article' => true,
  66                 'aside' => true,
  67                 'blockquote' => true,
  68                 'br' => true,
  69                 'canvas' => true,
  70                 'dd' => true,
  71                 'div' => true,
  72                 'dl' => true,
  73                 'dt' => true,
  74                 'fieldset' => true,
  75                 'figcaption' => true,
  76                 'figure' => true,
  77                 'footer' => true,
  78                 'form' => true,
  79                 'h1' => true,
  80                 'h2' => true,
  81                 'h3' => true,
  82                 'h4' => true,
  83                 'h5' => true,
  84                 'h6' => true,
  85                 'header' => true,
  86                 'hgroup' => true,
  87                 'hr' => true,
  88                 'li' => true,
  89                 'main' => true,
  90                 'nav' => true,
  91                 'noscript' => true,
  92                 'ol' => true,
  93                 'output' => true,
  94                 'p' => true,
  95                 'pre' => true,
  96                 'section' => true,
  97                 'table' => true,
  98                 'td' => true,
  99                 'tfoot' => true,
 100                 'th' => true,
 101                 'tr' => true,
 102                 'ul' => true,
 103                 'video' => true,
 104         ];
 105
 106         /**
 107          * Detect block level tags. Of course css can make anything a block
 108          * level tag, but this is still better than nothing.
 109          *
 110          * @param string $tagName HTML tag name
 111          * @return bool True when tag is an html block level element
 112          */
 113         private function isBlockLevelTag( $tagName ) {
 114                 $key = strtolower( trim( $tagName ) );
 115                 return isset( self::$BLOCK_LEVEL_TAGS[$key] );
 116         }
 117 }