vendor/wikimedia/remex-html/RemexHtml/TreeBuilder/InForeignContent.php

   1 <?php
   2
   3 namespace RemexHtml\TreeBuilder;
   4 use RemexHtml\HTMLData;
   5 use RemexHtml\Tokenizer\Attributes;
   6
   7 /**
   8  * The rules for parsing tokens in foreign content.
   9  *
  10  * This is not referred to as an insertion mode in the spec, but is
  11  * sufficiently similar to one that we can inherit from InsertionMode here.
  12  */
  13 class InForeignContent extends InsertionMode {
  14         /**
  15          * The list of tag names which unconditionally generate a parse error when
  16          * seen in foreign content.
  17          * @var bool[string]
  18          */
  19         private static $notAllowed = [
  20                 'b' => true,
  21                 'big' => true,
  22                 'blockquote' => true,
  23                 'body' => true,
  24                 'br' => true,
  25                 'center' => true,
  26                 'code' => true,
  27                 'dd' => true,
  28                 'div' => true,
  29                 'dl' => true,
  30                 'dt' => true,
  31                 'em' => true,
  32                 'embed' => true,
  33                 'h1' => true,
  34                 'h2' => true,
  35                 'h3' => true,
  36                 'h4' => true,
  37                 'h5' => true,
  38                 'h6' => true,
  39                 'head' => true,
  40                 'hr' => true,
  41                 'i' => true,
  42                 'img' => true,
  43                 'li' => true,
  44                 'listing' => true,
  45                 'menu' => true,
  46                 'meta' => true,
  47                 'nobr' => true,
  48                 'ol' => true,
  49                 'p' => true,
  50                 'pre' => true,
  51                 'ruby' => true,
  52                 's' => true,
  53                 'small' => true,
  54                 'span' => true,
  55                 'strong' => true,
  56                 'strike' => true,
  57                 'sub' => true,
  58                 'sup' => true,
  59                 'table' => true,
  60                 'tt' => true,
  61                 'u' => true,
  62                 'ul' => true,
  63                 'var' => true,
  64         ];
  65
  66         /**
  67          * The table for correcting the tag names of SVG elements, given in the
  68          * "Any other start tag" section of the spec.
  69          */
  70         private static $svgElementCase = [
  71                 'altglyph' => 'altGlyph',
  72                 'altglyphdef' => 'altGlyphDef',
  73                 'altglyphitem' => 'altGlyphItem',
  74                 'animatecolor' => 'animateColor',
  75                 'animatemotion' => 'animateMotion',
  76                 'animatetransform' => 'animateTransform',
  77                 'clippath' => 'clipPath',
  78                 'feblend' => 'feBlend',
  79                 'fecolormatrix' => 'feColorMatrix',
  80                 'fecomponenttransfer' => 'feComponentTransfer',
  81                 'fecomposite' => 'feComposite',
  82                 'feconvolvematrix' => 'feConvolveMatrix',
  83                 'fediffuselighting' => 'feDiffuseLighting',
  84                 'fedisplacementmap' => 'feDisplacementMap',
  85                 'fedistantlight' => 'feDistantLight',
  86                 'fedropshadow' => 'feDropShadow',
  87                 'feflood' => 'feFlood',
  88                 'fefunca' => 'feFuncA',
  89                 'fefuncb' => 'feFuncB',
  90                 'fefuncg' => 'feFuncG',
  91                 'fefuncr' => 'feFuncR',
  92                 'fegaussianblur' => 'feGaussianBlur',
  93                 'feimage' => 'feImage',
  94                 'femerge' => 'feMerge',
  95                 'femergenode' => 'feMergeNode',
  96                 'femorphology' => 'feMorphology',
  97                 'feoffset' => 'feOffset',
  98                 'fepointlight' => 'fePointLight',
  99                 'fespecularlighting' => 'feSpecularLighting',
 100                 'fespotlight' => 'feSpotLight',
 101                 'fetile' => 'feTile',
 102                 'feturbulence' => 'feTurbulence',
 103                 'foreignobject' => 'foreignObject',
 104                 'glyphref' => 'glyphRef',
 105                 'lineargradient' => 'linearGradient',
 106                 'radialgradient' => 'radialGradient',
 107                 'textpath' => 'textPath',
 108         ];
 109
 110         public function characters( $text, $start, $length, $sourceStart, $sourceLength ) {
 111                 $builder = $this->builder;
 112
 113                 while ( $length ) {
 114                         $normalLength = strcspn( $text, "\0\t\n\f\r ", $start, $length );
 115                         if ( $normalLength ) {
 116                                 $builder->framesetOK = false;
 117                                 $builder->insertCharacters( $text, $start, $normalLength,
 118                                         $sourceStart, $sourceLength );
 119                         }
 120                         $start += $normalLength;
 121                         $length -= $normalLength;
 122                         $sourceStart += $normalLength;
 123                         $sourceLength -= $normalLength;
 124                         if ( !$length ) {
 125                                 break;
 126                         }
 127
 128                         $char = $text[$start];
 129                         if ( $char === "\0" ) {
 130                                 $builder->error( "replaced null character", $sourceStart );
 131                                 $builder->insertCharacters( "\xef\xbf\xbd", 0, 3, $sourceStart, $sourceLength );
 132                                 $start++;
 133                                 $length--;
 134                                 $sourceStart++;
 135                                 $sourceLength--;
 136                         } else {
 137                                 // Whitespace
 138                                 $wsLength = strspn( $text, "\t\n\f\r ", $start, $length );
 139                                 $builder->insertCharacters( $text, $start, $wsLength, $sourceStart, $wsLength );
 140                                 $start += $wsLength;
 141                                 $length -= $wsLength;
 142                                 $sourceStart += $wsLength;
 143                                 $sourceLength -= $wsLength;
 144                         }
 145                 }
 146         }
 147
 148         private function isIntegrationPoint( Element $element ) {
 149                 return $element->namespace === HTMLData::NS_HTML
 150                         || $element->isMathmlTextIntegration()
 151                         || $element->isHtmlIntegration();
 152         }
 153
 154         public function startTag( $name, Attributes $attrs, $selfClose, $sourceStart, $sourceLength ) {
 155                 $builder = $this->builder;
 156                 $stack = $builder->stack;
 157                 $dispatcher = $this->dispatcher;
 158
 159                 if ( isset( self::$notAllowed[$name] ) ) {
 160                         $allowed = false;
 161                 } elseif ( $name === 'font' && (
 162                         isset( $attrs['color'] ) || isset( $attrs['face'] ) || isset( $attrs['size'] ) )
 163                 ) {
 164                         $allowed = false;
 165                 } else {
 166                         $allowed = true;
 167                 }
 168
 169                 if ( !$allowed ) {
 170                         $builder->error( "unexpected <$name> tag in foreign content", $sourceStart );
 171                         if ( !$builder->isFragment ) {
 172                                 do {
 173                                         $builder->pop( $sourceStart, 0 );
 174                                 } while ( $stack->current && !$this->isIntegrationPoint( $stack->current ) );
 175                                 $dispatcher->startTag( $name, $attrs, $selfClose, $sourceStart, $sourceLength );
 176                                 return;
 177                         }
 178                 }
 179
 180                 $acnNs = $builder->adjustedCurrentNode()->namespace;
 181                 if ( $acnNs === HTMLData::NS_MATHML ) {
 182                         $attrs = new ForeignAttributes( $attrs, 'math' );
 183                 } elseif ( $acnNs === HTMLData::NS_SVG ) {
 184                         $attrs = new ForeignAttributes( $attrs, 'svg' );
 185                         if ( isset( self::$svgElementCase[$name] ) ) {
 186                                 $name = self::$svgElementCase[$name];
 187                         }
 188                 } else {
 189                         $attrs = new ForeignAttributes( $attrs, 'other' );
 190                 }
 191                 $dispatcher->ack = true;
 192                 $builder->insertForeign( $acnNs, $name, $attrs, $selfClose, $sourceStart, $sourceLength );
 193         }
 194
 195         public function endTag( $name, $sourceStart, $sourceLength ) {
 196                 $builder = $this->builder;
 197                 $stack = $builder->stack;
 198                 $dispatcher = $this->dispatcher;
 199
 200                 $node = $stack->current;
 201                 if ( strcasecmp( $node->name, $name ) !== 0 ) {
 202                         $builder->error( "mismatched end tag in foreign content", $sourceStart );
 203                 }
 204                 for ( $idx = $stack->length() - 1; $idx > 0; $idx-- ) {
 205                         if ( strcasecmp( $node->name, $name ) === 0 ) {
 206                                 $builder->popAllUpToElement( $node, $sourceStart, $sourceLength );
 207                                 break;
 208                         }
 209                         $node = $stack->item( $idx - 1 );
 210                         if ( $node->namespace === HTMLData::NS_HTML ) {
 211                                 $dispatcher->getHandler()->endTag( $name, $sourceStart, $sourceLength );
 212                                 break;
 213                         }
 214                 }
 215         }
 216
 217         public function endDocument( $pos ) {
 218                 throw new TreeBuilderError( "unspecified, presumed unreachable" );
 219         }
 220 }