Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 41 |
|
0.00% |
0 / 5 |
CRAP | |
0.00% |
0 / 1 |
| RemexRemoveTagHandler | |
0.00% |
0 / 41 |
|
0.00% |
0 / 5 |
600 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
2 | |||
| comment | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
12 | |||
| validateTag | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
72 | |||
| startTag | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
110 | |||
| endTag | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Parser; |
| 4 | |
| 5 | use Wikimedia\RemexHtml\Tokenizer\Attributes; |
| 6 | use Wikimedia\RemexHtml\Tokenizer\PlainAttributes; |
| 7 | use Wikimedia\RemexHtml\Tokenizer\RelayTokenHandler; |
| 8 | use Wikimedia\RemexHtml\Tokenizer\TokenHandler; |
| 9 | |
| 10 | /** |
| 11 | * Helper class for Sanitizer::removeSomeTags(). |
| 12 | * @internal |
| 13 | */ |
| 14 | class RemexRemoveTagHandler extends RelayTokenHandler { |
| 15 | /** |
| 16 | * @var string The original HTML source string (used for fallback text |
| 17 | * when rejecting an HTML tag). |
| 18 | */ |
| 19 | private $source; |
| 20 | |
| 21 | /** |
| 22 | * @var array<string,true> Set of HTML tags which can be self-closed. |
| 23 | */ |
| 24 | private $htmlsingle; |
| 25 | |
| 26 | /** |
| 27 | * @var array<string,true> Self-closed tags which are on $htmlsingle |
| 28 | * but not on $htmlsingleonly will be emitted as an empty element. |
| 29 | */ |
| 30 | private $htmlsingleonly; |
| 31 | |
| 32 | /** |
| 33 | * @var array<string,true> Set of allowed HTML open/close tags. |
| 34 | */ |
| 35 | private $htmlelements; |
| 36 | |
| 37 | /** @var ?string */ |
| 38 | private $commentRegex; |
| 39 | |
| 40 | /** |
| 41 | * @var ?callable(Attributes,mixed...):Attributes Callback to mutate or |
| 42 | * sanitize attributes. |
| 43 | */ |
| 44 | private $attrCallback; |
| 45 | |
| 46 | /** |
| 47 | * @var ?array $args Optional extra arguments to provide to the |
| 48 | * $attrCallback. |
| 49 | */ |
| 50 | private $callbackArgs; |
| 51 | |
| 52 | /** |
| 53 | * @param TokenHandler $nextHandler Handler to relay accepted tokens. |
| 54 | * @param string $source Input source string. |
| 55 | * @param array $tagData Information about allowed/rejected tags. |
| 56 | * @param ?callable $attrCallback Attribute handler callback. |
| 57 | * The full signature is ?callable(Attributes,mixed...):Attributes |
| 58 | * @param ?array $callbackArgs Optional arguments to attribute handler. |
| 59 | * @param array $options Associative array of options: |
| 60 | * - commentRegex: If present, allow comments with inner text matching |
| 61 | * the specified regular expression. |
| 62 | */ |
| 63 | public function __construct( |
| 64 | TokenHandler $nextHandler, |
| 65 | string $source, |
| 66 | array $tagData, |
| 67 | ?callable $attrCallback, |
| 68 | ?array $callbackArgs, |
| 69 | array $options |
| 70 | ) { |
| 71 | parent::__construct( $nextHandler ); |
| 72 | $this->source = $source; |
| 73 | $this->htmlsingle = $tagData['htmlsingle']; |
| 74 | $this->htmlsingleonly = $tagData['htmlsingleonly']; |
| 75 | $this->htmlelements = $tagData['htmlelements']; |
| 76 | $this->attrCallback = $attrCallback; |
| 77 | $this->callbackArgs = $callbackArgs ?? []; |
| 78 | $this->commentRegex = $options['commentRegex'] ?? null; |
| 79 | } |
| 80 | |
| 81 | /** |
| 82 | * @inheritDoc |
| 83 | */ |
| 84 | public function comment( $text, $sourceStart, $sourceLength ) { |
| 85 | if ( $this->commentRegex !== null && preg_match( $this->commentRegex, $text ) ) { |
| 86 | $this->nextHandler->comment( $text, $sourceStart, $sourceLength ); |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * Takes attribute names and values for a tag and the tag name and |
| 92 | * validates that the tag is allowed to be present. |
| 93 | * This DOES NOT validate the attributes, nor does it validate the |
| 94 | * tags themselves. This method only handles the special circumstances |
| 95 | * where we may want to allow a tag within content but ONLY when it has |
| 96 | * specific attributes set. |
| 97 | * |
| 98 | * @param string $element |
| 99 | * @param Attributes $attrs |
| 100 | * @return bool |
| 101 | * |
| 102 | * @see Sanitizer::validateTag() |
| 103 | */ |
| 104 | private static function validateTag( string $element, Attributes $attrs ): bool { |
| 105 | if ( $element == 'meta' || $element == 'link' ) { |
| 106 | $params = $attrs->getValues(); |
| 107 | if ( !isset( $params['itemprop'] ) ) { |
| 108 | // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content |
| 109 | return false; |
| 110 | } |
| 111 | if ( $element == 'meta' && !isset( $params['content'] ) ) { |
| 112 | // <meta> must have a content="" for the itemprop |
| 113 | return false; |
| 114 | } |
| 115 | if ( $element == 'link' && !isset( $params['href'] ) ) { |
| 116 | // <link> must have an associated href="" |
| 117 | return false; |
| 118 | } |
| 119 | } |
| 120 | |
| 121 | return true; |
| 122 | } |
| 123 | |
| 124 | /** |
| 125 | * @inheritDoc |
| 126 | */ |
| 127 | public function startTag( $name, Attributes $attrs, $selfClose, $sourceStart, $sourceLength ) { |
| 128 | // Handle a start tag from the tokenizer: either relay it to the |
| 129 | // next stage, or re-emit it as raw text. |
| 130 | |
| 131 | $badtag = false; |
| 132 | $t = strtolower( $name ); |
| 133 | if ( isset( $this->htmlelements[$t] ) ) { |
| 134 | if ( $this->attrCallback ) { |
| 135 | $attrs = ( $this->attrCallback )( $attrs, ...$this->callbackArgs ); |
| 136 | } |
| 137 | if ( $selfClose && !( isset( $this->htmlsingle[$t] ) || isset( $this->htmlsingleonly[$t] ) ) ) { |
| 138 | // Remove the self-closing slash, to be consistent with |
| 139 | // HTML5 semantics. T134423 |
| 140 | $selfClose = false; |
| 141 | } |
| 142 | if ( !self::validateTag( $t, $attrs ) ) { |
| 143 | $badtag = true; |
| 144 | } |
| 145 | $fixedAttrs = Sanitizer::validateTagAttributes( $attrs->getValues(), $t ); |
| 146 | $attrs = new PlainAttributes( $fixedAttrs ); |
| 147 | if ( !$badtag ) { |
| 148 | if ( $selfClose && !isset( $this->htmlsingleonly[$t] ) ) { |
| 149 | // Interpret self-closing tags as empty tags even when |
| 150 | // HTML5 would interpret them as start tags. Such input |
| 151 | // is commonly seen on Wikimedia wikis with this intention. |
| 152 | $this->nextHandler->startTag( $name, $attrs, false, $sourceStart, $sourceLength ); |
| 153 | $this->nextHandler->endTag( $name, $sourceStart + $sourceLength, 0 ); |
| 154 | } else { |
| 155 | $this->nextHandler->startTag( $name, $attrs, $selfClose, $sourceStart, $sourceLength ); |
| 156 | } |
| 157 | return; |
| 158 | } |
| 159 | } |
| 160 | // Emit this as a text node instead. |
| 161 | $this->nextHandler->characters( $this->source, $sourceStart, $sourceLength, $sourceStart, $sourceLength ); |
| 162 | } |
| 163 | |
| 164 | /** |
| 165 | * @inheritDoc |
| 166 | */ |
| 167 | public function endTag( $name, $sourceStart, $sourceLength ) { |
| 168 | // Handle an end tag from the tokenizer: either relay it to the |
| 169 | // next stage, or re-emit it as raw text. |
| 170 | |
| 171 | $t = strtolower( $name ); |
| 172 | if ( isset( $this->htmlelements[$t] ) ) { |
| 173 | // This is a good tag, relay it. |
| 174 | $this->nextHandler->endTag( $name, $sourceStart, $sourceLength ); |
| 175 | } else { |
| 176 | // Emit this as a text node instead. |
| 177 | $this->nextHandler->characters( $this->source, $sourceStart, $sourceLength, $sourceStart, $sourceLength ); |
| 178 | } |
| 179 | } |
| 180 | |
| 181 | } |