[^:;]+):(?(?:[^;,\(]+(?:\([^\)]+\))?,?)+))+?/'; /** * Regular expression pattern to match individual actions within an event. * * @since 1.4.0 * @var string */ const AMP_ACTION_REGEX_PATTERN = '/(?[^(),\s]+(?:\([^\)]+\))?)+/'; /** * HTML elements that are self-closing. * * Not all are valid AMP, but we include them for completeness. * * @since 0.7 * @link https://www.w3.org/TR/html5/syntax.html#serializing-html-fragments * @var array */ private static $self_closing_tags = [ 'area', 'base', 'basefont', 'bgsound', 'br', 'col', 'embed', 'frame', 'hr', 'img', 'input', 'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr', ]; /** * List of elements allowed in head. * * @link https://github.com/ampproject/amphtml/blob/445d6e3be8a5063e2738c6f90fdcd57f2b6208be/validator/engine/htmlparser.js#L83-L100 * @link https://www.w3.org/TR/html5/document-metadata.html * @var array */ private static $elements_allowed_in_head = [ 'title', 'base', 'link', 'meta', 'style', 'noscript', 'script', ]; /** * Stored noscript/comment replacements for libxml<2.8. * * @since 0.7 * @var array */ public static $noscript_placeholder_comments = []; /** * Return a valid DOMDocument representing HTML document passed as a parameter. * * @since 0.7 * @see AMP_DOM_Utils::get_content_from_dom_node() * * @param string $document Valid HTML document to be represented by a DOMDocument. * @return DOMDocument|false Returns DOMDocument, or false if conversion failed. */ public static function get_dom( $document ) { $libxml_previous_state = libxml_use_internal_errors( true ); $dom = new DOMDocument(); // @todo In the future consider an AMP_DOMDocument subclass that does this automatically. See . $document = self::convert_amp_bind_attributes( $document ); // Force all self-closing tags to have closing tags since DOMDocument isn't fully aware. $document = preg_replace( '#<(' . implode( '|', self::$self_closing_tags ) . ')[^>]*>(?!)#', '$0', $document ); // Deal with bugs in older versions of libxml. $added_back_compat_meta_content_type = false; if ( version_compare( LIBXML_DOTTED_VERSION, '2.8', '<' ) ) { /* * Replace noscript elements with placeholders since libxml<2.8 can parse them incorrectly. * When appearing in the head element, a noscript can cause the head to close prematurely * and the noscript gets moved to the body and anything after it which was in the head. * See . * This is limited to only running in the head element because this is where the problem lies, * and it is important for the AMP_Script_Sanitizer to be able to access the noscript elements * in the body otherwise. */ $document = preg_replace_callback( '#^.+?(?=]*>.*?#si', static function( $noscript_matches ) { $placeholder = sprintf( '', (string) wp_rand() ); AMP_DOM_Utils::$noscript_placeholder_comments[ $placeholder ] = $noscript_matches[0]; return $placeholder; }, $head_matches[0] ); }, $document ); } /* * Add a pre-HTML5-style declaration of the encoding since libxml doesn't always recognize * HTML5's meta charset. In libxml<2.8 it never does, see . * In libxml>=2.8, if the meta charset does not appear at the beginning of the head then it fails to be understood. */ $document = preg_replace( '#(?=', $document, 1, $count ); if ( 1 === $count ) { $added_back_compat_meta_content_type = true; } /* * Wrap in dummy tags, since XML needs one parent node. * It also makes it easier to loop through nodes. * We can later use this to extract our nodes. * Add charset so loadHTML does not have problems parsing it. */ $result = $dom->loadHTML( $document ); libxml_clear_errors(); libxml_use_internal_errors( $libxml_previous_state ); if ( ! $result ) { return false; } // Remove pre-HTML5-style encoding declaration if added above. if ( $added_back_compat_meta_content_type ) { $meta_http_equiv_element = $dom->getElementById( 'meta-http-equiv-content-type' ); if ( $meta_http_equiv_element ) { $meta_http_equiv_element->parentNode->removeChild( $meta_http_equiv_element ); } } // Make sure there is a head and a body. $head = $dom->getElementsByTagName( 'head' )->item( 0 ); if ( ! $head ) { $head = $dom->createElement( 'head' ); $dom->documentElement->insertBefore( $head, $dom->documentElement->firstChild ); } $body = $dom->getElementsByTagName( 'body' )->item( 0 ); if ( ! $body ) { $body = $dom->createElement( 'body' ); $dom->documentElement->appendChild( $body ); } self::move_invalid_head_nodes_to_body( $head, $body ); return $dom; } /** * Move elements not allowed in the head to the body. * * Apparently PHP's DOM is more lenient when parsing HTML to allow nodes in the HEAD which do not belong. A proper * HTML5 parser should rather prematurely short-circuit the HEAD when it finds an illegal element. * * @param DOMElement $head HEAD element. * @param DOMElement $body BODY element. */ private static function move_invalid_head_nodes_to_body( $head, $body ) { // Walking backwards makes it easier to move elements in the expected order. $node = $head->lastChild; while ( $node ) { $next_sibling = $node->previousSibling; if ( ! self::is_valid_head_node( $node ) ) { $body->insertBefore( $head->removeChild( $node ), $body->firstChild ); } $node = $next_sibling; } } /** * Determine whether a node can be in the head. * * @link https://github.com/ampproject/amphtml/blob/445d6e3be8a5063e2738c6f90fdcd57f2b6208be/validator/engine/htmlparser.js#L83-L100 * @link https://www.w3.org/TR/html5/document-metadata.html * * @param DOMNode $node Node. * @return bool Whether valid head node. */ public static function is_valid_head_node( DOMNode $node ) { return ( ( $node instanceof DOMElement && in_array( $node->nodeName, self::$elements_allowed_in_head, true ) ) || ( $node instanceof DOMText && preg_match( '/^\s*$/', $node->nodeValue ) ) // Whitespace text nodes are OK. || $node instanceof DOMComment ); } /** * Get attribute prefix for converted amp-bind attributes. * * This contains a random string to prevent HTML content containing this data- attribute * originally from being mutated to contain an amp-bind attribute when attributes are restored. * * @since 0.7 * @see \AMP_DOM_Utils::convert_amp_bind_attributes() * @see \AMP_DOM_Utils::restore_amp_bind_attributes() * @deprecated Use AMP_DOM_Utils::AMP_BIND_DATA_ATTR_PREFIX alone. * @link https://www.ampproject.org/docs/reference/components/amp-bind * * @return string HTML5 data-* attribute name prefix for AMP binding attributes. */ public static function get_amp_bind_placeholder_prefix() { _deprecated_function( __METHOD__, '1.2.1' ); return self::AMP_BIND_DATA_ATTR_PREFIX; } /** * Get amp-mustache tag/placeholder mappings. * * @since 0.7 * @see \wpdb::placeholder_escape() * * @return array Mapping of mustache tag token to its placeholder. */ private static function get_mustache_tag_placeholders() { static $placeholders; if ( ! isset( $placeholders ) ) { $salt = wp_rand(); // Note: The order of these tokens is important, as it determines the order of the order of the replacements. $tokens = [ '{{{', '}}}', '{{#', '{{^', '{{/', '{{/', '{{', '}}', ]; $placeholders = []; foreach ( $tokens as $token ) { $placeholders[ $token ] = '_amp_mustache_' . md5( $salt . $token ); } } return $placeholders; } /** * Replace AMP binding attributes with something that libxml can parse (as HTML5 data-* attributes). * * This is necessary because attributes in square brackets are not understood in PHP and * get dropped with an error raised: * > Warning: DOMDocument::loadHTML(): error parsing attribute name * This is a reciprocal function of AMP_DOM_Utils::restore_amp_bind_attributes(). * * @since 0.7 * @see \AMP_DOM_Utils::convert_amp_bind_attributes() * @link https://www.ampproject.org/docs/reference/components/amp-bind * * @param string $html HTML containing amp-bind attributes. * @return string HTML with AMP binding attributes replaced with HTML5 data-* attributes. */ public static function convert_amp_bind_attributes( $html ) { // Pattern for HTML attribute accounting for binding attr name, boolean attribute, single/double-quoted attribute value, and unquoted attribute values. $attr_regex = '#^\s+(?P\[?[a-zA-Z0-9_\-]+\]?)(?P=(?:"[^"]*+"|\'[^\']*+\'|[^\'"\s]+))?#'; /** * Replace callback. * * @param array $tag_matches Tag matches. * @return string Replacement. */ $replace_callback = static function( $tag_matches ) use ( $attr_regex ) { // Strip the self-closing slash as long as it is not an attribute value, like for the href attribute (). $old_attrs = preg_replace( '#(?'; }; // Match all start tags that contain a binding attribute. $pattern = implode( '', [ '#<', '(?P[a-zA-Z0-9_\-]+)', // Tag name. '(?P\s', // Attributes. '(?:[^>"\'\[\]]+|"[^"]*+"|\'[^\']*+\')*+', // Non-binding attributes tokens. '\[[a-zA-Z0-9_\-]+\]', // One binding attribute key. '(?:[^>"\']+|"[^"]*+"|\'[^\']*+\')*+', // Any attribute tokens, including binding ones. ')>#s', ] ); $converted = preg_replace_callback( $pattern, $replace_callback, $html ); /** * If the regex engine incurred an error during processing, for example exceeding the backtrack * limit, $converted will be null. In this case we return the originally passed document to allow * DOMDocument to attempt to load it. If the AMP HTML doesn't make use of amp-bind or similar * attributes, then everything should still work. * * See https://github.com/ampproject/amp-wp/issues/993 for additional context on this issue. * See http://php.net/manual/en/pcre.constants.php for additional info on PCRE errors. */ return ( null !== $converted ) ? $converted : $html; } /** * Convert AMP bind-attributes back to their original syntax. * * This is a reciprocal function of AMP_DOM_Utils::convert_amp_bind_attributes(). * * @since 0.7 * @see \AMP_DOM_Utils::convert_amp_bind_attributes() * @deprecated Allow the data-amp-bind-* attributes to be used instead. * @link https://www.ampproject.org/docs/reference/components/amp-bind * * @param string $html HTML with amp-bind attributes converted. * @return string HTML with amp-bind attributes restored. */ public static function restore_amp_bind_attributes( $html ) { _deprecated_function( __METHOD__, '1.2.1' ); $html = preg_replace( '#\s' . self::AMP_BIND_DATA_ATTR_PREFIX . '([a-zA-Z0-9_\-]+)#', ' [$1]', $html ); return $html; } /** * Return a valid DOMDocument representing arbitrary HTML content passed as a parameter. * * @see Reciprocal function get_content_from_dom() * * @since 0.2 * * @param string $content Valid HTML content to be represented by a DOMDocument. * * @return DOMDocument|false Returns DOMDocument, or false if conversion failed. */ public static function get_dom_from_content( $content ) { /* * Wrap in dummy tags, since XML needs one parent node. * It also makes it easier to loop through nodes. * We can later use this to extract our nodes. * Add utf-8 charset so loadHTML does not have problems parsing it. * See: http://php.net/manual/en/domdocument.loadhtml.php#78243 */ $document = sprintf( '%s', get_bloginfo( 'charset' ), $content ); return self::get_dom( $document ); } /** * Return valid HTML *body* content extracted from the DOMDocument passed as a parameter. * * @since 0.2 * @see AMP_DOM_Utils::get_content_from_dom_node() Reciprocal function. * * @param DOMDocument $dom Represents an HTML document from which to extract HTML content. * @return string Returns the HTML content of the body element represented in the DOMDocument. */ public static function get_content_from_dom( $dom ) { $body = $dom->getElementsByTagName( 'body' )->item( 0 ); // The DOMDocument may contain no body. In which case return nothing. if ( null === $body ) { return ''; } return preg_replace( '#^.*?(.*).*?$#si', '$1', self::get_content_from_dom_node( $dom, $body ) ); } /** * Return valid HTML content extracted from the DOMNode passed as a parameter. * * @since 0.6 * @see AMP_DOM_Utils::get_dom() Where the operations in this method are mirrored. * @see AMP_DOM_Utils::get_content_from_dom() Reciprocal function. * @todo In the future consider an AMP_DOMDocument subclass that does this automatically at saveHTML(). See . * * @param DOMDocument $dom Represents an HTML document. * @param DOMElement $node Represents an HTML element of the $dom from which to extract HTML content. * @return string Returns the HTML content represented in the DOMNode */ public static function get_content_from_dom_node( $dom, $node ) { /** * Self closing tags regex. * * @var string Regular expression to match self-closing tags * that saveXML() has generated a closing tag for. */ static $self_closing_tags_regex; /* * Cache this regex so we don't have to recreate it every call. */ if ( ! isset( $self_closing_tags_regex ) ) { $self_closing_tags = implode( '|', self::$self_closing_tags ); $self_closing_tags_regex = "##i"; } /* * Prevent amp-mustache syntax from getting URL-encoded in attributes when saveHTML is done. * While this is applying to the entire document, it only really matters inside of