. * @type string|null $title Document title text, null when there is no <title>. * @type bool $skip_link Whether the page already has a skip link. * @type array|null $main Skip-link target: ordinal, id and source (main, role or id) of the main region. * @type bool $complete False when the markup ended inside an unfinished tag. * } */ public static function analyze( $html ) { $doc = array( 'elements' => array(), 'ids' => array(), 'label_for' => array(), 'referenced' => array(), 'tables' => array(), 'videos' => array(), 'headings' => array(), 'positive_tabindex' => array(), 'html_ord' => null, 'head_ord' => null, 'body_ord' => null, 'title_ord' => null, 'title' => null, 'skip_link' => false, 'main' => null, 'complete' => true, ); $processor = new \WP_HTML_Tag_Processor( $html ); $ordinal = -1; $collectors = array(); // Stack of ordinals whose text we are collecting. $hidden_stack = array(); // Open aria-hidden subtrees: tag name and how many same-name tags are open inside. $svg_depth = 0; $label_depth = 0; $table_stack = array(); $media_stack = array(); $fallback_ids = array(); $first_role_main = null; $open_stack = array(); // Open elements (tag, class, id), for each element's parent. $css_hidden = array(); // Open subtrees a compatibility pack says its product hides with CSS. $packs = Compat::packs(); while ( $processor->next_token() ) { $type = $processor->get_token_type(); if ( '#text' === $type ) { if ( $collectors && ! $hidden_stack ) { $text = $processor->get_modifiable_text(); $field = $css_hidden ? 'hidden_text' : 'text'; foreach ( $collectors as $collector ) { $doc['elements'][ $collector ][ $field ] .= $text; } } continue; } if ( '#tag' !== $type ) { continue; } $tag = $processor->get_tag(); if ( $processor->is_tag_closer() ) { $top = count( $hidden_stack ) - 1; if ( $top >= 0 && $hidden_stack[ $top ]['tag'] === $tag ) { if ( $hidden_stack[ $top ]['depth'] > 0 ) { --$hidden_stack[ $top ]['depth']; } else { array_pop( $hidden_stack ); } } $top = count( $css_hidden ) - 1; if ( $top >= 0 && $css_hidden[ $top ]['tag'] === $tag ) { if ( $css_hidden[ $top ]['depth'] > 0 ) { --$css_hidden[ $top ]['depth']; } else { array_pop( $css_hidden ); } } for ( $i = count( $open_stack ) - 1; $i >= 0; $i-- ) { if ( $open_stack[ $i ]['tag'] === $tag ) { array_splice( $open_stack, $i ); // Also closes elements left open inside it. break; } } if ( 'SVG' === $tag && $svg_depth > 0 ) { --$svg_depth; } elseif ( 'LABEL' === $tag && $label_depth > 0 ) { --$label_depth; } elseif ( 'TABLE' === $tag && $table_stack ) { array_pop( $table_stack ); } elseif ( ( 'VIDEO' === $tag || 'AUDIO' === $tag ) && $media_stack ) { array_pop( $media_stack ); } if ( in_array( $tag, self::NAMED_BY_CONTENT, true ) ) { // Close the innermost collector of this tag type. for ( $i = count( $collectors ) - 1; $i >= 0; $i-- ) { if ( $doc['elements'][ $collectors[ $i ] ]['tag'] === $tag ) { array_splice( $collectors, $i, 1 ); break; } } } continue; } ++$ordinal; // Read attribute names once; most tags have one or two, so only ask for those. $names = $processor->get_attribute_names_with_prefix( '' ); $names = $names ? array_flip( $names ) : array(); $attr = static function ( $name ) use ( $processor, $names ) { return isset( $names[ $name ] ) ? $processor->get_attribute( $name ) : null; }; // Every id on the page, for duplicate checks and skip-link targets. $id = $attr( 'id' ); if ( is_string( $id ) && '' !== $id ) { $doc['ids'][ $id ] = isset( $doc['ids'][ $id ] ) ? $doc['ids'][ $id ] + 1 : 1; if ( in_array( $id, array( 'main', 'content', 'primary', 'main-content', 'maincontent', 'site-content', 'wp--skip-link--target' ), true ) && ! isset( $fallback_ids[ $id ] ) ) { $fallback_ids[ $id ] = $ordinal; } if ( 0 === strpos( $id, 'wp-block-template-skip-link' ) ) { $doc['skip_link'] = true; // Block themes add their skip link with this script. } } foreach ( array( 'aria-labelledby', 'aria-describedby', 'aria-controls' ) as $ref_attribute ) { $refs = $attr( $ref_attribute ); if ( is_string( $refs ) ) { foreach ( preg_split( '/\s+/', trim( $refs ) ) as $ref ) { if ( '' !== $ref ) { $doc['referenced'][ $ref ] = true; } } } } if ( null === $first_role_main && 'main' === $attr( 'role' ) ) { $first_role_main = array( 'ord' => $ordinal, 'id' => is_string( $id ) ? $id : '', 'source' => 'role', ); } $element_hidden = 'true' === $attr( 'aria-hidden' ); $class = $attr( 'class' ); $class = is_string( $class ) ? $class : ''; $id_value = is_string( $id ) ? $id : ''; $parent = $open_stack ? $open_stack[ count( $open_stack ) - 1 ] : array( 'tag' => '', 'class' => '', 'id' => '', ); $watched = false; $hides_text = false; foreach ( $packs as $pack ) { $watched = $watched || $pack::watches( $tag, $class, $id_value ); $hides_text = $hides_text || $pack::hides_text( $tag, $class, $id_value, $parent ); } $tabindex = $attr( 'tabindex' ); if ( is_string( $tabindex ) && is_numeric( trim( $tabindex ) ) && (int) $tabindex > 0 ) { $doc['positive_tabindex'][] = $ordinal; } if ( isset( self::ATTRIBUTES[ $tag ] ) || $watched ) { $element = array( 'ord' => $ordinal, 'tag' => $tag, 'attrs' => array(), 'text' => '', 'hidden_text' => '', 'hint' => '', 'in_label' => $label_depth > 0, 'in_svg' => $svg_depth > 0, 'hidden' => $element_hidden || (bool) $hidden_stack, 'css_hidden' => (bool) $css_hidden, 'parent' => $parent, 'link_ord' => null, ); $recorded = $watched ? array_keys( $names ) : self::ATTRIBUTES[ $tag ]; foreach ( $recorded as $attribute ) { if ( isset( $names[ $attribute ] ) ) { $element['attrs'][ $attribute ] = $processor->get_attribute( $attribute ); } } if ( 'IMG' === $tag ) { // The innermost link around the image, so a linked image's alt can name the link. for ( $i = count( $collectors ) - 1; $i >= 0; $i-- ) { if ( 'A' === $doc['elements'][ $collectors[ $i ] ]['tag'] ) { $element['link_ord'] = $collectors[ $i ]; break; } } } $doc['elements'][ $ordinal ] = $element; } // Content inside named elements: image alt text, aria-label on children, svg titles. if ( $collectors && ! $hidden_stack && ! $element_hidden ) { $extra = ''; if ( 'IMG' === $tag ) { $alt = $attr( 'alt' ); $extra = is_string( $alt ) ? $alt : ''; } elseif ( 'TITLE' === $tag && $svg_depth > 0 ) { $extra = $processor->get_modifiable_text(); } $child_label = $attr( 'aria-label' ); if ( is_string( $child_label ) && ! in_array( $tag, self::NAMED_BY_CONTENT, true ) ) { $extra .= ' ' . $child_label; } if ( '' !== trim( $extra ) ) { $field = $css_hidden ? 'hidden_text' : 'text'; foreach ( $collectors as $collector ) { $doc['elements'][ $collector ][ $field ] .= ' ' . $extra; } } } // Class names of descendants hint at icon meaning (fa-search, dashicons-facebook). if ( $collectors ) { $hint = $attr( 'class' ); if ( is_string( $hint ) && '' !== $hint ) { foreach ( $collectors as $collector ) { if ( strlen( $doc['elements'][ $collector ]['hint'] ) < 300 ) { $doc['elements'][ $collector ]['hint'] .= ' ' . $hint; } } } } switch ( $tag ) { case 'HTML': if ( null === $doc['html_ord'] ) { $doc['html_ord'] = $ordinal; } break; case 'HEAD': if ( null === $doc['head_ord'] ) { $doc['head_ord'] = $ordinal; } break; case 'BODY': if ( null === $doc['body_ord'] ) { $doc['body_ord'] = $ordinal; } break; case 'TITLE': if ( 0 === $svg_depth && null === $doc['title_ord'] ) { $doc['title_ord'] = $ordinal; $doc['title'] = trim( (string) $processor->get_modifiable_text() ); } break; case 'MAIN': if ( null === $doc['main'] ) { $doc['main'] = array( 'ord' => $ordinal, 'id' => is_string( $id ) ? $id : '', 'source' => 'main', ); } break; case 'SVG': if ( ! $processor->has_self_closing_flag() ) { ++$svg_depth; } break; case 'LABEL': ++$label_depth; $for = $processor->get_attribute( 'for' ); if ( is_string( $for ) && '' !== $for ) { $doc['label_for'][ $for ] = true; $doc['referenced'][ $for ] = true; } break; case 'A': $href = $processor->get_attribute( 'href' ); if ( Compat::has_class( $class, 'fixlight-skip-link' ) ) { $doc['skip_link'] = true; // Fixlight's own, whatever form a pack gave its address. } if ( is_string( $href ) && 1 < strlen( $href ) && '#' === $href[0] ) { $fragment = substr( $href, 1 ); $doc['referenced'][ $fragment ] = true; $class = (string) $processor->get_attribute( 'class' ); if ( false !== stripos( $class, 'skip' ) || false !== stripos( $fragment, 'skip' ) ) { $doc['skip_link'] = true; } } break; case 'TABLE': $doc['tables'][] = array( 'ord' => $ordinal, 'has_th' => false, 'rows' => 0, 'presentation' => in_array( $processor->get_attribute( 'role' ), array( 'presentation', 'none' ), true ), ); $table_stack[] = count( $doc['tables'] ) - 1; break; case 'TH': if ( $table_stack ) { $doc['tables'][ end( $table_stack ) ]['has_th'] = true; } break; case 'TR': if ( $table_stack ) { ++$doc['tables'][ end( $table_stack ) ]['rows']; } break; case 'VIDEO': case 'AUDIO': $doc['videos'][] = array( 'ord' => $ordinal, 'tag' => $tag, 'captions' => false, ); $media_stack[] = count( $doc['videos'] ) - 1; break; case 'TRACK': $kind = $processor->get_attribute( 'kind' ); if ( $media_stack && in_array( $kind, array( 'captions', 'subtitles' ), true ) ) { $doc['videos'][ end( $media_stack ) ]['captions'] = true; } break; } if ( in_array( $tag, array( 'H1', 'H2', 'H3', 'H4', 'H5', 'H6' ), true ) && 0 === $svg_depth ) { $doc['headings'][] = $ordinal; } // Start collecting text for elements named by their content. if ( in_array( $tag, self::NAMED_BY_CONTENT, true ) && ! $processor->has_self_closing_flag() ) { if ( ! isset( $doc['elements'][ $ordinal ] ) ) { $doc['elements'][ $ordinal ] = array( 'ord' => $ordinal, 'tag' => $tag, 'attrs' => array(), 'text' => '', 'hidden_text' => '', 'hint' => '', 'in_label' => $label_depth > 1, 'in_svg' => $svg_depth > 0, 'hidden' => $element_hidden || (bool) $hidden_stack, 'css_hidden' => (bool) $css_hidden, 'parent' => $parent, 'link_ord' => null, ); } $collectors[] = $ordinal; } if ( ! in_array( $tag, self::VOID_TAGS, true ) && ! in_array( $tag, self::ATOMIC_TAGS, true ) && ! $processor->has_self_closing_flag() ) { $open_stack[] = array( 'tag' => $tag, 'class' => $class, 'id' => $id_value, ); $top = count( $css_hidden ) - 1; if ( $hides_text ) { $css_hidden[] = array( 'tag' => $tag, 'depth' => 0, ); } elseif ( $top >= 0 && $css_hidden[ $top ]['tag'] === $tag ) { ++$css_hidden[ $top ]['depth']; } $top = count( $hidden_stack ) - 1; if ( $element_hidden ) { $hidden_stack[] = array( 'tag' => $tag, 'depth' => 0, ); } elseif ( $top >= 0 && $hidden_stack[ $top ]['tag'] === $tag ) { // Same tag nested inside the hidden subtree: its closing tag does not end the subtree. ++$hidden_stack[ $top ]['depth']; } } } if ( method_exists( $processor, 'paused_at_incomplete_token' ) && $processor->paused_at_incomplete_token() ) { $doc['complete'] = false; } $first_link = true; foreach ( $doc['elements'] as $ord => $element ) { foreach ( array( 'text', 'hidden_text' ) as $field ) { if ( isset( $element[ $field ] ) && '' !== $element[ $field ] ) { $doc['elements'][ $ord ][ $field ] = self::clean_text( $element[ $field ] ); } } if ( $doc['skip_link'] || 'A' !== $element['tag'] || ! isset( $element['attrs']['href'] ) || ( null !== $doc['body_ord'] && $ord < $doc['body_ord'] ) ) { continue; } $href = (string) $element['attrs']['href']; if ( 1 < strlen( $href ) && '#' === $href[0] ) { // A skip link in any common language, or the page's first link jumping to an anchor on the page. if ( self::is_skip_text( $doc['elements'][ $ord ]['text'] ) || ( $first_link && isset( $doc['ids'][ substr( $href, 1 ) ] ) ) ) { $doc['skip_link'] = true; } } $first_link = false; } // Pick the skip-link target: <main>, then role="main", then a well-known id. if ( null === $doc['main'] && null !== $first_role_main ) { $doc['main'] = $first_role_main; } if ( null === $doc['main'] ) { foreach ( array( 'wp--skip-link--target', 'main', 'main-content', 'maincontent', 'content', 'primary', 'site-content' ) as $candidate ) { // A skip link to an id used twice may land on the wrong element. if ( isset( $fallback_ids[ $candidate ] ) && 1 === $doc['ids'][ $candidate ] ) { $doc['main'] = array( 'ord' => $fallback_ids[ $candidate ], 'id' => $candidate, 'source' => 'id', ); break; } } } foreach ( $doc['headings'] as $index => $ord ) { $doc['headings'][ $index ] = array( 'ord' => $ord, 'level' => (int) substr( $doc['elements'][ $ord ]['tag'], 1 ), 'text' => $doc['elements'][ $ord ]['text'], ); } return $doc; } /** * Collapses whitespace and drops icon-font glyphs (Unicode private use area), * which screen readers announce as nonsense or not at all. * * @param string $text Text. * @return string */ public static function clean_text( $text ) { $stripped = preg_replace( '/[\x{E000}-\x{F8FF}]/u', '', (string) $text ); if ( null !== $stripped ) { $text = $stripped; // Not valid UTF-8 (a Latin-1 page): it cannot hold these glyphs either. } return trim( self::text_replace( '/\s+/u', ' ', $text ) ); } /** * Whether a name is only numbers or symbols (a share count, "0"): technically * a name, but it says nothing about what the link or button does. * * @param string $text Visible text. * @return bool */ public static function is_weak_name( $text ) { // "1.2k" and "3M" are counts too. return '' !== $text && ! self::text_matches( '/\p{L}/u', self::text_replace( '/(?<=\d)[kmb]\b/iu', '', $text ) ); } /** * Whether link text reads like a skip link. * * @param string $text Link text. * @return bool */ public static function is_skip_text( $text ) { return self::text_matches( '/\b(skip|jump) to\b|\bskip (navigation|nav|links?)\b|saltar (al|a la|el)|ir al contenido|ir a(l)? (contenido|principal)|aller au contenu|passer au contenu|zum inhalt|inhalt springen|vai al contenuto|salta al contenuto|pular para|ir para o conte|naar (de )?inhoud/iu', $text ); } /** * Runs preg_replace() with a pattern ending in the u modifier. Text that is * not valid UTF-8 (a Latin-1 page, for example) makes PCRE fail, so the same * pattern then runs on bytes. * * @param string $pattern Pattern; its last modifier is u. * @param string $replacement Replacement. * @param string $text Text. * @return string */ public static function text_replace( $pattern, $replacement, $text ) { $result = preg_replace( $pattern, $replacement, (string) $text ); if ( null === $result ) { $result = preg_replace( substr( $pattern, 0, -1 ), $replacement, (string) $text ); } return (string) $result; } /** * Runs preg_match() with a pattern ending in the u modifier, on bytes when * the text is not valid UTF-8. * * @param string $pattern Pattern; its last modifier is u. * @param string $text Text. * @return bool */ public static function text_matches( $pattern, $text ) { $result = preg_match( $pattern, (string) $text ); if ( false === $result ) { $result = preg_match( substr( $pattern, 0, -1 ), (string) $text ); } return 1 === $result; } /** * Whether a form control already has an accessible name. * * @param array $element Recorded INPUT, SELECT or TEXTAREA. * @param array $doc Analysis result. * @return bool */ public static function control_has_name( array $element, array $doc ) { $attrs = $element['attrs']; if ( $element['in_label'] ) { return true; } foreach ( array( 'aria-label', 'aria-labelledby', 'title' ) as $attribute ) { if ( isset( $attrs[ $attribute ] ) && is_string( $attrs[ $attribute ] ) && '' !== trim( $attrs[ $attribute ] ) ) { return true; } } return isset( $attrs['id'] ) && is_string( $attrs['id'] ) && isset( $doc['label_for'][ $attrs['id'] ] ); } /** * Whether a form control needs a label at all. * * @param array $element Recorded INPUT, SELECT or TEXTAREA. * @return bool */ public static function control_needs_name( array $element ) { if ( $element['hidden'] ) { return false; } if ( 'INPUT' !== $element['tag'] ) { return true; } $type = isset( $element['attrs']['type'] ) && is_string( $element['attrs']['type'] ) ? strtolower( $element['attrs']['type'] ) : 'text'; return ! in_array( $type, array( 'hidden', 'submit', 'reset', 'button', 'image' ), true ); } /** * Whether a link or button has an accessible name. * * @param array $element Recorded A or BUTTON. * @return bool */ public static function has_accessible_name( array $element ) { if ( '' !== $element['text'] ) { return true; } foreach ( array( 'aria-label', 'aria-labelledby', 'title' ) as $attribute ) { if ( isset( $element['attrs'][ $attribute ] ) && is_string( $element['attrs'][ $attribute ] ) && '' !== trim( $element['attrs'][ $attribute ] ) ) { return true; } } return false; } /** * Whether a link is a real, focusable link (has an href and is not hidden). * * @param array $element Recorded A. * @return bool */ public static function is_live_link( array $element ) { if ( $element['hidden'] || ! isset( $element['attrs']['href'] ) ) { return false; } $role = isset( $element['attrs']['role'] ) ? $element['attrs']['role'] : ''; if ( in_array( $role, array( 'presentation', 'none' ), true ) ) { return false; } return ! ( isset( $element['attrs']['tabindex'] ) && '-1' === $element['attrs']['tabindex'] ); } }

*/ const ATTRIBUTES = array( 'HTML' => array( 'lang' ), 'BODY' => array(), 'HEAD' => array(), 'MAIN' => array( 'id' ), 'ARTICLE' => array( 'id' ), 'A' => array( 'href', 'target', 'rel', 'class', 'title', 'aria-label', 'aria-labelledby', 'aria-describedby', 'aria-hidden', 'role', 'tabindex', 'id' ), 'BUTTON' => array( 'class', 'title', 'aria-label', 'aria-labelledby', 'aria-hidden', 'type', 'tabindex', 'id' ), 'IMG' => array( 'alt', 'class', 'src', 'width', 'height', 'role', 'aria-hidden', 'aria-label', 'aria-labelledby' ), 'INPUT' => array( 'type', 'id', 'name', 'placeholder', 'aria-label', 'aria-labelledby', 'title', 'aria-hidden', 'tabindex' ), 'SELECT' => array( 'id', 'name', 'aria-label', 'aria-labelledby', 'title', 'aria-hidden' ), 'TEXTAREA' => array( 'id', 'name', 'placeholder', 'aria-label', 'aria-labelledby', 'title', 'aria-hidden' ), 'LABEL' => array( 'for' ), 'IFRAME' => array( 'title', 'src', 'aria-hidden', 'aria-label', 'width', 'height' ), 'META' => array( 'name', 'content' ), 'VIDEO' => array( 'autoplay', 'controls', 'muted' ), 'AUDIO' => array( 'autoplay', 'controls', 'muted' ), 'TRACK' => array( 'kind' ), 'TABLE' => array( 'role' ), 'SVG' => array( 'aria-label', 'aria-hidden', 'role' ), ); /** * Analyzes a page. * * @param string $html Full HTML document. * @return array { * @type array[] $elements Recorded elements, keyed by ordinal. * @type array $ids Count of each id value. * @type array $label_for Set of ids that a points at. * @type array $referenced Set of ids referenced by for, aria-* or #fragment links. * @type array $tables One entry per table: ordinal, has_th, rows, presentation. * @type array $videos One entry per video/audio: ordinal, captions. * @type array $headings Heading levels in document order with ordinal and text. * @type int[] $positive_tabindex Ordinals of elements with tabindex above 0. * @type int|null $html_ord Ordinal of . * @type int|null $head_ord Ordinal of . * @type int|null $body_ord Ordinal of . * @type int|null $title_ord Ordinal of the document . * @type string|null $title Document title text, null when there is no <title>. * @type bool $skip_link Whether the page already has a skip link. * @type array|null $main Skip-link target: ordinal, id and source (main, role or id) of the main region. * @type bool $complete False when the markup ended inside an unfinished tag. * } */ public static function analyze( $html ) { $doc = array( 'elements' => array(), 'ids' => array(), 'label_for' => array(), 'referenced' => array(), 'tables' => array(), 'videos' => array(), 'headings' => array(), 'positive_tabindex' => array(), 'html_ord' => null, 'head_ord' => null, 'body_ord' => null, 'title_ord' => null, 'title' => null, 'skip_link' => false, 'main' => null, 'complete' => true, ); $processor = new \WP_HTML_Tag_Processor( $html ); $ordinal = -1; $collectors = array(); // Stack of ordinals whose text we are collecting. $hidden_stack = array(); // Open aria-hidden subtrees: tag name and how many same-name tags are open inside. $svg_depth = 0; $label_depth = 0; $table_stack = array(); $media_stack = array(); $fallback_ids = array(); $first_role_main = null; $open_stack = array(); // Open elements (tag, class, id), for each element's parent. $css_hidden = array(); // Open subtrees a compatibility pack says its product hides with CSS. $packs = Compat::packs(); while ( $processor->next_token() ) { $type = $processor->get_token_type(); if ( '#text' === $type ) { if ( $collectors && ! $hidden_stack ) { $text = $processor->get_modifiable_text(); $field = $css_hidden ? 'hidden_text' : 'text'; foreach ( $collectors as $collector ) { $doc['elements'][ $collector ][ $field ] .= $text; } } continue; } if ( '#tag' !== $type ) { continue; } $tag = $processor->get_tag(); if ( $processor->is_tag_closer() ) { $top = count( $hidden_stack ) - 1; if ( $top >= 0 && $hidden_stack[ $top ]['tag'] === $tag ) { if ( $hidden_stack[ $top ]['depth'] > 0 ) { --$hidden_stack[ $top ]['depth']; } else { array_pop( $hidden_stack ); } } $top = count( $css_hidden ) - 1; if ( $top >= 0 && $css_hidden[ $top ]['tag'] === $tag ) { if ( $css_hidden[ $top ]['depth'] > 0 ) { --$css_hidden[ $top ]['depth']; } else { array_pop( $css_hidden ); } } for ( $i = count( $open_stack ) - 1; $i >= 0; $i-- ) { if ( $open_stack[ $i ]['tag'] === $tag ) { array_splice( $open_stack, $i ); // Also closes elements left open inside it. break; } } if ( 'SVG' === $tag && $svg_depth > 0 ) { --$svg_depth; } elseif ( 'LABEL' === $tag && $label_depth > 0 ) { --$label_depth; } elseif ( 'TABLE' === $tag && $table_stack ) { array_pop( $table_stack ); } elseif ( ( 'VIDEO' === $tag || 'AUDIO' === $tag ) && $media_stack ) { array_pop( $media_stack ); } if ( in_array( $tag, self::NAMED_BY_CONTENT, true ) ) { // Close the innermost collector of this tag type. for ( $i = count( $collectors ) - 1; $i >= 0; $i-- ) { if ( $doc['elements'][ $collectors[ $i ] ]['tag'] === $tag ) { array_splice( $collectors, $i, 1 ); break; } } } continue; } ++$ordinal; // Read attribute names once; most tags have one or two, so only ask for those. $names = $processor->get_attribute_names_with_prefix( '' ); $names = $names ? array_flip( $names ) : array(); $attr = static function ( $name ) use ( $processor, $names ) { return isset( $names[ $name ] ) ? $processor->get_attribute( $name ) : null; }; // Every id on the page, for duplicate checks and skip-link targets. $id = $attr( 'id' ); if ( is_string( $id ) && '' !== $id ) { $doc['ids'][ $id ] = isset( $doc['ids'][ $id ] ) ? $doc['ids'][ $id ] + 1 : 1; if ( in_array( $id, array( 'main', 'content', 'primary', 'main-content', 'maincontent', 'site-content', 'wp--skip-link--target' ), true ) && ! isset( $fallback_ids[ $id ] ) ) { $fallback_ids[ $id ] = $ordinal; } if ( 0 === strpos( $id, 'wp-block-template-skip-link' ) ) { $doc['skip_link'] = true; // Block themes add their skip link with this script. } } foreach ( array( 'aria-labelledby', 'aria-describedby', 'aria-controls' ) as $ref_attribute ) { $refs = $attr( $ref_attribute ); if ( is_string( $refs ) ) { foreach ( preg_split( '/\s+/', trim( $refs ) ) as $ref ) { if ( '' !== $ref ) { $doc['referenced'][ $ref ] = true; } } } } if ( null === $first_role_main && 'main' === $attr( 'role' ) ) { $first_role_main = array( 'ord' => $ordinal, 'id' => is_string( $id ) ? $id : '', 'source' => 'role', ); } $element_hidden = 'true' === $attr( 'aria-hidden' ); $class = $attr( 'class' ); $class = is_string( $class ) ? $class : ''; $id_value = is_string( $id ) ? $id : ''; $parent = $open_stack ? $open_stack[ count( $open_stack ) - 1 ] : array( 'tag' => '', 'class' => '', 'id' => '', ); $watched = false; $hides_text = false; foreach ( $packs as $pack ) { $watched = $watched || $pack::watches( $tag, $class, $id_value ); $hides_text = $hides_text || $pack::hides_text( $tag, $class, $id_value, $parent ); } $tabindex = $attr( 'tabindex' ); if ( is_string( $tabindex ) && is_numeric( trim( $tabindex ) ) && (int) $tabindex > 0 ) { $doc['positive_tabindex'][] = $ordinal; } if ( isset( self::ATTRIBUTES[ $tag ] ) || $watched ) { $element = array( 'ord' => $ordinal, 'tag' => $tag, 'attrs' => array(), 'text' => '', 'hidden_text' => '', 'hint' => '', 'in_label' => $label_depth > 0, 'in_svg' => $svg_depth > 0, 'hidden' => $element_hidden || (bool) $hidden_stack, 'css_hidden' => (bool) $css_hidden, 'parent' => $parent, 'link_ord' => null, ); $recorded = $watched ? array_keys( $names ) : self::ATTRIBUTES[ $tag ]; foreach ( $recorded as $attribute ) { if ( isset( $names[ $attribute ] ) ) { $element['attrs'][ $attribute ] = $processor->get_attribute( $attribute ); } } if ( 'IMG' === $tag ) { // The innermost link around the image, so a linked image's alt can name the link. for ( $i = count( $collectors ) - 1; $i >= 0; $i-- ) { if ( 'A' === $doc['elements'][ $collectors[ $i ] ]['tag'] ) { $element['link_ord'] = $collectors[ $i ]; break; } } } $doc['elements'][ $ordinal ] = $element; } // Content inside named elements: image alt text, aria-label on children, svg titles. if ( $collectors && ! $hidden_stack && ! $element_hidden ) { $extra = ''; if ( 'IMG' === $tag ) { $alt = $attr( 'alt' ); $extra = is_string( $alt ) ? $alt : ''; } elseif ( 'TITLE' === $tag && $svg_depth > 0 ) { $extra = $processor->get_modifiable_text(); } $child_label = $attr( 'aria-label' ); if ( is_string( $child_label ) && ! in_array( $tag, self::NAMED_BY_CONTENT, true ) ) { $extra .= ' ' . $child_label; } if ( '' !== trim( $extra ) ) { $field = $css_hidden ? 'hidden_text' : 'text'; foreach ( $collectors as $collector ) { $doc['elements'][ $collector ][ $field ] .= ' ' . $extra; } } } // Class names of descendants hint at icon meaning (fa-search, dashicons-facebook). if ( $collectors ) { $hint = $attr( 'class' ); if ( is_string( $hint ) && '' !== $hint ) { foreach ( $collectors as $collector ) { if ( strlen( $doc['elements'][ $collector ]['hint'] ) < 300 ) { $doc['elements'][ $collector ]['hint'] .= ' ' . $hint; } } } } switch ( $tag ) { case 'HTML': if ( null === $doc['html_ord'] ) { $doc['html_ord'] = $ordinal; } break; case 'HEAD': if ( null === $doc['head_ord'] ) { $doc['head_ord'] = $ordinal; } break; case 'BODY': if ( null === $doc['body_ord'] ) { $doc['body_ord'] = $ordinal; } break; case 'TITLE': if ( 0 === $svg_depth && null === $doc['title_ord'] ) { $doc['title_ord'] = $ordinal; $doc['title'] = trim( (string) $processor->get_modifiable_text() ); } break; case 'MAIN': if ( null === $doc['main'] ) { $doc['main'] = array( 'ord' => $ordinal, 'id' => is_string( $id ) ? $id : '', 'source' => 'main', ); } break; case 'SVG': if ( ! $processor->has_self_closing_flag() ) { ++$svg_depth; } break; case 'LABEL': ++$label_depth; $for = $processor->get_attribute( 'for' ); if ( is_string( $for ) && '' !== $for ) { $doc['label_for'][ $for ] = true; $doc['referenced'][ $for ] = true; } break; case 'A': $href = $processor->get_attribute( 'href' ); if ( Compat::has_class( $class, 'fixlight-skip-link' ) ) { $doc['skip_link'] = true; // Fixlight's own, whatever form a pack gave its address. } if ( is_string( $href ) && 1 < strlen( $href ) && '#' === $href[0] ) { $fragment = substr( $href, 1 ); $doc['referenced'][ $fragment ] = true; $class = (string) $processor->get_attribute( 'class' ); if ( false !== stripos( $class, 'skip' ) || false !== stripos( $fragment, 'skip' ) ) { $doc['skip_link'] = true; } } break; case 'TABLE': $doc['tables'][] = array( 'ord' => $ordinal, 'has_th' => false, 'rows' => 0, 'presentation' => in_array( $processor->get_attribute( 'role' ), array( 'presentation', 'none' ), true ), ); $table_stack[] = count( $doc['tables'] ) - 1; break; case 'TH': if ( $table_stack ) { $doc['tables'][ end( $table_stack ) ]['has_th'] = true; } break; case 'TR': if ( $table_stack ) { ++$doc['tables'][ end( $table_stack ) ]['rows']; } break; case 'VIDEO': case 'AUDIO': $doc['videos'][] = array( 'ord' => $ordinal, 'tag' => $tag, 'captions' => false, ); $media_stack[] = count( $doc['videos'] ) - 1; break; case 'TRACK': $kind = $processor->get_attribute( 'kind' ); if ( $media_stack && in_array( $kind, array( 'captions', 'subtitles' ), true ) ) { $doc['videos'][ end( $media_stack ) ]['captions'] = true; } break; } if ( in_array( $tag, array( 'H1', 'H2', 'H3', 'H4', 'H5', 'H6' ), true ) && 0 === $svg_depth ) { $doc['headings'][] = $ordinal; } // Start collecting text for elements named by their content. if ( in_array( $tag, self::NAMED_BY_CONTENT, true ) && ! $processor->has_self_closing_flag() ) { if ( ! isset( $doc['elements'][ $ordinal ] ) ) { $doc['elements'][ $ordinal ] = array( 'ord' => $ordinal, 'tag' => $tag, 'attrs' => array(), 'text' => '', 'hidden_text' => '', 'hint' => '', 'in_label' => $label_depth > 1, 'in_svg' => $svg_depth > 0, 'hidden' => $element_hidden || (bool) $hidden_stack, 'css_hidden' => (bool) $css_hidden, 'parent' => $parent, 'link_ord' => null, ); } $collectors[] = $ordinal; } if ( ! in_array( $tag, self::VOID_TAGS, true ) && ! in_array( $tag, self::ATOMIC_TAGS, true ) && ! $processor->has_self_closing_flag() ) { $open_stack[] = array( 'tag' => $tag, 'class' => $class, 'id' => $id_value, ); $top = count( $css_hidden ) - 1; if ( $hides_text ) { $css_hidden[] = array( 'tag' => $tag, 'depth' => 0, ); } elseif ( $top >= 0 && $css_hidden[ $top ]['tag'] === $tag ) { ++$css_hidden[ $top ]['depth']; } $top = count( $hidden_stack ) - 1; if ( $element_hidden ) { $hidden_stack[] = array( 'tag' => $tag, 'depth' => 0, ); } elseif ( $top >= 0 && $hidden_stack[ $top ]['tag'] === $tag ) { // Same tag nested inside the hidden subtree: its closing tag does not end the subtree. ++$hidden_stack[ $top ]['depth']; } } } if ( method_exists( $processor, 'paused_at_incomplete_token' ) && $processor->paused_at_incomplete_token() ) { $doc['complete'] = false; } $first_link = true; foreach ( $doc['elements'] as $ord => $element ) { foreach ( array( 'text', 'hidden_text' ) as $field ) { if ( isset( $element[ $field ] ) && '' !== $element[ $field ] ) { $doc['elements'][ $ord ][ $field ] = self::clean_text( $element[ $field ] ); } } if ( $doc['skip_link'] || 'A' !== $element['tag'] || ! isset( $element['attrs']['href'] ) || ( null !== $doc['body_ord'] && $ord < $doc['body_ord'] ) ) { continue; } $href = (string) $element['attrs']['href']; if ( 1 < strlen( $href ) && '#' === $href[0] ) { // A skip link in any common language, or the page's first link jumping to an anchor on the page. if ( self::is_skip_text( $doc['elements'][ $ord ]['text'] ) || ( $first_link && isset( $doc['ids'][ substr( $href, 1 ) ] ) ) ) { $doc['skip_link'] = true; } } $first_link = false; } // Pick the skip-link target: <main>, then role="main", then a well-known id. if ( null === $doc['main'] && null !== $first_role_main ) { $doc['main'] = $first_role_main; } if ( null === $doc['main'] ) { foreach ( array( 'wp--skip-link--target', 'main', 'main-content', 'maincontent', 'content', 'primary', 'site-content' ) as $candidate ) { // A skip link to an id used twice may land on the wrong element. if ( isset( $fallback_ids[ $candidate ] ) && 1 === $doc['ids'][ $candidate ] ) { $doc['main'] = array( 'ord' => $fallback_ids[ $candidate ], 'id' => $candidate, 'source' => 'id', ); break; } } } foreach ( $doc['headings'] as $index => $ord ) { $doc['headings'][ $index ] = array( 'ord' => $ord, 'level' => (int) substr( $doc['elements'][ $ord ]['tag'], 1 ), 'text' => $doc['elements'][ $ord ]['text'], ); } return $doc; } /** * Collapses whitespace and drops icon-font glyphs (Unicode private use area), * which screen readers announce as nonsense or not at all. * * @param string $text Text. * @return string */ public static function clean_text( $text ) { $stripped = preg_replace( '/[\x{E000}-\x{F8FF}]/u', '', (string) $text ); if ( null !== $stripped ) { $text = $stripped; // Not valid UTF-8 (a Latin-1 page): it cannot hold these glyphs either. } return trim( self::text_replace( '/\s+/u', ' ', $text ) ); } /** * Whether a name is only numbers or symbols (a share count, "0"): technically * a name, but it says nothing about what the link or button does. * * @param string $text Visible text. * @return bool */ public static function is_weak_name( $text ) { // "1.2k" and "3M" are counts too. return '' !== $text && ! self::text_matches( '/\p{L}/u', self::text_replace( '/(?<=\d)[kmb]\b/iu', '', $text ) ); } /** * Whether link text reads like a skip link. * * @param string $text Link text. * @return bool */ public static function is_skip_text( $text ) { return self::text_matches( '/\b(skip|jump) to\b|\bskip (navigation|nav|links?)\b|saltar (al|a la|el)|ir al contenido|ir a(l)? (contenido|principal)|aller au contenu|passer au contenu|zum inhalt|inhalt springen|vai al contenuto|salta al contenuto|pular para|ir para o conte|naar (de )?inhoud/iu', $text ); } /** * Runs preg_replace() with a pattern ending in the u modifier. Text that is * not valid UTF-8 (a Latin-1 page, for example) makes PCRE fail, so the same * pattern then runs on bytes. * * @param string $pattern Pattern; its last modifier is u. * @param string $replacement Replacement. * @param string $text Text. * @return string */ public static function text_replace( $pattern, $replacement, $text ) { $result = preg_replace( $pattern, $replacement, (string) $text ); if ( null === $result ) { $result = preg_replace( substr( $pattern, 0, -1 ), $replacement, (string) $text ); } return (string) $result; } /** * Runs preg_match() with a pattern ending in the u modifier, on bytes when * the text is not valid UTF-8. * * @param string $pattern Pattern; its last modifier is u. * @param string $text Text. * @return bool */ public static function text_matches( $pattern, $text ) { $result = preg_match( $pattern, (string) $text ); if ( false === $result ) { $result = preg_match( substr( $pattern, 0, -1 ), (string) $text ); } return 1 === $result; } /** * Whether a form control already has an accessible name. * * @param array $element Recorded INPUT, SELECT or TEXTAREA. * @param array $doc Analysis result. * @return bool */ public static function control_has_name( array $element, array $doc ) { $attrs = $element['attrs']; if ( $element['in_label'] ) { return true; } foreach ( array( 'aria-label', 'aria-labelledby', 'title' ) as $attribute ) { if ( isset( $attrs[ $attribute ] ) && is_string( $attrs[ $attribute ] ) && '' !== trim( $attrs[ $attribute ] ) ) { return true; } } return isset( $attrs['id'] ) && is_string( $attrs['id'] ) && isset( $doc['label_for'][ $attrs['id'] ] ); } /** * Whether a form control needs a label at all. * * @param array $element Recorded INPUT, SELECT or TEXTAREA. * @return bool */ public static function control_needs_name( array $element ) { if ( $element['hidden'] ) { return false; } if ( 'INPUT' !== $element['tag'] ) { return true; } $type = isset( $element['attrs']['type'] ) && is_string( $element['attrs']['type'] ) ? strtolower( $element['attrs']['type'] ) : 'text'; return ! in_array( $type, array( 'hidden', 'submit', 'reset', 'button', 'image' ), true ); } /** * Whether a link or button has an accessible name. * * @param array $element Recorded A or BUTTON. * @return bool */ public static function has_accessible_name( array $element ) { if ( '' !== $element['text'] ) { return true; } foreach ( array( 'aria-label', 'aria-labelledby', 'title' ) as $attribute ) { if ( isset( $element['attrs'][ $attribute ] ) && is_string( $element['attrs'][ $attribute ] ) && '' !== trim( $element['attrs'][ $attribute ] ) ) { return true; } } return false; } /** * Whether a link is a real, focusable link (has an href and is not hidden). * * @param array $element Recorded A. * @return bool */ public static function is_live_link( array $element ) { if ( $element['hidden'] || ! isset( $element['attrs']['href'] ) ) { return false; } $role = isset( $element['attrs']['role'] ) ? $element['attrs']['role'] : ''; if ( in_array( $role, array( 'presentation', 'none' ), true ) ) { return false; } return ! ( isset( $element['attrs']['tabindex'] ) && '-1' === $element['attrs']['tabindex'] ); } }