Html.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Html;
 
use MediaWiki\Json\FormatJson;
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use MediaWiki\Parser\Sanitizer;
use MediaWiki\Request\ContentSecurityPolicy;
use UnexpectedValueException;
 
class Html {
    private static $voidElements = [
        'area' => true,
        'base' => true,
        'br' => true,
        'col' => true,
        'embed' => true,
        'hr' => true,
        'img' => true,
        'input' => true,
        'keygen' => true,
        'link' => true,
        'meta' => true,
        'param' => true,
        'source' => true,
        'track' => true,
        'wbr' => true,
    ];
 
    private static $boolAttribs = [
        'async' => true,
        'autofocus' => true,
        'autoplay' => true,
        'checked' => true,
        'controls' => true,
        'default' => true,
        'defer' => true,
        'disabled' => true,
        'formnovalidate' => true,
        'hidden' => true,
        'ismap' => true,
        'itemscope' => true,
        'loop' => true,
        'multiple' => true,
        'muted' => true,
        'novalidate' => true,
        'open' => true,
        'pubdate' => true,
        'readonly' => true,
        'required' => true,
        'reversed' => true,
        'scoped' => true,
        'seamless' => true,
        'selected' => true,
        'truespeed' => true,
        'typemustmatch' => true,
    ];
 
    public static function buttonAttributes( array $attrs, array $modifiers = [] ) {
        wfDeprecated( __METHOD__, '1.42' );
        return $attrs;
    }
 
    public static function getTextInputAttributes( array $attrs ) {
        wfDeprecated( __METHOD__, '1.42' );
        return $attrs;
    }
 
    public static function linkButton( $text, array $attrs, array $modifiers = [] ) {
        return self::element(
            'a',
            $attrs,
            $text
        );
    }
 
    public static function submitButton( $contents, array $attrs = [], array $modifiers = [] ) {
        $attrs['type'] = 'submit';
        $attrs['value'] = $contents;
        return self::element( 'input', $attrs );
    }
 
    public static function rawElement( $element, $attribs = [], $contents = '' ) {
        $start = self::openElement( $element, $attribs );
        if ( isset( self::$voidElements[$element] ) ) {
            return $start;
        } else {
            $contents = Sanitizer::escapeCombiningChar( $contents ?? '' );
            return $start . $contents . self::closeElement( $element );
        }
    }
 
    public static function element( $element, $attribs = [], $contents = '' ) {
        return self::rawElement(
            $element,
            $attribs,
            strtr( $contents ?? '', [
                // There's no point in escaping quotes, >, etc. in the contents of
                // elements.
                '&' => '&amp;',
                '<' => '&lt;',
            ] )
        );
    }
 
    public static function openElement( $element, $attribs = [] ) {
        $attribs = (array)$attribs;
        // This is not required in HTML5, but let's do it anyway, for
        // consistency and better compression.
        $element = strtolower( $element );
 
        // Some people were abusing this by passing things like
        // 'h1 id="foo" to $element, which we don't want.
        if ( str_contains( $element, ' ' ) ) {
            wfWarn( __METHOD__ . " given element name with space '$element'" );
        }
 
        // Remove invalid input types
        if ( $element == 'input' ) {
            $validTypes = [
                'hidden' => true,
                'text' => true,
                'password' => true,
                'checkbox' => true,
                'radio' => true,
                'file' => true,
                'submit' => true,
                'image' => true,
                'reset' => true,
                'button' => true,
 
                // HTML input types
                'datetime' => true,
                'datetime-local' => true,
                'date' => true,
                'month' => true,
                'time' => true,
                'week' => true,
                'number' => true,
                'range' => true,
                'email' => true,
                'url' => true,
                'search' => true,
                'tel' => true,
                'color' => true,
            ];
            if ( isset( $attribs['type'] ) && !isset( $validTypes[$attribs['type']] ) ) {
                unset( $attribs['type'] );
            }
        }
 
        // According to standard the default type for <button> elements is "submit".
        // Depending on compatibility mode IE might use "button", instead.
        // We enforce the standard "submit".
        if ( $element == 'button' && !isset( $attribs['type'] ) ) {
            $attribs['type'] = 'submit';
        }
 
        return "<$element" . self::expandAttributes(
            self::dropDefaults( $element, $attribs ) ) . '>';
    }
 
    public static function closeElement( $element ) {
        $element = strtolower( $element );
 
        return "</$element>";
    }
 
    private static function dropDefaults( $element, array $attribs ) {
        // Whenever altering this array, please provide a covering test case
        // in HtmlTest::provideElementsWithAttributesHavingDefaultValues
        static $attribDefaults = [
            'area' => [ 'shape' => 'rect' ],
            'button' => [
                'formaction' => 'GET',
                'formenctype' => 'application/x-www-form-urlencoded',
            ],
            'canvas' => [
                'height' => '150',
                'width' => '300',
            ],
            'form' => [
                'action' => 'GET',
                'autocomplete' => 'on',
                'enctype' => 'application/x-www-form-urlencoded',
            ],
            'input' => [
                'formaction' => 'GET',
                'type' => 'text',
            ],
            'keygen' => [ 'keytype' => 'rsa' ],
            'link' => [ 'media' => 'all' ],
            'menu' => [ 'type' => 'list' ],
            'script' => [ 'type' => 'text/javascript' ],
            'style' => [
                'media' => 'all',
                'type' => 'text/css',
            ],
            'textarea' => [ 'wrap' => 'soft' ],
        ];
 
        foreach ( $attribs as $attrib => $value ) {
            if ( $attrib === 'class' ) {
                if ( $value === '' || $value === [] || $value === [ '' ] ) {
                    unset( $attribs[$attrib] );
                }
            } elseif ( isset( $attribDefaults[$element][$attrib] ) ) {
                if ( is_array( $value ) ) {
                    $value = implode( ' ', $value );
                } else {
                    $value = strval( $value );
                }
                if ( $attribDefaults[$element][$attrib] == $value ) {
                    unset( $attribs[$attrib] );
                }
            }
        }
 
        // More subtle checks
        if ( $element === 'link'
            && isset( $attribs['type'] ) && strval( $attribs['type'] ) == 'text/css'
        ) {
            unset( $attribs['type'] );
        }
        if ( $element === 'input' ) {
            $type = $attribs['type'] ?? null;
            $value = $attribs['value'] ?? null;
            if ( $type === 'checkbox' || $type === 'radio' ) {
                // The default value for checkboxes and radio buttons is 'on'
                // not ''. By stripping value="" we break radio boxes that
                // actually wants empty values.
                if ( $value === 'on' ) {
                    unset( $attribs['value'] );
                }
            } elseif ( $type === 'submit' ) {
                // The default value for submit appears to be "Submit" but
                // let's not bother stripping out localized text that matches
                // that.
            } else {
                // The default value for nearly every other field type is ''
                // The 'range' and 'color' types use different defaults but
                // stripping a value="" does not hurt them.
                if ( $value === '' ) {
                    unset( $attribs['value'] );
                }
            }
        }
        if ( $element === 'select' && isset( $attribs['size'] ) ) {
            $multiple = ( $attribs['multiple'] ?? false ) !== false ||
                in_array( 'multiple', $attribs );
            $default = $multiple ? 4 : 1;
            if ( (int)$attribs['size'] === $default ) {
                unset( $attribs['size'] );
            }
        }
 
        return $attribs;
    }
 
    public static function expandAttributes( array $attribs ) {
        $ret = '';
        foreach ( $attribs as $key => $value ) {
            // Support intuitive [ 'checked' => true/false ] form
            if ( $value === false || $value === null ) {
                continue;
            }
 
            // For boolean attributes, support [ 'foo' ] instead of
            // requiring [ 'foo' => 'meaningless' ].
            if ( is_int( $key ) && isset( self::$boolAttribs[strtolower( $value )] ) ) {
                $key = $value;
            }
 
            // Not technically required in HTML5 but we'd like consistency
            // and better compression anyway.
            $key = strtolower( $key );
 
            // https://www.w3.org/TR/html401/index/attributes.html ("space-separated")
            // https://www.w3.org/TR/html5/index.html#attributes-1 ("space-separated")
            $spaceSeparatedListAttributes = [
                'class' => true, // html4, html5
                'accesskey' => true, // as of html5, multiple space-separated values allowed
                // html4-spec doesn't document rel= as space-separated
                // but has been used like that and is now documented as such
                // in the html5-spec.
                'rel' => true,
            ];
 
            // Specific features for attributes that allow a list of space-separated values
            if ( isset( $spaceSeparatedListAttributes[$key] ) ) {
                // Apply some normalization and remove duplicates
 
                // Convert into correct array. Array can contain space-separated
                // values. Implode/explode to get those into the main array as well.
                if ( is_array( $value ) ) {
                    // If input wasn't an array, we can skip this step
                    $arrayValue = [];
                    foreach ( $value as $k => $v ) {
                        if ( is_string( $v ) ) {
                            // String values should be normal `[ 'foo' ]`
                            // Just append them
                            if ( !isset( $value[$v] ) ) {
                                // As a special case don't set 'foo' if a
                                // separate 'foo' => true/false exists in the array
                                // keys should be authoritative
                                foreach ( explode( ' ', $v ) as $part ) {
                                    // Normalize spacing by fixing up cases where people used
                                    // more than 1 space and/or a trailing/leading space
                                    if ( $part !== '' && $part !== ' ' ) {
                                        $arrayValue[] = $part;
                                    }
                                }
                            }
                        } elseif ( $v ) {
                            // If the value is truthy but not a string this is likely
                            // an [ 'foo' => true ], falsy values don't add strings
                            $arrayValue[] = $k;
                        }
                    }
                } else {
                    $arrayValue = explode( ' ', $value );
                    // Normalize spacing by fixing up cases where people used
                    // more than 1 space and/or a trailing/leading space
                    $arrayValue = array_diff( $arrayValue, [ '', ' ' ] );
                }
 
                // Remove duplicates and create the string
                $value = implode( ' ', array_unique( $arrayValue ) );
 
                // Optimization: Skip below boolAttribs check and jump straight
                // to its `else` block. The current $spaceSeparatedListAttributes
                // block is mutually exclusive with $boolAttribs.
                // phpcs:ignore Generic.PHP.DiscourageGoto
                goto not_bool; // NOSONAR
            } elseif ( is_array( $value ) ) {
                throw new UnexpectedValueException( "HTML attribute $key can not contain a list of values" );
            }
 
            if ( isset( self::$boolAttribs[$key] ) ) {
                $ret .= " $key=\"\"";
            } else {
                // phpcs:ignore Generic.PHP.DiscourageGoto
                not_bool:
                // Inlined from Sanitizer::encodeAttribute() for improved performance
                $encValue = htmlspecialchars( $value, ENT_QUOTES );
                // Whitespace is normalized during attribute decoding,
                // so if we've been passed non-spaces we must encode them
                // ahead of time or they won't be preserved.
                $encValue = strtr( $encValue, [
                    "\n" => '&#10;',
                    "\r" => '&#13;',
                    "\t" => '&#9;',
                ] );
                $ret .= " $key=\"$encValue\"";
            }
        }
        return $ret;
    }
 
    public static function inlineScript( $contents, $nonce = null ) {
        if ( preg_match( '/<\/?script/i', $contents ) ) {
            wfLogWarning( __METHOD__ . ': Illegal character sequence found in inline script.' );
            $contents = '/* ERROR: Invalid script */';
        }
 
        return self::rawElement( 'script', [], $contents );
    }
 
    public static function linkedScript( $url, $nonce = null ) {
        $attrs = [ 'src' => $url ];
        if ( $nonce !== null ) {
            $attrs['nonce'] = $nonce;
        } elseif ( ContentSecurityPolicy::isNonceRequired( MediaWikiServices::getInstance()->getMainConfig() ) ) {
            wfWarn( "no nonce set on script. CSP will break it" );
        }
 
        return self::element( 'script', $attrs );
    }
 
    public static function inlineStyle( $contents, $media = 'all', $attribs = [] ) {
        // Don't escape '>' since that is used
        // as direct child selector.
        // Remember, in css, there is no "x" for hexadecimal escapes, and
        // the space immediately after an escape sequence is swallowed.
        $contents = strtr( $contents, [
            '<' => '\3C ',
            // CDATA end tag for good measure, but the main security
            // is from escaping the '<'.
            ']]>' => '\5D\5D\3E '
        ] );
 
        if ( preg_match( '/[<&]/', $contents ) ) {
            $contents = "/*<![CDATA[*/$contents/*]]>*/";
        }
 
        return self::rawElement( 'style', [
            'media' => $media,
        ] + $attribs, $contents );
    }
 
    public static function linkedStyle( $url, $media = 'all' ) {
        return self::element( 'link', [
            'rel' => 'stylesheet',
            'href' => $url,
            'media' => $media,
        ] );
    }
 
    public static function input( $name, $value = '', $type = 'text', array $attribs = [] ) {
        $attribs['type'] = $type;
        $attribs['value'] = $value;
        $attribs['name'] = $name;
        return self::element( 'input', $attribs );
    }
 
    public static function check( $name, $checked = false, array $attribs = [] ) {
        if ( isset( $attribs['value'] ) ) {
            $value = $attribs['value'];
            unset( $attribs['value'] );
        } else {
            $value = 1;
        }
 
        if ( $checked ) {
            $attribs[] = 'checked';
        }
 
        return self::input( $name, $value, 'checkbox', $attribs );
    }
 
    private static function messageBox( $html, $className, $heading = '', $iconClassName = '' ) {
        if ( $heading !== '' ) {
            $html = self::element( 'h2', [], $heading ) . $html;
        }
        $coreClasses = [
            'cdx-message',
            'cdx-message--block'
        ];
        if ( is_array( $className ) ) {
            $className = array_merge(
                $coreClasses,
                $className
            );
        } else {
            $className .= ' ' . implode( ' ', $coreClasses );
        }
        return self::rawElement( 'div', [ 'class' => $className ],
            self::element( 'span', [ 'class' => [
                'cdx-message__icon',
                $iconClassName
            ] ] ) .
            self::rawElement( 'div', [
                'class' => 'cdx-message__content'
            ], $html )
        );
    }
 
    public static function noticeBox( $html, $className, $heading = '', $iconClassName = '' ) {
        return self::messageBox( $html, [
            'cdx-message--notice',
            $className
        ], $heading, $iconClassName );
    }
 
    public static function warningBox( $html, $className = '' ) {
        return self::messageBox( $html, [
            'cdx-message--warning', $className ] );
    }
 
    public static function errorBox( $html, $heading = '', $className = '' ) {
        return self::messageBox( $html, [
            'cdx-message--error', $className ], $heading );
    }
 
    public static function successBox( $html, $className = '' ) {
        return self::messageBox( $html, [
            'cdx-message--success', $className ] );
    }
 
    public static function radio( $name, $checked = false, array $attribs = [] ) {
        if ( isset( $attribs['value'] ) ) {
            $value = $attribs['value'];
            unset( $attribs['value'] );
        } else {
            $value = 1;
        }
 
        if ( $checked ) {
            $attribs[] = 'checked';
        }
 
        return self::input( $name, $value, 'radio', $attribs );
    }
 
    public static function label( $label, $id, array $attribs = [] ) {
        $attribs += [
            'for' => $id,
        ];
        return self::element( 'label', $attribs, $label );
    }
 
    public static function hidden( $name, $value, array $attribs = [] ) {
        return self::input( $name, $value, 'hidden', $attribs );
    }
 
    public static function textarea( $name, $value = '', array $attribs = [] ) {
        $attribs['name'] = $name;
 
        if ( substr( $value ?? '', 0, 1 ) == "\n" ) {
            // Workaround for T14130: browsers eat the initial newline
            // assuming that it's just for show, but they do keep the later
            // newlines, which we may want to preserve during editing.
            // Prepending a single newline
            $spacedValue = "\n" . $value;
        } else {
            $spacedValue = $value;
        }
        return self::element( 'textarea', $attribs, $spacedValue );
    }
 
    public static function namespaceSelectorOptions( array $params = [] ) {
        if ( !isset( $params['exclude'] ) || !is_array( $params['exclude'] ) ) {
            $params['exclude'] = [];
        }
 
        if ( $params['in-user-lang'] ?? false ) {
            global $wgLang;
            $lang = $wgLang;
        } else {
            $lang = MediaWikiServices::getInstance()->getContentLanguage();
        }
 
        $optionsOut = [];
        if ( isset( $params['all'] ) ) {
            // add an option that would let the user select all namespaces.
            // Value is provided by user, the name shown is localized for the user.
            $optionsOut[$params['all']] = wfMessage( 'namespacesall' )->text();
        }
        // Add all namespaces as options
        $options = $lang->getFormattedNamespaces();
        // Filter out namespaces below 0 and massage labels
        foreach ( $options as $nsId => $nsName ) {
            if ( $nsId < NS_MAIN || in_array( $nsId, $params['exclude'] ) ) {
                continue;
            }
            if ( $nsId === NS_MAIN ) {
                // For other namespaces use the namespace prefix as label, but for
                // main we don't use "" but the user message describing it (e.g. "(Main)" or "(Article)")
                $nsName = wfMessage( 'blanknamespace' )->text();
            } elseif ( is_int( $nsId ) ) {
                $converter = MediaWikiServices::getInstance()->getLanguageConverterFactory()
                    ->getLanguageConverter( $lang );
                $nsName = $converter->convertNamespace( $nsId );
            }
            $optionsOut[$nsId] = $nsName;
        }
 
        return $optionsOut;
    }
 
    public static function namespaceSelector(
        array $params = [],
        array $selectAttribs = []
    ) {
        ksort( $selectAttribs );
 
        // Is a namespace selected?
        if ( isset( $params['selected'] ) ) {
            // If string only contains digits, convert to clean int. Selected could also
            // be "all" or "" etc. which needs to be left untouched.
            if ( !is_int( $params['selected'] ) && ctype_digit( (string)$params['selected'] ) ) {
                $params['selected'] = (int)$params['selected'];
            }
            // else: leaves it untouched for later processing
        } else {
            $params['selected'] = '';
        }
 
        if ( !isset( $params['disable'] ) || !is_array( $params['disable'] ) ) {
            $params['disable'] = [];
        }
 
        // Associative array between option-values and option-labels
        $options = self::namespaceSelectorOptions( $params );
 
        // Convert $options to HTML
        $optionsHtml = [];
        foreach ( $options as $nsId => $nsName ) {
            $optionsHtml[] = self::element(
                'option',
                [
                    'disabled' => in_array( $nsId, $params['disable'] ),
                    'value' => $nsId,
                    'selected' => $nsId === $params['selected'],
                ],
                $nsName
            );
        }
 
        $selectAttribs['id'] ??= 'namespace';
        $selectAttribs['name'] ??= 'namespace';
 
        $ret = '';
        if ( isset( $params['label'] ) ) {
            $ret .= self::element(
                'label', [
                    'for' => $selectAttribs['id'],
                ], $params['label']
            ) . "\u{00A0}";
        }
 
        // Wrap options in a <select>
        $ret .= self::openElement( 'select', $selectAttribs )
            . "\n"
            . implode( "\n", $optionsHtml )
            . "\n"
            . self::closeElement( 'select' );
 
        return $ret;
    }
 
    public static function htmlHeader( array $attribs = [] ) {
        $ret = '';
        $mainConfig = MediaWikiServices::getInstance()->getMainConfig();
        $html5Version = $mainConfig->get( MainConfigNames::Html5Version );
        $mimeType = $mainConfig->get( MainConfigNames::MimeType );
        $xhtmlNamespaces = $mainConfig->get( MainConfigNames::XhtmlNamespaces );
 
        $isXHTML = self::isXmlMimeType( $mimeType );
 
        if ( $isXHTML ) { // XHTML5
            // XML MIME-typed markup should have an xml header.
            // However a DOCTYPE is not needed.
            $ret .= "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
 
            // Add the standard xmlns
            $attribs['xmlns'] = 'http://www.w3.org/1999/xhtml';
 
            // And support custom namespaces
            foreach ( $xhtmlNamespaces as $tag => $ns ) {
                $attribs["xmlns:$tag"] = $ns;
            }
        } else { // HTML5
            $ret .= "<!DOCTYPE html>\n";
        }
 
        if ( $html5Version ) {
            $attribs['version'] = $html5Version;
        }
 
        $ret .= self::openElement( 'html', $attribs );
 
        return $ret;
    }
 
    public static function isXmlMimeType( $mimetype ) {
        # https://html.spec.whatwg.org/multipage/infrastructure.html#xml-mime-type
        # * text/xml
        # * application/xml
        # * Any MIME type with a subtype ending in +xml (this implicitly includes application/xhtml+xml)
        return (bool)preg_match( '!^(text|application)/xml$|^.+/.+\+xml$!', $mimetype );
    }
 
    public static function srcSet( array $urls ) {
        $candidates = [];
        foreach ( $urls as $density => $url ) {
            // Cast density to float to strip 'x', then back to string to serve
            // as array index.
            $density = (string)(float)$density;
            $candidates[$density] = $url;
        }
 
        // Remove duplicates that are the same as a smaller value
        ksort( $candidates, SORT_NUMERIC );
        $candidates = array_unique( $candidates );
 
        // Append density info to the url
        foreach ( $candidates as $density => $url ) {
            $candidates[$density] = $url . ' ' . $density . 'x';
        }
 
        return implode( ", ", $candidates );
    }
 
    public static function encodeJsVar( $value, $pretty = false ) {
        if ( $value instanceof HtmlJsCode ) {
            return $value->value;
        }
        return FormatJson::encode( $value, $pretty, FormatJson::UTF8_OK );
    }
 
    public static function encodeJsCall( $name, $args, $pretty = false ) {
        $encodedArgs = self::encodeJsList( $args, $pretty );
        if ( $encodedArgs === false ) {
            return false;
        }
        return "$name($encodedArgs);";
    }
 
    public static function encodeJsList( $args, $pretty = false ) {
        foreach ( $args as &$arg ) {
            $arg = self::encodeJsVar( $arg, $pretty );
            if ( $arg === false ) {
                return false;
            }
        }
        if ( $pretty ) {
            return ' ' . implode( ', ', $args ) . ' ';
        } else {
            return implode( ',', $args );
        }
    }
 
    public static function listDropdownOptions( $list, $params = [] ) {
        $options = [];
 
        if ( isset( $params['other'] ) ) {
            $options[ $params['other'] ] = 'other';
        }
 
        $optgroup = false;
        foreach ( explode( "\n", $list ) as $option ) {
            $value = trim( $option );
            if ( $value == '' ) {
                continue;
            }
            if ( substr( $value, 0, 1 ) == '*' && substr( $value, 1, 1 ) != '*' ) {
                # A new group is starting...
                $value = trim( substr( $value, 1 ) );
                if ( $value !== '' &&
                    // Do not use the value for 'other' as option group - T251351
                    ( !isset( $params['other'] ) || $value !== $params['other'] )
                ) {
                    $optgroup = $value;
                } else {
                    $optgroup = false;
                }
            } elseif ( substr( $value, 0, 2 ) == '**' ) {
                # groupmember
                $opt = trim( substr( $value, 2 ) );
                if ( $optgroup === false ) {
                    $options[$opt] = $opt;
                } else {
                    $options[$optgroup][$opt] = $opt;
                }
            } else {
                # groupless reason list
                $optgroup = false;
                $options[$option] = $option;
            }
        }
 
        return $options;
    }
 
    public static function listDropdownOptionsOoui( $options ) {
        $optionsOoui = [];
 
        foreach ( $options as $text => $value ) {
            if ( is_array( $value ) ) {
                $optionsOoui[] = [ 'optgroup' => (string)$text ];
                foreach ( $value as $text2 => $value2 ) {
                    $optionsOoui[] = [ 'data' => (string)$value2, 'label' => (string)$text2 ];
                }
            } else {
                $optionsOoui[] = [ 'data' => (string)$value, 'label' => (string)$text ];
            }
        }
 
        return $optionsOoui;
    }
 
    public static function listDropdownOptionsCodex( $options ) {
        $optionsCodex = [];
 
        foreach ( $options as $text => $value ) {
            if ( is_array( $value ) ) {
                // No support for optgroups in Codex yet (T367241)
                $optionsCodex[] = [ 'label' => (string)$text, 'value' => '', 'disabled' => true ];
                foreach ( $value as $text2 => $value2 ) {
                    $optionsCodex[] = [ 'label' => (string)$text2, 'value' => (string)$value2 ];
                }
            } else {
                $optionsCodex[] = [ 'label' => (string)$text, 'value' => (string)$value ];
            }
        }
        return $optionsCodex;
    }
}
 
class_alias( Html::class, 'Html' );