X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FHtml.php;h=fb3f16789ae7217f8322faf08af85635fd2a1def;hb=957c3b5b683bcb27c5ae1bfcfec3985530698210;hp=075cb3b8b690fb043e89862553b4609889fffa86;hpb=f61d9e089d691203543a523d29e43b216caad184;p=lhc%2Fweb%2Fwiklou.git diff --git a/includes/Html.php b/includes/Html.php index 075cb3b8b6..fb3f16789a 100644 --- a/includes/Html.php +++ b/includes/Html.php @@ -1,26 +1,32 @@ elements. + * + * @since 1.16 */ class Html { - # List of void elements from HTML 5, section 9.1.2 as of 2009-08-10 + # List of void elements from HTML5, section 8.1.2 as of 2011-08-12 private static $voidElements = array( 'area', 'base', @@ -56,116 +64,99 @@ class Html { 'meta', 'param', 'source', + 'track', + 'wbr', ); # Boolean attributes, which may have the value omitted entirely. Manually - # collected from the HTML 5 spec as of 2009-08-10. + # collected from the HTML5 spec as of 2011-08-12. private static $boolAttribs = array( 'async', - 'autobuffer', 'autofocus', 'autoplay', 'checked', 'controls', + 'default', 'defer', 'disabled', 'formnovalidate', 'hidden', 'ismap', + 'itemscope', 'loop', 'multiple', + 'muted', 'novalidate', 'open', + 'pubdate', 'readonly', 'required', 'reversed', 'scoped', 'seamless', + 'selected', + 'truespeed', + 'typemustmatch', + # HTML5 Microdata + 'itemscope', + ); + + private static $HTMLFiveOnlyAttribs = array( + 'autocomplete', + 'autofocus', + 'max', + 'min', + 'multiple', + 'pattern', + 'placeholder', + 'required', + 'step', + 'spellcheck', ); /** * Returns an HTML element in a string. The major advantage here over * manually typing out the HTML is that it will escape all attribute * values. If you're hardcoding all the attributes, or there are none, you - * should probably type out the string yourself. + * should probably just type out the html element yourself. * * This is quite similar to Xml::tags(), but it implements some useful * HTML-specific logic. For instance, there is no $allowShortTag * parameter: the closing tag is magically omitted if $element has an empty * content model. If $wgWellFormedXml is false, then a few bytes will be - * shaved off the HTML output as well. In the future, other HTML-specific - * features might be added, like allowing arrays for the values of - * attributes like class= and media=. + * shaved off the HTML output as well. * - * @param $element string The element's name, e.g., 'a' - * @param $attribs array Associative array of attributes, e.g., array( - * 'href' => 'http://www.mediawiki.org/' ). Values will be HTML-escaped. + * @param $element string The element's name, e.g., 'a' + * @param $attribs array Associative array of attributes, e.g., array( + * 'href' => 'http://www.mediawiki.org/' ). See expandAttributes() for + * further documentation. * @param $contents string The raw HTML contents of the element: *not* * escaped! * @return string Raw HTML */ public static function rawElement( $element, $attribs = array(), $contents = '' ) { - global $wgHtml5, $wgWellFormedXml; - # This is not required in HTML 5, but let's do it anyway, for - # consistency and better compression. - $element = strtolower( $element ); - - # Element-specific hacks to slim down output and ensure validity - if ( $element == 'input' ) { - if ( !$wgHtml5 ) { - # With $wgHtml5 off we want to validate as XHTML 1, so we - # strip out any fancy HTML 5-only input types for now. - # - # Whitelist of valid types: - $validTypes = array( - 'hidden', - 'text', - 'password', - 'checkbox', - 'radio', - 'file', - 'submit', - 'image', - 'reset', - 'button', - ); - if ( isset( $attribs['type'] ) - && !in_array( $attribs['type'], $validTypes ) ) { - # Fall back to type=text, the default - unset( $attribs['type'] ); - } - # Here we're blacklisting some HTML5-only attributes... - $html5attribs = array( - 'autocomplete', - 'autofocus', - 'max', - 'min', - 'multiple', - 'pattern', - 'placeholder', - 'required', - 'step', - ); - foreach ( $html5attribs as $badAttr ) { - unset( $attribs[$badAttr] ); - } - } - } - - $start = "<$element" . self::expandAttributes( $attribs ); + global $wgWellFormedXml; + $start = self::openElement( $element, $attribs ); if ( in_array( $element, self::$voidElements ) ) { if ( $wgWellFormedXml ) { - return "$start />"; + # Silly XML. + return substr( $start, 0, -1 ) . ' />'; } - return "$start>"; + return $start; } else { - return "$start>$contents"; + return "$start$contents" . self::closeElement( $element ); } } /** * Identical to rawElement(), but HTML-escapes $contents (like * Xml::element()). + * + * @param $element string + * @param $attribs array + * @param $contents string + * + * @return string */ public static function element( $element, $attribs = array(), $contents = '' ) { return self::rawElement( $element, $attribs, strtr( $contents, array( @@ -176,6 +167,201 @@ class Html { ) ) ); } + /** + * Identical to rawElement(), but has no third parameter and omits the end + * tag (and the self-closing '/' in XML mode for empty elements). + * + * @param $element string + * @param $attribs array + * + * @return string + */ + public static function openElement( $element, $attribs = array() ) { + global $wgHtml5, $wgWellFormedXml; + $attribs = (array)$attribs; + # This is not required in HTML5, but let's do it anyway, for + # consistency and better compression. + $element = strtolower( $element ); + + # In text/html, initial and tags can be omitted under + # pretty much any sane circumstances, if they have no attributes. See: + # + if ( !$wgWellFormedXml && !$attribs + && in_array( $element, array( 'html', 'head' ) ) ) { + return ''; + } + + # Remove HTML5-only attributes if we aren't doing HTML5, and disable + # form validation regardless (see bug 23769 and the more detailed + # comment in expandAttributes()) + if ( $element == 'input' ) { + # Whitelist of types that don't cause validation. All except + # 'search' are valid in XHTML1. + $validTypes = array( + 'hidden', + 'text', + 'password', + 'checkbox', + 'radio', + 'file', + 'submit', + 'image', + 'reset', + 'button', + 'search', + ); + + if ( isset( $attribs['type'] ) + && !in_array( $attribs['type'], $validTypes ) ) { + unset( $attribs['type'] ); + } + + if ( isset( $attribs['type'] ) && $attribs['type'] == 'search' + && !$wgHtml5 ) { + unset( $attribs['type'] ); + } + } + + if ( !$wgHtml5 && $element == 'textarea' && isset( $attribs['maxlength'] ) ) { + unset( $attribs['maxlength'] ); + } + + return "<$element" . self::expandAttributes( + self::dropDefaults( $element, $attribs ) ) . '>'; + } + + /** + * Returns "", except if $wgWellFormedXml is off, in which case + * it returns the empty string when that's guaranteed to be safe. + * + * @since 1.17 + * @param $element string Name of the element, e.g., 'a' + * @return string A closing tag, if required + */ + public static function closeElement( $element ) { + global $wgWellFormedXml; + + $element = strtolower( $element ); + + # Reference: + # http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#optional-tags + if ( !$wgWellFormedXml && in_array( $element, array( + 'html', + 'head', + 'body', + 'li', + 'dt', + 'dd', + 'tr', + 'td', + 'th', + ) ) ) { + return ''; + } + return ""; + } + + /** + * Given an element name and an associative array of element attributes, + * return an array that is functionally identical to the input array, but + * possibly smaller. In particular, attributes might be stripped if they + * are given their default values. + * + * This method is not guaranteed to remove all redundant attributes, only + * some common ones and some others selected arbitrarily at random. It + * only guarantees that the output array should be functionally identical + * to the input array (currently per the HTML 5 draft as of 2009-09-06). + * + * @param $element string Name of the element, e.g., 'a' + * @param $attribs array Associative array of attributes, e.g., array( + * 'href' => 'http://www.mediawiki.org/' ). See expandAttributes() for + * further documentation. + * @return array An array of attributes functionally identical to $attribs + */ + private static function dropDefaults( $element, $attribs ) { + # Don't bother doing anything if we aren't outputting HTML5; it's too + # much of a pain to maintain two sets of defaults. + global $wgHtml5; + if ( !$wgHtml5 ) { + return $attribs; + } + + static $attribDefaults = array( + 'area' => array( 'shape' => 'rect' ), + 'button' => array( + 'formaction' => 'GET', + 'formenctype' => 'application/x-www-form-urlencoded', + 'type' => 'submit', + ), + 'canvas' => array( + 'height' => '150', + 'width' => '300', + ), + 'command' => array( 'type' => 'command' ), + 'form' => array( + 'action' => 'GET', + 'autocomplete' => 'on', + 'enctype' => 'application/x-www-form-urlencoded', + ), + 'input' => array( + 'formaction' => 'GET', + 'type' => 'text', + 'value' => '', + ), + 'keygen' => array( 'keytype' => 'rsa' ), + 'link' => array( 'media' => 'all' ), + 'menu' => array( 'type' => 'list' ), + # Note: the use of text/javascript here instead of other JavaScript + # MIME types follows the HTML5 spec. + 'script' => array( 'type' => 'text/javascript' ), + 'style' => array( + 'media' => 'all', + 'type' => 'text/css', + ), + 'textarea' => array( 'wrap' => 'soft' ), + ); + + $element = strtolower( $element ); + + foreach ( $attribs as $attrib => $value ) { + $lcattrib = strtolower( $attrib ); + $value = strval( $value ); + + # Simple checks using $attribDefaults + if ( isset( $attribDefaults[$element][$lcattrib] ) && + $attribDefaults[$element][$lcattrib] == $value ) { + unset( $attribs[$attrib] ); + } + + if ( $lcattrib == 'class' && $value == '' ) { + unset( $attribs[$attrib] ); + } + } + + # More subtle checks + if ( $element === 'link' && isset( $attribs['type'] ) + && strval( $attribs['type'] ) == 'text/css' ) { + unset( $attribs['type'] ); + } + if ( $element === 'select' && isset( $attribs['size'] ) ) { + if ( in_array( 'multiple', $attribs ) + || ( isset( $attribs['multiple'] ) && $attribs['multiple'] !== false ) + ) { + # A multi-select + if ( strval( $attribs['size'] ) == '4' ) { + unset( $attribs['size'] ); + } + } else { + # Single select + if ( strval( $attribs['size'] ) == '1' ) { + unset( $attribs['size'] ); + } + } + } + + return $attribs; + } + /** * Given an associative array of element attributes, generate a string * to stick after the element name in HTML output. Like array( 'href' => @@ -185,8 +371,31 @@ class Html { * For instance, it will omit quotation marks if $wgWellFormedXml is false, * and will treat boolean attributes specially. * + * Attributes that should contain space-separated lists (such as 'class') array + * values are allowed as well, which will automagically be normalized + * and converted to a space-separated string. In addition to a numerical + * array, the attribute value may also be an associative array. See the + * example below for how that works. + * @example Numerical array + * + * Html::element( 'em', array( + * 'class' => array( 'foo', 'bar' ) + * ) ); + * // gives '' + * + * @example Associative array + * + * Html::element( 'em', array( + * 'class' => array( 'foo', 'bar', 'foo' => false, 'quux' => true ) + * ) ); + * // gives '' + * + * * @param $attribs array Associative array of attributes, e.g., array( * 'href' => 'http://www.mediawiki.org/' ). Values will be HTML-escaped. + * A value of false means to omit the attribute. For boolean attributes, + * you can omit the key, e.g., array( 'checked' ) instead of + * array( 'checked' => 'checked' ) or such. * @return string HTML fragment that goes between element name and '>' * (starting with a space if at least one attribute is output) */ @@ -194,7 +403,12 @@ class Html { global $wgHtml5, $wgWellFormedXml; $ret = ''; + $attribs = (array)$attribs; foreach ( $attribs as $key => $value ) { + if ( $value === false || is_null( $value ) ) { + continue; + } + # For boolean attributes, support array( 'foo' ) instead of # requiring array( 'foo' => 'meaningless' ). if ( is_int( $key ) @@ -202,17 +416,88 @@ class Html { $key = $value; } - # Not technically required in HTML 5, but required in XHTML 1.0, + # Not technically required in HTML5, but required in XHTML 1.0, # and we'd like consistency and better compression anyway. $key = strtolower( $key ); - # See the "Attributes" section in the HTML syntax part of HTML 5, + # Here we're blacklisting some HTML5-only attributes... + if ( !$wgHtml5 && in_array( $key, self::$HTMLFiveOnlyAttribs ) + ) { + continue; + } + + # Bug 23769: Blacklist all form validation attributes for now. Current + # (June 2010) WebKit has no UI, so the form just refuses to submit + # without telling the user why, which is much worse than failing + # server-side validation. Opera is the only other implementation at + # this time, and has ugly UI, so just kill the feature entirely until + # we have at least one good implementation. + if ( in_array( $key, array( 'max', 'min', 'pattern', 'required', 'step' ) ) ) { + continue; + } + + // http://www.w3.org/TR/html401/index/attributes.html ("space-separated") + // http://www.w3.org/TR/html5/index.html#attributes-1 ("space-separated") + $spaceSeparatedListAttributes = array( + 'class', // html4, html5 + 'accesskey', // 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', + ); + + # Specific features for attributes that allow a list of space-separated values + if ( in_array( $key, $spaceSeparatedListAttributes ) ) { + // Apply some normalization and remove duplicates + + // Convert into correct array. Array can contain space-seperated + // 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 + + $newValue = array(); + foreach ( $value as $k => $v ) { + if ( is_string( $v ) ) { + // String values should be normal `array( '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 authoritive + $newValue[] = $v; + } + } elseif ( $v ) { + // If the value is truthy but not a string this is likely + // an array( 'foo' => true ), falsy values don't add strings + $newValue[] = $k; + } + } + $value = implode( ' ', $newValue ); + } + $value = explode( ' ', $value ); + + // Normalize spacing by fixing up cases where people used + // more than 1 space and/or a trailing/leading space + $value = array_diff( $value, array( '', ' ' ) ); + + // Remove duplicates and create the string + $value = implode( ' ', array_unique( $value ) ); + } + + # See the "Attributes" section in the HTML syntax part of HTML5, # 9.1.2.3 as of 2009-08-10. Most attributes can have quotation # marks omitted, but not all. (Although a literal " is not # permitted, we don't check for that, since it will be escaped # anyway.) - if ( $wgWellFormedXml || $value == '' - || preg_match( "/[ '=<>]/", $value ) ) { + # + # See also research done on further characters that need to be + # escaped: http://code.google.com/p/html5lib/issues/detail?id=93 + $badChars = "\\x00- '=<>`/\x{00a0}\x{1680}\x{180e}\x{180F}\x{2000}\x{2001}" + . "\x{2002}\x{2003}\x{2004}\x{2005}\x{2006}\x{2007}\x{2008}\x{2009}" + . "\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}"; + if ( $wgWellFormedXml || $value === '' + || preg_match( "![$badChars]!u", $value ) ) { $quote = '"'; } else { $quote = ''; @@ -220,7 +505,7 @@ class Html { if ( in_array( $key, self::$boolAttribs ) ) { # In XHTML 1.0 Transitional, the value needs to be equal to the - # key. In HTML 5, we can leave the value empty instead. If we + # key. In HTML5, we can leave the value empty instead. If we # don't need well-formed XML, we can omit the = entirely. if ( !$wgWellFormedXml ) { $ret .= " $key"; @@ -233,15 +518,28 @@ class Html { # Apparently we need to entity-encode \n, \r, \t, although the # spec doesn't mention that. Since we're doing strtr() anyway, # and we don't need <> escaped here, we may as well not call - # htmlspecialchars(). FIXME: verify that we actually need to + # htmlspecialchars(). + # @todo FIXME: Verify that we actually need to # escape \n\r\t here, and explain why, exactly. - $ret .= " $key=$quote" . strtr( $value, array( + # + # We could call Sanitizer::encodeAttribute() for this, but we + # don't because we're stubborn and like our marginal savings on + # byte size from not having to encode unnecessary quotes. + $map = array( '&' => '&', '"' => '"', "\n" => ' ', "\r" => ' ', "\t" => ' ' - ) ) . $quote; + ); + if ( $wgWellFormedXml ) { + # This is allowed per spec: + # But reportedly it breaks some XML tools? + # @todo FIXME: Is this really true? + $map['<'] = '<'; + } + + $ret .= " $key=$quote" . strtr( $value, $map ) . $quote; } } return $ret; @@ -256,13 +554,18 @@ class Html { * @return string Raw HTML */ public static function inlineScript( $contents ) { - global $wgHtml5, $wgJsMimeType; + global $wgHtml5, $wgJsMimeType, $wgWellFormedXml; $attrs = array(); + if ( !$wgHtml5 ) { $attrs['type'] = $wgJsMimeType; + } + + if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) { $contents = "/**/"; } + return self::rawElement( 'script', $attrs, $contents ); } @@ -277,9 +580,11 @@ class Html { global $wgHtml5, $wgJsMimeType; $attrs = array( 'src' => $url ); + if ( !$wgHtml5 ) { $attrs['type'] = $wgJsMimeType; } + return self::element( 'script', $attrs ); } @@ -289,24 +594,20 @@ class Html { * contains literal '' (admittedly unlikely). * * @param $contents string CSS - * @param $media mixed A media type string, like 'screen', or null for all - * media + * @param $media mixed A media type string, like 'screen' * @return string Raw HTML */ - public static function inlineStyle( $contents, $media = null ) { - global $wgHtml5; + public static function inlineStyle( $contents, $media = 'all' ) { + global $wgWellFormedXml; - $attrs = array(); - if ( !$wgHtml5 ) { - # Technically we should probably add CDATA stuff here like with - # scripts, but in practice, stylesheets tend not to have - # problematic characters anyway. - $attrs['type'] = 'text/css'; - } - if ( $media !== null ) { - $attrs['media'] = $media; + if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) { + $contents = "/**/"; } - return self::rawElement( 'style', $attrs, $contents ); + + return self::rawElement( 'style', array( + 'type' => 'text/css', + 'media' => $media, + ), $contents ); } /** @@ -314,44 +615,192 @@ class Html { * media type (if any). * * @param $url string - * @param $media mixed A media type string, like 'screen', or null for all - * media + * @param $media mixed A media type string, like 'screen' * @return string Raw HTML */ - public static function linkedStyle( $url, $media = null ) { - global $wgHtml5; - - $attrs = array( 'rel' => 'stylesheet', 'href' => $url ); - if ( !$wgHtml5 ) { - $attrs['type'] = 'text/css'; - } - if ( $media !== null ) { - $attrs['media'] = $media; - } - return self::element( 'link', $attrs ); + public static function linkedStyle( $url, $media = 'all' ) { + return self::element( 'link', array( + 'rel' => 'stylesheet', + 'href' => $url, + 'type' => 'text/css', + 'media' => $media, + ) ); } /** * Convenience function to produce an element. This supports the - * new HTML 5 input types and attributes, and will silently strip them if + * new HTML5 input types and attributes, and will silently strip them if * $wgHtml5 is false. * * @param $name string name attribute - * @param $value mixed value attribute (null = omit) + * @param $value mixed value attribute * @param $type string type attribute - * @param $attribs array Assocative array of miscellaneous extra attributes, - * passed to Html::element() + * @param $attribs array Associative array of miscellaneous extra + * attributes, passed to Html::element() * @return string Raw HTML */ - public static function input( $name, $value = null, $type = 'text', $attribs = array() ) { - if ( $type != 'text' ) { - $attribs['type'] = $type; - } - if ( $value !== null && $value !== '' ) { - $attribs['value'] = $value; - } + public static function input( $name, $value = '', $type = 'text', $attribs = array() ) { + $attribs['type'] = $type; + $attribs['value'] = $value; $attribs['name'] = $name; return self::element( 'input', $attribs ); } + + /** + * Convenience function to produce an input element with type=hidden + * + * @param $name string name attribute + * @param $value string value attribute + * @param $attribs array Associative array of miscellaneous extra + * attributes, passed to Html::element() + * @return string Raw HTML + */ + public static function hidden( $name, $value, $attribs = array() ) { + return self::input( $name, $value, 'hidden', $attribs ); + } + + /** + * Convenience function to produce an element. This supports leaving + * out the cols= and rows= which Xml requires and are required by HTML4/XHTML + * but not required by HTML5 and will silently set cols="" and rows="" if + * $wgHtml5 is false and cols and rows are omitted (HTML4 validates present + * but empty cols="" and rows="" as valid). + * + * @param $name string name attribute + * @param $value string value attribute + * @param $attribs array Associative array of miscellaneous extra + * attributes, passed to Html::element() + * @return string Raw HTML + */ + public static function textarea( $name, $value = '', $attribs = array() ) { + global $wgHtml5; + + $attribs['name'] = $name; + + if ( !$wgHtml5 ) { + if ( !isset( $attribs['cols'] ) ) { + $attribs['cols'] = ""; + } + + if ( !isset( $attribs['rows'] ) ) { + $attribs['rows'] = ""; + } + } + + if (substr($value, 0, 1) == "\n") { + // Workaround for bug 12130: 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 ); + } + + /** + * Constructs the opening html-tag with necessary doctypes depending on + * global variables. + * + * @param $attribs array Associative array of miscellaneous extra + * attributes, passed to Html::element() of html tag. + * @return string Raw HTML + */ + public static function htmlHeader( $attribs = array() ) { + $ret = ''; + + global $wgMimeType; + + if ( self::isXmlMimeType( $wgMimeType ) ) { + $ret .= "\n"; + } + + global $wgHtml5, $wgHtml5Version, $wgDocType, $wgDTD; + global $wgXhtmlNamespaces, $wgXhtmlDefaultNamespace; + + if ( $wgHtml5 ) { + $ret .= "\n"; + + if ( $wgHtml5Version ) { + $attribs['version'] = $wgHtml5Version; + } + } else { + $ret .= "\n"; + $attribs['xmlns'] = $wgXhtmlDefaultNamespace; + + foreach ( $wgXhtmlNamespaces as $tag => $ns ) { + $attribs["xmlns:$tag"] = $ns; + } + } + + $html = Html::openElement( 'html', $attribs ); + + if ( $html ) { + $html .= "\n"; + } + + $ret .= $html; + + return $ret; + } + + /** + * Determines if the given mime type is xml. + * + * @param $mimetype string MimeType + * @return Boolean + */ + public static function isXmlMimeType( $mimetype ) { + switch ( $mimetype ) { + case 'text/xml': + case 'application/xhtml+xml': + case 'application/xml': + return true; + default: + return false; + } + } + + /** + * Get HTML for an info box with an icon. + * + * @param $text String: wikitext, get this with wfMsgNoTrans() + * @param $icon String: icon name, file in skins/common/images + * @param $alt String: alternate text for the icon + * @param $class String: additional class name to add to the wrapper div + * @param $useStylePath + * + * @return string + */ + static function infoBox( $text, $icon, $alt, $class = false, $useStylePath = true ) { + global $wgStylePath; + + if ( $useStylePath ) { + $icon = $wgStylePath.'/common/images/'.$icon; + } + + $s = Html::openElement( 'div', array( 'class' => "mw-infobox $class") ); + + $s .= Html::openElement( 'div', array( 'class' => 'mw-infobox-left' ) ). + Html::element( 'img', + array( + 'src' => $icon, + 'alt' => $alt, + ) + ). + Html::closeElement( 'div' ); + + $s .= Html::openElement( 'div', array( 'class' => 'mw-infobox-right' ) ). + $text. + Html::closeElement( 'div' ); + $s .= Html::element( 'div', array( 'style' => 'clear: left;' ), ' ' ); + + $s .= Html::closeElement( 'div' ); + + $s .= Html::element( 'div', array( 'style' => 'clear: left;' ), ' ' ); + + return $s; + } }