diff options
Diffstat (limited to 'includes/libs/CSSMin.php')
| -rw-r--r-- | includes/libs/CSSMin.php | 300 |
1 files changed, 205 insertions, 95 deletions
diff --git a/includes/libs/CSSMin.php b/includes/libs/CSSMin.php index 4f142fc7..c69e79f5 100644 --- a/includes/libs/CSSMin.php +++ b/includes/libs/CSSMin.php @@ -38,11 +38,13 @@ class CSSMin { * which when base64 encoded will result in a 1/3 increase in size. */ const EMBED_SIZE_LIMIT = 24576; - const URL_REGEX = 'url\(\s*[\'"]?(?P<file>[^\?\)\'"]*)(?P<query>\??[^\)\'"]*)[\'"]?\s*\)'; + const URL_REGEX = 'url\(\s*[\'"]?(?P<file>[^\?\)\'"]*?)(?P<query>\?[^\)\'"]*?|)[\'"]?\s*\)'; + const EMBED_REGEX = '\/\*\s*\@embed\s*\*\/'; + const COMMENT_REGEX = '\/\*.*?\*\/'; /* Protected Static Members */ - /** @var array List of common image files extensions and mime-types */ + /** @var array List of common image files extensions and MIME-types */ protected static $mimeTypes = array( 'gif' => 'image/gif', 'jpe' => 'image/jpeg', @@ -52,6 +54,7 @@ class CSSMin { 'tif' => 'image/tiff', 'tiff' => 'image/tiff', 'xbm' => 'image/x-xbitmap', + 'svg' => 'image/svg+xml', ); /* Static Methods */ @@ -59,23 +62,38 @@ class CSSMin { /** * Gets a list of local file paths which are referenced in a CSS style sheet * + * This function will always return an empty array if the second parameter is not given or null + * for backwards-compatibility. + * * @param string $source CSS data to remap * @param string $path File path where the source was read from (optional) * @return array List of local file references */ public static function getLocalFileReferences( $source, $path = null ) { + if ( $path === null ) { + return array(); + } + + $path = rtrim( $path, '/' ) . '/'; $files = array(); + $rFlags = PREG_OFFSET_CAPTURE | PREG_SET_ORDER; if ( preg_match_all( '/' . self::URL_REGEX . '/', $source, $matches, $rFlags ) ) { foreach ( $matches as $match ) { - $file = ( isset( $path ) - ? rtrim( $path, '/' ) . '/' - : '' ) . "{$match['file'][0]}"; + $url = $match['file'][0]; - // Only proceed if we can access the file - if ( !is_null( $path ) && file_exists( $file ) ) { - $files[] = $file; + // Skip fully-qualified and protocol-relative URLs and data URIs + if ( substr( $url, 0, 2 ) === '//' || parse_url( $url, PHP_URL_SCHEME ) ) { + break; } + + $file = $path . $url; + // Skip non-existent files + if ( file_exists( $file ) ) { + break; + } + + $files[] = $file; } } return $files; @@ -95,7 +113,9 @@ class CSSMin { * instead. If $sizeLimit is false, no limit is enforced. * @return string|bool: Image contents encoded as a data URI or false. */ - public static function encodeImageAsDataURI( $file, $type = null, $sizeLimit = self::EMBED_SIZE_LIMIT ) { + public static function encodeImageAsDataURI( $file, $type = null, + $sizeLimit = self::EMBED_SIZE_LIMIT + ) { if ( $sizeLimit !== false && filesize( $file ) >= $sizeLimit ) { return false; } @@ -115,124 +135,214 @@ class CSSMin { */ public static function getMimeType( $file ) { $realpath = realpath( $file ); - // Try a couple of different ways to get the mime-type of a file, in order of - // preference if ( $realpath && function_exists( 'finfo_file' ) && function_exists( 'finfo_open' ) && defined( 'FILEINFO_MIME_TYPE' ) ) { - // As of PHP 5.3, this is how you get the mime-type of a file; it uses the Fileinfo - // PECL extension return finfo_file( finfo_open( FILEINFO_MIME_TYPE ), $realpath ); - } elseif ( function_exists( 'mime_content_type' ) ) { - // Before this was deprecated in PHP 5.3, this was how you got the mime-type of a file - return mime_content_type( $file ); - } else { - // Worst-case scenario has happened, use the file extension to infer the mime-type - $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) ); - if ( isset( self::$mimeTypes[$ext] ) ) { - return self::$mimeTypes[$ext]; - } } + + // Infer the MIME-type from the file extension + $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) ); + if ( isset( self::$mimeTypes[$ext] ) ) { + return self::$mimeTypes[$ext]; + } + return false; } /** - * Remaps CSS URL paths and automatically embeds data URIs for URL rules - * preceded by an /* @embed * / comment + * Build a CSS 'url()' value for the given URL, quoting parentheses (and other funny characters) + * and escaping quotes as necessary. + * + * See http://www.w3.org/TR/css-syntax-3/#consume-a-url-token + * + * @param string $url URL to process + * @return string 'url()' value, usually just `"url($url)"`, quoted/escaped if necessary + */ + public static function buildUrlValue( $url ) { + // The list below has been crafted to match URLs such as: + // scheme://user@domain:port/~user/fi%20le.png?query=yes&really=y+s + // data:image/png;base64,R0lGODlh/+== + if ( preg_match( '!^[\w\d:@/~.%+;,?&=-]+$!', $url ) ) { + return "url($url)"; + } else { + return 'url("' . strtr( $url, array( '\\' => '\\\\', '"' => '\\"' ) ) . '")'; + } + } + + /** + * Remaps CSS URL paths and automatically embeds data URIs for CSS rules + * or url() values preceded by an / * @embed * / comment. * * @param string $source CSS data to remap * @param string $local File path where the source was read from * @param string $remote URL path to the file - * @param bool $embedData If false, never do any data URI embedding, even if / * @embed * / is found + * @param bool $embedData If false, never do any data URI embedding, + * even if / * @embed * / is found. * @return string Remapped CSS data */ public static function remap( $source, $local, $remote, $embedData = true ) { - $pattern = '/((?P<embed>\s*\/\*\s*\@embed\s*\*\/)(?P<pre>[^\;\}]*))?' . - self::URL_REGEX . '(?P<post>[^;]*)[\;]?/'; - $offset = 0; - while ( preg_match( $pattern, $source, $match, PREG_OFFSET_CAPTURE, $offset ) ) { - // Skip fully-qualified URLs and data URIs - $urlScheme = parse_url( $match['file'][0], PHP_URL_SCHEME ); - if ( $urlScheme ) { - // Move the offset to the end of the match, leaving it alone - $offset = $match[0][1] + strlen( $match[0][0] ); - continue; - } - // URLs with absolute paths like /w/index.php need to be expanded - // to absolute URLs but otherwise left alone - if ( $match['file'][0] !== '' && $match['file'][0][0] === '/' ) { - // Replace the file path with an expanded (possibly protocol-relative) URL - // ...but only if wfExpandUrl() is even available. - // This will not be the case if we're running outside of MW - $lengthIncrease = 0; - if ( function_exists( 'wfExpandUrl' ) ) { - $expanded = wfExpandUrl( $match['file'][0], PROTO_RELATIVE ); - $origLength = strlen( $match['file'][0] ); - $lengthIncrease = strlen( $expanded ) - $origLength; - $source = substr_replace( $source, $expanded, - $match['file'][1], $origLength + // High-level overview: + // * For each CSS rule in $source that includes at least one url() value: + // * Check for an @embed comment at the start indicating that all URIs should be embedded + // * For each url() value: + // * Check for an @embed comment directly preceding the value + // * If either @embed comment exists: + // * Embedding the URL as data: URI, if it's possible / allowed + // * Otherwise remap the URL to work in generated stylesheets + + // Guard against trailing slashes, because "some/remote/../foo.png" + // resolves to "some/remote/foo.png" on (some?) clients (bug 27052). + if ( substr( $remote, -1 ) == '/' ) { + $remote = substr( $remote, 0, -1 ); + } + + // Replace all comments by a placeholder so they will not interfere with the remapping. + // Warning: This will also catch on anything looking like the start of a comment between + // quotation marks (e.g. "foo /* bar"). + $comments = array(); + $placeholder = uniqid( '', true ); + + $pattern = '/(?!' . CSSMin::EMBED_REGEX . ')(' . CSSMin::COMMENT_REGEX . ')/s'; + + $source = preg_replace_callback( + $pattern, + function ( $match ) use ( &$comments, $placeholder ) { + $comments[] = $match[ 0 ]; + return $placeholder . ( count( $comments ) - 1 ) . 'x'; + }, + $source + ); + + // Note: This will not correctly handle cases where ';', '{' or '}' + // appears in the rule itself, e.g. in a quoted string. You are advised + // not to use such characters in file names. We also match start/end of + // the string to be consistent in edge-cases ('@import url(…)'). + $pattern = '/(?:^|[;{])\K[^;{}]*' . CSSMin::URL_REGEX . '[^;}]*(?=[;}]|$)/'; + + $source = preg_replace_callback( + $pattern, + function ( $matchOuter ) use ( $local, $remote, $embedData, $placeholder ) { + $rule = $matchOuter[0]; + + // Check for global @embed comment and remove it. Allow other comments to be present + // before @embed (they have been replaced with placeholders at this point). + $embedAll = false; + $rule = preg_replace( '/^((?:\s+|' . $placeholder . '(\d+)x)*)' . CSSMin::EMBED_REGEX . '\s*/', '$1', $rule, 1, $embedAll ); + + // Build two versions of current rule: with remapped URLs + // and with embedded data: URIs (where possible). + $pattern = '/(?P<embed>' . CSSMin::EMBED_REGEX . '\s*|)' . CSSMin::URL_REGEX . '/'; + + $ruleWithRemapped = preg_replace_callback( + $pattern, + function ( $match ) use ( $local, $remote ) { + $remapped = CSSMin::remapOne( $match['file'], $match['query'], $local, $remote, false ); + + return CSSMin::buildUrlValue( $remapped ); + }, + $rule + ); + + if ( $embedData ) { + $ruleWithEmbedded = preg_replace_callback( + $pattern, + function ( $match ) use ( $embedAll, $local, $remote ) { + $embed = $embedAll || $match['embed']; + $embedded = CSSMin::remapOne( + $match['file'], + $match['query'], + $local, + $remote, + $embed + ); + + return CSSMin::buildUrlValue( $embedded ); + }, + $rule ); } - // Move the offset to the end of the match, leaving it alone - $offset = $match[0][1] + strlen( $match[0][0] ) + $lengthIncrease; - continue; - } - // Guard against double slashes, because "some/remote/../foo.png" - // resolves to "some/remote/foo.png" on (some?) clients (bug 27052). - if ( substr( $remote, -1 ) == '/' ) { - $remote = substr( $remote, 0, -1 ); - } + if ( $embedData && $ruleWithEmbedded !== $ruleWithRemapped ) { + // Build 2 CSS properties; one which uses a base64 encoded data URI in place + // of the @embed comment to try and retain line-number integrity, and the + // other with a remapped an versioned URL and an Internet Explorer hack + // making it ignored in all browsers that support data URIs + return "$ruleWithEmbedded;$ruleWithRemapped!ie"; + } else { + // No reason to repeat twice + return $ruleWithRemapped; + } + }, $source ); - // Shortcuts - $embed = $match['embed'][0]; - $pre = $match['pre'][0]; - $post = $match['post'][0]; - $query = $match['query'][0]; - $url = "{$remote}/{$match['file'][0]}"; - $file = "{$local}/{$match['file'][0]}"; + // Re-insert comments + $pattern = '/' . $placeholder . '(\d+)x/'; + $source = preg_replace_callback( $pattern, function( $match ) use ( &$comments ) { + return $comments[ $match[1] ]; + }, $source ); - $replacement = false; + return $source; - if ( $local !== false && file_exists( $file ) ) { + } + + /** + * Remap or embed a CSS URL path. + * + * @param string $file URL to remap/embed + * @param string $query + * @param string $local File path where the source was read from + * @param string $remote URL path to the file + * @param bool $embed Whether to do any data URI embedding + * @return string Remapped/embedded URL data + */ + public static function remapOne( $file, $query, $local, $remote, $embed ) { + // The full URL possibly with query, as passed to the 'url()' value in CSS + $url = $file . $query; + + // Skip fully-qualified and protocol-relative URLs and data URIs + if ( substr( $url, 0, 2 ) === '//' || parse_url( $url, PHP_URL_SCHEME ) ) { + return $url; + } + + // URLs with absolute paths like /w/index.php need to be expanded + // to absolute URLs but otherwise left alone + if ( $url !== '' && $url[0] === '/' ) { + // Replace the file path with an expanded (possibly protocol-relative) URL + // ...but only if wfExpandUrl() is even available. + // This will not be the case if we're running outside of MW + if ( function_exists( 'wfExpandUrl' ) ) { + return wfExpandUrl( $url, PROTO_RELATIVE ); + } else { + return $url; + } + } + + if ( $local === false ) { + // Assume that all paths are relative to $remote, and make them absolute + return $remote . '/' . $url; + } else { + // We drop the query part here and instead make the path relative to $remote + $url = "{$remote}/{$file}"; + // Path to the actual file on the filesystem + $localFile = "{$local}/{$file}"; + if ( file_exists( $localFile ) ) { // Add version parameter as a time-stamp in ISO 8601 format, // using Z for the timezone, meaning GMT - $url .= '?' . gmdate( 'Y-m-d\TH:i:s\Z', round( filemtime( $file ), -2 ) ); - // Embedding requires a bit of extra processing, so let's skip that if we can - if ( $embedData && $embed && $match['embed'][1] > 0 ) { - $data = self::encodeImageAsDataURI( $file ); + $url .= '?' . gmdate( 'Y-m-d\TH:i:s\Z', round( filemtime( $localFile ), -2 ) ); + if ( $embed ) { + $data = self::encodeImageAsDataURI( $localFile ); if ( $data !== false ) { - // Build 2 CSS properties; one which uses a base64 encoded data URI in place - // of the @embed comment to try and retain line-number integrity, and the - // other with a remapped an versioned URL and an Internet Explorer hack - // making it ignored in all browsers that support data URIs - $replacement = "{$pre}url({$data}){$post};{$pre}url({$url}){$post}!ie;"; + return $data; } } - if ( $replacement === false ) { - // Assume that all paths are relative to $remote, and make them absolute - $replacement = "{$embed}{$pre}url({$url}){$post};"; - } - } elseif ( $local === false ) { - // Assume that all paths are relative to $remote, and make them absolute - $replacement = "{$embed}{$pre}url({$url}{$query}){$post};"; } - if ( $replacement !== false ) { - // Perform replacement on the source - $source = substr_replace( - $source, $replacement, $match[0][1], strlen( $match[0][0] ) - ); - // Move the offset to the end of the replacement in the source - $offset = $match[0][1] + strlen( $replacement ); - continue; - } - // Move the offset to the end of the match, leaving it alone - $offset = $match[0][1] + strlen( $match[0][0] ); + // If any of these conditions failed (file missing, we don't want to embed it + // or it's not embeddable), return the URL (possibly with ?timestamp part) + return $url; } - return $source; } /** |