includes/libs/CSSMin.php

   1 <?php
   2 /**
   3  * Minification of CSS stylesheets.
   4  *
   5  * Copyright 2010 Wikimedia Foundation
   6  *
   7  * Licensed under the Apache License, Version 2.0 (the "License"); you may
   8  * not use this file except in compliance with the License.
   9  * You may obtain a copy of the License at
  10  *
  11  *              http://www.apache.org/licenses/LICENSE-2.0
  12  *
  13  * Unless required by applicable law or agreed to in writing, software distributed
  14  * under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS
  15  * OF ANY KIND, either express or implied. See the License for the
  16  * specific language governing permissions and limitations under the License.
  17  *
  18  * @file
  19  * @version 0.1.1 -- 2010-09-11
  20  * @author Trevor Parscal <tparscal@wikimedia.org>
  21  * @copyright Copyright 2010 Wikimedia Foundation
  22  * @license http://www.apache.org/licenses/LICENSE-2.0
  23  */
  24
  25 /**
  26  * Transforms CSS data
  27  *
  28  * This class provides minification, URL remapping, URL extracting, and data-URL embedding.
  29  */
  30 class CSSMin {
  31
  32         /* Constants */
  33
  34         /** @var string Strip marker for comments. **/
  35         const PLACEHOLDER = "\x7fPLACEHOLDER\x7f";
  36
  37         /**
  38          * Internet Explorer data URI length limit. See encodeImageAsDataURI().
  39          */
  40         const DATA_URI_SIZE_LIMIT = 32768;
  41         const URL_REGEX = 'url\(\s*[\'"]?(?P<file>[^\?\)\'"]*?)(?P<query>\?[^\)\'"]*?|)[\'"]?\s*\)';
  42         const EMBED_REGEX = '\/\*\s*\@embed\s*\*\/';
  43         const COMMENT_REGEX = '\/\*.*?\*\/';
  44
  45         /* Protected Static Members */
  46
  47         /** @var array List of common image files extensions and MIME-types */
  48         protected static $mimeTypes = array(
  49                 'gif' => 'image/gif',
  50                 'jpe' => 'image/jpeg',
  51                 'jpeg' => 'image/jpeg',
  52                 'jpg' => 'image/jpeg',
  53                 'png' => 'image/png',
  54                 'tif' => 'image/tiff',
  55                 'tiff' => 'image/tiff',
  56                 'xbm' => 'image/x-xbitmap',
  57                 'svg' => 'image/svg+xml',
  58         );
  59
  60         /* Static Methods */
  61
  62         /**
  63          * Gets a list of local file paths which are referenced in a CSS style sheet.
  64          *
  65          * If you wish non-existent files to be listed too, use getAllLocalFileReferences().
  66          *
  67          * For backwards-compatibility, if the second parameter is not given or null,
  68          * this function will return an empty array instead of erroring out.
  69          *
  70          * @param string $source CSS stylesheet source to process
  71          * @param string $path File path where the source was read from
  72          * @return array List of local file references
  73          */
  74         public static function getLocalFileReferences( $source, $path = null ) {
  75                 if ( $path === null ) {
  76                         return array();
  77                 }
  78
  79                 $files = self::getAllLocalFileReferences( $source, $path );
  80
  81                 // Skip non-existent files
  82                 $files = array_filter( $files, function ( $file ) {
  83                         return file_exists( $file );
  84                 } );
  85
  86                 return $files;
  87         }
  88
  89         /**
  90          * Gets a list of local file paths which are referenced in a CSS style sheet, including
  91          * non-existent files.
  92          *
  93          * @param string $source CSS stylesheet source to process
  94          * @param string $path File path where the source wa
  95          * @return array List of local file references
  96          */
  97         public static function getAllLocalFileReferences( $source, $path ) {
  98                 $path = rtrim( $path, '/' ) . '/';
  99                 $files = array();
 100
 101                 $rFlags = PREG_OFFSET_CAPTURE | PREG_SET_ORDER;
 102                 if ( preg_match_all( '/' . self::URL_REGEX . '/', $source, $matches, $rFlags ) ) {
 103                         foreach ( $matches as $match ) {
 104                                 $url = $match['file'][0];
 105
 106                                 // Skip fully-qualified and protocol-relative URLs and data URIs
 107                                 if ( substr( $url, 0, 2 ) === '//' || parse_url( $url, PHP_URL_SCHEME ) ) {
 108                                         break;
 109                                 }
 110
 111                                 $files[] = $path . $url;
 112                         }
 113                 }
 114                 return $files;
 115         }
 116
 117         /**
 118          * Encode an image file as a data URI.
 119          *
 120          * If the image file has a suitable MIME type and size, encode it as a data URI, base64-encoded
 121          * for binary files or just percent-encoded otherwise. Return false if the image type is
 122          * unfamiliar or file exceeds the size limit.
 123          *
 124          * @param string $file Image file to encode.
 125          * @param string|null $type File's MIME type or null. If null, CSSMin will
 126          *     try to autodetect the type.
 127          * @param bool $ie8Compat By default, a data URI will only be produced if it can be made short
 128          *     enough to fit in Internet Explorer 8 (and earlier) URI length limit (32,768 bytes). Pass
 129          *     `false` to remove this limitation.
 130          * @return string|bool Image contents encoded as a data URI or false.
 131          */
 132         public static function encodeImageAsDataURI( $file, $type = null, $ie8Compat = true ) {
 133                 // Fast-fail for files that definitely exceed the maximum data URI length
 134                 if ( $ie8Compat && filesize( $file ) >= self::DATA_URI_SIZE_LIMIT ) {
 135                         return false;
 136                 }
 137
 138                 if ( $type === null ) {
 139                         $type = self::getMimeType( $file );
 140                 }
 141                 if ( !$type ) {
 142                         return false;
 143                 }
 144
 145                 return self::encodeStringAsDataURI( file_get_contents( $file ), $type, $ie8Compat );
 146         }
 147
 148         /**
 149          * Encode file contents as a data URI with chosen MIME type.
 150          *
 151          * The URI will be base64-encoded for binary files or just percent-encoded otherwise.
 152          *
 153          * @since 1.25
 154          *
 155          * @param string $contents File contents to encode.
 156          * @param string $type File's MIME type.
 157          * @param bool $ie8Compat See encodeImageAsDataURI().
 158          * @return string|bool Image contents encoded as a data URI or false.
 159          */
 160         public static function encodeStringAsDataURI( $contents, $type, $ie8Compat = true ) {
 161                 // Try #1: Non-encoded data URI
 162                 // The regular expression matches ASCII whitespace and printable characters.
 163                 if ( preg_match( '/^[\r\n\t\x20-\x7e]+$/', $contents ) ) {
 164                         // Do not base64-encode non-binary files (sane SVGs).
 165                         // (This often produces longer URLs, but they compress better, yielding a net smaller size.)
 166                         $uri = 'data:' . $type . ',' . rawurlencode( $contents );
 167                         if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 168                                 return $uri;
 169                         }
 170                 }
 171
 172                 // Try #2: Encoded data URI
 173                 $uri = 'data:' . $type . ';base64,' . base64_encode( $contents );
 174                 if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 175                         return $uri;
 176                 }
 177
 178                 // A data URI couldn't be produced
 179                 return false;
 180         }
 181
 182         /**
 183          * @param $file string
 184          * @return bool|string
 185          */
 186         public static function getMimeType( $file ) {
 187                 $realpath = realpath( $file );
 188                 if (
 189                         $realpath
 190                         && function_exists( 'finfo_file' )
 191                         && function_exists( 'finfo_open' )
 192                         && defined( 'FILEINFO_MIME_TYPE' )
 193                 ) {
 194                         return finfo_file( finfo_open( FILEINFO_MIME_TYPE ), $realpath );
 195                 }
 196
 197                 // Infer the MIME-type from the file extension
 198                 $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) );
 199                 if ( isset( self::$mimeTypes[$ext] ) ) {
 200                         return self::$mimeTypes[$ext];
 201                 }
 202
 203                 return false;
 204         }
 205
 206         /**
 207          * Build a CSS 'url()' value for the given URL, quoting parentheses (and other funny characters)
 208          * and escaping quotes as necessary.
 209          *
 210          * See http://www.w3.org/TR/css-syntax-3/#consume-a-url-token
 211          *
 212          * @param string $url URL to process
 213          * @return string 'url()' value, usually just `"url($url)"`, quoted/escaped if necessary
 214          */
 215         public static function buildUrlValue( $url ) {
 216                 // The list below has been crafted to match URLs such as:
 217                 //   scheme://user@domain:port/~user/fi%20le.png?query=yes&really=y+s
 218                 //   data:image/png;base64,R0lGODlh/+==
 219                 if ( preg_match( '!^[\w\d:@/~.%+;,?&=-]+$!', $url ) ) {
 220                         return "url($url)";
 221                 } else {
 222                         return 'url("' . strtr( $url, array( '\\' => '\\\\', '"' => '\\"' ) ) . '")';
 223                 }
 224         }
 225
 226         /**
 227          * Remaps CSS URL paths and automatically embeds data URIs for CSS rules
 228          * or url() values preceded by an / * @embed * / comment.
 229          *
 230          * @param string $source CSS data to remap
 231          * @param string $local File path where the source was read from
 232          * @param string $remote URL path to the file
 233          * @param bool $embedData If false, never do any data URI embedding,
 234          *   even if / * @embed * / is found.
 235          * @return string Remapped CSS data
 236          */
 237         public static function remap( $source, $local, $remote, $embedData = true ) {
 238                 // High-level overview:
 239                 // * For each CSS rule in $source that includes at least one url() value:
 240                 //   * Check for an @embed comment at the start indicating that all URIs should be embedded
 241                 //   * For each url() value:
 242                 //     * Check for an @embed comment directly preceding the value
 243                 //     * If either @embed comment exists:
 244                 //       * Embedding the URL as data: URI, if it's possible / allowed
 245                 //       * Otherwise remap the URL to work in generated stylesheets
 246
 247                 // Guard against trailing slashes, because "some/remote/../foo.png"
 248                 // resolves to "some/remote/foo.png" on (some?) clients (bug 27052).
 249                 if ( substr( $remote, -1 ) == '/' ) {
 250                         $remote = substr( $remote, 0, -1 );
 251                 }
 252
 253                 // Disallow U+007F DELETE, which is illegal anyway, and which
 254                 // we use for comment placeholders.
 255                 $source = str_replace( "\x7f", "?", $source );
 256
 257                 // Replace all comments by a placeholder so they will not interfere with the remapping.
 258                 // Warning: This will also catch on anything looking like the start of a comment between
 259                 // quotation marks (e.g. "foo /* bar").
 260                 $comments = array();
 261
 262                 $pattern = '/(?!' . CSSMin::EMBED_REGEX . ')(' . CSSMin::COMMENT_REGEX . ')/s';
 263
 264                 $source = preg_replace_callback(
 265                         $pattern,
 266                         function ( $match ) use ( &$comments ) {
 267                                 $comments[] = $match[ 0 ];
 268                                 return CSSMin::PLACEHOLDER . ( count( $comments ) - 1 ) . 'x';
 269                         },
 270                         $source
 271                 );
 272
 273                 // Note: This will not correctly handle cases where ';', '{' or '}'
 274                 // appears in the rule itself, e.g. in a quoted string. You are advised
 275                 // not to use such characters in file names. We also match start/end of
 276                 // the string to be consistent in edge-cases ('@import url(…)').
 277                 $pattern = '/(?:^|[;{])\K[^;{}]*' . CSSMin::URL_REGEX . '[^;}]*(?=[;}]|$)/';
 278
 279                 $source = preg_replace_callback(
 280                         $pattern,
 281                         function ( $matchOuter ) use ( $local, $remote, $embedData ) {
 282                                 $rule = $matchOuter[0];
 283
 284                                 // Check for global @embed comment and remove it. Allow other comments to be present
 285                                 // before @embed (they have been replaced with placeholders at this point).
 286                                 $embedAll = false;
 287                                 $rule = preg_replace( '/^((?:\s+|' . CSSMin::PLACEHOLDER . '(\d+)x)*)' . CSSMin::EMBED_REGEX . '\s*/', '$1', $rule, 1, $embedAll );
 288
 289                                 // Build two versions of current rule: with remapped URLs
 290                                 // and with embedded data: URIs (where possible).
 291                                 $pattern = '/(?P<embed>' . CSSMin::EMBED_REGEX . '\s*|)' . CSSMin::URL_REGEX . '/';
 292
 293                                 $ruleWithRemapped = preg_replace_callback(
 294                                         $pattern,
 295                                         function ( $match ) use ( $local, $remote ) {
 296                                                 $remapped = CSSMin::remapOne( $match['file'], $match['query'], $local, $remote, false );
 297
 298                                                 return CSSMin::buildUrlValue( $remapped );
 299                                         },
 300                                         $rule
 301                                 );
 302
 303                                 if ( $embedData ) {
 304                                         // Remember the occurring MIME types to avoid fallbacks when embedding some files.
 305                                         $mimeTypes = array();
 306
 307                                         $ruleWithEmbedded = preg_replace_callback(
 308                                                 $pattern,
 309                                                 function ( $match ) use ( $embedAll, $local, $remote, &$mimeTypes ) {
 310                                                         $embed = $embedAll || $match['embed'];
 311                                                         $embedded = CSSMin::remapOne(
 312                                                                 $match['file'],
 313                                                                 $match['query'],
 314                                                                 $local,
 315                                                                 $remote,
 316                                                                 $embed
 317                                                         );
 318
 319                                                         $url = $match['file'] . $match['query'];
 320                                                         $file = $local . $match['file'];
 321                                                         if (
 322                                                                 !CSSMin::isRemoteUrl( $url ) && !CSSMin::isLocalUrl( $url )
 323                                                                 && file_exists( $file )
 324                                                         ) {
 325                                                                 $mimeTypes[ CSSMin::getMimeType( $file ) ] = true;
 326                                                         }
 327
 328                                                         return CSSMin::buildUrlValue( $embedded );
 329                                                 },
 330                                                 $rule
 331                                         );
 332
 333                                         // Are all referenced images SVGs?
 334                                         $needsEmbedFallback = $mimeTypes !== array( 'image/svg+xml' => true );
 335                                 }
 336
 337                                 if ( !$embedData || $ruleWithEmbedded === $ruleWithRemapped ) {
 338                                         // We're not embedding anything, or we tried to but the file is not embeddable
 339                                         return $ruleWithRemapped;
 340                                 } elseif ( $embedData && $needsEmbedFallback ) {
 341                                         // Build 2 CSS properties; one which uses a data URI in place of the @embed comment, and
 342                                         // the other with a remapped and versioned URL with an Internet Explorer 6 and 7 hack
 343                                         // making it ignored in all browsers that support data URIs
 344                                         return "$ruleWithEmbedded;$ruleWithRemapped!ie";
 345                                 } else {
 346                                         // Look ma, no fallbacks! This is for files which IE 6 and 7 don't support anyway: SVG.
 347                                         return $ruleWithEmbedded;
 348                                 }
 349                         }, $source );
 350
 351                 // Re-insert comments
 352                 $pattern = '/' . CSSMin::PLACEHOLDER . '(\d+)x/';
 353                 $source = preg_replace_callback( $pattern, function( $match ) use ( &$comments ) {
 354                         return $comments[ $match[1] ];
 355                 }, $source );
 356
 357                 return $source;
 358
 359         }
 360
 361         /**
 362          * Is this CSS rule referencing a remote URL?
 363          *
 364          * @private Until we require PHP 5.5 and we can access self:: from closures.
 365          * @param string $maybeUrl
 366          * @return bool
 367          */
 368         public static function isRemoteUrl( $maybeUrl ) {
 369                 if ( substr( $maybeUrl, 0, 2 ) === '//' || parse_url( $maybeUrl, PHP_URL_SCHEME ) ) {
 370                         return true;
 371                 }
 372                 return false;
 373         }
 374
 375         /**
 376          * Is this CSS rule referencing a local URL?
 377          *
 378          * @private Until we require PHP 5.5 and we can access self:: from closures.
 379          * @param string $maybeUrl
 380          * @return bool
 381          */
 382         public static function isLocalUrl( $maybeUrl ) {
 383                 if ( $maybeUrl !== '' && $maybeUrl[0] === '/' && !self::isRemoteUrl( $maybeUrl ) ) {
 384                         return true;
 385                 }
 386                 return false;
 387         }
 388
 389         /**
 390          * Remap or embed a CSS URL path.
 391          *
 392          * @param string $file URL to remap/embed
 393          * @param string $query
 394          * @param string $local File path where the source was read from
 395          * @param string $remote URL path to the file
 396          * @param bool $embed Whether to do any data URI embedding
 397          * @return string Remapped/embedded URL data
 398          */
 399         public static function remapOne( $file, $query, $local, $remote, $embed ) {
 400                 // The full URL possibly with query, as passed to the 'url()' value in CSS
 401                 $url = $file . $query;
 402
 403                 // Expand local URLs with absolute paths like /w/index.php to possibly protocol-relative URL, if
 404                 // wfExpandUrl() is available. (This will not be the case if we're running outside of MW.)
 405                 if ( self::isLocalUrl( $url ) && function_exists( 'wfExpandUrl' ) ) {
 406                         return wfExpandUrl( $url, PROTO_RELATIVE );
 407                 }
 408
 409                 // Pass thru fully-qualified and protocol-relative URLs and data URIs, as well as local URLs if
 410                 // we can't expand them.
 411                 if ( self::isRemoteUrl( $url ) || self::isLocalUrl( $url ) ) {
 412                         return $url;
 413                 }
 414
 415                 if ( $local === false ) {
 416                         // Assume that all paths are relative to $remote, and make them absolute
 417                         $url = $remote . '/' . $url;
 418                 } else {
 419                         // We drop the query part here and instead make the path relative to $remote
 420                         $url = "{$remote}/{$file}";
 421                         // Path to the actual file on the filesystem
 422                         $localFile = "{$local}/{$file}";
 423                         if ( file_exists( $localFile ) ) {
 424                                 // Add version parameter as the first five hex digits
 425                                 // of the MD5 hash of the file's contents.
 426                                 $url .= '?' . substr( md5_file( $localFile ), 0, 5 );
 427                                 if ( $embed ) {
 428                                         $data = self::encodeImageAsDataURI( $localFile );
 429                                         if ( $data !== false ) {
 430                                                 return $data;
 431                                         }
 432                                 }
 433                         }
 434                         // If any of these conditions failed (file missing, we don't want to embed it
 435                         // or it's not embeddable), return the URL (possibly with ?timestamp part)
 436                 }
 437                 if ( function_exists( 'wfRemoveDotSegments' ) ) {
 438                         $url = wfRemoveDotSegments( $url );
 439                 }
 440                 return $url;
 441         }
 442
 443         /**
 444          * Removes whitespace from CSS data
 445          *
 446          * @param string $css CSS data to minify
 447          * @return string Minified CSS data
 448          */
 449         public static function minify( $css ) {
 450                 return trim(
 451                         str_replace(
 452                                 array( '; ', ': ', ' {', '{ ', ', ', '} ', ';}' ),
 453                                 array( ';', ':', '{', '{', ',', '}', '}' ),
 454                                 preg_replace( array( '/\s+/', '/\/\*.*?\*\//s' ), array( ' ', '' ), $css )
 455                         )
 456                 );
 457         }
 458 }