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 = [
  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 [];
  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 was read from
  95          * @return array List of local file references
  96          */
  97         public static function getAllLocalFileReferences( $source, $path ) {
  98                 $stripped = preg_replace( '/' . self::COMMENT_REGEX . '/s', '', $source );
  99                 $path = rtrim( $path, '/' ) . '/';
 100                 $files = [];
 101
 102                 $rFlags = PREG_OFFSET_CAPTURE | PREG_SET_ORDER;
 103                 if ( preg_match_all( '/' . self::URL_REGEX . '/', $stripped, $matches, $rFlags ) ) {
 104                         foreach ( $matches as $match ) {
 105                                 $url = $match['file'][0];
 106
 107                                 // Skip fully-qualified and protocol-relative URLs and data URIs
 108                                 if ( substr( $url, 0, 2 ) === '//' || parse_url( $url, PHP_URL_SCHEME ) ) {
 109                                         break;
 110                                 }
 111
 112                                 $files[] = $path . $url;
 113                         }
 114                 }
 115                 return $files;
 116         }
 117
 118         /**
 119          * Encode an image file as a data URI.
 120          *
 121          * If the image file has a suitable MIME type and size, encode it as a data URI, base64-encoded
 122          * for binary files or just percent-encoded otherwise. Return false if the image type is
 123          * unfamiliar or file exceeds the size limit.
 124          *
 125          * @param string $file Image file to encode.
 126          * @param string|null $type File's MIME type or null. If null, CSSMin will
 127          *     try to autodetect the type.
 128          * @param bool $ie8Compat By default, a data URI will only be produced if it can be made short
 129          *     enough to fit in Internet Explorer 8 (and earlier) URI length limit (32,768 bytes). Pass
 130          *     `false` to remove this limitation.
 131          * @return string|bool Image contents encoded as a data URI or false.
 132          */
 133         public static function encodeImageAsDataURI( $file, $type = null, $ie8Compat = true ) {
 134                 // Fast-fail for files that definitely exceed the maximum data URI length
 135                 if ( $ie8Compat && filesize( $file ) >= self::DATA_URI_SIZE_LIMIT ) {
 136                         return false;
 137                 }
 138
 139                 if ( $type === null ) {
 140                         $type = self::getMimeType( $file );
 141                 }
 142                 if ( !$type ) {
 143                         return false;
 144                 }
 145
 146                 return self::encodeStringAsDataURI( file_get_contents( $file ), $type, $ie8Compat );
 147         }
 148
 149         /**
 150          * Encode file contents as a data URI with chosen MIME type.
 151          *
 152          * The URI will be base64-encoded for binary files or just percent-encoded otherwise.
 153          *
 154          * @since 1.25
 155          *
 156          * @param string $contents File contents to encode.
 157          * @param string $type File's MIME type.
 158          * @param bool $ie8Compat See encodeImageAsDataURI().
 159          * @return string|bool Image contents encoded as a data URI or false.
 160          */
 161         public static function encodeStringAsDataURI( $contents, $type, $ie8Compat = true ) {
 162                 // Try #1: Non-encoded data URI
 163                 // The regular expression matches ASCII whitespace and printable characters.
 164                 if ( preg_match( '/^[\r\n\t\x20-\x7e]+$/', $contents ) ) {
 165                         // Do not base64-encode non-binary files (sane SVGs).
 166                         // (This often produces longer URLs, but they compress better, yielding a net smaller size.)
 167                         $uri = 'data:' . $type . ',' . rawurlencode( $contents );
 168                         if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 169                                 return $uri;
 170                         }
 171                 }
 172
 173                 // Try #2: Encoded data URI
 174                 $uri = 'data:' . $type . ';base64,' . base64_encode( $contents );
 175                 if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 176                         return $uri;
 177                 }
 178
 179                 // A data URI couldn't be produced
 180                 return false;
 181         }
 182
 183         /**
 184          * Serialize a string (escape and quote) for use as a CSS string value.
 185          * http://www.w3.org/TR/2013/WD-cssom-20131205/#serialize-a-string
 186          *
 187          * @param string $value
 188          * @return string
 189          * @throws Exception
 190          */
 191         public static function serializeStringValue( $value ) {
 192                 if ( strstr( $value, "\0" ) ) {
 193                         throw new Exception( "Invalid character in CSS string" );
 194                 }
 195                 $value = strtr( $value, [ '\\' => '\\\\', '"' => '\\"' ] );
 196                 $value = preg_replace_callback( '/[\x01-\x1f\x7f-\x9f]/', function ( $match ) {
 197                         return '\\' . base_convert( ord( $match[0] ), 10, 16 ) . ' ';
 198                 }, $value );
 199                 return '"' . $value . '"';
 200         }
 201
 202         /**
 203          * @param $file string
 204          * @return bool|string
 205          */
 206         public static function getMimeType( $file ) {
 207                 $realpath = realpath( $file );
 208                 if (
 209                         $realpath
 210                         && function_exists( 'finfo_file' )
 211                         && function_exists( 'finfo_open' )
 212                         && defined( 'FILEINFO_MIME_TYPE' )
 213                 ) {
 214                         return finfo_file( finfo_open( FILEINFO_MIME_TYPE ), $realpath );
 215                 }
 216
 217                 // Infer the MIME-type from the file extension
 218                 $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) );
 219                 if ( isset( self::$mimeTypes[$ext] ) ) {
 220                         return self::$mimeTypes[$ext];
 221                 }
 222
 223                 return false;
 224         }
 225
 226         /**
 227          * Build a CSS 'url()' value for the given URL, quoting parentheses (and other funny characters)
 228          * and escaping quotes as necessary.
 229          *
 230          * See http://www.w3.org/TR/css-syntax-3/#consume-a-url-token
 231          *
 232          * @param string $url URL to process
 233          * @return string 'url()' value, usually just `"url($url)"`, quoted/escaped if necessary
 234          */
 235         public static function buildUrlValue( $url ) {
 236                 // The list below has been crafted to match URLs such as:
 237                 //   scheme://user@domain:port/~user/fi%20le.png?query=yes&really=y+s
 238                 //   data:image/png;base64,R0lGODlh/+==
 239                 if ( preg_match( '!^[\w\d:@/~.%+;,?&=-]+$!', $url ) ) {
 240                         return "url($url)";
 241                 } else {
 242                         return 'url("' . strtr( $url, [ '\\' => '\\\\', '"' => '\\"' ] ) . '")';
 243                 }
 244         }
 245
 246         /**
 247          * Remaps CSS URL paths and automatically embeds data URIs for CSS rules
 248          * or url() values preceded by an / * @embed * / comment.
 249          *
 250          * @param string $source CSS data to remap
 251          * @param string $local File path where the source was read from
 252          * @param string $remote URL path to the file
 253          * @param bool $embedData If false, never do any data URI embedding,
 254          *   even if / * @embed * / is found.
 255          * @return string Remapped CSS data
 256          */
 257         public static function remap( $source, $local, $remote, $embedData = true ) {
 258                 // High-level overview:
 259                 // * For each CSS rule in $source that includes at least one url() value:
 260                 //   * Check for an @embed comment at the start indicating that all URIs should be embedded
 261                 //   * For each url() value:
 262                 //     * Check for an @embed comment directly preceding the value
 263                 //     * If either @embed comment exists:
 264                 //       * Embedding the URL as data: URI, if it's possible / allowed
 265                 //       * Otherwise remap the URL to work in generated stylesheets
 266
 267                 // Guard against trailing slashes, because "some/remote/../foo.png"
 268                 // resolves to "some/remote/foo.png" on (some?) clients (bug 27052).
 269                 if ( substr( $remote, -1 ) == '/' ) {
 270                         $remote = substr( $remote, 0, -1 );
 271                 }
 272
 273                 // Disallow U+007F DELETE, which is illegal anyway, and which
 274                 // we use for comment placeholders.
 275                 $source = str_replace( "\x7f", "?", $source );
 276
 277                 // Replace all comments by a placeholder so they will not interfere with the remapping.
 278                 // Warning: This will also catch on anything looking like the start of a comment between
 279                 // quotation marks (e.g. "foo /* bar").
 280                 $comments = [];
 281
 282                 $pattern = '/(?!' . CSSMin::EMBED_REGEX . ')(' . CSSMin::COMMENT_REGEX . ')/s';
 283
 284                 $source = preg_replace_callback(
 285                         $pattern,
 286                         function ( $match ) use ( &$comments ) {
 287                                 $comments[] = $match[ 0 ];
 288                                 return CSSMin::PLACEHOLDER . ( count( $comments ) - 1 ) . 'x';
 289                         },
 290                         $source
 291                 );
 292
 293                 // Note: This will not correctly handle cases where ';', '{' or '}'
 294                 // appears in the rule itself, e.g. in a quoted string. You are advised
 295                 // not to use such characters in file names. We also match start/end of
 296                 // the string to be consistent in edge-cases ('@import url(…)').
 297                 $pattern = '/(?:^|[;{])\K[^;{}]*' . CSSMin::URL_REGEX . '[^;}]*(?=[;}]|$)/';
 298
 299                 $source = preg_replace_callback(
 300                         $pattern,
 301                         function ( $matchOuter ) use ( $local, $remote, $embedData ) {
 302                                 $rule = $matchOuter[0];
 303
 304                                 // Check for global @embed comment and remove it. Allow other comments to be present
 305                                 // before @embed (they have been replaced with placeholders at this point).
 306                                 $embedAll = false;
 307                                 $rule = preg_replace(
 308                                         '/^((?:\s+|' .
 309                                                 CSSMin::PLACEHOLDER .
 310                                                 '(\d+)x)*)' .
 311                                                 CSSMin::EMBED_REGEX .
 312                                                 '\s*/',
 313                                         '$1',
 314                                         $rule,
 315                                         1,
 316                                         $embedAll
 317                                 );
 318
 319                                 // Build two versions of current rule: with remapped URLs
 320                                 // and with embedded data: URIs (where possible).
 321                                 $pattern = '/(?P<embed>' . CSSMin::EMBED_REGEX . '\s*|)' . CSSMin::URL_REGEX . '/';
 322
 323                                 $ruleWithRemapped = preg_replace_callback(
 324                                         $pattern,
 325                                         function ( $match ) use ( $local, $remote ) {
 326                                                 $remapped = CSSMin::remapOne( $match['file'], $match['query'], $local, $remote, false );
 327
 328                                                 return CSSMin::buildUrlValue( $remapped );
 329                                         },
 330                                         $rule
 331                                 );
 332
 333                                 if ( $embedData ) {
 334                                         // Remember the occurring MIME types to avoid fallbacks when embedding some files.
 335                                         $mimeTypes = [];
 336
 337                                         $ruleWithEmbedded = preg_replace_callback(
 338                                                 $pattern,
 339                                                 function ( $match ) use ( $embedAll, $local, $remote, &$mimeTypes ) {
 340                                                         $embed = $embedAll || $match['embed'];
 341                                                         $embedded = CSSMin::remapOne(
 342                                                                 $match['file'],
 343                                                                 $match['query'],
 344                                                                 $local,
 345                                                                 $remote,
 346                                                                 $embed
 347                                                         );
 348
 349                                                         $url = $match['file'] . $match['query'];
 350                                                         $file = $local . $match['file'];
 351                                                         if (
 352                                                                 !self::isRemoteUrl( $url ) && !self::isLocalUrl( $url )
 353                                                                 && file_exists( $file )
 354                                                         ) {
 355                                                                 $mimeTypes[ CSSMin::getMimeType( $file ) ] = true;
 356                                                         }
 357
 358                                                         return CSSMin::buildUrlValue( $embedded );
 359                                                 },
 360                                                 $rule
 361                                         );
 362
 363                                         // Are all referenced images SVGs?
 364                                         $needsEmbedFallback = $mimeTypes !== [ 'image/svg+xml' => true ];
 365                                 }
 366
 367                                 if ( !$embedData || $ruleWithEmbedded === $ruleWithRemapped ) {
 368                                         // We're not embedding anything, or we tried to but the file is not embeddable
 369                                         return $ruleWithRemapped;
 370                                 } elseif ( $embedData && $needsEmbedFallback ) {
 371                                         // Build 2 CSS properties; one which uses a data URI in place of the @embed comment, and
 372                                         // the other with a remapped and versioned URL with an Internet Explorer 6 and 7 hack
 373                                         // making it ignored in all browsers that support data URIs
 374                                         return "$ruleWithEmbedded;$ruleWithRemapped!ie";
 375                                 } else {
 376                                         // Look ma, no fallbacks! This is for files which IE 6 and 7 don't support anyway: SVG.
 377                                         return $ruleWithEmbedded;
 378                                 }
 379                         }, $source );
 380
 381                 // Re-insert comments
 382                 $pattern = '/' . CSSMin::PLACEHOLDER . '(\d+)x/';
 383                 $source = preg_replace_callback( $pattern, function( $match ) use ( &$comments ) {
 384                         return $comments[ $match[1] ];
 385                 }, $source );
 386
 387                 return $source;
 388
 389         }
 390
 391         /**
 392          * Is this CSS rule referencing a remote URL?
 393          *
 394          * @param string $maybeUrl
 395          * @return bool
 396          */
 397         protected static function isRemoteUrl( $maybeUrl ) {
 398                 if ( substr( $maybeUrl, 0, 2 ) === '//' || parse_url( $maybeUrl, PHP_URL_SCHEME ) ) {
 399                         return true;
 400                 }
 401                 return false;
 402         }
 403
 404         /**
 405          * Is this CSS rule referencing a local URL?
 406          *
 407          * @param string $maybeUrl
 408          * @return bool
 409          */
 410         protected static function isLocalUrl( $maybeUrl ) {
 411                 if ( $maybeUrl !== '' && $maybeUrl[0] === '/' && !self::isRemoteUrl( $maybeUrl ) ) {
 412                         return true;
 413                 }
 414                 return false;
 415         }
 416
 417         /**
 418          * Remap or embed a CSS URL path.
 419          *
 420          * @param string $file URL to remap/embed
 421          * @param string $query
 422          * @param string $local File path where the source was read from
 423          * @param string $remote URL path to the file
 424          * @param bool $embed Whether to do any data URI embedding
 425          * @return string Remapped/embedded URL data
 426          */
 427         public static function remapOne( $file, $query, $local, $remote, $embed ) {
 428                 // The full URL possibly with query, as passed to the 'url()' value in CSS
 429                 $url = $file . $query;
 430
 431                 // Expand local URLs with absolute paths like /w/index.php to possibly protocol-relative URL, if
 432                 // wfExpandUrl() is available. (This will not be the case if we're running outside of MW.)
 433                 if ( self::isLocalUrl( $url ) && function_exists( 'wfExpandUrl' ) ) {
 434                         return wfExpandUrl( $url, PROTO_RELATIVE );
 435                 }
 436
 437                 // Pass thru fully-qualified and protocol-relative URLs and data URIs, as well as local URLs if
 438                 // we can't expand them.
 439                 if ( self::isRemoteUrl( $url ) || self::isLocalUrl( $url ) ) {
 440                         return $url;
 441                 }
 442
 443                 if ( $local === false ) {
 444                         // Assume that all paths are relative to $remote, and make them absolute
 445                         $url = $remote . '/' . $url;
 446                 } else {
 447                         // We drop the query part here and instead make the path relative to $remote
 448                         $url = "{$remote}/{$file}";
 449                         // Path to the actual file on the filesystem
 450                         $localFile = "{$local}/{$file}";
 451                         if ( file_exists( $localFile ) ) {
 452                                 if ( $embed ) {
 453                                         $data = self::encodeImageAsDataURI( $localFile );
 454                                         if ( $data !== false ) {
 455                                                 return $data;
 456                                         }
 457                                 }
 458                                 if ( method_exists( 'OutputPage', 'transformFilePath' ) ) {
 459                                         $url = OutputPage::transformFilePath( $remote, $local, $file );
 460                                 } else {
 461                                         // Add version parameter as the first five hex digits
 462                                         // of the MD5 hash of the file's contents.
 463                                         $url .= '?' . substr( md5_file( $localFile ), 0, 5 );
 464                                 }
 465                         }
 466                         // If any of these conditions failed (file missing, we don't want to embed it
 467                         // or it's not embeddable), return the URL (possibly with ?timestamp part)
 468                 }
 469                 if ( function_exists( 'wfRemoveDotSegments' ) ) {
 470                         $url = wfRemoveDotSegments( $url );
 471                 }
 472                 return $url;
 473         }
 474
 475         /**
 476          * Removes whitespace from CSS data
 477          *
 478          * @param string $css CSS data to minify
 479          * @return string Minified CSS data
 480          */
 481         public static function minify( $css ) {
 482                 return trim(
 483                         str_replace(
 484                                 [ '; ', ': ', ' {', '{ ', ', ', '} ', ';}' ],
 485                                 [ ';', ':', '{', '{', ',', '}', '}' ],
 486                                 preg_replace( [ '/\s+/', '/\/\*.*?\*\//s' ], [ ' ', '' ], $css )
 487                         )
 488                 );
 489         }
 490 }