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         /**
  35          * Internet Explorer data URI length limit. See encodeImageAsDataURI().
  36          */
  37         const DATA_URI_SIZE_LIMIT = 32768;
  38         const URL_REGEX = 'url\(\s*[\'"]?(?P<file>[^\?\)\'"]*?)(?P<query>\?[^\)\'"]*?|)[\'"]?\s*\)';
  39         const EMBED_REGEX = '\/\*\s*\@embed\s*\*\/';
  40         const COMMENT_REGEX = '\/\*.*?\*\/';
  41
  42         /* Protected Static Members */
  43
  44         /** @var array List of common image files extensions and MIME-types */
  45         protected static $mimeTypes = array(
  46                 'gif' => 'image/gif',
  47                 'jpe' => 'image/jpeg',
  48                 'jpeg' => 'image/jpeg',
  49                 'jpg' => 'image/jpeg',
  50                 'png' => 'image/png',
  51                 'tif' => 'image/tiff',
  52                 'tiff' => 'image/tiff',
  53                 'xbm' => 'image/x-xbitmap',
  54                 'svg' => 'image/svg+xml',
  55         );
  56
  57         /* Static Methods */
  58
  59         /**
  60          * Gets a list of local file paths which are referenced in a CSS style sheet
  61          *
  62          * This function will always return an empty array if the second parameter is not given or null
  63          * for backwards-compatibility.
  64          *
  65          * @param string $source CSS data to remap
  66          * @param string $path File path where the source was read from (optional)
  67          * @return array List of local file references
  68          */
  69         public static function getLocalFileReferences( $source, $path = null ) {
  70                 if ( $path === null ) {
  71                         return array();
  72                 }
  73
  74                 $path = rtrim( $path, '/' ) . '/';
  75                 $files = array();
  76
  77                 $rFlags = PREG_OFFSET_CAPTURE | PREG_SET_ORDER;
  78                 if ( preg_match_all( '/' . self::URL_REGEX . '/', $source, $matches, $rFlags ) ) {
  79                         foreach ( $matches as $match ) {
  80                                 $url = $match['file'][0];
  81
  82                                 // Skip fully-qualified and protocol-relative URLs and data URIs
  83                                 if ( substr( $url, 0, 2 ) === '//' || parse_url( $url, PHP_URL_SCHEME ) ) {
  84                                         break;
  85                                 }
  86
  87                                 $file = $path . $url;
  88                                 // Skip non-existent files
  89                                 if ( file_exists( $file ) ) {
  90                                         break;
  91                                 }
  92
  93                                 $files[] = $file;
  94                         }
  95                 }
  96                 return $files;
  97         }
  98
  99         /**
 100          * Encode an image file as a data URI.
 101          *
 102          * If the image file has a suitable MIME type and size, encode it as a data URI, base64-encoded
 103          * for binary files or just percent-encoded otherwise. Return false if the image type is
 104          * unfamiliar or file exceeds the size limit.
 105          *
 106          * @param string $file Image file to encode.
 107          * @param string|null $type File's MIME type or null. If null, CSSMin will
 108          *     try to autodetect the type.
 109          * @param bool $ie8Compat By default, a data URI will only be produced if it can be made short
 110          *     enough to fit in Internet Explorer 8 (and earlier) URI length limit (32,768 bytes). Pass
 111          *     `false` to remove this limitation.
 112          * @return string|bool Image contents encoded as a data URI or false.
 113          */
 114         public static function encodeImageAsDataURI( $file, $type = null, $ie8Compat = true ) {
 115                 // Fast-fail for files that definitely exceed the maximum data URI length
 116                 if ( $ie8Compat && filesize( $file ) >= self::DATA_URI_SIZE_LIMIT ) {
 117                         return false;
 118                 }
 119
 120                 if ( $type === null ) {
 121                         $type = self::getMimeType( $file );
 122                 }
 123                 if ( !$type ) {
 124                         return false;
 125                 }
 126
 127                 return self::encodeStringAsDataURI( file_get_contents( $file ), $type, $ie8Compat );
 128         }
 129
 130         /**
 131          * Encode file contents as a data URI with chosen MIME type.
 132          *
 133          * The URI will be base64-encoded for binary files or just percent-encoded otherwise.
 134          *
 135          * @since 1.25
 136          *
 137          * @param string $contents File contents to encode.
 138          * @param string $type File's MIME type.
 139          * @param bool $ie8Compat See encodeImageAsDataURI().
 140          * @return string|bool Image contents encoded as a data URI or false.
 141          */
 142         public static function encodeStringAsDataURI( $contents, $type, $ie8Compat = true ) {
 143                 // Try #1: Non-encoded data URI
 144                 // The regular expression matches ASCII whitespace and printable characters.
 145                 if ( preg_match( '/^[\r\n\t\x20-\x7e]+$/', $contents ) ) {
 146                         // Do not base64-encode non-binary files (sane SVGs).
 147                         // (This often produces longer URLs, but they compress better, yielding a net smaller size.)
 148                         $uri = 'data:' . $type . ',' . rawurlencode( $contents );
 149                         if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 150                                 return $uri;
 151                         }
 152                 }
 153
 154                 // Try #2: Encoded data URI
 155                 $uri = 'data:' . $type . ';base64,' . base64_encode( $contents );
 156                 if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 157                         return $uri;
 158                 }
 159
 160                 // A data URI couldn't be produced
 161                 return false;
 162         }
 163
 164         /**
 165          * @param $file string
 166          * @return bool|string
 167          */
 168         public static function getMimeType( $file ) {
 169                 $realpath = realpath( $file );
 170                 if (
 171                         $realpath
 172                         && function_exists( 'finfo_file' )
 173                         && function_exists( 'finfo_open' )
 174                         && defined( 'FILEINFO_MIME_TYPE' )
 175                 ) {
 176                         return finfo_file( finfo_open( FILEINFO_MIME_TYPE ), $realpath );
 177                 }
 178
 179                 // Infer the MIME-type from the file extension
 180                 $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) );
 181                 if ( isset( self::$mimeTypes[$ext] ) ) {
 182                         return self::$mimeTypes[$ext];
 183                 }
 184
 185                 return false;
 186         }
 187
 188         /**
 189          * Build a CSS 'url()' value for the given URL, quoting parentheses (and other funny characters)
 190          * and escaping quotes as necessary.
 191          *
 192          * See http://www.w3.org/TR/css-syntax-3/#consume-a-url-token
 193          *
 194          * @param string $url URL to process
 195          * @return string 'url()' value, usually just `"url($url)"`, quoted/escaped if necessary
 196          */
 197         public static function buildUrlValue( $url ) {
 198                 // The list below has been crafted to match URLs such as:
 199                 //   scheme://user@domain:port/~user/fi%20le.png?query=yes&really=y+s
 200                 //   data:image/png;base64,R0lGODlh/+==
 201                 if ( preg_match( '!^[\w\d:@/~.%+;,?&=-]+$!', $url ) ) {
 202                         return "url($url)";
 203                 } else {
 204                         return 'url("' . strtr( $url, array( '\\' => '\\\\', '"' => '\\"' ) ) . '")';
 205                 }
 206         }
 207
 208         /**
 209          * Remaps CSS URL paths and automatically embeds data URIs for CSS rules
 210          * or url() values preceded by an / * @embed * / comment.
 211          *
 212          * @param string $source CSS data to remap
 213          * @param string $local File path where the source was read from
 214          * @param string $remote URL path to the file
 215          * @param bool $embedData If false, never do any data URI embedding,
 216          *   even if / * @embed * / is found.
 217          * @return string Remapped CSS data
 218          */
 219         public static function remap( $source, $local, $remote, $embedData = true ) {
 220                 // High-level overview:
 221                 // * For each CSS rule in $source that includes at least one url() value:
 222                 //   * Check for an @embed comment at the start indicating that all URIs should be embedded
 223                 //   * For each url() value:
 224                 //     * Check for an @embed comment directly preceding the value
 225                 //     * If either @embed comment exists:
 226                 //       * Embedding the URL as data: URI, if it's possible / allowed
 227                 //       * Otherwise remap the URL to work in generated stylesheets
 228
 229                 // Guard against trailing slashes, because "some/remote/../foo.png"
 230                 // resolves to "some/remote/foo.png" on (some?) clients (bug 27052).
 231                 if ( substr( $remote, -1 ) == '/' ) {
 232                         $remote = substr( $remote, 0, -1 );
 233                 }
 234
 235                 // Replace all comments by a placeholder so they will not interfere with the remapping.
 236                 // Warning: This will also catch on anything looking like the start of a comment between
 237                 // quotation marks (e.g. "foo /* bar").
 238                 $comments = array();
 239                 $placeholder = uniqid( '', true );
 240
 241                 $pattern = '/(?!' . CSSMin::EMBED_REGEX . ')(' . CSSMin::COMMENT_REGEX . ')/s';
 242
 243                 $source = preg_replace_callback(
 244                         $pattern,
 245                         function ( $match ) use ( &$comments, $placeholder ) {
 246                                 $comments[] = $match[ 0 ];
 247                                 return $placeholder . ( count( $comments ) - 1 ) . 'x';
 248                         },
 249                         $source
 250                 );
 251
 252                 // Note: This will not correctly handle cases where ';', '{' or '}'
 253                 // appears in the rule itself, e.g. in a quoted string. You are advised
 254                 // not to use such characters in file names. We also match start/end of
 255                 // the string to be consistent in edge-cases ('@import url(…)').
 256                 $pattern = '/(?:^|[;{])\K[^;{}]*' . CSSMin::URL_REGEX . '[^;}]*(?=[;}]|$)/';
 257
 258                 $source = preg_replace_callback(
 259                         $pattern,
 260                         function ( $matchOuter ) use ( $local, $remote, $embedData, $placeholder ) {
 261                                 $rule = $matchOuter[0];
 262
 263                                 // Check for global @embed comment and remove it. Allow other comments to be present
 264                                 // before @embed (they have been replaced with placeholders at this point).
 265                                 $embedAll = false;
 266                                 $rule = preg_replace( '/^((?:\s+|' . $placeholder . '(\d+)x)*)' . CSSMin::EMBED_REGEX . '\s*/', '$1', $rule, 1, $embedAll );
 267
 268                                 // Build two versions of current rule: with remapped URLs
 269                                 // and with embedded data: URIs (where possible).
 270                                 $pattern = '/(?P<embed>' . CSSMin::EMBED_REGEX . '\s*|)' . CSSMin::URL_REGEX . '/';
 271
 272                                 $ruleWithRemapped = preg_replace_callback(
 273                                         $pattern,
 274                                         function ( $match ) use ( $local, $remote ) {
 275                                                 $remapped = CSSMin::remapOne( $match['file'], $match['query'], $local, $remote, false );
 276
 277                                                 return CSSMin::buildUrlValue( $remapped );
 278                                         },
 279                                         $rule
 280                                 );
 281
 282                                 if ( $embedData ) {
 283                                         // Remember the occurring MIME types to avoid fallbacks when embedding some files.
 284                                         $mimeTypes = array();
 285
 286                                         $ruleWithEmbedded = preg_replace_callback(
 287                                                 $pattern,
 288                                                 function ( $match ) use ( $embedAll, $local, $remote, &$mimeTypes ) {
 289                                                         $embed = $embedAll || $match['embed'];
 290                                                         $embedded = CSSMin::remapOne(
 291                                                                 $match['file'],
 292                                                                 $match['query'],
 293                                                                 $local,
 294                                                                 $remote,
 295                                                                 $embed
 296                                                         );
 297
 298                                                         $url = $match['file'] . $match['query'];
 299                                                         $file = $local . $match['file'];
 300                                                         if (
 301                                                                 !CSSMin::isRemoteUrl( $url ) && !CSSMin::isLocalUrl( $url )
 302                                                                 && file_exists( $file )
 303                                                         ) {
 304                                                                 $mimeTypes[ CSSMin::getMimeType( $file ) ] = true;
 305                                                         }
 306
 307                                                         return CSSMin::buildUrlValue( $embedded );
 308                                                 },
 309                                                 $rule
 310                                         );
 311
 312                                         // Are all referenced images SVGs?
 313                                         $needsEmbedFallback = $mimeTypes !== array( 'image/svg+xml' => true );
 314                                 }
 315
 316                                 if ( !$embedData || $ruleWithEmbedded === $ruleWithRemapped ) {
 317                                         // We're not embedding anything, or we tried to but the file is not embeddable
 318                                         return $ruleWithRemapped;
 319                                 } elseif ( $embedData && $needsEmbedFallback ) {
 320                                         // Build 2 CSS properties; one which uses a data URI in place of the @embed comment, and
 321                                         // the other with a remapped and versioned URL with an Internet Explorer 6 and 7 hack
 322                                         // making it ignored in all browsers that support data URIs
 323                                         return "$ruleWithEmbedded;$ruleWithRemapped!ie";
 324                                 } else {
 325                                         // Look ma, no fallbacks! This is for files which IE 6 and 7 don't support anyway: SVG.
 326                                         return $ruleWithEmbedded;
 327                                 }
 328                         }, $source );
 329
 330                 // Re-insert comments
 331                 $pattern = '/' . $placeholder . '(\d+)x/';
 332                 $source = preg_replace_callback( $pattern, function( $match ) use ( &$comments ) {
 333                         return $comments[ $match[1] ];
 334                 }, $source );
 335
 336                 return $source;
 337
 338         }
 339
 340         /**
 341          * Is this CSS rule referencing a remote URL?
 342          *
 343          * @private Until we require PHP 5.5 and we can access self:: from closures.
 344          * @param string $maybeUrl
 345          * @return bool
 346          */
 347         public static function isRemoteUrl( $maybeUrl ) {
 348                 if ( substr( $maybeUrl, 0, 2 ) === '//' || parse_url( $maybeUrl, PHP_URL_SCHEME ) ) {
 349                         return true;
 350                 }
 351                 return false;
 352         }
 353
 354         /**
 355          * Is this CSS rule referencing a local URL?
 356          *
 357          * @private Until we require PHP 5.5 and we can access self:: from closures.
 358          * @param string $maybeUrl
 359          * @return bool
 360          */
 361         public static function isLocalUrl( $maybeUrl ) {
 362                 if ( $maybeUrl !== '' && $maybeUrl[0] === '/' && !self::isRemoteUrl( $maybeUrl ) ) {
 363                         return true;
 364                 }
 365                 return false;
 366         }
 367
 368         /**
 369          * Remap or embed a CSS URL path.
 370          *
 371          * @param string $file URL to remap/embed
 372          * @param string $query
 373          * @param string $local File path where the source was read from
 374          * @param string $remote URL path to the file
 375          * @param bool $embed Whether to do any data URI embedding
 376          * @return string Remapped/embedded URL data
 377          */
 378         public static function remapOne( $file, $query, $local, $remote, $embed ) {
 379                 // The full URL possibly with query, as passed to the 'url()' value in CSS
 380                 $url = $file . $query;
 381
 382                 // Expand local URLs with absolute paths like /w/index.php to possibly protocol-relative URL, if
 383                 // wfExpandUrl() is available. (This will not be the case if we're running outside of MW.)
 384                 if ( self::isLocalUrl( $url ) && function_exists( 'wfExpandUrl' ) ) {
 385                         return wfExpandUrl( $url, PROTO_RELATIVE );
 386                 }
 387
 388                 // Pass thru fully-qualified and protocol-relative URLs and data URIs, as well as local URLs if
 389                 // we can't expand them.
 390                 if ( self::isRemoteUrl( $url ) || self::isLocalUrl( $url ) ) {
 391                         return $url;
 392                 }
 393
 394                 if ( $local === false ) {
 395                         // Assume that all paths are relative to $remote, and make them absolute
 396                         return $remote . '/' . $url;
 397                 } else {
 398                         // We drop the query part here and instead make the path relative to $remote
 399                         $url = "{$remote}/{$file}";
 400                         // Path to the actual file on the filesystem
 401                         $localFile = "{$local}/{$file}";
 402                         if ( file_exists( $localFile ) ) {
 403                                 // Add version parameter as a time-stamp in ISO 8601 format,
 404                                 // using Z for the timezone, meaning GMT
 405                                 $url .= '?' . gmdate( 'Y-m-d\TH:i:s\Z', round( filemtime( $localFile ), -2 ) );
 406                                 if ( $embed ) {
 407                                         $data = self::encodeImageAsDataURI( $localFile );
 408                                         if ( $data !== false ) {
 409                                                 return $data;
 410                                         }
 411                                 }
 412                         }
 413                         // If any of these conditions failed (file missing, we don't want to embed it
 414                         // or it's not embeddable), return the URL (possibly with ?timestamp part)
 415                         return $url;
 416                 }
 417         }
 418
 419         /**
 420          * Removes whitespace from CSS data
 421          *
 422          * @param string $css CSS data to minify
 423          * @return string Minified CSS data
 424          */
 425         public static function minify( $css ) {
 426                 return trim(
 427                         str_replace(
 428                                 array( '; ', ': ', ' {', '{ ', ', ', '} ', ';}' ),
 429                                 array( ';', ':', '{', '{', ',', '}', '}' ),
 430                                 preg_replace( array( '/\s+/', '/\/\*.*?\*\//s' ), array( ' ', '' ), $css )
 431                         )
 432                 );
 433         }
 434 }