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
  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          * Get a list of local files referenced in a stylesheet (includes non-existent files).
  64          *
  65          * @param string $source CSS stylesheet source to process
  66          * @param string $path File path where the source was read from
  67          * @return array List of local file references
  68          */
  69         public static function getLocalFileReferences( $source, $path ) {
  70                 $stripped = preg_replace( '/' . self::COMMENT_REGEX . '/s', '', $source );
  71                 $path = rtrim( $path, '/' ) . '/';
  72                 $files = [];
  73
  74                 $rFlags = PREG_OFFSET_CAPTURE | PREG_SET_ORDER;
  75                 if ( preg_match_all( '/' . self::getUrlRegex() . '/', $stripped, $matches, $rFlags ) ) {
  76                         foreach ( $matches as $match ) {
  77                                 self::processUrlMatch( $match, $rFlags );
  78                                 $url = $match['file'][0];
  79
  80                                 // Skip fully-qualified and protocol-relative URLs and data URIs
  81                                 // Also skips the rare `behavior` property specifying application's default behavior
  82                                 if (
  83                                         substr( $url, 0, 2 ) === '//' ||
  84                                         parse_url( $url, PHP_URL_SCHEME ) ||
  85                                         substr( $url, 0, 9 ) === '#default#'
  86                                 ) {
  87                                         break;
  88                                 }
  89
  90                                 $files[] = $path . $url;
  91                         }
  92                 }
  93                 return $files;
  94         }
  95
  96         /**
  97          * Encode an image file as a data URI.
  98          *
  99          * If the image file has a suitable MIME type and size, encode it as a data URI, base64-encoded
 100          * for binary files or just percent-encoded otherwise. Return false if the image type is
 101          * unfamiliar or file exceeds the size limit.
 102          *
 103          * @param string $file Image file to encode.
 104          * @param string|null $type File's MIME type or null. If null, CSSMin will
 105          *     try to autodetect the type.
 106          * @param bool $ie8Compat By default, a data URI will only be produced if it can be made short
 107          *     enough to fit in Internet Explorer 8 (and earlier) URI length limit (32,768 bytes). Pass
 108          *     `false` to remove this limitation.
 109          * @return string|bool Image contents encoded as a data URI or false.
 110          */
 111         public static function encodeImageAsDataURI( $file, $type = null, $ie8Compat = true ) {
 112                 // Fast-fail for files that definitely exceed the maximum data URI length
 113                 if ( $ie8Compat && filesize( $file ) >= self::DATA_URI_SIZE_LIMIT ) {
 114                         return false;
 115                 }
 116
 117                 if ( $type === null ) {
 118                         $type = self::getMimeType( $file );
 119                 }
 120                 if ( !$type ) {
 121                         return false;
 122                 }
 123
 124                 return self::encodeStringAsDataURI( file_get_contents( $file ), $type, $ie8Compat );
 125         }
 126
 127         /**
 128          * Encode file contents as a data URI with chosen MIME type.
 129          *
 130          * The URI will be base64-encoded for binary files or just percent-encoded otherwise.
 131          *
 132          * @since 1.25
 133          *
 134          * @param string $contents File contents to encode.
 135          * @param string $type File's MIME type.
 136          * @param bool $ie8Compat See encodeImageAsDataURI().
 137          * @return string|bool Image contents encoded as a data URI or false.
 138          */
 139         public static function encodeStringAsDataURI( $contents, $type, $ie8Compat = true ) {
 140                 // Try #1: Non-encoded data URI
 141                 // The regular expression matches ASCII whitespace and printable characters.
 142                 if ( preg_match( '/^[\r\n\t\x20-\x7e]+$/', $contents ) ) {
 143                         // Do not base64-encode non-binary files (sane SVGs).
 144                         // (This often produces longer URLs, but they compress better, yielding a net smaller size.)
 145                         $encoded = rawurlencode( $contents );
 146                         // Unencode some things that don't need to be encoded, to make the encoding smaller
 147                         $encoded = strtr( $encoded, [
 148                                 '%20' => ' ', // Unencode spaces
 149                                 '%2F' => '/', // Unencode slashes
 150                                 '%3A' => ':', // Unencode colons
 151                                 '%3D' => '=', // Unencode equals signs
 152                                 '%22' => '"', // Unencode double quotes
 153                                 '%0A' => ' ', // Change newlines to spaces
 154                                 '%0D' => ' ', // Change carriage returns to spaces
 155                                 '%09' => ' ', // Change tabs to spaces
 156                         ] );
 157                         // Consolidate runs of multiple spaces in a row
 158                         $encoded = preg_replace( '/ {2,}/', ' ', $encoded );
 159                         // Remove leading and trailing spaces
 160                         $encoded = preg_replace( '/^ | $/', '', $encoded );
 161                         $uri = 'data:' . $type . ',' . $encoded;
 162                         if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 163                                 return $uri;
 164                         }
 165                 }
 166
 167                 // Try #2: Encoded data URI
 168                 $uri = 'data:' . $type . ';base64,' . base64_encode( $contents );
 169                 if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 170                         return $uri;
 171                 }
 172
 173                 // A data URI couldn't be produced
 174                 return false;
 175         }
 176
 177         /**
 178          * Serialize a string (escape and quote) for use as a CSS string value.
 179          * http://www.w3.org/TR/2013/WD-cssom-20131205/#serialize-a-string
 180          *
 181          * @param string $value
 182          * @return string
 183          * @throws Exception
 184          */
 185         public static function serializeStringValue( $value ) {
 186                 if ( strstr( $value, "\0" ) ) {
 187                         throw new Exception( "Invalid character in CSS string" );
 188                 }
 189                 $value = strtr( $value, [ '\\' => '\\\\', '"' => '\\"' ] );
 190                 $value = preg_replace_callback( '/[\x01-\x1f\x7f-\x9f]/', function ( $match ) {
 191                         return '\\' . base_convert( ord( $match[0] ), 10, 16 ) . ' ';
 192                 }, $value );
 193                 return '"' . $value . '"';
 194         }
 195
 196         /**
 197          * @param string $file
 198          * @return bool|string
 199          */
 200         public static function getMimeType( $file ) {
 201                 // Infer the MIME-type from the file extension
 202                 $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) );
 203                 if ( isset( self::$mimeTypes[$ext] ) ) {
 204                         return self::$mimeTypes[$ext];
 205                 }
 206
 207                 return mime_content_type( realpath( $file ) );
 208         }
 209
 210         /**
 211          * Build a CSS 'url()' value for the given URL, quoting parentheses (and other funny characters)
 212          * and escaping quotes as necessary.
 213          *
 214          * See http://www.w3.org/TR/css-syntax-3/#consume-a-url-token
 215          *
 216          * @param string $url URL to process
 217          * @return string 'url()' value, usually just `"url($url)"`, quoted/escaped if necessary
 218          */
 219         public static function buildUrlValue( $url ) {
 220                 // The list below has been crafted to match URLs such as:
 221                 //   scheme://user@domain:port/~user/fi%20le.png?query=yes&really=y+s
 222                 //   data:image/png;base64,R0lGODlh/+==
 223                 if ( preg_match( '!^[\w\d:@/~.%+;,?&=-]+$!', $url ) ) {
 224                         return "url($url)";
 225                 } else {
 226                         return "url('" . strtr( $url, [ '\\' => '\\\\', "'" => "\\'" ] ) . "')";
 227                 }
 228         }
 229
 230         /**
 231          * Remaps CSS URL paths and automatically embeds data URIs for CSS rules
 232          * or url() values preceded by an / * @embed * / comment.
 233          *
 234          * @param string $source CSS data to remap
 235          * @param string $local File path where the source was read from
 236          * @param string $remote URL path to the file
 237          * @param bool $embedData If false, never do any data URI embedding,
 238          *   even if / * @embed * / is found.
 239          * @return string Remapped CSS data
 240          */
 241         public static function remap( $source, $local, $remote, $embedData = true ) {
 242                 // High-level overview:
 243                 // * For each CSS rule in $source that includes at least one url() value:
 244                 //   * Check for an @embed comment at the start indicating that all URIs should be embedded
 245                 //   * For each url() value:
 246                 //     * Check for an @embed comment directly preceding the value
 247                 //     * If either @embed comment exists:
 248                 //       * Embedding the URL as data: URI, if it's possible / allowed
 249                 //       * Otherwise remap the URL to work in generated stylesheets
 250
 251                 // Guard against trailing slashes, because "some/remote/../foo.png"
 252                 // resolves to "some/remote/foo.png" on (some?) clients (T29052).
 253                 if ( substr( $remote, -1 ) == '/' ) {
 254                         $remote = substr( $remote, 0, -1 );
 255                 }
 256
 257                 // Disallow U+007F DELETE, which is illegal anyway, and which
 258                 // we use for comment placeholders.
 259                 $source = str_replace( "\x7f", "?", $source );
 260
 261                 // Replace all comments by a placeholder so they will not interfere with the remapping.
 262                 // Warning: This will also catch on anything looking like the start of a comment between
 263                 // quotation marks (e.g. "foo /* bar").
 264                 $comments = [];
 265
 266                 $pattern = '/(?!' . self::EMBED_REGEX . ')(' . self::COMMENT_REGEX . ')/s';
 267
 268                 $source = preg_replace_callback(
 269                         $pattern,
 270                         function ( $match ) use ( &$comments ) {
 271                                 $comments[] = $match[ 0 ];
 272                                 return CSSMin::PLACEHOLDER . ( count( $comments ) - 1 ) . 'x';
 273                         },
 274                         $source
 275                 );
 276
 277                 // Note: This will not correctly handle cases where ';', '{' or '}'
 278                 // appears in the rule itself, e.g. in a quoted string. You are advised
 279                 // not to use such characters in file names. We also match start/end of
 280                 // the string to be consistent in edge-cases ('@import url(…)').
 281                 $pattern = '/(?:^|[;{])\K[^;{}]*' . self::getUrlRegex() . '[^;}]*(?=[;}]|$)/';
 282
 283                 $source = preg_replace_callback(
 284                         $pattern,
 285                         function ( $matchOuter ) use ( $local, $remote, $embedData ) {
 286                                 $rule = $matchOuter[0];
 287
 288                                 // Check for global @embed comment and remove it. Allow other comments to be present
 289                                 // before @embed (they have been replaced with placeholders at this point).
 290                                 $embedAll = false;
 291                                 $rule = preg_replace(
 292                                         '/^((?:\s+|' .
 293                                                 CSSMin::PLACEHOLDER .
 294                                                 '(\d+)x)*)' .
 295                                                 CSSMin::EMBED_REGEX .
 296                                                 '\s*/',
 297                                         '$1',
 298                                         $rule,
 299                                         1,
 300                                         $embedAll
 301                                 );
 302
 303                                 // Build two versions of current rule: with remapped URLs
 304                                 // and with embedded data: URIs (where possible).
 305                                 $pattern = '/(?P<embed>' . CSSMin::EMBED_REGEX . '\s*|)' . self::getUrlRegex() . '/';
 306
 307                                 $ruleWithRemapped = preg_replace_callback(
 308                                         $pattern,
 309                                         function ( $match ) use ( $local, $remote ) {
 310                                                 self::processUrlMatch( $match );
 311
 312                                                 $remapped = CSSMin::remapOne( $match['file'], $match['query'], $local, $remote, false );
 313                                                 return CSSMin::buildUrlValue( $remapped );
 314                                         },
 315                                         $rule
 316                                 );
 317
 318                                 if ( $embedData ) {
 319                                         // Remember the occurring MIME types to avoid fallbacks when embedding some files.
 320                                         $mimeTypes = [];
 321
 322                                         $ruleWithEmbedded = preg_replace_callback(
 323                                                 $pattern,
 324                                                 function ( $match ) use ( $embedAll, $local, $remote, &$mimeTypes ) {
 325                                                         self::processUrlMatch( $match );
 326
 327                                                         $embed = $embedAll || $match['embed'];
 328                                                         $embedded = CSSMin::remapOne(
 329                                                                 $match['file'],
 330                                                                 $match['query'],
 331                                                                 $local,
 332                                                                 $remote,
 333                                                                 $embed
 334                                                         );
 335
 336                                                         $url = $match['file'] . $match['query'];
 337                                                         $file = "{$local}/{$match['file']}";
 338                                                         if (
 339                                                                 !self::isRemoteUrl( $url ) && !self::isLocalUrl( $url )
 340                                                                 && file_exists( $file )
 341                                                         ) {
 342                                                                 $mimeTypes[ CSSMin::getMimeType( $file ) ] = true;
 343                                                         }
 344
 345                                                         return CSSMin::buildUrlValue( $embedded );
 346                                                 },
 347                                                 $rule
 348                                         );
 349
 350                                         // Are all referenced images SVGs?
 351                                         $needsEmbedFallback = $mimeTypes !== [ 'image/svg+xml' => true ];
 352                                 }
 353
 354                                 if ( !$embedData || $ruleWithEmbedded === $ruleWithRemapped ) {
 355                                         // We're not embedding anything, or we tried to but the file is not embeddable
 356                                         return $ruleWithRemapped;
 357                                 } elseif ( $embedData && $needsEmbedFallback ) {
 358                                         // Build 2 CSS properties; one which uses a data URI in place of the @embed comment, and
 359                                         // the other with a remapped and versioned URL with an Internet Explorer 6 and 7 hack
 360                                         // making it ignored in all browsers that support data URIs
 361                                         return "$ruleWithEmbedded;$ruleWithRemapped!ie";
 362                                 } else {
 363                                         // Look ma, no fallbacks! This is for files which IE 6 and 7 don't support anyway: SVG.
 364                                         return $ruleWithEmbedded;
 365                                 }
 366                         }, $source );
 367
 368                 // Re-insert comments
 369                 $pattern = '/' . self::PLACEHOLDER . '(\d+)x/';
 370                 $source = preg_replace_callback( $pattern, function ( $match ) use ( &$comments ) {
 371                         return $comments[ $match[1] ];
 372                 }, $source );
 373
 374                 return $source;
 375         }
 376
 377         /**
 378          * Is this CSS rule referencing a remote URL?
 379          *
 380          * @param string $maybeUrl
 381          * @return bool
 382          */
 383         protected static function isRemoteUrl( $maybeUrl ) {
 384                 if ( substr( $maybeUrl, 0, 2 ) === '//' || parse_url( $maybeUrl, PHP_URL_SCHEME ) ) {
 385                         return true;
 386                 }
 387                 return false;
 388         }
 389
 390         /**
 391          * Is this CSS rule referencing a local URL?
 392          *
 393          * @param string $maybeUrl
 394          * @return bool
 395          */
 396         protected static function isLocalUrl( $maybeUrl ) {
 397                 if ( $maybeUrl !== '' && $maybeUrl[0] === '/' && !self::isRemoteUrl( $maybeUrl ) ) {
 398                         return true;
 399                 }
 400                 return false;
 401         }
 402
 403         /**
 404          * @codeCoverageIgnore
 405          */
 406         private static function getUrlRegex() {
 407                 static $urlRegex;
 408                 if ( $urlRegex === null ) {
 409                         // Match these three variants separately to avoid broken urls when
 410                         // e.g. a double quoted url contains a parenthesis, or when a
 411                         // single quoted url contains a double quote, etc.
 412                         // Note: PCRE doesn't support multiple capture groups with the same name by default.
 413                         // - PCRE 6.7 introduced the "J" modifier (PCRE_INFO_JCHANGED for PCRE_DUPNAMES).
 414                         //   https://secure.php.net/manual/en/reference.pcre.pattern.modifiers.php
 415                         //   However this isn't useful since it just ignores all but the first one.
 416                         //   Also, while the modifier was introduced in PCRE 6.7 (PHP 5.2+) it was
 417                         //   not exposed to public preg_* functions until PHP 5.6.0.
 418                         // - PCRE 8.36 fixed this to work as expected (e.g. merge conceptually to
 419                         //   only return the one matched in the part that actually matched).
 420                         //   However MediaWiki supports 5.5.9, which has PCRE 8.32
 421                         //   Per https://secure.php.net/manual/en/pcre.installation.php:
 422                         //   - PCRE 8.32 (PHP 5.5.0)
 423                         //   - PCRE 8.34 (PHP 5.5.10, PHP 5.6.0)
 424                         //   - PCRE 8.37 (PHP 5.5.26, PHP 5.6.9, PHP 7.0.0)
 425                         //   Workaround by using different groups and merge via processUrlMatch().
 426                         // - Using string concatenation for class constant or member assignments
 427                         //   is only supported in PHP 5.6. Use a getter method for now.
 428                         $urlRegex = '(' .
 429                                 // Unquoted url
 430                                 'url\(\s*(?P<file0>[^\'"][^\?\)]*?)(?P<query0>\?[^\)]*?|)\s*\)' .
 431                                 // Single quoted url
 432                                 '|url\(\s*\'(?P<file1>[^\?\']*?)(?P<query1>\?[^\']*?|)\'\s*\)' .
 433                                 // Double quoted url
 434                                 '|url\(\s*"(?P<file2>[^\?"]*?)(?P<query2>\?[^"]*?|)"\s*\)' .
 435                                 ')';
 436                 }
 437                 return $urlRegex;
 438         }
 439
 440         private static function processUrlMatch( array &$match, $flags = 0 ) {
 441                 if ( $flags & PREG_SET_ORDER ) {
 442                         // preg_match_all with PREG_SET_ORDER will return each group in each
 443                         // match array, and if it didn't match, instead of the sub array
 444                         // being an empty array it is `[ '', -1 ]`...
 445                         if ( isset( $match['file0'] ) && $match['file0'][1] !== -1 ) {
 446                                 $match['file'] = $match['file0'];
 447                                 $match['query'] = $match['query0'];
 448                         } elseif ( isset( $match['file1'] ) && $match['file1'][1] !== -1 ) {
 449                                 $match['file'] = $match['file1'];
 450                                 $match['query'] = $match['query1'];
 451                         } else {
 452                                 $match['file'] = $match['file2'];
 453                                 $match['query'] = $match['query2'];
 454                         }
 455                 } else {
 456                         if ( isset( $match['file0'] ) && $match['file0'] !== '' ) {
 457                                 $match['file'] = $match['file0'];
 458                                 $match['query'] = $match['query0'];
 459                         } elseif ( isset( $match['file1'] ) && $match['file1'] !== '' ) {
 460                                 $match['file'] = $match['file1'];
 461                                 $match['query'] = $match['query1'];
 462                         } else {
 463                                 $match['file'] = $match['file2'];
 464                                 $match['query'] = $match['query2'];
 465                         }
 466                 }
 467         }
 468
 469         /**
 470          * Remap or embed a CSS URL path.
 471          *
 472          * @param string $file URL to remap/embed
 473          * @param string $query
 474          * @param string $local File path where the source was read from
 475          * @param string $remote URL path to the file
 476          * @param bool $embed Whether to do any data URI embedding
 477          * @return string Remapped/embedded URL data
 478          */
 479         public static function remapOne( $file, $query, $local, $remote, $embed ) {
 480                 // The full URL possibly with query, as passed to the 'url()' value in CSS
 481                 $url = $file . $query;
 482
 483                 // Expand local URLs with absolute paths like /w/index.php to possibly protocol-relative URL, if
 484                 // wfExpandUrl() is available. (This will not be the case if we're running outside of MW.)
 485                 if ( self::isLocalUrl( $url ) && function_exists( 'wfExpandUrl' ) ) {
 486                         return wfExpandUrl( $url, PROTO_RELATIVE );
 487                 }
 488
 489                 // Pass thru fully-qualified and protocol-relative URLs and data URIs, as well as local URLs if
 490                 // we can't expand them.
 491                 // Also skips the rare `behavior` property specifying application's default behavior
 492                 if (
 493                         self::isRemoteUrl( $url ) ||
 494                         self::isLocalUrl( $url ) ||
 495                         substr( $url, 0, 9 ) === '#default#'
 496                 ) {
 497                         return $url;
 498                 }
 499
 500                 if ( $local === false ) {
 501                         // Assume that all paths are relative to $remote, and make them absolute
 502                         $url = $remote . '/' . $url;
 503                 } else {
 504                         // We drop the query part here and instead make the path relative to $remote
 505                         $url = "{$remote}/{$file}";
 506                         // Path to the actual file on the filesystem
 507                         $localFile = "{$local}/{$file}";
 508                         if ( file_exists( $localFile ) ) {
 509                                 if ( $embed ) {
 510                                         $data = self::encodeImageAsDataURI( $localFile );
 511                                         if ( $data !== false ) {
 512                                                 return $data;
 513                                         }
 514                                 }
 515                                 if ( method_exists( 'OutputPage', 'transformFilePath' ) ) {
 516                                         $url = OutputPage::transformFilePath( $remote, $local, $file );
 517                                 } else {
 518                                         // Add version parameter as the first five hex digits
 519                                         // of the MD5 hash of the file's contents.
 520                                         $url .= '?' . substr( md5_file( $localFile ), 0, 5 );
 521                                 }
 522                         }
 523                         // If any of these conditions failed (file missing, we don't want to embed it
 524                         // or it's not embeddable), return the URL (possibly with ?timestamp part)
 525                 }
 526                 if ( function_exists( 'wfRemoveDotSegments' ) ) {
 527                         $url = wfRemoveDotSegments( $url );
 528                 }
 529                 return $url;
 530         }
 531
 532         /**
 533          * Removes whitespace from CSS data
 534          *
 535          * @param string $css CSS data to minify
 536          * @return string Minified CSS data
 537          */
 538         public static function minify( $css ) {
 539                 return trim(
 540                         str_replace(
 541                                 [ '; ', ': ', ' {', '{ ', ', ', '} ', ';}' ],
 542                                 [ ';', ':', '{', '{', ',', '}', '}' ],
 543                                 preg_replace( [ '/\s+/', '/\/\*.*?\*\//s' ], [ ' ', '' ], $css )
 544                         )
 545                 );
 546         }
 547 }