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         /** @var string Strip marker for comments. **/
  33         const PLACEHOLDER = "\x7fPLACEHOLDER\x7f";
  34
  35         /**
  36          * Internet Explorer data URI length limit. See encodeImageAsDataURI().
  37          */
  38         const DATA_URI_SIZE_LIMIT = 32768;
  39
  40         const EMBED_REGEX = '\/\*\s*\@embed\s*\*\/';
  41         const COMMENT_REGEX = '\/\*.*?\*\/';
  42
  43         /** @var array List of common image files extensions and MIME-types */
  44         protected static $mimeTypes = [
  45                 'gif' => 'image/gif',
  46                 'jpe' => 'image/jpeg',
  47                 'jpeg' => 'image/jpeg',
  48                 'jpg' => 'image/jpeg',
  49                 'png' => 'image/png',
  50                 'tif' => 'image/tiff',
  51                 'tiff' => 'image/tiff',
  52                 'xbm' => 'image/x-xbitmap',
  53                 'svg' => 'image/svg+xml',
  54         ];
  55
  56         /**
  57          * Get a list of local files referenced in a stylesheet (includes non-existent files).
  58          *
  59          * @param string $source CSS stylesheet source to process
  60          * @param string $path File path where the source was read from
  61          * @return array List of local file references
  62          */
  63         public static function getLocalFileReferences( $source, $path ) {
  64                 $stripped = preg_replace( '/' . self::COMMENT_REGEX . '/s', '', $source );
  65                 $path = rtrim( $path, '/' ) . '/';
  66                 $files = [];
  67
  68                 $rFlags = PREG_OFFSET_CAPTURE | PREG_SET_ORDER;
  69                 if ( preg_match_all( '/' . self::getUrlRegex() . '/', $stripped, $matches, $rFlags ) ) {
  70                         foreach ( $matches as $match ) {
  71                                 self::processUrlMatch( $match, $rFlags );
  72                                 $url = $match['file'][0];
  73
  74                                 // Skip fully-qualified and protocol-relative URLs and data URIs
  75                                 // Also skips the rare `behavior` property specifying application's default behavior
  76                                 if (
  77                                         substr( $url, 0, 2 ) === '//' ||
  78                                         parse_url( $url, PHP_URL_SCHEME ) ||
  79                                         substr( $url, 0, 9 ) === '#default#'
  80                                 ) {
  81                                         break;
  82                                 }
  83
  84                                 $files[] = $path . $url;
  85                         }
  86                 }
  87                 return $files;
  88         }
  89
  90         /**
  91          * Encode an image file as a data URI.
  92          *
  93          * If the image file has a suitable MIME type and size, encode it as a data URI, base64-encoded
  94          * for binary files or just percent-encoded otherwise. Return false if the image type is
  95          * unfamiliar or file exceeds the size limit.
  96          *
  97          * @param string $file Image file to encode.
  98          * @param string|null $type File's MIME type or null. If null, CSSMin will
  99          *     try to autodetect the type.
 100          * @param bool $ie8Compat By default, a data URI will only be produced if it can be made short
 101          *     enough to fit in Internet Explorer 8 (and earlier) URI length limit (32,768 bytes). Pass
 102          *     `false` to remove this limitation.
 103          * @return string|bool Image contents encoded as a data URI or false.
 104          */
 105         public static function encodeImageAsDataURI( $file, $type = null, $ie8Compat = true ) {
 106                 // Fast-fail for files that definitely exceed the maximum data URI length
 107                 if ( $ie8Compat && filesize( $file ) >= self::DATA_URI_SIZE_LIMIT ) {
 108                         return false;
 109                 }
 110
 111                 if ( $type === null ) {
 112                         $type = self::getMimeType( $file );
 113                 }
 114                 if ( !$type ) {
 115                         return false;
 116                 }
 117
 118                 return self::encodeStringAsDataURI( file_get_contents( $file ), $type, $ie8Compat );
 119         }
 120
 121         /**
 122          * Encode file contents as a data URI with chosen MIME type.
 123          *
 124          * The URI will be base64-encoded for binary files or just percent-encoded otherwise.
 125          *
 126          * @since 1.25
 127          *
 128          * @param string $contents File contents to encode.
 129          * @param string $type File's MIME type.
 130          * @param bool $ie8Compat See encodeImageAsDataURI().
 131          * @return string|bool Image contents encoded as a data URI or false.
 132          */
 133         public static function encodeStringAsDataURI( $contents, $type, $ie8Compat = true ) {
 134                 // Try #1: Non-encoded data URI
 135
 136                 // Remove XML declaration, it's not needed with data URI usage
 137                 $contents = preg_replace( "/<\\?xml.*?\\?>/", '', $contents );
 138                 // The regular expression matches ASCII whitespace and printable characters.
 139                 if ( preg_match( '/^[\r\n\t\x20-\x7e]+$/', $contents ) ) {
 140                         // Do not base64-encode non-binary files (sane SVGs).
 141                         // (This often produces longer URLs, but they compress better, yielding a net smaller size.)
 142                         $encoded = rawurlencode( $contents );
 143                         // Unencode some things that don't need to be encoded, to make the encoding smaller
 144                         $encoded = strtr( $encoded, [
 145                                 '%20' => ' ', // Unencode spaces
 146                                 '%2F' => '/', // Unencode slashes
 147                                 '%3A' => ':', // Unencode colons
 148                                 '%3D' => '=', // Unencode equals signs
 149                                 '%0A' => ' ', // Change newlines to spaces
 150                                 '%0D' => ' ', // Change carriage returns to spaces
 151                                 '%09' => ' ', // Change tabs to spaces
 152                         ] );
 153                         // Consolidate runs of multiple spaces in a row
 154                         $encoded = preg_replace( '/ {2,}/', ' ', $encoded );
 155                         // Remove leading and trailing spaces
 156                         $encoded = preg_replace( '/^ | $/', '', $encoded );
 157
 158                         $uri = 'data:' . $type . ',' . $encoded;
 159                         if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 160                                 return $uri;
 161                         }
 162                 }
 163
 164                 // Try #2: Encoded data URI
 165                 $uri = 'data:' . $type . ';base64,' . base64_encode( $contents );
 166                 if ( !$ie8Compat || strlen( $uri ) < self::DATA_URI_SIZE_LIMIT ) {
 167                         return $uri;
 168                 }
 169
 170                 // A data URI couldn't be produced
 171                 return false;
 172         }
 173
 174         /**
 175          * Serialize a string (escape and quote) for use as a CSS string value.
 176          * http://www.w3.org/TR/2013/WD-cssom-20131205/#serialize-a-string
 177          *
 178          * @param string $value
 179          * @return string
 180          * @throws Exception
 181          */
 182         public static function serializeStringValue( $value ) {
 183                 if ( strstr( $value, "\0" ) ) {
 184                         throw new Exception( "Invalid character in CSS string" );
 185                 }
 186                 $value = strtr( $value, [ '\\' => '\\\\', '"' => '\\"' ] );
 187                 $value = preg_replace_callback( '/[\x01-\x1f\x7f-\x9f]/', function ( $match ) {
 188                         return '\\' . base_convert( ord( $match[0] ), 10, 16 ) . ' ';
 189                 }, $value );
 190                 return '"' . $value . '"';
 191         }
 192
 193         /**
 194          * @param string $file
 195          * @return bool|string
 196          */
 197         public static function getMimeType( $file ) {
 198                 // Infer the MIME-type from the file extension
 199                 $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) );
 200                 if ( isset( self::$mimeTypes[$ext] ) ) {
 201                         return self::$mimeTypes[$ext];
 202                 }
 203
 204                 return mime_content_type( realpath( $file ) );
 205         }
 206
 207         /**
 208          * Build a CSS 'url()' value for the given URL, quoting parentheses (and other funny characters)
 209          * and escaping quotes as necessary.
 210          *
 211          * See http://www.w3.org/TR/css-syntax-3/#consume-a-url-token
 212          *
 213          * @param string $url URL to process
 214          * @return string 'url()' value, usually just `"url($url)"`, quoted/escaped if necessary
 215          */
 216         public static function buildUrlValue( $url ) {
 217                 // The list below has been crafted to match URLs such as:
 218                 //   scheme://user@domain:port/~user/fi%20le.png?query=yes&really=y+s
 219                 //   data:image/png;base64,R0lGODlh/+==
 220                 if ( preg_match( '!^[\w\d:@/~.%+;,?&=-]+$!', $url ) ) {
 221                         return "url($url)";
 222                 } else {
 223                         return 'url("' . strtr( $url, [ '\\' => '\\\\', '"' => '\\"' ] ) . '")';
 224                 }
 225         }
 226
 227         /**
 228          * Remaps CSS URL paths and automatically embeds data URIs for CSS rules
 229          * or url() values preceded by an / * @embed * / comment.
 230          *
 231          * @param string $source CSS data to remap
 232          * @param string $local File path where the source was read from
 233          * @param string $remote URL path to the file
 234          * @param bool $embedData If false, never do any data URI embedding,
 235          *   even if / * @embed * / is found.
 236          * @return string Remapped CSS data
 237          */
 238         public static function remap( $source, $local, $remote, $embedData = true ) {
 239                 // High-level overview:
 240                 // * For each CSS rule in $source that includes at least one url() value:
 241                 //   * Check for an @embed comment at the start indicating that all URIs should be embedded
 242                 //   * For each url() value:
 243                 //     * Check for an @embed comment directly preceding the value
 244                 //     * If either @embed comment exists:
 245                 //       * Embedding the URL as data: URI, if it's possible / allowed
 246                 //       * Otherwise remap the URL to work in generated stylesheets
 247
 248                 // Guard against trailing slashes, because "some/remote/../foo.png"
 249                 // resolves to "some/remote/foo.png" on (some?) clients (T29052).
 250                 if ( substr( $remote, -1 ) == '/' ) {
 251                         $remote = substr( $remote, 0, -1 );
 252                 }
 253
 254                 // Disallow U+007F DELETE, which is illegal anyway, and which
 255                 // we use for comment placeholders.
 256                 $source = str_replace( "\x7f", "?", $source );
 257
 258                 // Replace all comments by a placeholder so they will not interfere with the remapping.
 259                 // Warning: This will also catch on anything looking like the start of a comment between
 260                 // quotation marks (e.g. "foo /* bar").
 261                 $comments = [];
 262
 263                 $pattern = '/(?!' . self::EMBED_REGEX . ')(' . self::COMMENT_REGEX . ')/s';
 264
 265                 $source = preg_replace_callback(
 266                         $pattern,
 267                         function ( $match ) use ( &$comments ) {
 268                                 $comments[] = $match[ 0 ];
 269                                 return CSSMin::PLACEHOLDER . ( count( $comments ) - 1 ) . 'x';
 270                         },
 271                         $source
 272                 );
 273
 274                 // Note: This will not correctly handle cases where ';', '{' or '}'
 275                 // appears in the rule itself, e.g. in a quoted string. You are advised
 276                 // not to use such characters in file names. We also match start/end of
 277                 // the string to be consistent in edge-cases ('@import url(…)').
 278                 $pattern = '/(?:^|[;{])\K[^;{}]*' . self::getUrlRegex() . '[^;}]*(?=[;}]|$)/';
 279
 280                 $source = preg_replace_callback(
 281                         $pattern,
 282                         function ( $matchOuter ) use ( $local, $remote, $embedData ) {
 283                                 $rule = $matchOuter[0];
 284
 285                                 // Check for global @embed comment and remove it. Allow other comments to be present
 286                                 // before @embed (they have been replaced with placeholders at this point).
 287                                 $embedAll = false;
 288                                 $rule = preg_replace(
 289                                         '/^((?:\s+|' .
 290                                                 CSSMin::PLACEHOLDER .
 291                                                 '(\d+)x)*)' .
 292                                                 CSSMin::EMBED_REGEX .
 293                                                 '\s*/',
 294                                         '$1',
 295                                         $rule,
 296                                         1,
 297                                         $embedAll
 298                                 );
 299
 300                                 // Build two versions of current rule: with remapped URLs
 301                                 // and with embedded data: URIs (where possible).
 302                                 $pattern = '/(?P<embed>' . CSSMin::EMBED_REGEX . '\s*|)' . self::getUrlRegex() . '/';
 303
 304                                 $ruleWithRemapped = preg_replace_callback(
 305                                         $pattern,
 306                                         function ( $match ) use ( $local, $remote ) {
 307                                                 self::processUrlMatch( $match );
 308
 309                                                 $remapped = CSSMin::remapOne( $match['file'], $match['query'], $local, $remote, false );
 310                                                 return CSSMin::buildUrlValue( $remapped );
 311                                         },
 312                                         $rule
 313                                 );
 314
 315                                 if ( $embedData ) {
 316                                         // Remember the occurring MIME types to avoid fallbacks when embedding some files.
 317                                         $mimeTypes = [];
 318
 319                                         $ruleWithEmbedded = preg_replace_callback(
 320                                                 $pattern,
 321                                                 function ( $match ) use ( $embedAll, $local, $remote, &$mimeTypes ) {
 322                                                         self::processUrlMatch( $match );
 323
 324                                                         $embed = $embedAll || $match['embed'];
 325                                                         $embedded = CSSMin::remapOne(
 326                                                                 $match['file'],
 327                                                                 $match['query'],
 328                                                                 $local,
 329                                                                 $remote,
 330                                                                 $embed
 331                                                         );
 332
 333                                                         $url = $match['file'] . $match['query'];
 334                                                         $file = "{$local}/{$match['file']}";
 335                                                         if (
 336                                                                 !self::isRemoteUrl( $url ) && !self::isLocalUrl( $url )
 337                                                                 && file_exists( $file )
 338                                                         ) {
 339                                                                 $mimeTypes[ CSSMin::getMimeType( $file ) ] = true;
 340                                                         }
 341
 342                                                         return CSSMin::buildUrlValue( $embedded );
 343                                                 },
 344                                                 $rule
 345                                         );
 346
 347                                         // Are all referenced images SVGs?
 348                                         $needsEmbedFallback = $mimeTypes !== [ 'image/svg+xml' => true ];
 349                                 }
 350
 351                                 if ( !$embedData || $ruleWithEmbedded === $ruleWithRemapped ) {
 352                                         // We're not embedding anything, or we tried to but the file is not embeddable
 353                                         return $ruleWithRemapped;
 354                                 } elseif ( $embedData && $needsEmbedFallback ) {
 355                                         // Build 2 CSS properties; one which uses a data URI in place of the @embed comment, and
 356                                         // the other with a remapped and versioned URL with an Internet Explorer 6 and 7 hack
 357                                         // making it ignored in all browsers that support data URIs
 358                                         return "$ruleWithEmbedded;$ruleWithRemapped!ie";
 359                                 } else {
 360                                         // Look ma, no fallbacks! This is for files which IE 6 and 7 don't support anyway: SVG.
 361                                         return $ruleWithEmbedded;
 362                                 }
 363                         }, $source );
 364
 365                 // Re-insert comments
 366                 $pattern = '/' . self::PLACEHOLDER . '(\d+)x/';
 367                 $source = preg_replace_callback( $pattern, function ( $match ) use ( &$comments ) {
 368                         return $comments[ $match[1] ];
 369                 }, $source );
 370
 371                 return $source;
 372         }
 373
 374         /**
 375          * Is this CSS rule referencing a remote URL?
 376          *
 377          * @param string $maybeUrl
 378          * @return bool
 379          */
 380         protected static function isRemoteUrl( $maybeUrl ) {
 381                 if ( substr( $maybeUrl, 0, 2 ) === '//' || parse_url( $maybeUrl, PHP_URL_SCHEME ) ) {
 382                         return true;
 383                 }
 384                 return false;
 385         }
 386
 387         /**
 388          * Is this CSS rule referencing a local URL?
 389          *
 390          * @param string $maybeUrl
 391          * @return bool
 392          */
 393         protected static function isLocalUrl( $maybeUrl ) {
 394                 if ( $maybeUrl !== '' && $maybeUrl[0] === '/' && !self::isRemoteUrl( $maybeUrl ) ) {
 395                         return true;
 396                 }
 397                 return false;
 398         }
 399
 400         /**
 401          * @codeCoverageIgnore
 402          */
 403         private static function getUrlRegex() {
 404                 static $urlRegex;
 405                 if ( $urlRegex === null ) {
 406                         // Match these three variants separately to avoid broken urls when
 407                         // e.g. a double quoted url contains a parenthesis, or when a
 408                         // single quoted url contains a double quote, etc.
 409                         // Note: PCRE doesn't support multiple capture groups with the same name by default.
 410                         // - PCRE 6.7 introduced the "J" modifier (PCRE_INFO_JCHANGED for PCRE_DUPNAMES).
 411                         //   https://secure.php.net/manual/en/reference.pcre.pattern.modifiers.php
 412                         //   However this isn't useful since it just ignores all but the first one.
 413                         //   Also, while the modifier was introduced in PCRE 6.7 (PHP 5.2+) it was
 414                         //   not exposed to public preg_* functions until PHP 5.6.0.
 415                         // - PCRE 8.36 fixed this to work as expected (e.g. merge conceptually to
 416                         //   only return the one matched in the part that actually matched).
 417                         //   However MediaWiki supports 5.5.9, which has PCRE 8.32
 418                         //   Per https://secure.php.net/manual/en/pcre.installation.php:
 419                         //   - PCRE 8.32 (PHP 5.5.0)
 420                         //   - PCRE 8.34 (PHP 5.5.10, PHP 5.6.0)
 421                         //   - PCRE 8.37 (PHP 5.5.26, PHP 5.6.9, PHP 7.0.0)
 422                         //   Workaround by using different groups and merge via processUrlMatch().
 423                         // - Using string concatenation for class constant or member assignments
 424                         //   is only supported in PHP 5.6. Use a getter method for now.
 425                         $urlRegex = '(' .
 426                                 // Unquoted url
 427                                 'url\(\s*(?P<file0>[^\'"][^\?\)]+?)(?P<query0>\?[^\)]*?|)\s*\)' .
 428                                 // Single quoted url
 429                                 '|url\(\s*\'(?P<file1>[^\?\']+?)(?P<query1>\?[^\']*?|)\'\s*\)' .
 430                                 // Double quoted url
 431                                 '|url\(\s*"(?P<file2>[^\?"]+?)(?P<query2>\?[^"]*?|)"\s*\)' .
 432                                 ')';
 433                 }
 434                 return $urlRegex;
 435         }
 436
 437         private static function processUrlMatch( array &$match, $flags = 0 ) {
 438                 if ( $flags & PREG_SET_ORDER ) {
 439                         // preg_match_all with PREG_SET_ORDER will return each group in each
 440                         // match array, and if it didn't match, instead of the sub array
 441                         // being an empty array it is `[ '', -1 ]`...
 442                         if ( isset( $match['file0'] ) && $match['file0'][1] !== -1 ) {
 443                                 $match['file'] = $match['file0'];
 444                                 $match['query'] = $match['query0'];
 445                         } elseif ( isset( $match['file1'] ) && $match['file1'][1] !== -1 ) {
 446                                 $match['file'] = $match['file1'];
 447                                 $match['query'] = $match['query1'];
 448                         } else {
 449                                 if ( !isset( $match['file2'] ) || $match['file2'][1] === -1 ) {
 450                                         throw new Exception( 'URL must be non-empty' );
 451                                 }
 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                                 if ( !isset( $match['file2'] ) || $match['file2'] === '' ) {
 464                                         throw new Exception( 'URL must be non-empty' );
 465                                 }
 466                                 $match['file'] = $match['file2'];
 467                                 $match['query'] = $match['query2'];
 468                         }
 469                 }
 470         }
 471
 472         /**
 473          * Remap or embed a CSS URL path.
 474          *
 475          * @param string $file URL to remap/embed
 476          * @param string $query
 477          * @param string $local File path where the source was read from
 478          * @param string $remote URL path to the file
 479          * @param bool $embed Whether to do any data URI embedding
 480          * @return string Remapped/embedded URL data
 481          */
 482         public static function remapOne( $file, $query, $local, $remote, $embed ) {
 483                 // The full URL possibly with query, as passed to the 'url()' value in CSS
 484                 $url = $file . $query;
 485
 486                 // Expand local URLs with absolute paths like /w/index.php to possibly protocol-relative URL, if
 487                 // wfExpandUrl() is available. (This will not be the case if we're running outside of MW.)
 488                 if ( self::isLocalUrl( $url ) && function_exists( 'wfExpandUrl' ) ) {
 489                         return wfExpandUrl( $url, PROTO_RELATIVE );
 490                 }
 491
 492                 // Pass thru fully-qualified and protocol-relative URLs and data URIs, as well as local URLs if
 493                 // we can't expand them.
 494                 // Also skips the rare `behavior` property specifying application's default behavior
 495                 if (
 496                         self::isRemoteUrl( $url ) ||
 497                         self::isLocalUrl( $url ) ||
 498                         substr( $url, 0, 9 ) === '#default#'
 499                 ) {
 500                         return $url;
 501                 }
 502
 503                 if ( $local === false ) {
 504                         // Assume that all paths are relative to $remote, and make them absolute
 505                         $url = $remote . '/' . $url;
 506                 } else {
 507                         // We drop the query part here and instead make the path relative to $remote
 508                         $url = "{$remote}/{$file}";
 509                         // Path to the actual file on the filesystem
 510                         $localFile = "{$local}/{$file}";
 511                         if ( file_exists( $localFile ) ) {
 512                                 if ( $embed ) {
 513                                         $data = self::encodeImageAsDataURI( $localFile );
 514                                         if ( $data !== false ) {
 515                                                 return $data;
 516                                         }
 517                                 }
 518                                 if ( method_exists( 'OutputPage', 'transformFilePath' ) ) {
 519                                         $url = OutputPage::transformFilePath( $remote, $local, $file );
 520                                 } else {
 521                                         // Add version parameter as the first five hex digits
 522                                         // of the MD5 hash of the file's contents.
 523                                         $url .= '?' . substr( md5_file( $localFile ), 0, 5 );
 524                                 }
 525                         }
 526                         // If any of these conditions failed (file missing, we don't want to embed it
 527                         // or it's not embeddable), return the URL (possibly with ?timestamp part)
 528                 }
 529                 if ( function_exists( 'wfRemoveDotSegments' ) ) {
 530                         $url = wfRemoveDotSegments( $url );
 531                 }
 532                 return $url;
 533         }
 534
 535         /**
 536          * Removes whitespace from CSS data
 537          *
 538          * @param string $css CSS data to minify
 539          * @return string Minified CSS data
 540          */
 541         public static function minify( $css ) {
 542                 return trim(
 543                         str_replace(
 544                                 [ '; ', ': ', ' {', '{ ', ', ', '} ', ';}', '( ', ' )', '[ ', ' ]' ],
 545                                 [ ';', ':', '{', '{', ',', '}', '}', '(', ')', '[', ']' ],
 546                                 preg_replace( [ '/\s+/', '/\/\*.*?\*\//s' ], [ ' ', '' ], $css )
 547                         )
 548                 );
 549         }
 550 }