MediaWiki  REL1_31
CSSMin.php
Go to the documentation of this file.
1 <?php
30 class CSSMin {
31 
33  const PLACEHOLDER = "\x7fPLACEHOLDER\x7f";
34 
38  const DATA_URI_SIZE_LIMIT = 32768;
39 
40  const EMBED_REGEX = '\/\*\s*\@embed\s*\*\/';
41  const COMMENT_REGEX = '\/\*.*?\*\/';
42 
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 
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  if (
76  substr( $url, 0, 2 ) === '//' ||
77  parse_url( $url, PHP_URL_SCHEME )
78  ) {
79  break;
80  }
81 
82  // Strip trailing anchors - T115436
83  $anchor = strpos( $url, '#' );
84  if ( $anchor !== false ) {
85  $url = substr( $url, 0, $anchor );
86 
87  // '#some-anchors' is not a file
88  if ( $url === '' ) {
89  break;
90  }
91  }
92 
93  $files[] = $path . $url;
94  }
95  }
96  return $files;
97  }
98 
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 
142  public static function encodeStringAsDataURI( $contents, $type, $ie8Compat = true ) {
143  // Try #1: Non-encoded data URI
144 
145  // Remove XML declaration, it's not needed with data URI usage
146  $contents = preg_replace( "/<\\?xml.*?\\?>/", '', $contents );
147  // The regular expression matches ASCII whitespace and printable characters.
148  if ( preg_match( '/^[\r\n\t\x20-\x7e]+$/', $contents ) ) {
149  // Do not base64-encode non-binary files (sane SVGs).
150  // (This often produces longer URLs, but they compress better, yielding a net smaller size.)
151  $encoded = rawurlencode( $contents );
152  // Unencode some things that don't need to be encoded, to make the encoding smaller
153  $encoded = strtr( $encoded, [
154  '%20' => ' ', // Unencode spaces
155  '%2F' => '/', // Unencode slashes
156  '%3A' => ':', // Unencode colons
157  '%3D' => '=', // Unencode equals signs
158  '%0A' => ' ', // Change newlines to spaces
159  '%0D' => ' ', // Change carriage returns to spaces
160  '%09' => ' ', // Change tabs to spaces
161  ] );
162  // Consolidate runs of multiple spaces in a row
163  $encoded = preg_replace( '/ {2,}/', ' ', $encoded );
164  // Remove leading and trailing spaces
165  $encoded = preg_replace( '/^ | $/', '', $encoded );
166 
167  $uri = 'data:' . $type . ',' . $encoded;
168  if ( !$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 
190  public static function serializeStringValue( $value ) {
191  $value = strtr( $value, [ "\0" => "\\fffd ", '\\' => '\\\\', '"' => '\\"' ] );
192  $value = preg_replace_callback( '/[\x01-\x1f\x7f]/', function ( $match ) {
193  return '\\' . base_convert( ord( $match[0] ), 10, 16 ) . ' ';
194  }, $value );
195  return '"' . $value . '"';
196  }
197 
202  public static function getMimeType( $file ) {
203  // Infer the MIME-type from the file extension
204  $ext = strtolower( pathinfo( $file, PATHINFO_EXTENSION ) );
205  if ( isset( self::$mimeTypes[$ext] ) ) {
206  return self::$mimeTypes[$ext];
207  }
208 
209  return mime_content_type( realpath( $file ) );
210  }
211 
221  public static function buildUrlValue( $url ) {
222  // The list below has been crafted to match URLs such as:
223  // scheme://user@domain:port/~user/fi%20le.png?query=yes&really=y+s
224  // data:image/png;base64,R0lGODlh/+==
225  if ( preg_match( '!^[\w\d:@/~.%+;,?&=-]+$!', $url ) ) {
226  return "url($url)";
227  } else {
228  return 'url("' . strtr( $url, [ '\\' => '\\\\', '"' => '\\"' ] ) . '")';
229  }
230  }
231 
243  public static function remap( $source, $local, $remote, $embedData = true ) {
244  // High-level overview:
245  // * For each CSS rule in $source that includes at least one url() value:
246  // * Check for an @embed comment at the start indicating that all URIs should be embedded
247  // * For each url() value:
248  // * Check for an @embed comment directly preceding the value
249  // * If either @embed comment exists:
250  // * Embedding the URL as data: URI, if it's possible / allowed
251  // * Otherwise remap the URL to work in generated stylesheets
252 
253  // Guard against trailing slashes, because "some/remote/../foo.png"
254  // resolves to "some/remote/foo.png" on (some?) clients (T29052).
255  if ( substr( $remote, -1 ) == '/' ) {
256  $remote = substr( $remote, 0, -1 );
257  }
258 
259  // Disallow U+007F DELETE, which is illegal anyway, and which
260  // we use for comment placeholders.
261  $source = str_replace( "\x7f", "?", $source );
262 
263  // Replace all comments by a placeholder so they will not interfere with the remapping.
264  // Warning: This will also catch on anything looking like the start of a comment between
265  // quotation marks (e.g. "foo /* bar").
266  $comments = [];
267 
268  $pattern = '/(?!' . self::EMBED_REGEX . ')(' . self::COMMENT_REGEX . ')/s';
269 
270  $source = preg_replace_callback(
271  $pattern,
272  function ( $match ) use ( &$comments ) {
273  $comments[] = $match[ 0 ];
274  return CSSMin::PLACEHOLDER . ( count( $comments ) - 1 ) . 'x';
275  },
276  $source
277  );
278 
279  // Note: This will not correctly handle cases where ';', '{' or '}'
280  // appears in the rule itself, e.g. in a quoted string. You are advised
281  // not to use such characters in file names. We also match start/end of
282  // the string to be consistent in edge-cases ('@import url(…)').
283  $pattern = '/(?:^|[;{])\K[^;{}]*' . self::getUrlRegex() . '[^;}]*(?=[;}]|$)/';
284 
285  $source = preg_replace_callback(
286  $pattern,
287  function ( $matchOuter ) use ( $local, $remote, $embedData ) {
288  $rule = $matchOuter[0];
289 
290  // Check for global @embed comment and remove it. Allow other comments to be present
291  // before @embed (they have been replaced with placeholders at this point).
292  $embedAll = false;
293  $rule = preg_replace(
294  '/^((?:\s+|' .
295  CSSMin::PLACEHOLDER .
296  '(\d+)x)*)' .
298  '\s*/',
299  '$1',
300  $rule,
301  1,
302  $embedAll
303  );
304 
305  // Build two versions of current rule: with remapped URLs
306  // and with embedded data: URIs (where possible).
307  $pattern = '/(?P<embed>' . CSSMin::EMBED_REGEX . '\s*|)' . self::getUrlRegex() . '/';
308 
309  $ruleWithRemapped = preg_replace_callback(
310  $pattern,
311  function ( $match ) use ( $local, $remote ) {
312  self::processUrlMatch( $match );
313 
314  $remapped = CSSMin::remapOne( $match['file'], $match['query'], $local, $remote, false );
315  return CSSMin::buildUrlValue( $remapped );
316  },
317  $rule
318  );
319 
320  if ( $embedData ) {
321  // Remember the occurring MIME types to avoid fallbacks when embedding some files.
322  $mimeTypes = [];
323 
324  $ruleWithEmbedded = preg_replace_callback(
325  $pattern,
326  function ( $match ) use ( $embedAll, $local, $remote, &$mimeTypes ) {
327  self::processUrlMatch( $match );
328 
329  $embed = $embedAll || $match['embed'];
330  $embedded = CSSMin::remapOne(
331  $match['file'],
332  $match['query'],
333  $local,
334  $remote,
335  $embed
336  );
337 
338  $url = $match['file'] . $match['query'];
339  $file = "{$local}/{$match['file']}";
340  if (
341  !self::isRemoteUrl( $url ) && !self::isLocalUrl( $url )
342  && file_exists( $file )
343  ) {
344  $mimeTypes[ CSSMin::getMimeType( $file ) ] = true;
345  }
346 
347  return CSSMin::buildUrlValue( $embedded );
348  },
349  $rule
350  );
351 
352  // Are all referenced images SVGs?
353  $needsEmbedFallback = $mimeTypes !== [ 'image/svg+xml' => true ];
354  }
355 
356  if ( !$embedData || $ruleWithEmbedded === $ruleWithRemapped ) {
357  // We're not embedding anything, or we tried to but the file is not embeddable
358  return $ruleWithRemapped;
359  } elseif ( $embedData && $needsEmbedFallback ) {
360  // Build 2 CSS properties; one which uses a data URI in place of the @embed comment, and
361  // the other with a remapped and versioned URL with an Internet Explorer 6 and 7 hack
362  // making it ignored in all browsers that support data URIs
363  return "$ruleWithEmbedded;$ruleWithRemapped!ie";
364  } else {
365  // Look ma, no fallbacks! This is for files which IE 6 and 7 don't support anyway: SVG.
366  return $ruleWithEmbedded;
367  }
368  }, $source );
369 
370  // Re-insert comments
371  $pattern = '/' . self::PLACEHOLDER . '(\d+)x/';
372  $source = preg_replace_callback( $pattern, function ( $match ) use ( &$comments ) {
373  return $comments[ $match[1] ];
374  }, $source );
375 
376  return $source;
377  }
378 
385  protected static function isRemoteUrl( $maybeUrl ) {
386  if ( substr( $maybeUrl, 0, 2 ) === '//' || parse_url( $maybeUrl, PHP_URL_SCHEME ) ) {
387  return true;
388  }
389  return false;
390  }
391 
398  protected static function isLocalUrl( $maybeUrl ) {
399  if ( $maybeUrl !== '' && $maybeUrl[0] === '/' && !self::isRemoteUrl( $maybeUrl ) ) {
400  return true;
401  }
402  return false;
403  }
404 
408  private static function getUrlRegex() {
409  static $urlRegex;
410  if ( $urlRegex === null ) {
411  // Match these three variants separately to avoid broken urls when
412  // e.g. a double quoted url contains a parenthesis, or when a
413  // single quoted url contains a double quote, etc.
414  // FIXME: Simplify now we only support PHP 7.0.0+
415  // Note: PCRE doesn't support multiple capture groups with the same name by default.
416  // - PCRE 6.7 introduced the "J" modifier (PCRE_INFO_JCHANGED for PCRE_DUPNAMES).
417  // https://secure.php.net/manual/en/reference.pcre.pattern.modifiers.php
418  // However this isn't useful since it just ignores all but the first one.
419  // Also, while the modifier was introduced in PCRE 6.7 (PHP 5.2+) it was
420  // not exposed to public preg_* functions until PHP 5.6.0.
421  // - PCRE 8.36 fixed this to work as expected (e.g. merge conceptually to
422  // only return the one matched in the part that actually matched).
423  // However MediaWiki supports 5.5.9, which has PCRE 8.32
424  // Per https://secure.php.net/manual/en/pcre.installation.php:
425  // - PCRE 8.32 (PHP 5.5.0)
426  // - PCRE 8.34 (PHP 5.5.10, PHP 5.6.0)
427  // - PCRE 8.37 (PHP 5.5.26, PHP 5.6.9, PHP 7.0.0)
428  // Workaround by using different groups and merge via processUrlMatch().
429  // - Using string concatenation for class constant or member assignments
430  // is only supported in PHP 5.6. Use a getter method for now.
431  $urlRegex = '(' .
432  // Unquoted url
433  'url\‍(\s*(?P<file0>[^\s\'"][^\?\‍)]+?)(?P<query0>\?[^\‍)]*?|)\s*\‍)' .
434  // Single quoted url
435  '|url\‍(\s*\'(?P<file1>[^\?\']+?)(?P<query1>\?[^\']*?|)\'\s*\‍)' .
436  // Double quoted url
437  '|url\‍(\s*"(?P<file2>[^\?"]+?)(?P<query2>\?[^"]*?|)"\s*\‍)' .
438  ')';
439  }
440  return $urlRegex;
441  }
442 
443  private static function processUrlMatch( array &$match, $flags = 0 ) {
444  if ( $flags & PREG_SET_ORDER ) {
445  // preg_match_all with PREG_SET_ORDER will return each group in each
446  // match array, and if it didn't match, instead of the sub array
447  // being an empty array it is `[ '', -1 ]`...
448  if ( isset( $match['file0'] ) && $match['file0'][1] !== -1 ) {
449  $match['file'] = $match['file0'];
450  $match['query'] = $match['query0'];
451  } elseif ( isset( $match['file1'] ) && $match['file1'][1] !== -1 ) {
452  $match['file'] = $match['file1'];
453  $match['query'] = $match['query1'];
454  } else {
455  if ( !isset( $match['file2'] ) || $match['file2'][1] === -1 ) {
456  throw new Exception( 'URL must be non-empty' );
457  }
458  $match['file'] = $match['file2'];
459  $match['query'] = $match['query2'];
460  }
461  } else {
462  if ( isset( $match['file0'] ) && $match['file0'] !== '' ) {
463  $match['file'] = $match['file0'];
464  $match['query'] = $match['query0'];
465  } elseif ( isset( $match['file1'] ) && $match['file1'] !== '' ) {
466  $match['file'] = $match['file1'];
467  $match['query'] = $match['query1'];
468  } else {
469  if ( !isset( $match['file2'] ) || $match['file2'] === '' ) {
470  throw new Exception( 'URL must be non-empty' );
471  }
472  $match['file'] = $match['file2'];
473  $match['query'] = $match['query2'];
474  }
475  }
476  }
477 
488  public static function remapOne( $file, $query, $local, $remote, $embed ) {
489  // The full URL possibly with query, as passed to the 'url()' value in CSS
490  $url = $file . $query;
491 
492  // Expand local URLs with absolute paths like /w/index.php to possibly protocol-relative URL, if
493  // wfExpandUrl() is available. (This will not be the case if we're running outside of MW.)
494  if ( self::isLocalUrl( $url ) && function_exists( 'wfExpandUrl' ) ) {
495  return wfExpandUrl( $url, PROTO_RELATIVE );
496  }
497 
498  // Pass thru fully-qualified and protocol-relative URLs and data URIs, as well as local URLs if
499  // we can't expand them.
500  // Also skips anchors or the rare `behavior` property specifying application's default behavior
501  if (
502  self::isRemoteUrl( $url ) ||
503  self::isLocalUrl( $url ) ||
504  substr( $url, 0, 1 ) === '#'
505  ) {
506  return $url;
507  }
508 
509  if ( $local === false ) {
510  // Assume that all paths are relative to $remote, and make them absolute
511  $url = $remote . '/' . $url;
512  } else {
513  // We drop the query part here and instead make the path relative to $remote
514  $url = "{$remote}/{$file}";
515  // Path to the actual file on the filesystem
516  $localFile = "{$local}/{$file}";
517  if ( file_exists( $localFile ) ) {
518  if ( $embed ) {
519  $data = self::encodeImageAsDataURI( $localFile );
520  if ( $data !== false ) {
521  return $data;
522  }
523  }
524  if ( method_exists( 'OutputPage', 'transformFilePath' ) ) {
525  $url = OutputPage::transformFilePath( $remote, $local, $file );
526  } else {
527  // Add version parameter as the first five hex digits
528  // of the MD5 hash of the file's contents.
529  $url .= '?' . substr( md5_file( $localFile ), 0, 5 );
530  }
531  }
532  // If any of these conditions failed (file missing, we don't want to embed it
533  // or it's not embeddable), return the URL (possibly with ?timestamp part)
534  }
535  if ( function_exists( 'wfRemoveDotSegments' ) ) {
536  $url = wfRemoveDotSegments( $url );
537  }
538  return $url;
539  }
540 
547  public static function minify( $css ) {
548  return trim(
549  str_replace(
550  [ '; ', ': ', ' {', '{ ', ', ', '} ', ';}', '( ', ' )', '[ ', ' ]' ],
551  [ ';', ':', '{', '{', ',', '}', '}', '(', ')', '[', ']' ],
552  preg_replace( [ '/\s+/', '/\/\*.*?\*\//s' ], [ ' ', '' ], $css )
553  )
554  );
555  }
556 }
CSSMin\isRemoteUrl
static isRemoteUrl( $maybeUrl)
Is this CSS rule referencing a remote URL?
Definition: CSSMin.php:385
CSSMin\buildUrlValue
static buildUrlValue( $url)
Build a CSS 'url()' value for the given URL, quoting parentheses (and other funny characters) and esc...
Definition: CSSMin.php:221
CSSMin\minify
static minify( $css)
Removes whitespace from CSS data.
Definition: CSSMin.php:547
use
Apache License January AND DISTRIBUTION Definitions License shall mean the terms and conditions for use
Definition: APACHE-LICENSE-2.0.txt:10
CSSMin\encodeStringAsDataURI
static encodeStringAsDataURI( $contents, $type, $ie8Compat=true)
Encode file contents as a data URI with chosen MIME type.
Definition: CSSMin.php:142
array
the array() calling protocol came about after MediaWiki 1.4rc1.
CSSMin\remap
static remap( $source, $local, $remote, $embedData=true)
Remaps CSS URL paths and automatically embeds data URIs for CSS rules or url() values preceded by an ...
Definition: CSSMin.php:243
wfRemoveDotSegments
wfRemoveDotSegments( $urlPath)
Remove all dot-segments in the provided URL path.
Definition: GlobalFunctions.php:649
CSSMin\$mimeTypes
static array $mimeTypes
List of common image files extensions and MIME-types.
Definition: CSSMin.php:44
CSSMin\serializeStringValue
static serializeStringValue( $value)
Serialize a string (escape and quote) for use as a CSS string value.
Definition: CSSMin.php:190
CSSMin\encodeImageAsDataURI
static encodeImageAsDataURI( $file, $type=null, $ie8Compat=true)
Encode an image file as a data URI.
Definition: CSSMin.php:114
true
null means default in associative array with keys and values unescaped Should be merged with default with a value of false meaning to suppress the attribute in associative array with keys and values unescaped noclasses just before the function returns a value If you return true
Definition: hooks.txt:2006
php
injection txt This is an overview of how MediaWiki makes use of dependency injection The design described here grew from the discussion of RFC T384 The term dependency this means that anything an object needs to operate should be injected from the the object itself should only know narrow no concrete implementation of the logic it relies on The requirement to inject everything typically results in an architecture that based on two main types of and essentially stateless service objects that use other service objects to operate on the value objects As of the beginning MediaWiki is only starting to use the DI approach Much of the code still relies on global state or direct resulting in a highly cyclical dependency which acts as the top level factory for services in MediaWiki which can be used to gain access to default instances of various services MediaWikiServices however also allows new services to be defined and default services to be redefined Services are defined or redefined by providing a callback the instantiator that will return a new instance of the service When it will create an instance of MediaWikiServices and populate it with the services defined in the files listed by thereby bootstrapping the DI framework Per $wgServiceWiringFiles lists includes ServiceWiring php
Definition: injection.txt:37
CSSMin\EMBED_REGEX
const EMBED_REGEX
Definition: CSSMin.php:40
$css
$css
Definition: styleTest.css.php:50
$matches
$matches
Definition: NoLocalSettings.php:24
CSSMin\processUrlMatch
static processUrlMatch(array &$match, $flags=0)
Definition: CSSMin.php:443
CSSMin\getUrlRegex
static getUrlRegex()
Definition: CSSMin.php:408
CSSMin\isLocalUrl
static isLocalUrl( $maybeUrl)
Is this CSS rule referencing a local URL?
Definition: CSSMin.php:398
$value
$value
Definition: styleTest.css.php:45
CSSMin\getLocalFileReferences
static getLocalFileReferences( $source, $path)
Get a list of local files referenced in a stylesheet (includes non-existent files).
Definition: CSSMin.php:63
PROTO_RELATIVE
const PROTO_RELATIVE
Definition: Defines.php:231
CSSMin\DATA_URI_SIZE_LIMIT
const DATA_URI_SIZE_LIMIT
Internet Explorer data URI length limit.
Definition: CSSMin.php:38
CSSMin\COMMENT_REGEX
const COMMENT_REGEX
Definition: CSSMin.php:41
CSSMin\getMimeType
static getMimeType( $file)
Definition: CSSMin.php:202
CSSMin
Transforms CSS data.
Definition: CSSMin.php:30
$path
$path
Definition: NoLocalSettings.php:25
as
This document is intended to provide useful advice for parties seeking to redistribute MediaWiki to end users It s targeted particularly at maintainers for Linux since it s been observed that distribution packages of MediaWiki often break We ve consistently had to recommend that users seeking support use official tarballs instead of their distribution s and this often solves whatever problem the user is having It would be nice if this could such as
Definition: distributors.txt:22
$source
$source
Definition: mwdoc-filter.php:46
CSSMin\remapOne
static remapOne( $file, $query, $local, $remote, $embed)
Remap or embed a CSS URL path.
Definition: CSSMin.php:488
$query
null for the wiki Added should default to null in handler for backwards compatibility add a value to it if you want to add a cookie that have to vary cache options can modify $query
Definition: hooks.txt:1620
$ext
$ext
Definition: router.php:55
wfExpandUrl
wfExpandUrl( $url, $defaultProto=PROTO_CURRENT)
Expand a potentially local URL to a fully-qualified URL.
Definition: GlobalFunctions.php:521
OutputPage\transformFilePath
static transformFilePath( $remotePathPrefix, $localPath, $file)
Utility method for transformResourceFilePath().
Definition: OutputPage.php:3741
$type
$type
Definition: testCompression.php:48