Improve documentation for wfParseUrl
[lhc/web/wiklou.git] / includes / GlobalFunctions.php
1 <?php
2 /**
3 * Global functions used everywhere.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 */
22
23 if ( !defined( 'MEDIAWIKI' ) ) {
24 die( "This file is part of MediaWiki, it is not a valid entry point" );
25 }
26
27 use Liuggio\StatsdClient\Sender\SocketSender;
28 use MediaWiki\Logger\LoggerFactory;
29 use MediaWiki\Session\SessionManager;
30 use MediaWiki\MediaWikiServices;
31 use Wikimedia\ScopedCallback;
32 use Wikimedia\Rdbms\DBReplicationWaitError;
33
34 // Hide compatibility functions from Doxygen
35 /// @cond
36 /**
37 * Compatibility functions
38 *
39 * We support PHP 5.5.9 and up.
40 * Re-implementations of newer functions or functions in non-standard
41 * PHP extensions may be included here.
42 */
43
44 // hash_equals function only exists in PHP >= 5.6.0
45 // https://secure.php.net/hash_equals
46 if ( !function_exists( 'hash_equals' ) ) {
47 /**
48 * Check whether a user-provided string is equal to a fixed-length secret string
49 * without revealing bytes of the secret string through timing differences.
50 *
51 * The usual way to compare strings (PHP's === operator or the underlying memcmp()
52 * function in C) is to compare corresponding bytes and stop at the first difference,
53 * which would take longer for a partial match than for a complete mismatch. This
54 * is not secure when one of the strings (e.g. an HMAC or token) must remain secret
55 * and the other may come from an attacker. Statistical analysis of timing measurements
56 * over many requests may allow the attacker to guess the string's bytes one at a time
57 * (and check his guesses) even if the timing differences are extremely small.
58 *
59 * When making such a security-sensitive comparison, it is essential that the sequence
60 * in which instructions are executed and memory locations are accessed not depend on
61 * the secret string's value. HOWEVER, for simplicity, we do not attempt to minimize
62 * the inevitable leakage of the string's length. That is generally known anyway as
63 * a chararacteristic of the hash function used to compute the secret value.
64 *
65 * Longer explanation: http://www.emerose.com/timing-attacks-explained
66 *
67 * @codeCoverageIgnore
68 * @param string $known_string Fixed-length secret string to compare against
69 * @param string $user_string User-provided string
70 * @return bool True if the strings are the same, false otherwise
71 */
72 function hash_equals( $known_string, $user_string ) {
73 // Strict type checking as in PHP's native implementation
74 if ( !is_string( $known_string ) ) {
75 trigger_error( 'hash_equals(): Expected known_string to be a string, ' .
76 gettype( $known_string ) . ' given', E_USER_WARNING );
77
78 return false;
79 }
80
81 if ( !is_string( $user_string ) ) {
82 trigger_error( 'hash_equals(): Expected user_string to be a string, ' .
83 gettype( $user_string ) . ' given', E_USER_WARNING );
84
85 return false;
86 }
87
88 $known_string_len = strlen( $known_string );
89 if ( $known_string_len !== strlen( $user_string ) ) {
90 return false;
91 }
92
93 $result = 0;
94 for ( $i = 0; $i < $known_string_len; $i++ ) {
95 $result |= ord( $known_string[$i] ) ^ ord( $user_string[$i] );
96 }
97
98 return ( $result === 0 );
99 }
100 }
101 /// @endcond
102
103 /**
104 * Load an extension
105 *
106 * This queues an extension to be loaded through
107 * the ExtensionRegistry system.
108 *
109 * @param string $ext Name of the extension to load
110 * @param string|null $path Absolute path of where to find the extension.json file
111 * @since 1.25
112 */
113 function wfLoadExtension( $ext, $path = null ) {
114 if ( !$path ) {
115 global $wgExtensionDirectory;
116 $path = "$wgExtensionDirectory/$ext/extension.json";
117 }
118 ExtensionRegistry::getInstance()->queue( $path );
119 }
120
121 /**
122 * Load multiple extensions at once
123 *
124 * Same as wfLoadExtension, but more efficient if you
125 * are loading multiple extensions.
126 *
127 * If you want to specify custom paths, you should interact with
128 * ExtensionRegistry directly.
129 *
130 * @see wfLoadExtension
131 * @param string[] $exts Array of extension names to load
132 * @since 1.25
133 */
134 function wfLoadExtensions( array $exts ) {
135 global $wgExtensionDirectory;
136 $registry = ExtensionRegistry::getInstance();
137 foreach ( $exts as $ext ) {
138 $registry->queue( "$wgExtensionDirectory/$ext/extension.json" );
139 }
140 }
141
142 /**
143 * Load a skin
144 *
145 * @see wfLoadExtension
146 * @param string $skin Name of the extension to load
147 * @param string|null $path Absolute path of where to find the skin.json file
148 * @since 1.25
149 */
150 function wfLoadSkin( $skin, $path = null ) {
151 if ( !$path ) {
152 global $wgStyleDirectory;
153 $path = "$wgStyleDirectory/$skin/skin.json";
154 }
155 ExtensionRegistry::getInstance()->queue( $path );
156 }
157
158 /**
159 * Load multiple skins at once
160 *
161 * @see wfLoadExtensions
162 * @param string[] $skins Array of extension names to load
163 * @since 1.25
164 */
165 function wfLoadSkins( array $skins ) {
166 global $wgStyleDirectory;
167 $registry = ExtensionRegistry::getInstance();
168 foreach ( $skins as $skin ) {
169 $registry->queue( "$wgStyleDirectory/$skin/skin.json" );
170 }
171 }
172
173 /**
174 * Like array_diff( $a, $b ) except that it works with two-dimensional arrays.
175 * @param array $a
176 * @param array $b
177 * @return array
178 */
179 function wfArrayDiff2( $a, $b ) {
180 return array_udiff( $a, $b, 'wfArrayDiff2_cmp' );
181 }
182
183 /**
184 * @param array|string $a
185 * @param array|string $b
186 * @return int
187 */
188 function wfArrayDiff2_cmp( $a, $b ) {
189 if ( is_string( $a ) && is_string( $b ) ) {
190 return strcmp( $a, $b );
191 } elseif ( count( $a ) !== count( $b ) ) {
192 return count( $a ) < count( $b ) ? -1 : 1;
193 } else {
194 reset( $a );
195 reset( $b );
196 while ( ( list( , $valueA ) = each( $a ) ) && ( list( , $valueB ) = each( $b ) ) ) {
197 $cmp = strcmp( $valueA, $valueB );
198 if ( $cmp !== 0 ) {
199 return $cmp;
200 }
201 }
202 return 0;
203 }
204 }
205
206 /**
207 * Appends to second array if $value differs from that in $default
208 *
209 * @param string|int $key
210 * @param mixed $value
211 * @param mixed $default
212 * @param array $changed Array to alter
213 * @throws MWException
214 */
215 function wfAppendToArrayIfNotDefault( $key, $value, $default, &$changed ) {
216 if ( is_null( $changed ) ) {
217 throw new MWException( 'GlobalFunctions::wfAppendToArrayIfNotDefault got null' );
218 }
219 if ( $default[$key] !== $value ) {
220 $changed[$key] = $value;
221 }
222 }
223
224 /**
225 * Merge arrays in the style of getUserPermissionsErrors, with duplicate removal
226 * e.g.
227 * wfMergeErrorArrays(
228 * [ [ 'x' ] ],
229 * [ [ 'x', '2' ] ],
230 * [ [ 'x' ] ],
231 * [ [ 'y' ] ]
232 * );
233 * returns:
234 * [
235 * [ 'x', '2' ],
236 * [ 'x' ],
237 * [ 'y' ]
238 * ]
239 *
240 * @param array $array1,...
241 * @return array
242 */
243 function wfMergeErrorArrays( /*...*/ ) {
244 $args = func_get_args();
245 $out = [];
246 foreach ( $args as $errors ) {
247 foreach ( $errors as $params ) {
248 $originalParams = $params;
249 if ( $params[0] instanceof MessageSpecifier ) {
250 $msg = $params[0];
251 $params = array_merge( [ $msg->getKey() ], $msg->getParams() );
252 }
253 # @todo FIXME: Sometimes get nested arrays for $params,
254 # which leads to E_NOTICEs
255 $spec = implode( "\t", $params );
256 $out[$spec] = $originalParams;
257 }
258 }
259 return array_values( $out );
260 }
261
262 /**
263 * Insert array into another array after the specified *KEY*
264 *
265 * @param array $array The array.
266 * @param array $insert The array to insert.
267 * @param mixed $after The key to insert after
268 * @return array
269 */
270 function wfArrayInsertAfter( array $array, array $insert, $after ) {
271 // Find the offset of the element to insert after.
272 $keys = array_keys( $array );
273 $offsetByKey = array_flip( $keys );
274
275 $offset = $offsetByKey[$after];
276
277 // Insert at the specified offset
278 $before = array_slice( $array, 0, $offset + 1, true );
279 $after = array_slice( $array, $offset + 1, count( $array ) - $offset, true );
280
281 $output = $before + $insert + $after;
282
283 return $output;
284 }
285
286 /**
287 * Recursively converts the parameter (an object) to an array with the same data
288 *
289 * @param object|array $objOrArray
290 * @param bool $recursive
291 * @return array
292 */
293 function wfObjectToArray( $objOrArray, $recursive = true ) {
294 $array = [];
295 if ( is_object( $objOrArray ) ) {
296 $objOrArray = get_object_vars( $objOrArray );
297 }
298 foreach ( $objOrArray as $key => $value ) {
299 if ( $recursive && ( is_object( $value ) || is_array( $value ) ) ) {
300 $value = wfObjectToArray( $value );
301 }
302
303 $array[$key] = $value;
304 }
305
306 return $array;
307 }
308
309 /**
310 * Get a random decimal value between 0 and 1, in a way
311 * not likely to give duplicate values for any realistic
312 * number of articles.
313 *
314 * @note This is designed for use in relation to Special:RandomPage
315 * and the page_random database field.
316 *
317 * @return string
318 */
319 function wfRandom() {
320 // The maximum random value is "only" 2^31-1, so get two random
321 // values to reduce the chance of dupes
322 $max = mt_getrandmax() + 1;
323 $rand = number_format( ( mt_rand() * $max + mt_rand() ) / $max / $max, 12, '.', '' );
324 return $rand;
325 }
326
327 /**
328 * Get a random string containing a number of pseudo-random hex characters.
329 *
330 * @note This is not secure, if you are trying to generate some sort
331 * of token please use MWCryptRand instead.
332 *
333 * @param int $length The length of the string to generate
334 * @return string
335 * @since 1.20
336 */
337 function wfRandomString( $length = 32 ) {
338 $str = '';
339 for ( $n = 0; $n < $length; $n += 7 ) {
340 $str .= sprintf( '%07x', mt_rand() & 0xfffffff );
341 }
342 return substr( $str, 0, $length );
343 }
344
345 /**
346 * We want some things to be included as literal characters in our title URLs
347 * for prettiness, which urlencode encodes by default. According to RFC 1738,
348 * all of the following should be safe:
349 *
350 * ;:@&=$-_.+!*'(),
351 *
352 * RFC 1738 says ~ is unsafe, however RFC 3986 considers it an unreserved
353 * character which should not be encoded. More importantly, google chrome
354 * always converts %7E back to ~, and converting it in this function can
355 * cause a redirect loop (T105265).
356 *
357 * But + is not safe because it's used to indicate a space; &= are only safe in
358 * paths and not in queries (and we don't distinguish here); ' seems kind of
359 * scary; and urlencode() doesn't touch -_. to begin with. Plus, although /
360 * is reserved, we don't care. So the list we unescape is:
361 *
362 * ;:@$!*(),/~
363 *
364 * However, IIS7 redirects fail when the url contains a colon (see T24709),
365 * so no fancy : for IIS7.
366 *
367 * %2F in the page titles seems to fatally break for some reason.
368 *
369 * @param string $s
370 * @return string
371 */
372 function wfUrlencode( $s ) {
373 static $needle;
374
375 if ( is_null( $s ) ) {
376 $needle = null;
377 return '';
378 }
379
380 if ( is_null( $needle ) ) {
381 $needle = [ '%3B', '%40', '%24', '%21', '%2A', '%28', '%29', '%2C', '%2F', '%7E' ];
382 if ( !isset( $_SERVER['SERVER_SOFTWARE'] ) ||
383 ( strpos( $_SERVER['SERVER_SOFTWARE'], 'Microsoft-IIS/7' ) === false )
384 ) {
385 $needle[] = '%3A';
386 }
387 }
388
389 $s = urlencode( $s );
390 $s = str_ireplace(
391 $needle,
392 [ ';', '@', '$', '!', '*', '(', ')', ',', '/', '~', ':' ],
393 $s
394 );
395
396 return $s;
397 }
398
399 /**
400 * This function takes one or two arrays as input, and returns a CGI-style string, e.g.
401 * "days=7&limit=100". Options in the first array override options in the second.
402 * Options set to null or false will not be output.
403 *
404 * @param array $array1 ( String|Array )
405 * @param array|null $array2 ( String|Array )
406 * @param string $prefix
407 * @return string
408 */
409 function wfArrayToCgi( $array1, $array2 = null, $prefix = '' ) {
410 if ( !is_null( $array2 ) ) {
411 $array1 = $array1 + $array2;
412 }
413
414 $cgi = '';
415 foreach ( $array1 as $key => $value ) {
416 if ( !is_null( $value ) && $value !== false ) {
417 if ( $cgi != '' ) {
418 $cgi .= '&';
419 }
420 if ( $prefix !== '' ) {
421 $key = $prefix . "[$key]";
422 }
423 if ( is_array( $value ) ) {
424 $firstTime = true;
425 foreach ( $value as $k => $v ) {
426 $cgi .= $firstTime ? '' : '&';
427 if ( is_array( $v ) ) {
428 $cgi .= wfArrayToCgi( $v, null, $key . "[$k]" );
429 } else {
430 $cgi .= urlencode( $key . "[$k]" ) . '=' . urlencode( $v );
431 }
432 $firstTime = false;
433 }
434 } else {
435 if ( is_object( $value ) ) {
436 $value = $value->__toString();
437 }
438 $cgi .= urlencode( $key ) . '=' . urlencode( $value );
439 }
440 }
441 }
442 return $cgi;
443 }
444
445 /**
446 * This is the logical opposite of wfArrayToCgi(): it accepts a query string as
447 * its argument and returns the same string in array form. This allows compatibility
448 * with legacy functions that accept raw query strings instead of nice
449 * arrays. Of course, keys and values are urldecode()d.
450 *
451 * @param string $query Query string
452 * @return string[] Array version of input
453 */
454 function wfCgiToArray( $query ) {
455 if ( isset( $query[0] ) && $query[0] == '?' ) {
456 $query = substr( $query, 1 );
457 }
458 $bits = explode( '&', $query );
459 $ret = [];
460 foreach ( $bits as $bit ) {
461 if ( $bit === '' ) {
462 continue;
463 }
464 if ( strpos( $bit, '=' ) === false ) {
465 // Pieces like &qwerty become 'qwerty' => '' (at least this is what php does)
466 $key = $bit;
467 $value = '';
468 } else {
469 list( $key, $value ) = explode( '=', $bit );
470 }
471 $key = urldecode( $key );
472 $value = urldecode( $value );
473 if ( strpos( $key, '[' ) !== false ) {
474 $keys = array_reverse( explode( '[', $key ) );
475 $key = array_pop( $keys );
476 $temp = $value;
477 foreach ( $keys as $k ) {
478 $k = substr( $k, 0, -1 );
479 $temp = [ $k => $temp ];
480 }
481 if ( isset( $ret[$key] ) ) {
482 $ret[$key] = array_merge( $ret[$key], $temp );
483 } else {
484 $ret[$key] = $temp;
485 }
486 } else {
487 $ret[$key] = $value;
488 }
489 }
490 return $ret;
491 }
492
493 /**
494 * Append a query string to an existing URL, which may or may not already
495 * have query string parameters already. If so, they will be combined.
496 *
497 * @param string $url
498 * @param string|string[] $query String or associative array
499 * @return string
500 */
501 function wfAppendQuery( $url, $query ) {
502 if ( is_array( $query ) ) {
503 $query = wfArrayToCgi( $query );
504 }
505 if ( $query != '' ) {
506 // Remove the fragment, if there is one
507 $fragment = false;
508 $hashPos = strpos( $url, '#' );
509 if ( $hashPos !== false ) {
510 $fragment = substr( $url, $hashPos );
511 $url = substr( $url, 0, $hashPos );
512 }
513
514 // Add parameter
515 if ( false === strpos( $url, '?' ) ) {
516 $url .= '?';
517 } else {
518 $url .= '&';
519 }
520 $url .= $query;
521
522 // Put the fragment back
523 if ( $fragment !== false ) {
524 $url .= $fragment;
525 }
526 }
527 return $url;
528 }
529
530 /**
531 * Expand a potentially local URL to a fully-qualified URL. Assumes $wgServer
532 * is correct.
533 *
534 * The meaning of the PROTO_* constants is as follows:
535 * PROTO_HTTP: Output a URL starting with http://
536 * PROTO_HTTPS: Output a URL starting with https://
537 * PROTO_RELATIVE: Output a URL starting with // (protocol-relative URL)
538 * PROTO_CURRENT: Output a URL starting with either http:// or https:// , depending
539 * on which protocol was used for the current incoming request
540 * PROTO_CANONICAL: For URLs without a domain, like /w/index.php , use $wgCanonicalServer.
541 * For protocol-relative URLs, use the protocol of $wgCanonicalServer
542 * PROTO_INTERNAL: Like PROTO_CANONICAL, but uses $wgInternalServer instead of $wgCanonicalServer
543 *
544 * @todo this won't work with current-path-relative URLs
545 * like "subdir/foo.html", etc.
546 *
547 * @param string $url Either fully-qualified or a local path + query
548 * @param string $defaultProto One of the PROTO_* constants. Determines the
549 * protocol to use if $url or $wgServer is protocol-relative
550 * @return string|false Fully-qualified URL, current-path-relative URL or false if
551 * no valid URL can be constructed
552 */
553 function wfExpandUrl( $url, $defaultProto = PROTO_CURRENT ) {
554 global $wgServer, $wgCanonicalServer, $wgInternalServer, $wgRequest,
555 $wgHttpsPort;
556 if ( $defaultProto === PROTO_CANONICAL ) {
557 $serverUrl = $wgCanonicalServer;
558 } elseif ( $defaultProto === PROTO_INTERNAL && $wgInternalServer !== false ) {
559 // Make $wgInternalServer fall back to $wgServer if not set
560 $serverUrl = $wgInternalServer;
561 } else {
562 $serverUrl = $wgServer;
563 if ( $defaultProto === PROTO_CURRENT ) {
564 $defaultProto = $wgRequest->getProtocol() . '://';
565 }
566 }
567
568 // Analyze $serverUrl to obtain its protocol
569 $bits = wfParseUrl( $serverUrl );
570 $serverHasProto = $bits && $bits['scheme'] != '';
571
572 if ( $defaultProto === PROTO_CANONICAL || $defaultProto === PROTO_INTERNAL ) {
573 if ( $serverHasProto ) {
574 $defaultProto = $bits['scheme'] . '://';
575 } else {
576 // $wgCanonicalServer or $wgInternalServer doesn't have a protocol.
577 // This really isn't supposed to happen. Fall back to HTTP in this
578 // ridiculous case.
579 $defaultProto = PROTO_HTTP;
580 }
581 }
582
583 $defaultProtoWithoutSlashes = substr( $defaultProto, 0, -2 );
584
585 if ( substr( $url, 0, 2 ) == '//' ) {
586 $url = $defaultProtoWithoutSlashes . $url;
587 } elseif ( substr( $url, 0, 1 ) == '/' ) {
588 // If $serverUrl is protocol-relative, prepend $defaultProtoWithoutSlashes,
589 // otherwise leave it alone.
590 $url = ( $serverHasProto ? '' : $defaultProtoWithoutSlashes ) . $serverUrl . $url;
591 }
592
593 $bits = wfParseUrl( $url );
594
595 // ensure proper port for HTTPS arrives in URL
596 // https://phabricator.wikimedia.org/T67184
597 if ( $defaultProto === PROTO_HTTPS && $wgHttpsPort != 443 ) {
598 $bits['port'] = $wgHttpsPort;
599 }
600
601 if ( $bits && isset( $bits['path'] ) ) {
602 $bits['path'] = wfRemoveDotSegments( $bits['path'] );
603 return wfAssembleUrl( $bits );
604 } elseif ( $bits ) {
605 # No path to expand
606 return $url;
607 } elseif ( substr( $url, 0, 1 ) != '/' ) {
608 # URL is a relative path
609 return wfRemoveDotSegments( $url );
610 }
611
612 # Expanded URL is not valid.
613 return false;
614 }
615
616 /**
617 * This function will reassemble a URL parsed with wfParseURL. This is useful
618 * if you need to edit part of a URL and put it back together.
619 *
620 * This is the basic structure used (brackets contain keys for $urlParts):
621 * [scheme][delimiter][user]:[pass]@[host]:[port][path]?[query]#[fragment]
622 *
623 * @todo Need to integrate this into wfExpandUrl (see T34168)
624 *
625 * @since 1.19
626 * @param array $urlParts URL parts, as output from wfParseUrl
627 * @return string URL assembled from its component parts
628 */
629 function wfAssembleUrl( $urlParts ) {
630 $result = '';
631
632 if ( isset( $urlParts['delimiter'] ) ) {
633 if ( isset( $urlParts['scheme'] ) ) {
634 $result .= $urlParts['scheme'];
635 }
636
637 $result .= $urlParts['delimiter'];
638 }
639
640 if ( isset( $urlParts['host'] ) ) {
641 if ( isset( $urlParts['user'] ) ) {
642 $result .= $urlParts['user'];
643 if ( isset( $urlParts['pass'] ) ) {
644 $result .= ':' . $urlParts['pass'];
645 }
646 $result .= '@';
647 }
648
649 $result .= $urlParts['host'];
650
651 if ( isset( $urlParts['port'] ) ) {
652 $result .= ':' . $urlParts['port'];
653 }
654 }
655
656 if ( isset( $urlParts['path'] ) ) {
657 $result .= $urlParts['path'];
658 }
659
660 if ( isset( $urlParts['query'] ) ) {
661 $result .= '?' . $urlParts['query'];
662 }
663
664 if ( isset( $urlParts['fragment'] ) ) {
665 $result .= '#' . $urlParts['fragment'];
666 }
667
668 return $result;
669 }
670
671 /**
672 * Remove all dot-segments in the provided URL path. For example,
673 * '/a/./b/../c/' becomes '/a/c/'. For details on the algorithm, please see
674 * RFC3986 section 5.2.4.
675 *
676 * @todo Need to integrate this into wfExpandUrl (see T34168)
677 *
678 * @param string $urlPath URL path, potentially containing dot-segments
679 * @return string URL path with all dot-segments removed
680 */
681 function wfRemoveDotSegments( $urlPath ) {
682 $output = '';
683 $inputOffset = 0;
684 $inputLength = strlen( $urlPath );
685
686 while ( $inputOffset < $inputLength ) {
687 $prefixLengthOne = substr( $urlPath, $inputOffset, 1 );
688 $prefixLengthTwo = substr( $urlPath, $inputOffset, 2 );
689 $prefixLengthThree = substr( $urlPath, $inputOffset, 3 );
690 $prefixLengthFour = substr( $urlPath, $inputOffset, 4 );
691 $trimOutput = false;
692
693 if ( $prefixLengthTwo == './' ) {
694 # Step A, remove leading "./"
695 $inputOffset += 2;
696 } elseif ( $prefixLengthThree == '../' ) {
697 # Step A, remove leading "../"
698 $inputOffset += 3;
699 } elseif ( ( $prefixLengthTwo == '/.' ) && ( $inputOffset + 2 == $inputLength ) ) {
700 # Step B, replace leading "/.$" with "/"
701 $inputOffset += 1;
702 $urlPath[$inputOffset] = '/';
703 } elseif ( $prefixLengthThree == '/./' ) {
704 # Step B, replace leading "/./" with "/"
705 $inputOffset += 2;
706 } elseif ( $prefixLengthThree == '/..' && ( $inputOffset + 3 == $inputLength ) ) {
707 # Step C, replace leading "/..$" with "/" and
708 # remove last path component in output
709 $inputOffset += 2;
710 $urlPath[$inputOffset] = '/';
711 $trimOutput = true;
712 } elseif ( $prefixLengthFour == '/../' ) {
713 # Step C, replace leading "/../" with "/" and
714 # remove last path component in output
715 $inputOffset += 3;
716 $trimOutput = true;
717 } elseif ( ( $prefixLengthOne == '.' ) && ( $inputOffset + 1 == $inputLength ) ) {
718 # Step D, remove "^.$"
719 $inputOffset += 1;
720 } elseif ( ( $prefixLengthTwo == '..' ) && ( $inputOffset + 2 == $inputLength ) ) {
721 # Step D, remove "^..$"
722 $inputOffset += 2;
723 } else {
724 # Step E, move leading path segment to output
725 if ( $prefixLengthOne == '/' ) {
726 $slashPos = strpos( $urlPath, '/', $inputOffset + 1 );
727 } else {
728 $slashPos = strpos( $urlPath, '/', $inputOffset );
729 }
730 if ( $slashPos === false ) {
731 $output .= substr( $urlPath, $inputOffset );
732 $inputOffset = $inputLength;
733 } else {
734 $output .= substr( $urlPath, $inputOffset, $slashPos - $inputOffset );
735 $inputOffset += $slashPos - $inputOffset;
736 }
737 }
738
739 if ( $trimOutput ) {
740 $slashPos = strrpos( $output, '/' );
741 if ( $slashPos === false ) {
742 $output = '';
743 } else {
744 $output = substr( $output, 0, $slashPos );
745 }
746 }
747 }
748
749 return $output;
750 }
751
752 /**
753 * Returns a regular expression of url protocols
754 *
755 * @param bool $includeProtocolRelative If false, remove '//' from the returned protocol list.
756 * DO NOT USE this directly, use wfUrlProtocolsWithoutProtRel() instead
757 * @return string
758 */
759 function wfUrlProtocols( $includeProtocolRelative = true ) {
760 global $wgUrlProtocols;
761
762 // Cache return values separately based on $includeProtocolRelative
763 static $withProtRel = null, $withoutProtRel = null;
764 $cachedValue = $includeProtocolRelative ? $withProtRel : $withoutProtRel;
765 if ( !is_null( $cachedValue ) ) {
766 return $cachedValue;
767 }
768
769 // Support old-style $wgUrlProtocols strings, for backwards compatibility
770 // with LocalSettings files from 1.5
771 if ( is_array( $wgUrlProtocols ) ) {
772 $protocols = [];
773 foreach ( $wgUrlProtocols as $protocol ) {
774 // Filter out '//' if !$includeProtocolRelative
775 if ( $includeProtocolRelative || $protocol !== '//' ) {
776 $protocols[] = preg_quote( $protocol, '/' );
777 }
778 }
779
780 $retval = implode( '|', $protocols );
781 } else {
782 // Ignore $includeProtocolRelative in this case
783 // This case exists for pre-1.6 compatibility, and we can safely assume
784 // that '//' won't appear in a pre-1.6 config because protocol-relative
785 // URLs weren't supported until 1.18
786 $retval = $wgUrlProtocols;
787 }
788
789 // Cache return value
790 if ( $includeProtocolRelative ) {
791 $withProtRel = $retval;
792 } else {
793 $withoutProtRel = $retval;
794 }
795 return $retval;
796 }
797
798 /**
799 * Like wfUrlProtocols(), but excludes '//' from the protocol list. Use this if
800 * you need a regex that matches all URL protocols but does not match protocol-
801 * relative URLs
802 * @return string
803 */
804 function wfUrlProtocolsWithoutProtRel() {
805 return wfUrlProtocols( false );
806 }
807
808 /**
809 * parse_url() work-alike, but non-broken. Differences:
810 *
811 * 1) Does not raise warnings on bad URLs (just returns false).
812 * 2) Handles protocols that don't use :// (e.g., mailto: and news:, as well as
813 * protocol-relative URLs) correctly.
814 * 3) Adds a "delimiter" element to the array (see (2)).
815 * 4) Verifies that the protocol is on the $wgUrlProtocols whitelist.
816 * 5) Rejects some invalid URLs that parse_url doesn't, e.g. the empty string or URLs starting with
817 * a line feed character.
818 *
819 * @param string $url A URL to parse
820 * @return string[]|bool Bits of the URL in an associative array, or false on failure.
821 * Possible fields:
822 * - scheme: URI scheme (protocol), e.g. 'http', 'mailto'. Lowercase, always present, but can
823 * be an empty string for protocol-relative URLs.
824 * - delimiter: either '://', ':' or '//'. Always present.
825 * - host: domain name / IP. Always present, but could be an empty string, e.g. for file: URLs.
826 * - user: user name, e.g. for HTTP Basic auth URLs such as http://user:pass@example.com/
827 * Missing when there is no username.
828 * - pass: password, same as above.
829 * - path: path including the leading /. Will be missing when empty (e.g. 'http://example.com')
830 * - query: query string (as a string; see wfCgiToArray() for parsing it), can be missing.
831 * - fragment: the part after #, can be missing.
832 */
833 function wfParseUrl( $url ) {
834 global $wgUrlProtocols; // Allow all protocols defined in DefaultSettings/LocalSettings.php
835
836 // Protocol-relative URLs are handled really badly by parse_url(). It's so
837 // bad that the easiest way to handle them is to just prepend 'http:' and
838 // strip the protocol out later.
839 $wasRelative = substr( $url, 0, 2 ) == '//';
840 if ( $wasRelative ) {
841 $url = "http:$url";
842 }
843 MediaWiki\suppressWarnings();
844 $bits = parse_url( $url );
845 MediaWiki\restoreWarnings();
846 // parse_url() returns an array without scheme for some invalid URLs, e.g.
847 // parse_url("%0Ahttp://example.com") == [ 'host' => '%0Ahttp', 'path' => 'example.com' ]
848 if ( !$bits || !isset( $bits['scheme'] ) ) {
849 return false;
850 }
851
852 // parse_url() incorrectly handles schemes case-sensitively. Convert it to lowercase.
853 $bits['scheme'] = strtolower( $bits['scheme'] );
854
855 // most of the protocols are followed by ://, but mailto: and sometimes news: not, check for it
856 if ( in_array( $bits['scheme'] . '://', $wgUrlProtocols ) ) {
857 $bits['delimiter'] = '://';
858 } elseif ( in_array( $bits['scheme'] . ':', $wgUrlProtocols ) ) {
859 $bits['delimiter'] = ':';
860 // parse_url detects for news: and mailto: the host part of an url as path
861 // We have to correct this wrong detection
862 if ( isset( $bits['path'] ) ) {
863 $bits['host'] = $bits['path'];
864 $bits['path'] = '';
865 }
866 } else {
867 return false;
868 }
869
870 /* Provide an empty host for eg. file:/// urls (see T30627) */
871 if ( !isset( $bits['host'] ) ) {
872 $bits['host'] = '';
873
874 // See T47069
875 if ( isset( $bits['path'] ) ) {
876 /* parse_url loses the third / for file:///c:/ urls (but not on variants) */
877 if ( substr( $bits['path'], 0, 1 ) !== '/' ) {
878 $bits['path'] = '/' . $bits['path'];
879 }
880 } else {
881 $bits['path'] = '';
882 }
883 }
884
885 // If the URL was protocol-relative, fix scheme and delimiter
886 if ( $wasRelative ) {
887 $bits['scheme'] = '';
888 $bits['delimiter'] = '//';
889 }
890 return $bits;
891 }
892
893 /**
894 * Take a URL, make sure it's expanded to fully qualified, and replace any
895 * encoded non-ASCII Unicode characters with their UTF-8 original forms
896 * for more compact display and legibility for local audiences.
897 *
898 * @todo handle punycode domains too
899 *
900 * @param string $url
901 * @return string
902 */
903 function wfExpandIRI( $url ) {
904 return preg_replace_callback(
905 '/((?:%[89A-F][0-9A-F])+)/i',
906 'wfExpandIRI_callback',
907 wfExpandUrl( $url )
908 );
909 }
910
911 /**
912 * Private callback for wfExpandIRI
913 * @param array $matches
914 * @return string
915 */
916 function wfExpandIRI_callback( $matches ) {
917 return urldecode( $matches[1] );
918 }
919
920 /**
921 * Make URL indexes, appropriate for the el_index field of externallinks.
922 *
923 * @param string $url
924 * @return array
925 */
926 function wfMakeUrlIndexes( $url ) {
927 $bits = wfParseUrl( $url );
928
929 // Reverse the labels in the hostname, convert to lower case
930 // For emails reverse domainpart only
931 if ( $bits['scheme'] == 'mailto' ) {
932 $mailparts = explode( '@', $bits['host'], 2 );
933 if ( count( $mailparts ) === 2 ) {
934 $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
935 } else {
936 // No domain specified, don't mangle it
937 $domainpart = '';
938 }
939 $reversedHost = $domainpart . '@' . $mailparts[0];
940 } else {
941 $reversedHost = strtolower( implode( '.', array_reverse( explode( '.', $bits['host'] ) ) ) );
942 }
943 // Add an extra dot to the end
944 // Why? Is it in wrong place in mailto links?
945 if ( substr( $reversedHost, -1, 1 ) !== '.' ) {
946 $reversedHost .= '.';
947 }
948 // Reconstruct the pseudo-URL
949 $prot = $bits['scheme'];
950 $index = $prot . $bits['delimiter'] . $reversedHost;
951 // Leave out user and password. Add the port, path, query and fragment
952 if ( isset( $bits['port'] ) ) {
953 $index .= ':' . $bits['port'];
954 }
955 if ( isset( $bits['path'] ) ) {
956 $index .= $bits['path'];
957 } else {
958 $index .= '/';
959 }
960 if ( isset( $bits['query'] ) ) {
961 $index .= '?' . $bits['query'];
962 }
963 if ( isset( $bits['fragment'] ) ) {
964 $index .= '#' . $bits['fragment'];
965 }
966
967 if ( $prot == '' ) {
968 return [ "http:$index", "https:$index" ];
969 } else {
970 return [ $index ];
971 }
972 }
973
974 /**
975 * Check whether a given URL has a domain that occurs in a given set of domains
976 * @param string $url URL
977 * @param array $domains Array of domains (strings)
978 * @return bool True if the host part of $url ends in one of the strings in $domains
979 */
980 function wfMatchesDomainList( $url, $domains ) {
981 $bits = wfParseUrl( $url );
982 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
983 $host = '.' . $bits['host'];
984 foreach ( (array)$domains as $domain ) {
985 $domain = '.' . $domain;
986 if ( substr( $host, -strlen( $domain ) ) === $domain ) {
987 return true;
988 }
989 }
990 }
991 return false;
992 }
993
994 /**
995 * Sends a line to the debug log if enabled or, optionally, to a comment in output.
996 * In normal operation this is a NOP.
997 *
998 * Controlling globals:
999 * $wgDebugLogFile - points to the log file
1000 * $wgDebugRawPage - if false, 'action=raw' hits will not result in debug output.
1001 * $wgDebugComments - if on, some debug items may appear in comments in the HTML output.
1002 *
1003 * @since 1.25 support for additional context data
1004 *
1005 * @param string $text
1006 * @param string|bool $dest Destination of the message:
1007 * - 'all': both to the log and HTML (debug toolbar or HTML comments)
1008 * - 'private': excluded from HTML output
1009 * For backward compatibility, it can also take a boolean:
1010 * - true: same as 'all'
1011 * - false: same as 'private'
1012 * @param array $context Additional logging context data
1013 */
1014 function wfDebug( $text, $dest = 'all', array $context = [] ) {
1015 global $wgDebugRawPage, $wgDebugLogPrefix;
1016 global $wgDebugTimestamps, $wgRequestTime;
1017
1018 if ( !$wgDebugRawPage && wfIsDebugRawPage() ) {
1019 return;
1020 }
1021
1022 $text = trim( $text );
1023
1024 if ( $wgDebugTimestamps ) {
1025 $context['seconds_elapsed'] = sprintf(
1026 '%6.4f',
1027 microtime( true ) - $wgRequestTime
1028 );
1029 $context['memory_used'] = sprintf(
1030 '%5.1fM',
1031 ( memory_get_usage( true ) / ( 1024 * 1024 ) )
1032 );
1033 }
1034
1035 if ( $wgDebugLogPrefix !== '' ) {
1036 $context['prefix'] = $wgDebugLogPrefix;
1037 }
1038 $context['private'] = ( $dest === false || $dest === 'private' );
1039
1040 $logger = LoggerFactory::getInstance( 'wfDebug' );
1041 $logger->debug( $text, $context );
1042 }
1043
1044 /**
1045 * Returns true if debug logging should be suppressed if $wgDebugRawPage = false
1046 * @return bool
1047 */
1048 function wfIsDebugRawPage() {
1049 static $cache;
1050 if ( $cache !== null ) {
1051 return $cache;
1052 }
1053 # Check for raw action using $_GET not $wgRequest, since the latter might not be initialised yet
1054 if ( ( isset( $_GET['action'] ) && $_GET['action'] == 'raw' )
1055 || (
1056 isset( $_SERVER['SCRIPT_NAME'] )
1057 && substr( $_SERVER['SCRIPT_NAME'], -8 ) == 'load.php'
1058 )
1059 ) {
1060 $cache = true;
1061 } else {
1062 $cache = false;
1063 }
1064 return $cache;
1065 }
1066
1067 /**
1068 * Send a line giving PHP memory usage.
1069 *
1070 * @param bool $exact Print exact byte values instead of kibibytes (default: false)
1071 */
1072 function wfDebugMem( $exact = false ) {
1073 $mem = memory_get_usage();
1074 if ( !$exact ) {
1075 $mem = floor( $mem / 1024 ) . ' KiB';
1076 } else {
1077 $mem .= ' B';
1078 }
1079 wfDebug( "Memory usage: $mem\n" );
1080 }
1081
1082 /**
1083 * Send a line to a supplementary debug log file, if configured, or main debug
1084 * log if not.
1085 *
1086 * To configure a supplementary log file, set $wgDebugLogGroups[$logGroup] to
1087 * a string filename or an associative array mapping 'destination' to the
1088 * desired filename. The associative array may also contain a 'sample' key
1089 * with an integer value, specifying a sampling factor. Sampled log events
1090 * will be emitted with a 1 in N random chance.
1091 *
1092 * @since 1.23 support for sampling log messages via $wgDebugLogGroups.
1093 * @since 1.25 support for additional context data
1094 * @since 1.25 sample behavior dependent on configured $wgMWLoggerDefaultSpi
1095 *
1096 * @param string $logGroup
1097 * @param string $text
1098 * @param string|bool $dest Destination of the message:
1099 * - 'all': both to the log and HTML (debug toolbar or HTML comments)
1100 * - 'private': only to the specific log if set in $wgDebugLogGroups and
1101 * discarded otherwise
1102 * For backward compatibility, it can also take a boolean:
1103 * - true: same as 'all'
1104 * - false: same as 'private'
1105 * @param array $context Additional logging context data
1106 */
1107 function wfDebugLog(
1108 $logGroup, $text, $dest = 'all', array $context = []
1109 ) {
1110 $text = trim( $text );
1111
1112 $logger = LoggerFactory::getInstance( $logGroup );
1113 $context['private'] = ( $dest === false || $dest === 'private' );
1114 $logger->info( $text, $context );
1115 }
1116
1117 /**
1118 * Log for database errors
1119 *
1120 * @since 1.25 support for additional context data
1121 *
1122 * @param string $text Database error message.
1123 * @param array $context Additional logging context data
1124 */
1125 function wfLogDBError( $text, array $context = [] ) {
1126 $logger = LoggerFactory::getInstance( 'wfLogDBError' );
1127 $logger->error( trim( $text ), $context );
1128 }
1129
1130 /**
1131 * Throws a warning that $function is deprecated
1132 *
1133 * @param string $function
1134 * @param string|bool $version Version of MediaWiki that the function
1135 * was deprecated in (Added in 1.19).
1136 * @param string|bool $component Added in 1.19.
1137 * @param int $callerOffset How far up the call stack is the original
1138 * caller. 2 = function that called the function that called
1139 * wfDeprecated (Added in 1.20)
1140 *
1141 * @return null
1142 */
1143 function wfDeprecated( $function, $version = false, $component = false, $callerOffset = 2 ) {
1144 MWDebug::deprecated( $function, $version, $component, $callerOffset + 1 );
1145 }
1146
1147 /**
1148 * Send a warning either to the debug log or in a PHP error depending on
1149 * $wgDevelopmentWarnings. To log warnings in production, use wfLogWarning() instead.
1150 *
1151 * @param string $msg Message to send
1152 * @param int $callerOffset Number of items to go back in the backtrace to
1153 * find the correct caller (1 = function calling wfWarn, ...)
1154 * @param int $level PHP error level; defaults to E_USER_NOTICE;
1155 * only used when $wgDevelopmentWarnings is true
1156 */
1157 function wfWarn( $msg, $callerOffset = 1, $level = E_USER_NOTICE ) {
1158 MWDebug::warning( $msg, $callerOffset + 1, $level, 'auto' );
1159 }
1160
1161 /**
1162 * Send a warning as a PHP error and the debug log. This is intended for logging
1163 * warnings in production. For logging development warnings, use WfWarn instead.
1164 *
1165 * @param string $msg Message to send
1166 * @param int $callerOffset Number of items to go back in the backtrace to
1167 * find the correct caller (1 = function calling wfLogWarning, ...)
1168 * @param int $level PHP error level; defaults to E_USER_WARNING
1169 */
1170 function wfLogWarning( $msg, $callerOffset = 1, $level = E_USER_WARNING ) {
1171 MWDebug::warning( $msg, $callerOffset + 1, $level, 'production' );
1172 }
1173
1174 /**
1175 * Log to a file without getting "file size exceeded" signals.
1176 *
1177 * Can also log to TCP or UDP with the syntax udp://host:port/prefix. This will
1178 * send lines to the specified port, prefixed by the specified prefix and a space.
1179 * @since 1.25 support for additional context data
1180 *
1181 * @param string $text
1182 * @param string $file Filename
1183 * @param array $context Additional logging context data
1184 * @throws MWException
1185 * @deprecated since 1.25 Use \MediaWiki\Logger\LegacyLogger::emit or UDPTransport
1186 */
1187 function wfErrorLog( $text, $file, array $context = [] ) {
1188 wfDeprecated( __METHOD__, '1.25' );
1189 $logger = LoggerFactory::getInstance( 'wfErrorLog' );
1190 $context['destination'] = $file;
1191 $logger->info( trim( $text ), $context );
1192 }
1193
1194 /**
1195 * @todo document
1196 */
1197 function wfLogProfilingData() {
1198 global $wgDebugLogGroups, $wgDebugRawPage;
1199
1200 $context = RequestContext::getMain();
1201 $request = $context->getRequest();
1202
1203 $profiler = Profiler::instance();
1204 $profiler->setContext( $context );
1205 $profiler->logData();
1206
1207 $config = $context->getConfig();
1208 $stats = MediaWikiServices::getInstance()->getStatsdDataFactory();
1209 if ( $config->get( 'StatsdServer' ) && $stats->hasData() ) {
1210 try {
1211 $statsdServer = explode( ':', $config->get( 'StatsdServer' ) );
1212 $statsdHost = $statsdServer[0];
1213 $statsdPort = isset( $statsdServer[1] ) ? $statsdServer[1] : 8125;
1214 $statsdSender = new SocketSender( $statsdHost, $statsdPort );
1215 $statsdClient = new SamplingStatsdClient( $statsdSender, true, false );
1216 $statsdClient->setSamplingRates( $config->get( 'StatsdSamplingRates' ) );
1217 $statsdClient->send( $stats->getData() );
1218 } catch ( Exception $ex ) {
1219 MWExceptionHandler::logException( $ex );
1220 }
1221 }
1222
1223 # Profiling must actually be enabled...
1224 if ( $profiler instanceof ProfilerStub ) {
1225 return;
1226 }
1227
1228 if ( isset( $wgDebugLogGroups['profileoutput'] )
1229 && $wgDebugLogGroups['profileoutput'] === false
1230 ) {
1231 // Explicitly disabled
1232 return;
1233 }
1234 if ( !$wgDebugRawPage && wfIsDebugRawPage() ) {
1235 return;
1236 }
1237
1238 $ctx = [ 'elapsed' => $request->getElapsedTime() ];
1239 if ( !empty( $_SERVER['HTTP_X_FORWARDED_FOR'] ) ) {
1240 $ctx['forwarded_for'] = $_SERVER['HTTP_X_FORWARDED_FOR'];
1241 }
1242 if ( !empty( $_SERVER['HTTP_CLIENT_IP'] ) ) {
1243 $ctx['client_ip'] = $_SERVER['HTTP_CLIENT_IP'];
1244 }
1245 if ( !empty( $_SERVER['HTTP_FROM'] ) ) {
1246 $ctx['from'] = $_SERVER['HTTP_FROM'];
1247 }
1248 if ( isset( $ctx['forwarded_for'] ) ||
1249 isset( $ctx['client_ip'] ) ||
1250 isset( $ctx['from'] ) ) {
1251 $ctx['proxy'] = $_SERVER['REMOTE_ADDR'];
1252 }
1253
1254 // Don't load $wgUser at this late stage just for statistics purposes
1255 // @todo FIXME: We can detect some anons even if it is not loaded.
1256 // See User::getId()
1257 $user = $context->getUser();
1258 $ctx['anon'] = $user->isItemLoaded( 'id' ) && $user->isAnon();
1259
1260 // Command line script uses a FauxRequest object which does not have
1261 // any knowledge about an URL and throw an exception instead.
1262 try {
1263 $ctx['url'] = urldecode( $request->getRequestURL() );
1264 } catch ( Exception $ignored ) {
1265 // no-op
1266 }
1267
1268 $ctx['output'] = $profiler->getOutput();
1269
1270 $log = LoggerFactory::getInstance( 'profileoutput' );
1271 $log->info( "Elapsed: {elapsed}; URL: <{url}>\n{output}", $ctx );
1272 }
1273
1274 /**
1275 * Increment a statistics counter
1276 *
1277 * @param string $key
1278 * @param int $count
1279 * @return void
1280 */
1281 function wfIncrStats( $key, $count = 1 ) {
1282 $stats = MediaWikiServices::getInstance()->getStatsdDataFactory();
1283 $stats->updateCount( $key, $count );
1284 }
1285
1286 /**
1287 * Check whether the wiki is in read-only mode.
1288 *
1289 * @return bool
1290 */
1291 function wfReadOnly() {
1292 return MediaWikiServices::getInstance()->getReadOnlyMode()
1293 ->isReadOnly();
1294 }
1295
1296 /**
1297 * Check if the site is in read-only mode and return the message if so
1298 *
1299 * This checks wfConfiguredReadOnlyReason() and the main load balancer
1300 * for replica DB lag. This may result in DB connection being made.
1301 *
1302 * @return string|bool String when in read-only mode; false otherwise
1303 */
1304 function wfReadOnlyReason() {
1305 return MediaWikiServices::getInstance()->getReadOnlyMode()
1306 ->getReason();
1307 }
1308
1309 /**
1310 * Get the value of $wgReadOnly or the contents of $wgReadOnlyFile.
1311 *
1312 * @return string|bool String when in read-only mode; false otherwise
1313 * @since 1.27
1314 */
1315 function wfConfiguredReadOnlyReason() {
1316 return MediaWikiServices::getInstance()->getConfiguredReadOnlyMode()
1317 ->getReason();
1318 }
1319
1320 /**
1321 * Return a Language object from $langcode
1322 *
1323 * @param Language|string|bool $langcode Either:
1324 * - a Language object
1325 * - code of the language to get the message for, if it is
1326 * a valid code create a language for that language, if
1327 * it is a string but not a valid code then make a basic
1328 * language object
1329 * - a boolean: if it's false then use the global object for
1330 * the current user's language (as a fallback for the old parameter
1331 * functionality), or if it is true then use global object
1332 * for the wiki's content language.
1333 * @return Language
1334 */
1335 function wfGetLangObj( $langcode = false ) {
1336 # Identify which language to get or create a language object for.
1337 # Using is_object here due to Stub objects.
1338 if ( is_object( $langcode ) ) {
1339 # Great, we already have the object (hopefully)!
1340 return $langcode;
1341 }
1342
1343 global $wgContLang, $wgLanguageCode;
1344 if ( $langcode === true || $langcode === $wgLanguageCode ) {
1345 # $langcode is the language code of the wikis content language object.
1346 # or it is a boolean and value is true
1347 return $wgContLang;
1348 }
1349
1350 global $wgLang;
1351 if ( $langcode === false || $langcode === $wgLang->getCode() ) {
1352 # $langcode is the language code of user language object.
1353 # or it was a boolean and value is false
1354 return $wgLang;
1355 }
1356
1357 $validCodes = array_keys( Language::fetchLanguageNames() );
1358 if ( in_array( $langcode, $validCodes ) ) {
1359 # $langcode corresponds to a valid language.
1360 return Language::factory( $langcode );
1361 }
1362
1363 # $langcode is a string, but not a valid language code; use content language.
1364 wfDebug( "Invalid language code passed to wfGetLangObj, falling back to content language.\n" );
1365 return $wgContLang;
1366 }
1367
1368 /**
1369 * This is the function for getting translated interface messages.
1370 *
1371 * @see Message class for documentation how to use them.
1372 * @see https://www.mediawiki.org/wiki/Manual:Messages_API
1373 *
1374 * This function replaces all old wfMsg* functions.
1375 *
1376 * @param string|string[]|MessageSpecifier $key Message key, or array of keys, or a MessageSpecifier
1377 * @param mixed $params,... Normal message parameters
1378 * @return Message
1379 *
1380 * @since 1.17
1381 *
1382 * @see Message::__construct
1383 */
1384 function wfMessage( $key /*...*/ ) {
1385 $message = new Message( $key );
1386
1387 // We call Message::params() to reduce code duplication
1388 $params = func_get_args();
1389 array_shift( $params );
1390 if ( $params ) {
1391 call_user_func_array( [ $message, 'params' ], $params );
1392 }
1393
1394 return $message;
1395 }
1396
1397 /**
1398 * This function accepts multiple message keys and returns a message instance
1399 * for the first message which is non-empty. If all messages are empty then an
1400 * instance of the first message key is returned.
1401 *
1402 * @param string|string[] $keys,... Message keys
1403 * @return Message
1404 *
1405 * @since 1.18
1406 *
1407 * @see Message::newFallbackSequence
1408 */
1409 function wfMessageFallback( /*...*/ ) {
1410 $args = func_get_args();
1411 return call_user_func_array( 'Message::newFallbackSequence', $args );
1412 }
1413
1414 /**
1415 * Replace message parameter keys on the given formatted output.
1416 *
1417 * @param string $message
1418 * @param array $args
1419 * @return string
1420 * @private
1421 */
1422 function wfMsgReplaceArgs( $message, $args ) {
1423 # Fix windows line-endings
1424 # Some messages are split with explode("\n", $msg)
1425 $message = str_replace( "\r", '', $message );
1426
1427 // Replace arguments
1428 if ( is_array( $args ) && $args ) {
1429 if ( is_array( $args[0] ) ) {
1430 $args = array_values( $args[0] );
1431 }
1432 $replacementKeys = [];
1433 foreach ( $args as $n => $param ) {
1434 $replacementKeys['$' . ( $n + 1 )] = $param;
1435 }
1436 $message = strtr( $message, $replacementKeys );
1437 }
1438
1439 return $message;
1440 }
1441
1442 /**
1443 * Fetch server name for use in error reporting etc.
1444 * Use real server name if available, so we know which machine
1445 * in a server farm generated the current page.
1446 *
1447 * @return string
1448 */
1449 function wfHostname() {
1450 static $host;
1451 if ( is_null( $host ) ) {
1452 # Hostname overriding
1453 global $wgOverrideHostname;
1454 if ( $wgOverrideHostname !== false ) {
1455 # Set static and skip any detection
1456 $host = $wgOverrideHostname;
1457 return $host;
1458 }
1459
1460 if ( function_exists( 'posix_uname' ) ) {
1461 // This function not present on Windows
1462 $uname = posix_uname();
1463 } else {
1464 $uname = false;
1465 }
1466 if ( is_array( $uname ) && isset( $uname['nodename'] ) ) {
1467 $host = $uname['nodename'];
1468 } elseif ( getenv( 'COMPUTERNAME' ) ) {
1469 # Windows computer name
1470 $host = getenv( 'COMPUTERNAME' );
1471 } else {
1472 # This may be a virtual server.
1473 $host = $_SERVER['SERVER_NAME'];
1474 }
1475 }
1476 return $host;
1477 }
1478
1479 /**
1480 * Returns a script tag that stores the amount of time it took MediaWiki to
1481 * handle the request in milliseconds as 'wgBackendResponseTime'.
1482 *
1483 * If $wgShowHostnames is true, the script will also set 'wgHostname' to the
1484 * hostname of the server handling the request.
1485 *
1486 * @return string
1487 */
1488 function wfReportTime() {
1489 global $wgRequestTime, $wgShowHostnames;
1490
1491 $responseTime = round( ( microtime( true ) - $wgRequestTime ) * 1000 );
1492 $reportVars = [ 'wgBackendResponseTime' => $responseTime ];
1493 if ( $wgShowHostnames ) {
1494 $reportVars['wgHostname'] = wfHostname();
1495 }
1496 return Skin::makeVariablesScript( $reportVars );
1497 }
1498
1499 /**
1500 * Safety wrapper for debug_backtrace().
1501 *
1502 * Will return an empty array if debug_backtrace is disabled, otherwise
1503 * the output from debug_backtrace() (trimmed).
1504 *
1505 * @param int $limit This parameter can be used to limit the number of stack frames returned
1506 *
1507 * @return array Array of backtrace information
1508 */
1509 function wfDebugBacktrace( $limit = 0 ) {
1510 static $disabled = null;
1511
1512 if ( is_null( $disabled ) ) {
1513 $disabled = !function_exists( 'debug_backtrace' );
1514 if ( $disabled ) {
1515 wfDebug( "debug_backtrace() is disabled\n" );
1516 }
1517 }
1518 if ( $disabled ) {
1519 return [];
1520 }
1521
1522 if ( $limit ) {
1523 return array_slice( debug_backtrace( DEBUG_BACKTRACE_PROVIDE_OBJECT, $limit + 1 ), 1 );
1524 } else {
1525 return array_slice( debug_backtrace(), 1 );
1526 }
1527 }
1528
1529 /**
1530 * Get a debug backtrace as a string
1531 *
1532 * @param bool|null $raw If true, the return value is plain text. If false, HTML.
1533 * Defaults to $wgCommandLineMode if unset.
1534 * @return string
1535 * @since 1.25 Supports $raw parameter.
1536 */
1537 function wfBacktrace( $raw = null ) {
1538 global $wgCommandLineMode;
1539
1540 if ( $raw === null ) {
1541 $raw = $wgCommandLineMode;
1542 }
1543
1544 if ( $raw ) {
1545 $frameFormat = "%s line %s calls %s()\n";
1546 $traceFormat = "%s";
1547 } else {
1548 $frameFormat = "<li>%s line %s calls %s()</li>\n";
1549 $traceFormat = "<ul>\n%s</ul>\n";
1550 }
1551
1552 $frames = array_map( function ( $frame ) use ( $frameFormat ) {
1553 $file = !empty( $frame['file'] ) ? basename( $frame['file'] ) : '-';
1554 $line = isset( $frame['line'] ) ? $frame['line'] : '-';
1555 $call = $frame['function'];
1556 if ( !empty( $frame['class'] ) ) {
1557 $call = $frame['class'] . $frame['type'] . $call;
1558 }
1559 return sprintf( $frameFormat, $file, $line, $call );
1560 }, wfDebugBacktrace() );
1561
1562 return sprintf( $traceFormat, implode( '', $frames ) );
1563 }
1564
1565 /**
1566 * Get the name of the function which called this function
1567 * wfGetCaller( 1 ) is the function with the wfGetCaller() call (ie. __FUNCTION__)
1568 * wfGetCaller( 2 ) [default] is the caller of the function running wfGetCaller()
1569 * wfGetCaller( 3 ) is the parent of that.
1570 *
1571 * @param int $level
1572 * @return string
1573 */
1574 function wfGetCaller( $level = 2 ) {
1575 $backtrace = wfDebugBacktrace( $level + 1 );
1576 if ( isset( $backtrace[$level] ) ) {
1577 return wfFormatStackFrame( $backtrace[$level] );
1578 } else {
1579 return 'unknown';
1580 }
1581 }
1582
1583 /**
1584 * Return a string consisting of callers in the stack. Useful sometimes
1585 * for profiling specific points.
1586 *
1587 * @param int $limit The maximum depth of the stack frame to return, or false for the entire stack.
1588 * @return string
1589 */
1590 function wfGetAllCallers( $limit = 3 ) {
1591 $trace = array_reverse( wfDebugBacktrace() );
1592 if ( !$limit || $limit > count( $trace ) - 1 ) {
1593 $limit = count( $trace ) - 1;
1594 }
1595 $trace = array_slice( $trace, -$limit - 1, $limit );
1596 return implode( '/', array_map( 'wfFormatStackFrame', $trace ) );
1597 }
1598
1599 /**
1600 * Return a string representation of frame
1601 *
1602 * @param array $frame
1603 * @return string
1604 */
1605 function wfFormatStackFrame( $frame ) {
1606 if ( !isset( $frame['function'] ) ) {
1607 return 'NO_FUNCTION_GIVEN';
1608 }
1609 return isset( $frame['class'] ) && isset( $frame['type'] ) ?
1610 $frame['class'] . $frame['type'] . $frame['function'] :
1611 $frame['function'];
1612 }
1613
1614 /* Some generic result counters, pulled out of SearchEngine */
1615
1616 /**
1617 * @todo document
1618 *
1619 * @param int $offset
1620 * @param int $limit
1621 * @return string
1622 */
1623 function wfShowingResults( $offset, $limit ) {
1624 return wfMessage( 'showingresults' )->numParams( $limit, $offset + 1 )->parse();
1625 }
1626
1627 /**
1628 * Whether the client accept gzip encoding
1629 *
1630 * Uses the Accept-Encoding header to check if the client supports gzip encoding.
1631 * Use this when considering to send a gzip-encoded response to the client.
1632 *
1633 * @param bool $force Forces another check even if we already have a cached result.
1634 * @return bool
1635 */
1636 function wfClientAcceptsGzip( $force = false ) {
1637 static $result = null;
1638 if ( $result === null || $force ) {
1639 $result = false;
1640 if ( isset( $_SERVER['HTTP_ACCEPT_ENCODING'] ) ) {
1641 # @todo FIXME: We may want to blacklist some broken browsers
1642 $m = [];
1643 if ( preg_match(
1644 '/\bgzip(?:;(q)=([0-9]+(?:\.[0-9]+)))?\b/',
1645 $_SERVER['HTTP_ACCEPT_ENCODING'],
1646 $m
1647 )
1648 ) {
1649 if ( isset( $m[2] ) && ( $m[1] == 'q' ) && ( $m[2] == 0 ) ) {
1650 $result = false;
1651 return $result;
1652 }
1653 wfDebug( "wfClientAcceptsGzip: client accepts gzip.\n" );
1654 $result = true;
1655 }
1656 }
1657 }
1658 return $result;
1659 }
1660
1661 /**
1662 * Escapes the given text so that it may be output using addWikiText()
1663 * without any linking, formatting, etc. making its way through. This
1664 * is achieved by substituting certain characters with HTML entities.
1665 * As required by the callers, "<nowiki>" is not used.
1666 *
1667 * @param string $text Text to be escaped
1668 * @return string
1669 */
1670 function wfEscapeWikiText( $text ) {
1671 global $wgEnableMagicLinks;
1672 static $repl = null, $repl2 = null;
1673 if ( $repl === null || defined( 'MW_PARSER_TEST' ) || defined( 'MW_PHPUNIT_TEST' ) ) {
1674 // Tests depend upon being able to change $wgEnableMagicLinks, so don't cache
1675 // in those situations
1676 $repl = [
1677 '"' => '&#34;', '&' => '&#38;', "'" => '&#39;', '<' => '&#60;',
1678 '=' => '&#61;', '>' => '&#62;', '[' => '&#91;', ']' => '&#93;',
1679 '{' => '&#123;', '|' => '&#124;', '}' => '&#125;', ';' => '&#59;',
1680 "\n#" => "\n&#35;", "\r#" => "\r&#35;",
1681 "\n*" => "\n&#42;", "\r*" => "\r&#42;",
1682 "\n:" => "\n&#58;", "\r:" => "\r&#58;",
1683 "\n " => "\n&#32;", "\r " => "\r&#32;",
1684 "\n\n" => "\n&#10;", "\r\n" => "&#13;\n",
1685 "\n\r" => "\n&#13;", "\r\r" => "\r&#13;",
1686 "\n\t" => "\n&#9;", "\r\t" => "\r&#9;", // "\n\t\n" is treated like "\n\n"
1687 "\n----" => "\n&#45;---", "\r----" => "\r&#45;---",
1688 '__' => '_&#95;', '://' => '&#58;//',
1689 ];
1690
1691 $magicLinks = array_keys( array_filter( $wgEnableMagicLinks ) );
1692 // We have to catch everything "\s" matches in PCRE
1693 foreach ( $magicLinks as $magic ) {
1694 $repl["$magic "] = "$magic&#32;";
1695 $repl["$magic\t"] = "$magic&#9;";
1696 $repl["$magic\r"] = "$magic&#13;";
1697 $repl["$magic\n"] = "$magic&#10;";
1698 $repl["$magic\f"] = "$magic&#12;";
1699 }
1700
1701 // And handle protocols that don't use "://"
1702 global $wgUrlProtocols;
1703 $repl2 = [];
1704 foreach ( $wgUrlProtocols as $prot ) {
1705 if ( substr( $prot, -1 ) === ':' ) {
1706 $repl2[] = preg_quote( substr( $prot, 0, -1 ), '/' );
1707 }
1708 }
1709 $repl2 = $repl2 ? '/\b(' . implode( '|', $repl2 ) . '):/i' : '/^(?!)/';
1710 }
1711 $text = substr( strtr( "\n$text", $repl ), 1 );
1712 $text = preg_replace( $repl2, '$1&#58;', $text );
1713 return $text;
1714 }
1715
1716 /**
1717 * Sets dest to source and returns the original value of dest
1718 * If source is NULL, it just returns the value, it doesn't set the variable
1719 * If force is true, it will set the value even if source is NULL
1720 *
1721 * @param mixed $dest
1722 * @param mixed $source
1723 * @param bool $force
1724 * @return mixed
1725 */
1726 function wfSetVar( &$dest, $source, $force = false ) {
1727 $temp = $dest;
1728 if ( !is_null( $source ) || $force ) {
1729 $dest = $source;
1730 }
1731 return $temp;
1732 }
1733
1734 /**
1735 * As for wfSetVar except setting a bit
1736 *
1737 * @param int $dest
1738 * @param int $bit
1739 * @param bool $state
1740 *
1741 * @return bool
1742 */
1743 function wfSetBit( &$dest, $bit, $state = true ) {
1744 $temp = (bool)( $dest & $bit );
1745 if ( !is_null( $state ) ) {
1746 if ( $state ) {
1747 $dest |= $bit;
1748 } else {
1749 $dest &= ~$bit;
1750 }
1751 }
1752 return $temp;
1753 }
1754
1755 /**
1756 * A wrapper around the PHP function var_export().
1757 * Either print it or add it to the regular output ($wgOut).
1758 *
1759 * @param mixed $var A PHP variable to dump.
1760 */
1761 function wfVarDump( $var ) {
1762 global $wgOut;
1763 $s = str_replace( "\n", "<br />\n", var_export( $var, true ) . "\n" );
1764 if ( headers_sent() || !isset( $wgOut ) || !is_object( $wgOut ) ) {
1765 print $s;
1766 } else {
1767 $wgOut->addHTML( $s );
1768 }
1769 }
1770
1771 /**
1772 * Provide a simple HTTP error.
1773 *
1774 * @param int|string $code
1775 * @param string $label
1776 * @param string $desc
1777 */
1778 function wfHttpError( $code, $label, $desc ) {
1779 global $wgOut;
1780 HttpStatus::header( $code );
1781 if ( $wgOut ) {
1782 $wgOut->disable();
1783 $wgOut->sendCacheControl();
1784 }
1785
1786 MediaWiki\HeaderCallback::warnIfHeadersSent();
1787 header( 'Content-type: text/html; charset=utf-8' );
1788 print '<!DOCTYPE html>' .
1789 '<html><head><title>' .
1790 htmlspecialchars( $label ) .
1791 '</title></head><body><h1>' .
1792 htmlspecialchars( $label ) .
1793 '</h1><p>' .
1794 nl2br( htmlspecialchars( $desc ) ) .
1795 "</p></body></html>\n";
1796 }
1797
1798 /**
1799 * Clear away any user-level output buffers, discarding contents.
1800 *
1801 * Suitable for 'starting afresh', for instance when streaming
1802 * relatively large amounts of data without buffering, or wanting to
1803 * output image files without ob_gzhandler's compression.
1804 *
1805 * The optional $resetGzipEncoding parameter controls suppression of
1806 * the Content-Encoding header sent by ob_gzhandler; by default it
1807 * is left. See comments for wfClearOutputBuffers() for why it would
1808 * be used.
1809 *
1810 * Note that some PHP configuration options may add output buffer
1811 * layers which cannot be removed; these are left in place.
1812 *
1813 * @param bool $resetGzipEncoding
1814 */
1815 function wfResetOutputBuffers( $resetGzipEncoding = true ) {
1816 if ( $resetGzipEncoding ) {
1817 // Suppress Content-Encoding and Content-Length
1818 // headers from 1.10+s wfOutputHandler
1819 global $wgDisableOutputCompression;
1820 $wgDisableOutputCompression = true;
1821 }
1822 while ( $status = ob_get_status() ) {
1823 if ( isset( $status['flags'] ) ) {
1824 $flags = PHP_OUTPUT_HANDLER_CLEANABLE | PHP_OUTPUT_HANDLER_REMOVABLE;
1825 $deleteable = ( $status['flags'] & $flags ) === $flags;
1826 } elseif ( isset( $status['del'] ) ) {
1827 $deleteable = $status['del'];
1828 } else {
1829 // Guess that any PHP-internal setting can't be removed.
1830 $deleteable = $status['type'] !== 0; /* PHP_OUTPUT_HANDLER_INTERNAL */
1831 }
1832 if ( !$deleteable ) {
1833 // Give up, and hope the result doesn't break
1834 // output behavior.
1835 break;
1836 }
1837 if ( $status['name'] === 'MediaWikiTestCase::wfResetOutputBuffersBarrier' ) {
1838 // Unit testing barrier to prevent this function from breaking PHPUnit.
1839 break;
1840 }
1841 if ( !ob_end_clean() ) {
1842 // Could not remove output buffer handler; abort now
1843 // to avoid getting in some kind of infinite loop.
1844 break;
1845 }
1846 if ( $resetGzipEncoding ) {
1847 if ( $status['name'] == 'ob_gzhandler' ) {
1848 // Reset the 'Content-Encoding' field set by this handler
1849 // so we can start fresh.
1850 header_remove( 'Content-Encoding' );
1851 break;
1852 }
1853 }
1854 }
1855 }
1856
1857 /**
1858 * More legible than passing a 'false' parameter to wfResetOutputBuffers():
1859 *
1860 * Clear away output buffers, but keep the Content-Encoding header
1861 * produced by ob_gzhandler, if any.
1862 *
1863 * This should be used for HTTP 304 responses, where you need to
1864 * preserve the Content-Encoding header of the real result, but
1865 * also need to suppress the output of ob_gzhandler to keep to spec
1866 * and avoid breaking Firefox in rare cases where the headers and
1867 * body are broken over two packets.
1868 */
1869 function wfClearOutputBuffers() {
1870 wfResetOutputBuffers( false );
1871 }
1872
1873 /**
1874 * Converts an Accept-* header into an array mapping string values to quality
1875 * factors
1876 *
1877 * @param string $accept
1878 * @param string $def Default
1879 * @return float[] Associative array of string => float pairs
1880 */
1881 function wfAcceptToPrefs( $accept, $def = '*/*' ) {
1882 # No arg means accept anything (per HTTP spec)
1883 if ( !$accept ) {
1884 return [ $def => 1.0 ];
1885 }
1886
1887 $prefs = [];
1888
1889 $parts = explode( ',', $accept );
1890
1891 foreach ( $parts as $part ) {
1892 # @todo FIXME: Doesn't deal with params like 'text/html; level=1'
1893 $values = explode( ';', trim( $part ) );
1894 $match = [];
1895 if ( count( $values ) == 1 ) {
1896 $prefs[$values[0]] = 1.0;
1897 } elseif ( preg_match( '/q\s*=\s*(\d*\.\d+)/', $values[1], $match ) ) {
1898 $prefs[$values[0]] = floatval( $match[1] );
1899 }
1900 }
1901
1902 return $prefs;
1903 }
1904
1905 /**
1906 * Checks if a given MIME type matches any of the keys in the given
1907 * array. Basic wildcards are accepted in the array keys.
1908 *
1909 * Returns the matching MIME type (or wildcard) if a match, otherwise
1910 * NULL if no match.
1911 *
1912 * @param string $type
1913 * @param array $avail
1914 * @return string
1915 * @private
1916 */
1917 function mimeTypeMatch( $type, $avail ) {
1918 if ( array_key_exists( $type, $avail ) ) {
1919 return $type;
1920 } else {
1921 $mainType = explode( '/', $type )[0];
1922 if ( array_key_exists( "$mainType/*", $avail ) ) {
1923 return "$mainType/*";
1924 } elseif ( array_key_exists( '*/*', $avail ) ) {
1925 return '*/*';
1926 } else {
1927 return null;
1928 }
1929 }
1930 }
1931
1932 /**
1933 * Returns the 'best' match between a client's requested internet media types
1934 * and the server's list of available types. Each list should be an associative
1935 * array of type to preference (preference is a float between 0.0 and 1.0).
1936 * Wildcards in the types are acceptable.
1937 *
1938 * @param array $cprefs Client's acceptable type list
1939 * @param array $sprefs Server's offered types
1940 * @return string
1941 *
1942 * @todo FIXME: Doesn't handle params like 'text/plain; charset=UTF-8'
1943 * XXX: generalize to negotiate other stuff
1944 */
1945 function wfNegotiateType( $cprefs, $sprefs ) {
1946 $combine = [];
1947
1948 foreach ( array_keys( $sprefs ) as $type ) {
1949 $subType = explode( '/', $type )[1];
1950 if ( $subType != '*' ) {
1951 $ckey = mimeTypeMatch( $type, $cprefs );
1952 if ( $ckey ) {
1953 $combine[$type] = $sprefs[$type] * $cprefs[$ckey];
1954 }
1955 }
1956 }
1957
1958 foreach ( array_keys( $cprefs ) as $type ) {
1959 $subType = explode( '/', $type )[1];
1960 if ( $subType != '*' && !array_key_exists( $type, $sprefs ) ) {
1961 $skey = mimeTypeMatch( $type, $sprefs );
1962 if ( $skey ) {
1963 $combine[$type] = $sprefs[$skey] * $cprefs[$type];
1964 }
1965 }
1966 }
1967
1968 $bestq = 0;
1969 $besttype = null;
1970
1971 foreach ( array_keys( $combine ) as $type ) {
1972 if ( $combine[$type] > $bestq ) {
1973 $besttype = $type;
1974 $bestq = $combine[$type];
1975 }
1976 }
1977
1978 return $besttype;
1979 }
1980
1981 /**
1982 * Reference-counted warning suppression
1983 *
1984 * @deprecated since 1.26, use MediaWiki\suppressWarnings() directly
1985 * @param bool $end
1986 */
1987 function wfSuppressWarnings( $end = false ) {
1988 MediaWiki\suppressWarnings( $end );
1989 }
1990
1991 /**
1992 * @deprecated since 1.26, use MediaWiki\restoreWarnings() directly
1993 * Restore error level to previous value
1994 */
1995 function wfRestoreWarnings() {
1996 MediaWiki\suppressWarnings( true );
1997 }
1998
1999 /**
2000 * Get a timestamp string in one of various formats
2001 *
2002 * @param mixed $outputtype A timestamp in one of the supported formats, the
2003 * function will autodetect which format is supplied and act accordingly.
2004 * @param mixed $ts Optional timestamp to convert, default 0 for the current time
2005 * @return string|bool String / false The same date in the format specified in $outputtype or false
2006 */
2007 function wfTimestamp( $outputtype = TS_UNIX, $ts = 0 ) {
2008 $ret = MWTimestamp::convert( $outputtype, $ts );
2009 if ( $ret === false ) {
2010 wfDebug( "wfTimestamp() fed bogus time value: TYPE=$outputtype; VALUE=$ts\n" );
2011 }
2012 return $ret;
2013 }
2014
2015 /**
2016 * Return a formatted timestamp, or null if input is null.
2017 * For dealing with nullable timestamp columns in the database.
2018 *
2019 * @param int $outputtype
2020 * @param string $ts
2021 * @return string
2022 */
2023 function wfTimestampOrNull( $outputtype = TS_UNIX, $ts = null ) {
2024 if ( is_null( $ts ) ) {
2025 return null;
2026 } else {
2027 return wfTimestamp( $outputtype, $ts );
2028 }
2029 }
2030
2031 /**
2032 * Convenience function; returns MediaWiki timestamp for the present time.
2033 *
2034 * @return string
2035 */
2036 function wfTimestampNow() {
2037 # return NOW
2038 return MWTimestamp::now( TS_MW );
2039 }
2040
2041 /**
2042 * Check if the operating system is Windows
2043 *
2044 * @return bool True if it's Windows, false otherwise.
2045 */
2046 function wfIsWindows() {
2047 static $isWindows = null;
2048 if ( $isWindows === null ) {
2049 $isWindows = strtoupper( substr( PHP_OS, 0, 3 ) ) === 'WIN';
2050 }
2051 return $isWindows;
2052 }
2053
2054 /**
2055 * Check if we are running under HHVM
2056 *
2057 * @return bool
2058 */
2059 function wfIsHHVM() {
2060 return defined( 'HHVM_VERSION' );
2061 }
2062
2063 /**
2064 * Tries to get the system directory for temporary files. First
2065 * $wgTmpDirectory is checked, and then the TMPDIR, TMP, and TEMP
2066 * environment variables are then checked in sequence, then
2067 * sys_get_temp_dir(), then upload_tmp_dir from php.ini.
2068 *
2069 * NOTE: When possible, use instead the tmpfile() function to create
2070 * temporary files to avoid race conditions on file creation, etc.
2071 *
2072 * @return string
2073 */
2074 function wfTempDir() {
2075 global $wgTmpDirectory;
2076
2077 if ( $wgTmpDirectory !== false ) {
2078 return $wgTmpDirectory;
2079 }
2080
2081 return TempFSFile::getUsableTempDirectory();
2082 }
2083
2084 /**
2085 * Make directory, and make all parent directories if they don't exist
2086 *
2087 * @param string $dir Full path to directory to create
2088 * @param int $mode Chmod value to use, default is $wgDirectoryMode
2089 * @param string $caller Optional caller param for debugging.
2090 * @throws MWException
2091 * @return bool
2092 */
2093 function wfMkdirParents( $dir, $mode = null, $caller = null ) {
2094 global $wgDirectoryMode;
2095
2096 if ( FileBackend::isStoragePath( $dir ) ) { // sanity
2097 throw new MWException( __FUNCTION__ . " given storage path '$dir'." );
2098 }
2099
2100 if ( !is_null( $caller ) ) {
2101 wfDebug( "$caller: called wfMkdirParents($dir)\n" );
2102 }
2103
2104 if ( strval( $dir ) === '' || is_dir( $dir ) ) {
2105 return true;
2106 }
2107
2108 $dir = str_replace( [ '\\', '/' ], DIRECTORY_SEPARATOR, $dir );
2109
2110 if ( is_null( $mode ) ) {
2111 $mode = $wgDirectoryMode;
2112 }
2113
2114 // Turn off the normal warning, we're doing our own below
2115 MediaWiki\suppressWarnings();
2116 $ok = mkdir( $dir, $mode, true ); // PHP5 <3
2117 MediaWiki\restoreWarnings();
2118
2119 if ( !$ok ) {
2120 // directory may have been created on another request since we last checked
2121 if ( is_dir( $dir ) ) {
2122 return true;
2123 }
2124
2125 // PHP doesn't report the path in its warning message, so add our own to aid in diagnosis.
2126 wfLogWarning( sprintf( "failed to mkdir \"%s\" mode 0%o", $dir, $mode ) );
2127 }
2128 return $ok;
2129 }
2130
2131 /**
2132 * Remove a directory and all its content.
2133 * Does not hide error.
2134 * @param string $dir
2135 */
2136 function wfRecursiveRemoveDir( $dir ) {
2137 wfDebug( __FUNCTION__ . "( $dir )\n" );
2138 // taken from https://secure.php.net/manual/en/function.rmdir.php#98622
2139 if ( is_dir( $dir ) ) {
2140 $objects = scandir( $dir );
2141 foreach ( $objects as $object ) {
2142 if ( $object != "." && $object != ".." ) {
2143 if ( filetype( $dir . '/' . $object ) == "dir" ) {
2144 wfRecursiveRemoveDir( $dir . '/' . $object );
2145 } else {
2146 unlink( $dir . '/' . $object );
2147 }
2148 }
2149 }
2150 reset( $objects );
2151 rmdir( $dir );
2152 }
2153 }
2154
2155 /**
2156 * @param int $nr The number to format
2157 * @param int $acc The number of digits after the decimal point, default 2
2158 * @param bool $round Whether or not to round the value, default true
2159 * @return string
2160 */
2161 function wfPercent( $nr, $acc = 2, $round = true ) {
2162 $ret = sprintf( "%.${acc}f", $nr );
2163 return $round ? round( $ret, $acc ) . '%' : "$ret%";
2164 }
2165
2166 /**
2167 * Safety wrapper around ini_get() for boolean settings.
2168 * The values returned from ini_get() are pre-normalized for settings
2169 * set via php.ini or php_flag/php_admin_flag... but *not*
2170 * for those set via php_value/php_admin_value.
2171 *
2172 * It's fairly common for people to use php_value instead of php_flag,
2173 * which can leave you with an 'off' setting giving a false positive
2174 * for code that just takes the ini_get() return value as a boolean.
2175 *
2176 * To make things extra interesting, setting via php_value accepts
2177 * "true" and "yes" as true, but php.ini and php_flag consider them false. :)
2178 * Unrecognized values go false... again opposite PHP's own coercion
2179 * from string to bool.
2180 *
2181 * Luckily, 'properly' set settings will always come back as '0' or '1',
2182 * so we only have to worry about them and the 'improper' settings.
2183 *
2184 * I frickin' hate PHP... :P
2185 *
2186 * @param string $setting
2187 * @return bool
2188 */
2189 function wfIniGetBool( $setting ) {
2190 $val = strtolower( ini_get( $setting ) );
2191 // 'on' and 'true' can't have whitespace around them, but '1' can.
2192 return $val == 'on'
2193 || $val == 'true'
2194 || $val == 'yes'
2195 || preg_match( "/^\s*[+-]?0*[1-9]/", $val ); // approx C atoi() function
2196 }
2197
2198 /**
2199 * Version of escapeshellarg() that works better on Windows.
2200 *
2201 * Originally, this fixed the incorrect use of single quotes on Windows
2202 * (https://bugs.php.net/bug.php?id=26285) and the locale problems on Linux in
2203 * PHP 5.2.6+ (bug backported to earlier distro releases of PHP).
2204 *
2205 * @param string ... strings to escape and glue together, or a single array of strings parameter
2206 * @return string
2207 */
2208 function wfEscapeShellArg( /*...*/ ) {
2209 $args = func_get_args();
2210 if ( count( $args ) === 1 && is_array( reset( $args ) ) ) {
2211 // If only one argument has been passed, and that argument is an array,
2212 // treat it as a list of arguments
2213 $args = reset( $args );
2214 }
2215
2216 $first = true;
2217 $retVal = '';
2218 foreach ( $args as $arg ) {
2219 if ( !$first ) {
2220 $retVal .= ' ';
2221 } else {
2222 $first = false;
2223 }
2224
2225 if ( wfIsWindows() ) {
2226 // Escaping for an MSVC-style command line parser and CMD.EXE
2227 // @codingStandardsIgnoreStart For long URLs
2228 // Refs:
2229 // * https://web.archive.org/web/20020708081031/http://mailman.lyra.org/pipermail/scite-interest/2002-March/000436.html
2230 // * https://technet.microsoft.com/en-us/library/cc723564.aspx
2231 // * T15518
2232 // * CR r63214
2233 // Double the backslashes before any double quotes. Escape the double quotes.
2234 // @codingStandardsIgnoreEnd
2235 $tokens = preg_split( '/(\\\\*")/', $arg, -1, PREG_SPLIT_DELIM_CAPTURE );
2236 $arg = '';
2237 $iteration = 0;
2238 foreach ( $tokens as $token ) {
2239 if ( $iteration % 2 == 1 ) {
2240 // Delimiter, a double quote preceded by zero or more slashes
2241 $arg .= str_replace( '\\', '\\\\', substr( $token, 0, -1 ) ) . '\\"';
2242 } elseif ( $iteration % 4 == 2 ) {
2243 // ^ in $token will be outside quotes, need to be escaped
2244 $arg .= str_replace( '^', '^^', $token );
2245 } else { // $iteration % 4 == 0
2246 // ^ in $token will appear inside double quotes, so leave as is
2247 $arg .= $token;
2248 }
2249 $iteration++;
2250 }
2251 // Double the backslashes before the end of the string, because
2252 // we will soon add a quote
2253 $m = [];
2254 if ( preg_match( '/^(.*?)(\\\\+)$/', $arg, $m ) ) {
2255 $arg = $m[1] . str_replace( '\\', '\\\\', $m[2] );
2256 }
2257
2258 // Add surrounding quotes
2259 $retVal .= '"' . $arg . '"';
2260 } else {
2261 $retVal .= escapeshellarg( $arg );
2262 }
2263 }
2264 return $retVal;
2265 }
2266
2267 /**
2268 * Check if wfShellExec() is effectively disabled via php.ini config
2269 *
2270 * @return bool|string False or 'disabled'
2271 * @since 1.22
2272 */
2273 function wfShellExecDisabled() {
2274 static $disabled = null;
2275 if ( is_null( $disabled ) ) {
2276 if ( !function_exists( 'proc_open' ) ) {
2277 wfDebug( "proc_open() is disabled\n" );
2278 $disabled = 'disabled';
2279 } else {
2280 $disabled = false;
2281 }
2282 }
2283 return $disabled;
2284 }
2285
2286 /**
2287 * Execute a shell command, with time and memory limits mirrored from the PHP
2288 * configuration if supported.
2289 *
2290 * @param string|string[] $cmd If string, a properly shell-escaped command line,
2291 * or an array of unescaped arguments, in which case each value will be escaped
2292 * Example: [ 'convert', '-font', 'font name' ] would produce "'convert' '-font' 'font name'"
2293 * @param null|mixed &$retval Optional, will receive the program's exit code.
2294 * (non-zero is usually failure). If there is an error from
2295 * read, select, or proc_open(), this will be set to -1.
2296 * @param array $environ Optional environment variables which should be
2297 * added to the executed command environment.
2298 * @param array $limits Optional array with limits(filesize, memory, time, walltime)
2299 * this overwrites the global wgMaxShell* limits.
2300 * @param array $options Array of options:
2301 * - duplicateStderr: Set this to true to duplicate stderr to stdout,
2302 * including errors from limit.sh
2303 * - profileMethod: By default this function will profile based on the calling
2304 * method. Set this to a string for an alternative method to profile from
2305 *
2306 * @return string Collected stdout as a string
2307 */
2308 function wfShellExec( $cmd, &$retval = null, $environ = [],
2309 $limits = [], $options = []
2310 ) {
2311 global $IP, $wgMaxShellMemory, $wgMaxShellFileSize, $wgMaxShellTime,
2312 $wgMaxShellWallClockTime, $wgShellCgroup;
2313
2314 $disabled = wfShellExecDisabled();
2315 if ( $disabled ) {
2316 $retval = 1;
2317 return 'Unable to run external programs, proc_open() is disabled.';
2318 }
2319
2320 $includeStderr = isset( $options['duplicateStderr'] ) && $options['duplicateStderr'];
2321 $profileMethod = isset( $options['profileMethod'] ) ? $options['profileMethod'] : wfGetCaller();
2322
2323 $envcmd = '';
2324 foreach ( $environ as $k => $v ) {
2325 if ( wfIsWindows() ) {
2326 /* Surrounding a set in quotes (method used by wfEscapeShellArg) makes the quotes themselves
2327 * appear in the environment variable, so we must use carat escaping as documented in
2328 * https://technet.microsoft.com/en-us/library/cc723564.aspx
2329 * Note however that the quote isn't listed there, but is needed, and the parentheses
2330 * are listed there but doesn't appear to need it.
2331 */
2332 $envcmd .= "set $k=" . preg_replace( '/([&|()<>^"])/', '^\\1', $v ) . '&& ';
2333 } else {
2334 /* Assume this is a POSIX shell, thus required to accept variable assignments before the command
2335 * http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_09_01
2336 */
2337 $envcmd .= "$k=" . escapeshellarg( $v ) . ' ';
2338 }
2339 }
2340 if ( is_array( $cmd ) ) {
2341 $cmd = wfEscapeShellArg( $cmd );
2342 }
2343
2344 $cmd = $envcmd . $cmd;
2345
2346 $useLogPipe = false;
2347 if ( is_executable( '/bin/bash' ) ) {
2348 $time = intval( isset( $limits['time'] ) ? $limits['time'] : $wgMaxShellTime );
2349 if ( isset( $limits['walltime'] ) ) {
2350 $wallTime = intval( $limits['walltime'] );
2351 } elseif ( isset( $limits['time'] ) ) {
2352 $wallTime = $time;
2353 } else {
2354 $wallTime = intval( $wgMaxShellWallClockTime );
2355 }
2356 $mem = intval( isset( $limits['memory'] ) ? $limits['memory'] : $wgMaxShellMemory );
2357 $filesize = intval( isset( $limits['filesize'] ) ? $limits['filesize'] : $wgMaxShellFileSize );
2358
2359 if ( $time > 0 || $mem > 0 || $filesize > 0 || $wallTime > 0 ) {
2360 $cmd = '/bin/bash ' . escapeshellarg( "$IP/includes/limit.sh" ) . ' ' .
2361 escapeshellarg( $cmd ) . ' ' .
2362 escapeshellarg(
2363 "MW_INCLUDE_STDERR=" . ( $includeStderr ? '1' : '' ) . ';' .
2364 "MW_CPU_LIMIT=$time; " .
2365 'MW_CGROUP=' . escapeshellarg( $wgShellCgroup ) . '; ' .
2366 "MW_MEM_LIMIT=$mem; " .
2367 "MW_FILE_SIZE_LIMIT=$filesize; " .
2368 "MW_WALL_CLOCK_LIMIT=$wallTime; " .
2369 "MW_USE_LOG_PIPE=yes"
2370 );
2371 $useLogPipe = true;
2372 } elseif ( $includeStderr ) {
2373 $cmd .= ' 2>&1';
2374 }
2375 } elseif ( $includeStderr ) {
2376 $cmd .= ' 2>&1';
2377 }
2378 wfDebug( "wfShellExec: $cmd\n" );
2379
2380 // Don't try to execute commands that exceed Linux's MAX_ARG_STRLEN.
2381 // Other platforms may be more accomodating, but we don't want to be
2382 // accomodating, because very long commands probably include user
2383 // input. See T129506.
2384 if ( strlen( $cmd ) > SHELL_MAX_ARG_STRLEN ) {
2385 throw new Exception( __METHOD__ .
2386 '(): total length of $cmd must not exceed SHELL_MAX_ARG_STRLEN' );
2387 }
2388
2389 $desc = [
2390 0 => [ 'file', 'php://stdin', 'r' ],
2391 1 => [ 'pipe', 'w' ],
2392 2 => [ 'file', 'php://stderr', 'w' ] ];
2393 if ( $useLogPipe ) {
2394 $desc[3] = [ 'pipe', 'w' ];
2395 }
2396 $pipes = null;
2397 $scoped = Profiler::instance()->scopedProfileIn( __FUNCTION__ . '-' . $profileMethod );
2398 $proc = proc_open( $cmd, $desc, $pipes );
2399 if ( !$proc ) {
2400 wfDebugLog( 'exec', "proc_open() failed: $cmd" );
2401 $retval = -1;
2402 return '';
2403 }
2404 $outBuffer = $logBuffer = '';
2405 $emptyArray = [];
2406 $status = false;
2407 $logMsg = false;
2408
2409 /* According to the documentation, it is possible for stream_select()
2410 * to fail due to EINTR. I haven't managed to induce this in testing
2411 * despite sending various signals. If it did happen, the error
2412 * message would take the form:
2413 *
2414 * stream_select(): unable to select [4]: Interrupted system call (max_fd=5)
2415 *
2416 * where [4] is the value of the macro EINTR and "Interrupted system
2417 * call" is string which according to the Linux manual is "possibly"
2418 * localised according to LC_MESSAGES.
2419 */
2420 $eintr = defined( 'SOCKET_EINTR' ) ? SOCKET_EINTR : 4;
2421 $eintrMessage = "stream_select(): unable to select [$eintr]";
2422
2423 $running = true;
2424 $timeout = null;
2425 $numReadyPipes = 0;
2426
2427 while ( $running === true || $numReadyPipes !== 0 ) {
2428 if ( $running ) {
2429 $status = proc_get_status( $proc );
2430 // If the process has terminated, switch to nonblocking selects
2431 // for getting any data still waiting to be read.
2432 if ( !$status['running'] ) {
2433 $running = false;
2434 $timeout = 0;
2435 }
2436 }
2437
2438 $readyPipes = $pipes;
2439
2440 // Clear last error
2441 // @codingStandardsIgnoreStart Generic.PHP.NoSilencedErrors.Discouraged
2442 @trigger_error( '' );
2443 $numReadyPipes = @stream_select( $readyPipes, $emptyArray, $emptyArray, $timeout );
2444 if ( $numReadyPipes === false ) {
2445 // @codingStandardsIgnoreEnd
2446 $error = error_get_last();
2447 if ( strncmp( $error['message'], $eintrMessage, strlen( $eintrMessage ) ) == 0 ) {
2448 continue;
2449 } else {
2450 trigger_error( $error['message'], E_USER_WARNING );
2451 $logMsg = $error['message'];
2452 break;
2453 }
2454 }
2455 foreach ( $readyPipes as $fd => $pipe ) {
2456 $block = fread( $pipe, 65536 );
2457 if ( $block === '' ) {
2458 // End of file
2459 fclose( $pipes[$fd] );
2460 unset( $pipes[$fd] );
2461 if ( !$pipes ) {
2462 break 2;
2463 }
2464 } elseif ( $block === false ) {
2465 // Read error
2466 $logMsg = "Error reading from pipe";
2467 break 2;
2468 } elseif ( $fd == 1 ) {
2469 // From stdout
2470 $outBuffer .= $block;
2471 } elseif ( $fd == 3 ) {
2472 // From log FD
2473 $logBuffer .= $block;
2474 if ( strpos( $block, "\n" ) !== false ) {
2475 $lines = explode( "\n", $logBuffer );
2476 $logBuffer = array_pop( $lines );
2477 foreach ( $lines as $line ) {
2478 wfDebugLog( 'exec', $line );
2479 }
2480 }
2481 }
2482 }
2483 }
2484
2485 foreach ( $pipes as $pipe ) {
2486 fclose( $pipe );
2487 }
2488
2489 // Use the status previously collected if possible, since proc_get_status()
2490 // just calls waitpid() which will not return anything useful the second time.
2491 if ( $running ) {
2492 $status = proc_get_status( $proc );
2493 }
2494
2495 if ( $logMsg !== false ) {
2496 // Read/select error
2497 $retval = -1;
2498 proc_close( $proc );
2499 } elseif ( $status['signaled'] ) {
2500 $logMsg = "Exited with signal {$status['termsig']}";
2501 $retval = 128 + $status['termsig'];
2502 proc_close( $proc );
2503 } else {
2504 if ( $status['running'] ) {
2505 $retval = proc_close( $proc );
2506 } else {
2507 $retval = $status['exitcode'];
2508 proc_close( $proc );
2509 }
2510 if ( $retval == 127 ) {
2511 $logMsg = "Possibly missing executable file";
2512 } elseif ( $retval >= 129 && $retval <= 192 ) {
2513 $logMsg = "Probably exited with signal " . ( $retval - 128 );
2514 }
2515 }
2516
2517 if ( $logMsg !== false ) {
2518 wfDebugLog( 'exec', "$logMsg: $cmd" );
2519 }
2520
2521 return $outBuffer;
2522 }
2523
2524 /**
2525 * Execute a shell command, returning both stdout and stderr. Convenience
2526 * function, as all the arguments to wfShellExec can become unwieldy.
2527 *
2528 * @note This also includes errors from limit.sh, e.g. if $wgMaxShellFileSize is exceeded.
2529 * @param string|string[] $cmd If string, a properly shell-escaped command line,
2530 * or an array of unescaped arguments, in which case each value will be escaped
2531 * Example: [ 'convert', '-font', 'font name' ] would produce "'convert' '-font' 'font name'"
2532 * @param null|mixed &$retval Optional, will receive the program's exit code.
2533 * (non-zero is usually failure)
2534 * @param array $environ Optional environment variables which should be
2535 * added to the executed command environment.
2536 * @param array $limits Optional array with limits(filesize, memory, time, walltime)
2537 * this overwrites the global wgMaxShell* limits.
2538 * @return string Collected stdout and stderr as a string
2539 */
2540 function wfShellExecWithStderr( $cmd, &$retval = null, $environ = [], $limits = [] ) {
2541 return wfShellExec( $cmd, $retval, $environ, $limits,
2542 [ 'duplicateStderr' => true, 'profileMethod' => wfGetCaller() ] );
2543 }
2544
2545 /**
2546 * Formerly set the locale for locale-sensitive operations
2547 *
2548 * This is now done in Setup.php.
2549 *
2550 * @deprecated since 1.30, no longer needed
2551 * @see $wgShellLocale
2552 */
2553 function wfInitShellLocale() {
2554 }
2555
2556 /**
2557 * Generate a shell-escaped command line string to run a MediaWiki cli script.
2558 * Note that $parameters should be a flat array and an option with an argument
2559 * should consist of two consecutive items in the array (do not use "--option value").
2560 *
2561 * @param string $script MediaWiki cli script path
2562 * @param array $parameters Arguments and options to the script
2563 * @param array $options Associative array of options:
2564 * 'php': The path to the php executable
2565 * 'wrapper': Path to a PHP wrapper to handle the maintenance script
2566 * @return string
2567 */
2568 function wfShellWikiCmd( $script, array $parameters = [], array $options = [] ) {
2569 global $wgPhpCli;
2570 // Give site config file a chance to run the script in a wrapper.
2571 // The caller may likely want to call wfBasename() on $script.
2572 Hooks::run( 'wfShellWikiCmd', [ &$script, &$parameters, &$options ] );
2573 $cmd = isset( $options['php'] ) ? [ $options['php'] ] : [ $wgPhpCli ];
2574 if ( isset( $options['wrapper'] ) ) {
2575 $cmd[] = $options['wrapper'];
2576 }
2577 $cmd[] = $script;
2578 // Escape each parameter for shell
2579 return wfEscapeShellArg( array_merge( $cmd, $parameters ) );
2580 }
2581
2582 /**
2583 * wfMerge attempts to merge differences between three texts.
2584 * Returns true for a clean merge and false for failure or a conflict.
2585 *
2586 * @param string $old
2587 * @param string $mine
2588 * @param string $yours
2589 * @param string $result
2590 * @return bool
2591 */
2592 function wfMerge( $old, $mine, $yours, &$result ) {
2593 global $wgDiff3;
2594
2595 # This check may also protect against code injection in
2596 # case of broken installations.
2597 MediaWiki\suppressWarnings();
2598 $haveDiff3 = $wgDiff3 && file_exists( $wgDiff3 );
2599 MediaWiki\restoreWarnings();
2600
2601 if ( !$haveDiff3 ) {
2602 wfDebug( "diff3 not found\n" );
2603 return false;
2604 }
2605
2606 # Make temporary files
2607 $td = wfTempDir();
2608 $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
2609 $mytextFile = fopen( $mytextName = tempnam( $td, 'merge-mine-' ), 'w' );
2610 $yourtextFile = fopen( $yourtextName = tempnam( $td, 'merge-your-' ), 'w' );
2611
2612 # NOTE: diff3 issues a warning to stderr if any of the files does not end with
2613 # a newline character. To avoid this, we normalize the trailing whitespace before
2614 # creating the diff.
2615
2616 fwrite( $oldtextFile, rtrim( $old ) . "\n" );
2617 fclose( $oldtextFile );
2618 fwrite( $mytextFile, rtrim( $mine ) . "\n" );
2619 fclose( $mytextFile );
2620 fwrite( $yourtextFile, rtrim( $yours ) . "\n" );
2621 fclose( $yourtextFile );
2622
2623 # Check for a conflict
2624 $cmd = wfEscapeShellArg( $wgDiff3, '-a', '--overlap-only', $mytextName,
2625 $oldtextName, $yourtextName );
2626 $handle = popen( $cmd, 'r' );
2627
2628 if ( fgets( $handle, 1024 ) ) {
2629 $conflict = true;
2630 } else {
2631 $conflict = false;
2632 }
2633 pclose( $handle );
2634
2635 # Merge differences
2636 $cmd = wfEscapeShellArg( $wgDiff3, '-a', '-e', '--merge', $mytextName,
2637 $oldtextName, $yourtextName );
2638 $handle = popen( $cmd, 'r' );
2639 $result = '';
2640 do {
2641 $data = fread( $handle, 8192 );
2642 if ( strlen( $data ) == 0 ) {
2643 break;
2644 }
2645 $result .= $data;
2646 } while ( true );
2647 pclose( $handle );
2648 unlink( $mytextName );
2649 unlink( $oldtextName );
2650 unlink( $yourtextName );
2651
2652 if ( $result === '' && $old !== '' && !$conflict ) {
2653 wfDebug( "Unexpected null result from diff3. Command: $cmd\n" );
2654 $conflict = true;
2655 }
2656 return !$conflict;
2657 }
2658
2659 /**
2660 * Returns unified plain-text diff of two texts.
2661 * "Useful" for machine processing of diffs.
2662 *
2663 * @deprecated since 1.25, use DiffEngine/UnifiedDiffFormatter directly
2664 *
2665 * @param string $before The text before the changes.
2666 * @param string $after The text after the changes.
2667 * @param string $params Command-line options for the diff command.
2668 * @return string Unified diff of $before and $after
2669 */
2670 function wfDiff( $before, $after, $params = '-u' ) {
2671 if ( $before == $after ) {
2672 return '';
2673 }
2674
2675 global $wgDiff;
2676 MediaWiki\suppressWarnings();
2677 $haveDiff = $wgDiff && file_exists( $wgDiff );
2678 MediaWiki\restoreWarnings();
2679
2680 # This check may also protect against code injection in
2681 # case of broken installations.
2682 if ( !$haveDiff ) {
2683 wfDebug( "diff executable not found\n" );
2684 $diffs = new Diff( explode( "\n", $before ), explode( "\n", $after ) );
2685 $format = new UnifiedDiffFormatter();
2686 return $format->format( $diffs );
2687 }
2688
2689 # Make temporary files
2690 $td = wfTempDir();
2691 $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
2692 $newtextFile = fopen( $newtextName = tempnam( $td, 'merge-your-' ), 'w' );
2693
2694 fwrite( $oldtextFile, $before );
2695 fclose( $oldtextFile );
2696 fwrite( $newtextFile, $after );
2697 fclose( $newtextFile );
2698
2699 // Get the diff of the two files
2700 $cmd = "$wgDiff " . $params . ' ' . wfEscapeShellArg( $oldtextName, $newtextName );
2701
2702 $h = popen( $cmd, 'r' );
2703 if ( !$h ) {
2704 unlink( $oldtextName );
2705 unlink( $newtextName );
2706 throw new Exception( __METHOD__ . '(): popen() failed' );
2707 }
2708
2709 $diff = '';
2710
2711 do {
2712 $data = fread( $h, 8192 );
2713 if ( strlen( $data ) == 0 ) {
2714 break;
2715 }
2716 $diff .= $data;
2717 } while ( true );
2718
2719 // Clean up
2720 pclose( $h );
2721 unlink( $oldtextName );
2722 unlink( $newtextName );
2723
2724 // Kill the --- and +++ lines. They're not useful.
2725 $diff_lines = explode( "\n", $diff );
2726 if ( isset( $diff_lines[0] ) && strpos( $diff_lines[0], '---' ) === 0 ) {
2727 unset( $diff_lines[0] );
2728 }
2729 if ( isset( $diff_lines[1] ) && strpos( $diff_lines[1], '+++' ) === 0 ) {
2730 unset( $diff_lines[1] );
2731 }
2732
2733 $diff = implode( "\n", $diff_lines );
2734
2735 return $diff;
2736 }
2737
2738 /**
2739 * This function works like "use VERSION" in Perl, the program will die with a
2740 * backtrace if the current version of PHP is less than the version provided
2741 *
2742 * This is useful for extensions which due to their nature are not kept in sync
2743 * with releases, and might depend on other versions of PHP than the main code
2744 *
2745 * Note: PHP might die due to parsing errors in some cases before it ever
2746 * manages to call this function, such is life
2747 *
2748 * @see perldoc -f use
2749 *
2750 * @param string|int|float $req_ver The version to check, can be a string, an integer, or a float
2751 * @throws MWException
2752 */
2753 function wfUsePHP( $req_ver ) {
2754 $php_ver = PHP_VERSION;
2755
2756 if ( version_compare( $php_ver, (string)$req_ver, '<' ) ) {
2757 throw new MWException( "PHP $req_ver required--this is only $php_ver" );
2758 }
2759 }
2760
2761 /**
2762 * This function works like "use VERSION" in Perl except it checks the version
2763 * of MediaWiki, the program will die with a backtrace if the current version
2764 * of MediaWiki is less than the version provided.
2765 *
2766 * This is useful for extensions which due to their nature are not kept in sync
2767 * with releases
2768 *
2769 * Note: Due to the behavior of PHP's version_compare() which is used in this
2770 * function, if you want to allow the 'wmf' development versions add a 'c' (or
2771 * any single letter other than 'a', 'b' or 'p') as a post-fix to your
2772 * targeted version number. For example if you wanted to allow any variation
2773 * of 1.22 use `wfUseMW( '1.22c' )`. Using an 'a' or 'b' instead of 'c' will
2774 * not result in the same comparison due to the internal logic of
2775 * version_compare().
2776 *
2777 * @see perldoc -f use
2778 *
2779 * @deprecated since 1.26, use the "requires' property of extension.json
2780 * @param string|int|float $req_ver The version to check, can be a string, an integer, or a float
2781 * @throws MWException
2782 */
2783 function wfUseMW( $req_ver ) {
2784 global $wgVersion;
2785
2786 if ( version_compare( $wgVersion, (string)$req_ver, '<' ) ) {
2787 throw new MWException( "MediaWiki $req_ver required--this is only $wgVersion" );
2788 }
2789 }
2790
2791 /**
2792 * Return the final portion of a pathname.
2793 * Reimplemented because PHP5's "basename()" is buggy with multibyte text.
2794 * https://bugs.php.net/bug.php?id=33898
2795 *
2796 * PHP's basename() only considers '\' a pathchar on Windows and Netware.
2797 * We'll consider it so always, as we don't want '\s' in our Unix paths either.
2798 *
2799 * @param string $path
2800 * @param string $suffix String to remove if present
2801 * @return string
2802 */
2803 function wfBaseName( $path, $suffix = '' ) {
2804 if ( $suffix == '' ) {
2805 $encSuffix = '';
2806 } else {
2807 $encSuffix = '(?:' . preg_quote( $suffix, '#' ) . ')?';
2808 }
2809
2810 $matches = [];
2811 if ( preg_match( "#([^/\\\\]*?){$encSuffix}[/\\\\]*$#", $path, $matches ) ) {
2812 return $matches[1];
2813 } else {
2814 return '';
2815 }
2816 }
2817
2818 /**
2819 * Generate a relative path name to the given file.
2820 * May explode on non-matching case-insensitive paths,
2821 * funky symlinks, etc.
2822 *
2823 * @param string $path Absolute destination path including target filename
2824 * @param string $from Absolute source path, directory only
2825 * @return string
2826 */
2827 function wfRelativePath( $path, $from ) {
2828 // Normalize mixed input on Windows...
2829 $path = str_replace( '/', DIRECTORY_SEPARATOR, $path );
2830 $from = str_replace( '/', DIRECTORY_SEPARATOR, $from );
2831
2832 // Trim trailing slashes -- fix for drive root
2833 $path = rtrim( $path, DIRECTORY_SEPARATOR );
2834 $from = rtrim( $from, DIRECTORY_SEPARATOR );
2835
2836 $pieces = explode( DIRECTORY_SEPARATOR, dirname( $path ) );
2837 $against = explode( DIRECTORY_SEPARATOR, $from );
2838
2839 if ( $pieces[0] !== $against[0] ) {
2840 // Non-matching Windows drive letters?
2841 // Return a full path.
2842 return $path;
2843 }
2844
2845 // Trim off common prefix
2846 while ( count( $pieces ) && count( $against )
2847 && $pieces[0] == $against[0] ) {
2848 array_shift( $pieces );
2849 array_shift( $against );
2850 }
2851
2852 // relative dots to bump us to the parent
2853 while ( count( $against ) ) {
2854 array_unshift( $pieces, '..' );
2855 array_shift( $against );
2856 }
2857
2858 array_push( $pieces, wfBaseName( $path ) );
2859
2860 return implode( DIRECTORY_SEPARATOR, $pieces );
2861 }
2862
2863 /**
2864 * Convert an arbitrarily-long digit string from one numeric base
2865 * to another, optionally zero-padding to a minimum column width.
2866 *
2867 * Supports base 2 through 36; digit values 10-36 are represented
2868 * as lowercase letters a-z. Input is case-insensitive.
2869 *
2870 * @deprecated since 1.27 Use Wikimedia\base_convert() directly
2871 *
2872 * @param string $input Input number
2873 * @param int $sourceBase Base of the input number
2874 * @param int $destBase Desired base of the output
2875 * @param int $pad Minimum number of digits in the output (pad with zeroes)
2876 * @param bool $lowercase Whether to output in lowercase or uppercase
2877 * @param string $engine Either "gmp", "bcmath", or "php"
2878 * @return string|bool The output number as a string, or false on error
2879 */
2880 function wfBaseConvert( $input, $sourceBase, $destBase, $pad = 1,
2881 $lowercase = true, $engine = 'auto'
2882 ) {
2883 return Wikimedia\base_convert( $input, $sourceBase, $destBase, $pad, $lowercase, $engine );
2884 }
2885
2886 /**
2887 * @deprecated since 1.27, PHP's session generation isn't used with
2888 * MediaWiki\Session\SessionManager
2889 */
2890 function wfFixSessionID() {
2891 wfDeprecated( __FUNCTION__, '1.27' );
2892 }
2893
2894 /**
2895 * Reset the session id
2896 *
2897 * @deprecated since 1.27, use MediaWiki\Session\SessionManager instead
2898 * @since 1.22
2899 */
2900 function wfResetSessionID() {
2901 wfDeprecated( __FUNCTION__, '1.27' );
2902 $session = SessionManager::getGlobalSession();
2903 $delay = $session->delaySave();
2904
2905 $session->resetId();
2906
2907 // Make sure a session is started, since that's what the old
2908 // wfResetSessionID() did.
2909 if ( session_id() !== $session->getId() ) {
2910 wfSetupSession( $session->getId() );
2911 }
2912
2913 ScopedCallback::consume( $delay );
2914 }
2915
2916 /**
2917 * Initialise php session
2918 *
2919 * @deprecated since 1.27, use MediaWiki\Session\SessionManager instead.
2920 * Generally, "using" SessionManager will be calling ->getSessionById() or
2921 * ::getGlobalSession() (depending on whether you were passing $sessionId
2922 * here), then calling $session->persist().
2923 * @param bool|string $sessionId
2924 */
2925 function wfSetupSession( $sessionId = false ) {
2926 wfDeprecated( __FUNCTION__, '1.27' );
2927
2928 if ( $sessionId ) {
2929 session_id( $sessionId );
2930 }
2931
2932 $session = SessionManager::getGlobalSession();
2933 $session->persist();
2934
2935 if ( session_id() !== $session->getId() ) {
2936 session_id( $session->getId() );
2937 }
2938 MediaWiki\quietCall( 'session_start' );
2939 }
2940
2941 /**
2942 * Get an object from the precompiled serialized directory
2943 *
2944 * @param string $name
2945 * @return mixed The variable on success, false on failure
2946 */
2947 function wfGetPrecompiledData( $name ) {
2948 global $IP;
2949
2950 $file = "$IP/serialized/$name";
2951 if ( file_exists( $file ) ) {
2952 $blob = file_get_contents( $file );
2953 if ( $blob ) {
2954 return unserialize( $blob );
2955 }
2956 }
2957 return false;
2958 }
2959
2960 /**
2961 * Make a cache key for the local wiki.
2962 *
2963 * @deprecated since 1.30 Call makeKey on a BagOStuff instance
2964 * @param string $args,...
2965 * @return string
2966 */
2967 function wfMemcKey( /*...*/ ) {
2968 return call_user_func_array(
2969 [ ObjectCache::getLocalClusterInstance(), 'makeKey' ],
2970 func_get_args()
2971 );
2972 }
2973
2974 /**
2975 * Make a cache key for a foreign DB.
2976 *
2977 * Must match what wfMemcKey() would produce in context of the foreign wiki.
2978 *
2979 * @param string $db
2980 * @param string $prefix
2981 * @param string $args,...
2982 * @return string
2983 */
2984 function wfForeignMemcKey( $db, $prefix /*...*/ ) {
2985 $args = array_slice( func_get_args(), 2 );
2986 $keyspace = $prefix ? "$db-$prefix" : $db;
2987 return call_user_func_array(
2988 [ ObjectCache::getLocalClusterInstance(), 'makeKeyInternal' ],
2989 [ $keyspace, $args ]
2990 );
2991 }
2992
2993 /**
2994 * Make a cache key with database-agnostic prefix.
2995 *
2996 * Doesn't have a wiki-specific namespace. Uses a generic 'global' prefix
2997 * instead. Must have a prefix as otherwise keys that use a database name
2998 * in the first segment will clash with wfMemcKey/wfForeignMemcKey.
2999 *
3000 * @deprecated since 1.30 Call makeGlobalKey on a BagOStuff instance
3001 * @since 1.26
3002 * @param string $args,...
3003 * @return string
3004 */
3005 function wfGlobalCacheKey( /*...*/ ) {
3006 return call_user_func_array(
3007 [ ObjectCache::getLocalClusterInstance(), 'makeGlobalKey' ],
3008 func_get_args()
3009 );
3010 }
3011
3012 /**
3013 * Get an ASCII string identifying this wiki
3014 * This is used as a prefix in memcached keys
3015 *
3016 * @return string
3017 */
3018 function wfWikiID() {
3019 global $wgDBprefix, $wgDBname;
3020 if ( $wgDBprefix ) {
3021 return "$wgDBname-$wgDBprefix";
3022 } else {
3023 return $wgDBname;
3024 }
3025 }
3026
3027 /**
3028 * Split a wiki ID into DB name and table prefix
3029 *
3030 * @param string $wiki
3031 *
3032 * @return array
3033 */
3034 function wfSplitWikiID( $wiki ) {
3035 $bits = explode( '-', $wiki, 2 );
3036 if ( count( $bits ) < 2 ) {
3037 $bits[] = '';
3038 }
3039 return $bits;
3040 }
3041
3042 /**
3043 * Get a Database object.
3044 *
3045 * @param int $db Index of the connection to get. May be DB_MASTER for the
3046 * master (for write queries), DB_REPLICA for potentially lagged read
3047 * queries, or an integer >= 0 for a particular server.
3048 *
3049 * @param string|string[] $groups Query groups. An array of group names that this query
3050 * belongs to. May contain a single string if the query is only
3051 * in one group.
3052 *
3053 * @param string|bool $wiki The wiki ID, or false for the current wiki
3054 *
3055 * Note: multiple calls to wfGetDB(DB_REPLICA) during the course of one request
3056 * will always return the same object, unless the underlying connection or load
3057 * balancer is manually destroyed.
3058 *
3059 * Note 2: use $this->getDB() in maintenance scripts that may be invoked by
3060 * updater to ensure that a proper database is being updated.
3061 *
3062 * @todo Replace calls to wfGetDB with calls to LoadBalancer::getConnection()
3063 * on an injected instance of LoadBalancer.
3064 *
3065 * @return \Wikimedia\Rdbms\Database
3066 */
3067 function wfGetDB( $db, $groups = [], $wiki = false ) {
3068 return wfGetLB( $wiki )->getConnection( $db, $groups, $wiki );
3069 }
3070
3071 /**
3072 * Get a load balancer object.
3073 *
3074 * @deprecated since 1.27, use MediaWikiServices::getDBLoadBalancer()
3075 * or MediaWikiServices::getDBLoadBalancerFactory() instead.
3076 *
3077 * @param string|bool $wiki Wiki ID, or false for the current wiki
3078 * @return \Wikimedia\Rdbms\LoadBalancer
3079 */
3080 function wfGetLB( $wiki = false ) {
3081 if ( $wiki === false ) {
3082 return MediaWikiServices::getInstance()->getDBLoadBalancer();
3083 } else {
3084 $factory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
3085 return $factory->getMainLB( $wiki );
3086 }
3087 }
3088
3089 /**
3090 * Get the load balancer factory object
3091 *
3092 * @deprecated since 1.27, use MediaWikiServices::getDBLoadBalancerFactory() instead.
3093 *
3094 * @return \Wikimedia\Rdbms\LBFactory
3095 */
3096 function wfGetLBFactory() {
3097 return MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
3098 }
3099
3100 /**
3101 * Find a file.
3102 * Shortcut for RepoGroup::singleton()->findFile()
3103 *
3104 * @param string $title String or Title object
3105 * @param array $options Associative array of options (see RepoGroup::findFile)
3106 * @return File|bool File, or false if the file does not exist
3107 */
3108 function wfFindFile( $title, $options = [] ) {
3109 return RepoGroup::singleton()->findFile( $title, $options );
3110 }
3111
3112 /**
3113 * Get an object referring to a locally registered file.
3114 * Returns a valid placeholder object if the file does not exist.
3115 *
3116 * @param Title|string $title
3117 * @return LocalFile|null A File, or null if passed an invalid Title
3118 */
3119 function wfLocalFile( $title ) {
3120 return RepoGroup::singleton()->getLocalRepo()->newFile( $title );
3121 }
3122
3123 /**
3124 * Should low-performance queries be disabled?
3125 *
3126 * @return bool
3127 * @codeCoverageIgnore
3128 */
3129 function wfQueriesMustScale() {
3130 global $wgMiserMode;
3131 return $wgMiserMode
3132 || ( SiteStats::pages() > 100000
3133 && SiteStats::edits() > 1000000
3134 && SiteStats::users() > 10000 );
3135 }
3136
3137 /**
3138 * Get the path to a specified script file, respecting file
3139 * extensions; this is a wrapper around $wgScriptPath etc.
3140 * except for 'index' and 'load' which use $wgScript/$wgLoadScript
3141 *
3142 * @param string $script Script filename, sans extension
3143 * @return string
3144 */
3145 function wfScript( $script = 'index' ) {
3146 global $wgScriptPath, $wgScript, $wgLoadScript;
3147 if ( $script === 'index' ) {
3148 return $wgScript;
3149 } elseif ( $script === 'load' ) {
3150 return $wgLoadScript;
3151 } else {
3152 return "{$wgScriptPath}/{$script}.php";
3153 }
3154 }
3155
3156 /**
3157 * Get the script URL.
3158 *
3159 * @return string Script URL
3160 */
3161 function wfGetScriptUrl() {
3162 if ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
3163 /* as it was called, minus the query string.
3164 *
3165 * Some sites use Apache rewrite rules to handle subdomains,
3166 * and have PHP set up in a weird way that causes PHP_SELF
3167 * to contain the rewritten URL instead of the one that the
3168 * outside world sees.
3169 *
3170 * If in this mode, use SCRIPT_URL instead, which mod_rewrite
3171 * provides containing the "before" URL.
3172 */
3173 return $_SERVER['SCRIPT_NAME'];
3174 } else {
3175 return $_SERVER['URL'];
3176 }
3177 }
3178
3179 /**
3180 * Convenience function converts boolean values into "true"
3181 * or "false" (string) values
3182 *
3183 * @param bool $value
3184 * @return string
3185 */
3186 function wfBoolToStr( $value ) {
3187 return $value ? 'true' : 'false';
3188 }
3189
3190 /**
3191 * Get a platform-independent path to the null file, e.g. /dev/null
3192 *
3193 * @return string
3194 */
3195 function wfGetNull() {
3196 return wfIsWindows() ? 'NUL' : '/dev/null';
3197 }
3198
3199 /**
3200 * Waits for the replica DBs to catch up to the master position
3201 *
3202 * Use this when updating very large numbers of rows, as in maintenance scripts,
3203 * to avoid causing too much lag. Of course, this is a no-op if there are no replica DBs.
3204 *
3205 * By default this waits on the main DB cluster of the current wiki.
3206 * If $cluster is set to "*" it will wait on all DB clusters, including
3207 * external ones. If the lag being waiting on is caused by the code that
3208 * does this check, it makes since to use $ifWritesSince, particularly if
3209 * cluster is "*", to avoid excess overhead.
3210 *
3211 * Never call this function after a big DB write that is still in a transaction.
3212 * This only makes sense after the possible lag inducing changes were committed.
3213 *
3214 * @param float|null $ifWritesSince Only wait if writes were done since this UNIX timestamp
3215 * @param string|bool $wiki Wiki identifier accepted by wfGetLB
3216 * @param string|bool $cluster Cluster name accepted by LBFactory. Default: false.
3217 * @param int|null $timeout Max wait time. Default: 1 day (cli), ~10 seconds (web)
3218 * @return bool Success (able to connect and no timeouts reached)
3219 * @deprecated since 1.27 Use LBFactory::waitForReplication
3220 */
3221 function wfWaitForSlaves(
3222 $ifWritesSince = null, $wiki = false, $cluster = false, $timeout = null
3223 ) {
3224 if ( $timeout === null ) {
3225 $timeout = ( PHP_SAPI === 'cli' ) ? 86400 : 10;
3226 }
3227
3228 if ( $cluster === '*' ) {
3229 $cluster = false;
3230 $wiki = false;
3231 } elseif ( $wiki === false ) {
3232 $wiki = wfWikiID();
3233 }
3234
3235 try {
3236 wfGetLBFactory()->waitForReplication( [
3237 'wiki' => $wiki,
3238 'cluster' => $cluster,
3239 'timeout' => $timeout,
3240 // B/C: first argument used to be "max seconds of lag"; ignore such values
3241 'ifWritesSince' => ( $ifWritesSince > 1e9 ) ? $ifWritesSince : null
3242 ] );
3243 } catch ( DBReplicationWaitError $e ) {
3244 return false;
3245 }
3246
3247 return true;
3248 }
3249
3250 /**
3251 * Count down from $seconds to zero on the terminal, with a one-second pause
3252 * between showing each number. For use in command-line scripts.
3253 *
3254 * @codeCoverageIgnore
3255 * @param int $seconds
3256 */
3257 function wfCountDown( $seconds ) {
3258 for ( $i = $seconds; $i >= 0; $i-- ) {
3259 if ( $i != $seconds ) {
3260 echo str_repeat( "\x08", strlen( $i + 1 ) );
3261 }
3262 echo $i;
3263 flush();
3264 if ( $i ) {
3265 sleep( 1 );
3266 }
3267 }
3268 echo "\n";
3269 }
3270
3271 /**
3272 * Replace all invalid characters with '-'.
3273 * Additional characters can be defined in $wgIllegalFileChars (see T22489).
3274 * By default, $wgIllegalFileChars includes ':', '/', '\'.
3275 *
3276 * @param string $name Filename to process
3277 * @return string
3278 */
3279 function wfStripIllegalFilenameChars( $name ) {
3280 global $wgIllegalFileChars;
3281 $illegalFileChars = $wgIllegalFileChars ? "|[" . $wgIllegalFileChars . "]" : '';
3282 $name = preg_replace(
3283 "/[^" . Title::legalChars() . "]" . $illegalFileChars . "/",
3284 '-',
3285 $name
3286 );
3287 // $wgIllegalFileChars may not include '/' and '\', so we still need to do this
3288 $name = wfBaseName( $name );
3289 return $name;
3290 }
3291
3292 /**
3293 * Set PHP's memory limit to the larger of php.ini or $wgMemoryLimit
3294 *
3295 * @return int Resulting value of the memory limit.
3296 */
3297 function wfMemoryLimit() {
3298 global $wgMemoryLimit;
3299 $memlimit = wfShorthandToInteger( ini_get( 'memory_limit' ) );
3300 if ( $memlimit != -1 ) {
3301 $conflimit = wfShorthandToInteger( $wgMemoryLimit );
3302 if ( $conflimit == -1 ) {
3303 wfDebug( "Removing PHP's memory limit\n" );
3304 MediaWiki\suppressWarnings();
3305 ini_set( 'memory_limit', $conflimit );
3306 MediaWiki\restoreWarnings();
3307 return $conflimit;
3308 } elseif ( $conflimit > $memlimit ) {
3309 wfDebug( "Raising PHP's memory limit to $conflimit bytes\n" );
3310 MediaWiki\suppressWarnings();
3311 ini_set( 'memory_limit', $conflimit );
3312 MediaWiki\restoreWarnings();
3313 return $conflimit;
3314 }
3315 }
3316 return $memlimit;
3317 }
3318
3319 /**
3320 * Set PHP's time limit to the larger of php.ini or $wgTransactionalTimeLimit
3321 *
3322 * @return int Prior time limit
3323 * @since 1.26
3324 */
3325 function wfTransactionalTimeLimit() {
3326 global $wgTransactionalTimeLimit;
3327
3328 $timeLimit = ini_get( 'max_execution_time' );
3329 // Note that CLI scripts use 0
3330 if ( $timeLimit > 0 && $wgTransactionalTimeLimit > $timeLimit ) {
3331 set_time_limit( $wgTransactionalTimeLimit );
3332 }
3333
3334 ignore_user_abort( true ); // ignore client disconnects
3335
3336 return $timeLimit;
3337 }
3338
3339 /**
3340 * Converts shorthand byte notation to integer form
3341 *
3342 * @param string $string
3343 * @param int $default Returned if $string is empty
3344 * @return int
3345 */
3346 function wfShorthandToInteger( $string = '', $default = -1 ) {
3347 $string = trim( $string );
3348 if ( $string === '' ) {
3349 return $default;
3350 }
3351 $last = $string[strlen( $string ) - 1];
3352 $val = intval( $string );
3353 switch ( $last ) {
3354 case 'g':
3355 case 'G':
3356 $val *= 1024;
3357 // break intentionally missing
3358 case 'm':
3359 case 'M':
3360 $val *= 1024;
3361 // break intentionally missing
3362 case 'k':
3363 case 'K':
3364 $val *= 1024;
3365 }
3366
3367 return $val;
3368 }
3369
3370 /**
3371 * Get the normalised IETF language tag
3372 * See unit test for examples.
3373 *
3374 * @param string $code The language code.
3375 * @return string The language code which complying with BCP 47 standards.
3376 */
3377 function wfBCP47( $code ) {
3378 $codeSegment = explode( '-', $code );
3379 $codeBCP = [];
3380 foreach ( $codeSegment as $segNo => $seg ) {
3381 // when previous segment is x, it is a private segment and should be lc
3382 if ( $segNo > 0 && strtolower( $codeSegment[( $segNo - 1 )] ) == 'x' ) {
3383 $codeBCP[$segNo] = strtolower( $seg );
3384 // ISO 3166 country code
3385 } elseif ( ( strlen( $seg ) == 2 ) && ( $segNo > 0 ) ) {
3386 $codeBCP[$segNo] = strtoupper( $seg );
3387 // ISO 15924 script code
3388 } elseif ( ( strlen( $seg ) == 4 ) && ( $segNo > 0 ) ) {
3389 $codeBCP[$segNo] = ucfirst( strtolower( $seg ) );
3390 // Use lowercase for other cases
3391 } else {
3392 $codeBCP[$segNo] = strtolower( $seg );
3393 }
3394 }
3395 $langCode = implode( '-', $codeBCP );
3396 return $langCode;
3397 }
3398
3399 /**
3400 * Get a specific cache object.
3401 *
3402 * @param int|string $cacheType A CACHE_* constants, or other key in $wgObjectCaches
3403 * @return BagOStuff
3404 */
3405 function wfGetCache( $cacheType ) {
3406 return ObjectCache::getInstance( $cacheType );
3407 }
3408
3409 /**
3410 * Get the main cache object
3411 *
3412 * @return BagOStuff
3413 */
3414 function wfGetMainCache() {
3415 global $wgMainCacheType;
3416 return ObjectCache::getInstance( $wgMainCacheType );
3417 }
3418
3419 /**
3420 * Get the cache object used by the message cache
3421 *
3422 * @return BagOStuff
3423 */
3424 function wfGetMessageCacheStorage() {
3425 global $wgMessageCacheType;
3426 return ObjectCache::getInstance( $wgMessageCacheType );
3427 }
3428
3429 /**
3430 * Get the cache object used by the parser cache
3431 *
3432 * @return BagOStuff
3433 */
3434 function wfGetParserCacheStorage() {
3435 global $wgParserCacheType;
3436 return ObjectCache::getInstance( $wgParserCacheType );
3437 }
3438
3439 /**
3440 * Call hook functions defined in $wgHooks
3441 *
3442 * @param string $event Event name
3443 * @param array $args Parameters passed to hook functions
3444 * @param string|null $deprecatedVersion Optionally mark hook as deprecated with version number
3445 *
3446 * @return bool True if no handler aborted the hook
3447 * @deprecated since 1.25 - use Hooks::run
3448 */
3449 function wfRunHooks( $event, array $args = [], $deprecatedVersion = null ) {
3450 return Hooks::run( $event, $args, $deprecatedVersion );
3451 }
3452
3453 /**
3454 * Wrapper around php's unpack.
3455 *
3456 * @param string $format The format string (See php's docs)
3457 * @param string $data A binary string of binary data
3458 * @param int|bool $length The minimum length of $data or false. This is to
3459 * prevent reading beyond the end of $data. false to disable the check.
3460 *
3461 * Also be careful when using this function to read unsigned 32 bit integer
3462 * because php might make it negative.
3463 *
3464 * @throws MWException If $data not long enough, or if unpack fails
3465 * @return array Associative array of the extracted data
3466 */
3467 function wfUnpack( $format, $data, $length = false ) {
3468 if ( $length !== false ) {
3469 $realLen = strlen( $data );
3470 if ( $realLen < $length ) {
3471 throw new MWException( "Tried to use wfUnpack on a "
3472 . "string of length $realLen, but needed one "
3473 . "of at least length $length."
3474 );
3475 }
3476 }
3477
3478 MediaWiki\suppressWarnings();
3479 $result = unpack( $format, $data );
3480 MediaWiki\restoreWarnings();
3481
3482 if ( $result === false ) {
3483 // If it cannot extract the packed data.
3484 throw new MWException( "unpack could not unpack binary data" );
3485 }
3486 return $result;
3487 }
3488
3489 /**
3490 * Determine if an image exists on the 'bad image list'.
3491 *
3492 * The format of MediaWiki:Bad_image_list is as follows:
3493 * * Only list items (lines starting with "*") are considered
3494 * * The first link on a line must be a link to a bad image
3495 * * Any subsequent links on the same line are considered to be exceptions,
3496 * i.e. articles where the image may occur inline.
3497 *
3498 * @param string $name The image name to check
3499 * @param Title|bool $contextTitle The page on which the image occurs, if known
3500 * @param string $blacklist Wikitext of a file blacklist
3501 * @return bool
3502 */
3503 function wfIsBadImage( $name, $contextTitle = false, $blacklist = null ) {
3504 # Handle redirects; callers almost always hit wfFindFile() anyway,
3505 # so just use that method because it has a fast process cache.
3506 $file = wfFindFile( $name ); // get the final name
3507 $name = $file ? $file->getTitle()->getDBkey() : $name;
3508
3509 # Run the extension hook
3510 $bad = false;
3511 if ( !Hooks::run( 'BadImage', [ $name, &$bad ] ) ) {
3512 return (bool)$bad;
3513 }
3514
3515 $cache = ObjectCache::getLocalServerInstance( 'hash' );
3516 $key = $cache->makeKey(
3517 'bad-image-list', ( $blacklist === null ) ? 'default' : md5( $blacklist )
3518 );
3519 $badImages = $cache->get( $key );
3520
3521 if ( $badImages === false ) { // cache miss
3522 if ( $blacklist === null ) {
3523 $blacklist = wfMessage( 'bad_image_list' )->inContentLanguage()->plain(); // site list
3524 }
3525 # Build the list now
3526 $badImages = [];
3527 $lines = explode( "\n", $blacklist );
3528 foreach ( $lines as $line ) {
3529 # List items only
3530 if ( substr( $line, 0, 1 ) !== '*' ) {
3531 continue;
3532 }
3533
3534 # Find all links
3535 $m = [];
3536 if ( !preg_match_all( '/\[\[:?(.*?)\]\]/', $line, $m ) ) {
3537 continue;
3538 }
3539
3540 $exceptions = [];
3541 $imageDBkey = false;
3542 foreach ( $m[1] as $i => $titleText ) {
3543 $title = Title::newFromText( $titleText );
3544 if ( !is_null( $title ) ) {
3545 if ( $i == 0 ) {
3546 $imageDBkey = $title->getDBkey();
3547 } else {
3548 $exceptions[$title->getPrefixedDBkey()] = true;
3549 }
3550 }
3551 }
3552
3553 if ( $imageDBkey !== false ) {
3554 $badImages[$imageDBkey] = $exceptions;
3555 }
3556 }
3557 $cache->set( $key, $badImages, 60 );
3558 }
3559
3560 $contextKey = $contextTitle ? $contextTitle->getPrefixedDBkey() : false;
3561 $bad = isset( $badImages[$name] ) && !isset( $badImages[$name][$contextKey] );
3562
3563 return $bad;
3564 }
3565
3566 /**
3567 * Determine whether the client at a given source IP is likely to be able to
3568 * access the wiki via HTTPS.
3569 *
3570 * @param string $ip The IPv4/6 address in the normal human-readable form
3571 * @return bool
3572 */
3573 function wfCanIPUseHTTPS( $ip ) {
3574 $canDo = true;
3575 Hooks::run( 'CanIPUseHTTPS', [ $ip, &$canDo ] );
3576 return !!$canDo;
3577 }
3578
3579 /**
3580 * Determine input string is represents as infinity
3581 *
3582 * @param string $str The string to determine
3583 * @return bool
3584 * @since 1.25
3585 */
3586 function wfIsInfinity( $str ) {
3587 $infinityValues = [ 'infinite', 'indefinite', 'infinity', 'never' ];
3588 return in_array( $str, $infinityValues );
3589 }
3590
3591 /**
3592 * Returns true if these thumbnail parameters match one that MediaWiki
3593 * requests from file description pages and/or parser output.
3594 *
3595 * $params is considered non-standard if they involve a non-standard
3596 * width or any non-default parameters aside from width and page number.
3597 * The number of possible files with standard parameters is far less than
3598 * that of all combinations; rate-limiting for them can thus be more generious.
3599 *
3600 * @param File $file
3601 * @param array $params
3602 * @return bool
3603 * @since 1.24 Moved from thumb.php to GlobalFunctions in 1.25
3604 */
3605 function wfThumbIsStandard( File $file, array $params ) {
3606 global $wgThumbLimits, $wgImageLimits, $wgResponsiveImages;
3607
3608 $multipliers = [ 1 ];
3609 if ( $wgResponsiveImages ) {
3610 // These available sizes are hardcoded currently elsewhere in MediaWiki.
3611 // @see Linker::processResponsiveImages
3612 $multipliers[] = 1.5;
3613 $multipliers[] = 2;
3614 }
3615
3616 $handler = $file->getHandler();
3617 if ( !$handler || !isset( $params['width'] ) ) {
3618 return false;
3619 }
3620
3621 $basicParams = [];
3622 if ( isset( $params['page'] ) ) {
3623 $basicParams['page'] = $params['page'];
3624 }
3625
3626 $thumbLimits = [];
3627 $imageLimits = [];
3628 // Expand limits to account for multipliers
3629 foreach ( $multipliers as $multiplier ) {
3630 $thumbLimits = array_merge( $thumbLimits, array_map(
3631 function ( $width ) use ( $multiplier ) {
3632 return round( $width * $multiplier );
3633 }, $wgThumbLimits )
3634 );
3635 $imageLimits = array_merge( $imageLimits, array_map(
3636 function ( $pair ) use ( $multiplier ) {
3637 return [
3638 round( $pair[0] * $multiplier ),
3639 round( $pair[1] * $multiplier ),
3640 ];
3641 }, $wgImageLimits )
3642 );
3643 }
3644
3645 // Check if the width matches one of $wgThumbLimits
3646 if ( in_array( $params['width'], $thumbLimits ) ) {
3647 $normalParams = $basicParams + [ 'width' => $params['width'] ];
3648 // Append any default values to the map (e.g. "lossy", "lossless", ...)
3649 $handler->normaliseParams( $file, $normalParams );
3650 } else {
3651 // If not, then check if the width matchs one of $wgImageLimits
3652 $match = false;
3653 foreach ( $imageLimits as $pair ) {
3654 $normalParams = $basicParams + [ 'width' => $pair[0], 'height' => $pair[1] ];
3655 // Decide whether the thumbnail should be scaled on width or height.
3656 // Also append any default values to the map (e.g. "lossy", "lossless", ...)
3657 $handler->normaliseParams( $file, $normalParams );
3658 // Check if this standard thumbnail size maps to the given width
3659 if ( $normalParams['width'] == $params['width'] ) {
3660 $match = true;
3661 break;
3662 }
3663 }
3664 if ( !$match ) {
3665 return false; // not standard for description pages
3666 }
3667 }
3668
3669 // Check that the given values for non-page, non-width, params are just defaults
3670 foreach ( $params as $key => $value ) {
3671 if ( !isset( $normalParams[$key] ) || $normalParams[$key] != $value ) {
3672 return false;
3673 }
3674 }
3675
3676 return true;
3677 }
3678
3679 /**
3680 * Merges two (possibly) 2 dimensional arrays into the target array ($baseArray).
3681 *
3682 * Values that exist in both values will be combined with += (all values of the array
3683 * of $newValues will be added to the values of the array of $baseArray, while values,
3684 * that exists in both, the value of $baseArray will be used).
3685 *
3686 * @param array $baseArray The array where you want to add the values of $newValues to
3687 * @param array $newValues An array with new values
3688 * @return array The combined array
3689 * @since 1.26
3690 */
3691 function wfArrayPlus2d( array $baseArray, array $newValues ) {
3692 // First merge items that are in both arrays
3693 foreach ( $baseArray as $name => &$groupVal ) {
3694 if ( isset( $newValues[$name] ) ) {
3695 $groupVal += $newValues[$name];
3696 }
3697 }
3698 // Now add items that didn't exist yet
3699 $baseArray += $newValues;
3700
3701 return $baseArray;
3702 }