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 // Hide compatibility functions from Doxygen
  28 /// @cond
  29
  30 /**
  31  * Compatibility functions
  32  *
  33  * We support PHP 5.3.2 and up.
  34  * Re-implementations of newer functions or functions in non-standard
  35  * PHP extensions may be included here.
  36  */
  37
  38 if ( !function_exists( 'iconv' ) ) {
  39         /**
  40          * @codeCoverageIgnore
  41          * @return string
  42          */
  43         function iconv( $from, $to, $string ) {
  44                 return Fallback::iconv( $from, $to, $string );
  45         }
  46 }
  47
  48 if ( !function_exists( 'mb_substr' ) ) {
  49         /**
  50          * @codeCoverageIgnore
  51          * @return string
  52          */
  53         function mb_substr( $str, $start, $count = 'end' ) {
  54                 return Fallback::mb_substr( $str, $start, $count );
  55         }
  56
  57         /**
  58          * @codeCoverageIgnore
  59          * @return int
  60          */
  61         function mb_substr_split_unicode( $str, $splitPos ) {
  62                 return Fallback::mb_substr_split_unicode( $str, $splitPos );
  63         }
  64 }
  65
  66 if ( !function_exists( 'mb_strlen' ) ) {
  67         /**
  68          * @codeCoverageIgnore
  69          * @return int
  70          */
  71         function mb_strlen( $str, $enc = '' ) {
  72                 return Fallback::mb_strlen( $str, $enc );
  73         }
  74 }
  75
  76 if ( !function_exists( 'mb_strpos' ) ) {
  77         /**
  78          * @codeCoverageIgnore
  79          * @return int
  80          */
  81         function mb_strpos( $haystack, $needle, $offset = 0, $encoding = '' ) {
  82                 return Fallback::mb_strpos( $haystack, $needle, $offset, $encoding );
  83         }
  84 }
  85
  86 if ( !function_exists( 'mb_strrpos' ) ) {
  87         /**
  88          * @codeCoverageIgnore
  89          * @return int
  90          */
  91         function mb_strrpos( $haystack, $needle, $offset = 0, $encoding = '' ) {
  92                 return Fallback::mb_strrpos( $haystack, $needle, $offset, $encoding );
  93         }
  94 }
  95
  96 // gzdecode function only exists in PHP >= 5.4.0
  97 // http://php.net/gzdecode
  98 if ( !function_exists( 'gzdecode' ) ) {
  99         /**
 100          * @codeCoverageIgnore
 101          * @return string
 102          */
 103         function gzdecode( $data ) {
 104                 return gzinflate( substr( $data, 10, -8 ) );
 105         }
 106 }
 107
 108 // hash_equals function only exists in PHP >= 5.6.0
 109 if ( !function_exists( 'hash_equals' ) ) {
 110         /**
 111          * Check whether a user-provided string is equal to a fixed-length secret without
 112          * revealing bytes of the secret through timing differences.
 113          *
 114          * This timing guarantee -- that a partial match takes the same time as a complete
 115          * mismatch -- is why this function is used in some security-sensitive parts of the code.
 116          * For example, it shouldn't be possible to guess an HMAC signature one byte at a time.
 117          *
 118          * Longer explanation: http://www.emerose.com/timing-attacks-explained
 119          *
 120          * @codeCoverageIgnore
 121          * @param string $known_string Fixed-length secret to compare against
 122          * @param string $user_string User-provided string
 123          * @return bool True if the strings are the same, false otherwise
 124          */
 125         function hash_equals( $known_string, $user_string ) {
 126                 // Strict type checking as in PHP's native implementation
 127                 if ( !is_string( $known_string ) ) {
 128                         trigger_error( 'hash_equals(): Expected known_string to be a string, ' .
 129                                 gettype( $known_string ) . ' given', E_USER_WARNING );
 130
 131                         return false;
 132                 }
 133
 134                 if ( !is_string( $user_string ) ) {
 135                         trigger_error( 'hash_equals(): Expected user_string to be a string, ' .
 136                                 gettype( $user_string ) . ' given', E_USER_WARNING );
 137
 138                         return false;
 139                 }
 140
 141                 // Note that we do one thing PHP doesn't: try to avoid leaking information about
 142                 // relative lengths of $known_string and $user_string, and of multiple $known_strings.
 143                 // However, lengths may still inevitably leak through, for example, CPU cache misses.
 144                 $known_string_len = strlen( $known_string );
 145                 $user_string_len = strlen( $user_string );
 146                 $result = $known_string_len ^ $user_string_len;
 147                 for ( $i = 0; $i < $user_string_len; $i++ ) {
 148                         $result |= ord( $known_string[$i % $known_string_len] ) ^ ord( $user_string[$i] );
 149                 }
 150
 151                 return ( $result === 0 );
 152         }
 153 }
 154 /// @endcond
 155
 156 /**
 157  * Like array_diff( $a, $b ) except that it works with two-dimensional arrays.
 158  * @param array $a
 159  * @param array $b
 160  * @return array
 161  */
 162 function wfArrayDiff2( $a, $b ) {
 163         return array_udiff( $a, $b, 'wfArrayDiff2_cmp' );
 164 }
 165
 166 /**
 167  * @param array|string $a
 168  * @param array|string $b
 169  * @return int
 170  */
 171 function wfArrayDiff2_cmp( $a, $b ) {
 172         if ( is_string( $a ) && is_string( $b ) ) {
 173                 return strcmp( $a, $b );
 174         } elseif ( count( $a ) !== count( $b ) ) {
 175                 return count( $a ) < count( $b ) ? -1 : 1;
 176         } else {
 177                 reset( $a );
 178                 reset( $b );
 179                 while ( ( list( , $valueA ) = each( $a ) ) && ( list( , $valueB ) = each( $b ) ) ) {
 180                         $cmp = strcmp( $valueA, $valueB );
 181                         if ( $cmp !== 0 ) {
 182                                 return $cmp;
 183                         }
 184                 }
 185                 return 0;
 186         }
 187 }
 188
 189 /**
 190  * Array lookup
 191  * Returns an array where the values in array $b are replaced by the
 192  * values in array $a with the corresponding keys
 193  *
 194  * @deprecated since 1.22; use array_intersect_key()
 195  * @param array $a
 196  * @param array $b
 197  * @return array
 198  */
 199 function wfArrayLookup( $a, $b ) {
 200         wfDeprecated( __FUNCTION__, '1.22' );
 201         return array_flip( array_intersect( array_flip( $a ), array_keys( $b ) ) );
 202 }
 203
 204 /**
 205  * Appends to second array if $value differs from that in $default
 206  *
 207  * @param string|int $key
 208  * @param mixed $value
 209  * @param mixed $default
 210  * @param array $changed Array to alter
 211  * @throws MWException
 212  */
 213 function wfAppendToArrayIfNotDefault( $key, $value, $default, &$changed ) {
 214         if ( is_null( $changed ) ) {
 215                 throw new MWException( 'GlobalFunctions::wfAppendToArrayIfNotDefault got null' );
 216         }
 217         if ( $default[$key] !== $value ) {
 218                 $changed[$key] = $value;
 219         }
 220 }
 221
 222 /**
 223  * Backwards array plus for people who haven't bothered to read the PHP manual
 224  * XXX: will not darn your socks for you.
 225  *
 226  * @deprecated since 1.22; use array_replace()
 227  *
 228  * @param array $array1 Initial array to merge.
 229  * @param array [$array2,...] Variable list of arrays to merge.
 230  * @return array
 231  */
 232 function wfArrayMerge( $array1 /*...*/ ) {
 233         wfDeprecated( __FUNCTION__, '1.22' );
 234         $args = func_get_args();
 235         $args = array_reverse( $args, true );
 236         $out = array();
 237         foreach ( $args as $arg ) {
 238                 $out += $arg;
 239         }
 240         return $out;
 241 }
 242
 243 /**
 244  * Merge arrays in the style of getUserPermissionsErrors, with duplicate removal
 245  * e.g.
 246  *      wfMergeErrorArrays(
 247  *              array( array( 'x' ) ),
 248  *              array( array( 'x', '2' ) ),
 249  *              array( array( 'x' ) ),
 250  *              array( array( 'y' ) )
 251  *      );
 252  * returns:
 253  *              array(
 254  *              array( 'x', '2' ),
 255  *              array( 'x' ),
 256  *              array( 'y' )
 257  *      )
 258  *
 259  * @param array [$array1,...]
 260  * @return array
 261  */
 262 function wfMergeErrorArrays( /*...*/ ) {
 263         $args = func_get_args();
 264         $out = array();
 265         foreach ( $args as $errors ) {
 266                 foreach ( $errors as $params ) {
 267                         # @todo FIXME: Sometimes get nested arrays for $params,
 268                         # which leads to E_NOTICEs
 269                         $spec = implode( "\t", $params );
 270                         $out[$spec] = $params;
 271                 }
 272         }
 273         return array_values( $out );
 274 }
 275
 276 /**
 277  * Insert array into another array after the specified *KEY*
 278  *
 279  * @param array $array The array.
 280  * @param array $insert The array to insert.
 281  * @param mixed $after The key to insert after
 282  * @return array
 283  */
 284 function wfArrayInsertAfter( array $array, array $insert, $after ) {
 285         // Find the offset of the element to insert after.
 286         $keys = array_keys( $array );
 287         $offsetByKey = array_flip( $keys );
 288
 289         $offset = $offsetByKey[$after];
 290
 291         // Insert at the specified offset
 292         $before = array_slice( $array, 0, $offset + 1, true );
 293         $after = array_slice( $array, $offset + 1, count( $array ) - $offset, true );
 294
 295         $output = $before + $insert + $after;
 296
 297         return $output;
 298 }
 299
 300 /**
 301  * Recursively converts the parameter (an object) to an array with the same data
 302  *
 303  * @param object|array $objOrArray
 304  * @param bool $recursive
 305  * @return array
 306  */
 307 function wfObjectToArray( $objOrArray, $recursive = true ) {
 308         $array = array();
 309         if ( is_object( $objOrArray ) ) {
 310                 $objOrArray = get_object_vars( $objOrArray );
 311         }
 312         foreach ( $objOrArray as $key => $value ) {
 313                 if ( $recursive && ( is_object( $value ) || is_array( $value ) ) ) {
 314                         $value = wfObjectToArray( $value );
 315                 }
 316
 317                 $array[$key] = $value;
 318         }
 319
 320         return $array;
 321 }
 322
 323 /**
 324  * Get a random decimal value between 0 and 1, in a way
 325  * not likely to give duplicate values for any realistic
 326  * number of articles.
 327  *
 328  * @return string
 329  */
 330 function wfRandom() {
 331         # The maximum random value is "only" 2^31-1, so get two random
 332         # values to reduce the chance of dupes
 333         $max = mt_getrandmax() + 1;
 334         $rand = number_format( ( mt_rand() * $max + mt_rand() ) / $max / $max, 12, '.', '' );
 335
 336         return $rand;
 337 }
 338
 339 /**
 340  * Get a random string containing a number of pseudo-random hex
 341  * characters.
 342  * @note This is not secure, if you are trying to generate some sort
 343  *       of token please use MWCryptRand instead.
 344  *
 345  * @param int $length The length of the string to generate
 346  * @return string
 347  * @since 1.20
 348  */
 349 function wfRandomString( $length = 32 ) {
 350         $str = '';
 351         for ( $n = 0; $n < $length; $n += 7 ) {
 352                 $str .= sprintf( '%07x', mt_rand() & 0xfffffff );
 353         }
 354         return substr( $str, 0, $length );
 355 }
 356
 357 /**
 358  * We want some things to be included as literal characters in our title URLs
 359  * for prettiness, which urlencode encodes by default.  According to RFC 1738,
 360  * all of the following should be safe:
 361  *
 362  * ;:@&=$-_.+!*'(),
 363  *
 364  * But + is not safe because it's used to indicate a space; &= are only safe in
 365  * paths and not in queries (and we don't distinguish here); ' seems kind of
 366  * scary; and urlencode() doesn't touch -_. to begin with.  Plus, although /
 367  * is reserved, we don't care.  So the list we unescape is:
 368  *
 369  * ;:@$!*(),/
 370  *
 371  * However, IIS7 redirects fail when the url contains a colon (Bug 22709),
 372  * so no fancy : for IIS7.
 373  *
 374  * %2F in the page titles seems to fatally break for some reason.
 375  *
 376  * @param string $s
 377  * @return string
 378  */
 379 function wfUrlencode( $s ) {
 380         static $needle;
 381
 382         if ( is_null( $s ) ) {
 383                 $needle = null;
 384                 return '';
 385         }
 386
 387         if ( is_null( $needle ) ) {
 388                 $needle = array( '%3B', '%40', '%24', '%21', '%2A', '%28', '%29', '%2C', '%2F' );
 389                 if ( !isset( $_SERVER['SERVER_SOFTWARE'] ) ||
 390                         ( strpos( $_SERVER['SERVER_SOFTWARE'], 'Microsoft-IIS/7' ) === false )
 391                 ) {
 392                         $needle[] = '%3A';
 393                 }
 394         }
 395
 396         $s = urlencode( $s );
 397         $s = str_ireplace(
 398                 $needle,
 399                 array( ';', '@', '$', '!', '*', '(', ')', ',', '/', ':' ),
 400                 $s
 401         );
 402
 403         return $s;
 404 }
 405
 406 /**
 407  * This function takes two arrays as input, and returns a CGI-style string, e.g.
 408  * "days=7&limit=100". Options in the first array override options in the second.
 409  * Options set to null or false will not be output.
 410  *
 411  * @param array $array1 ( String|Array )
 412  * @param array $array2 ( String|Array )
 413  * @param string $prefix
 414  * @return string
 415  */
 416 function wfArrayToCgi( $array1, $array2 = null, $prefix = '' ) {
 417         if ( !is_null( $array2 ) ) {
 418                 $array1 = $array1 + $array2;
 419         }
 420
 421         $cgi = '';
 422         foreach ( $array1 as $key => $value ) {
 423                 if ( !is_null( $value ) && $value !== false ) {
 424                         if ( $cgi != '' ) {
 425                                 $cgi .= '&';
 426                         }
 427                         if ( $prefix !== '' ) {
 428                                 $key = $prefix . "[$key]";
 429                         }
 430                         if ( is_array( $value ) ) {
 431                                 $firstTime = true;
 432                                 foreach ( $value as $k => $v ) {
 433                                         $cgi .= $firstTime ? '' : '&';
 434                                         if ( is_array( $v ) ) {
 435                                                 $cgi .= wfArrayToCgi( $v, null, $key . "[$k]" );
 436                                         } else {
 437                                                 $cgi .= urlencode( $key . "[$k]" ) . '=' . urlencode( $v );
 438                                         }
 439                                         $firstTime = false;
 440                                 }
 441                         } else {
 442                                 if ( is_object( $value ) ) {
 443                                         $value = $value->__toString();
 444                                 }
 445                                 $cgi .= urlencode( $key ) . '=' . urlencode( $value );
 446                         }
 447                 }
 448         }
 449         return $cgi;
 450 }
 451
 452 /**
 453  * This is the logical opposite of wfArrayToCgi(): it accepts a query string as
 454  * its argument and returns the same string in array form.  This allows compatibility
 455  * with legacy functions that accept raw query strings instead of nice
 456  * arrays.  Of course, keys and values are urldecode()d.
 457  *
 458  * @param string $query Query string
 459  * @return string[] Array version of input
 460  */
 461 function wfCgiToArray( $query ) {
 462         if ( isset( $query[0] ) && $query[0] == '?' ) {
 463                 $query = substr( $query, 1 );
 464         }
 465         $bits = explode( '&', $query );
 466         $ret = array();
 467         foreach ( $bits as $bit ) {
 468                 if ( $bit === '' ) {
 469                         continue;
 470                 }
 471                 if ( strpos( $bit, '=' ) === false ) {
 472                         // Pieces like &qwerty become 'qwerty' => '' (at least this is what php does)
 473                         $key = $bit;
 474                         $value = '';
 475                 } else {
 476                         list( $key, $value ) = explode( '=', $bit );
 477                 }
 478                 $key = urldecode( $key );
 479                 $value = urldecode( $value );
 480                 if ( strpos( $key, '[' ) !== false ) {
 481                         $keys = array_reverse( explode( '[', $key ) );
 482                         $key = array_pop( $keys );
 483                         $temp = $value;
 484                         foreach ( $keys as $k ) {
 485                                 $k = substr( $k, 0, -1 );
 486                                 $temp = array( $k => $temp );
 487                         }
 488                         if ( isset( $ret[$key] ) ) {
 489                                 $ret[$key] = array_merge( $ret[$key], $temp );
 490                         } else {
 491                                 $ret[$key] = $temp;
 492                         }
 493                 } else {
 494                         $ret[$key] = $value;
 495                 }
 496         }
 497         return $ret;
 498 }
 499
 500 /**
 501  * Append a query string to an existing URL, which may or may not already
 502  * have query string parameters already. If so, they will be combined.
 503  *
 504  * @param string $url
 505  * @param string|string[] $query String or associative array
 506  * @return string
 507  */
 508 function wfAppendQuery( $url, $query ) {
 509         if ( is_array( $query ) ) {
 510                 $query = wfArrayToCgi( $query );
 511         }
 512         if ( $query != '' ) {
 513                 if ( false === strpos( $url, '?' ) ) {
 514                         $url .= '?';
 515                 } else {
 516                         $url .= '&';
 517                 }
 518                 $url .= $query;
 519         }
 520         return $url;
 521 }
 522
 523 /**
 524  * Expand a potentially local URL to a fully-qualified URL. Assumes $wgServer
 525  * is correct.
 526  *
 527  * The meaning of the PROTO_* constants is as follows:
 528  * PROTO_HTTP: Output a URL starting with http://
 529  * PROTO_HTTPS: Output a URL starting with https://
 530  * PROTO_RELATIVE: Output a URL starting with // (protocol-relative URL)
 531  * PROTO_CURRENT: Output a URL starting with either http:// or https:// , depending
 532  *    on which protocol was used for the current incoming request
 533  * PROTO_CANONICAL: For URLs without a domain, like /w/index.php , use $wgCanonicalServer.
 534  *    For protocol-relative URLs, use the protocol of $wgCanonicalServer
 535  * PROTO_INTERNAL: Like PROTO_CANONICAL, but uses $wgInternalServer instead of $wgCanonicalServer
 536  *
 537  * @todo this won't work with current-path-relative URLs
 538  * like "subdir/foo.html", etc.
 539  *
 540  * @param string $url Either fully-qualified or a local path + query
 541  * @param string $defaultProto One of the PROTO_* constants. Determines the
 542  *    protocol to use if $url or $wgServer is protocol-relative
 543  * @return string Fully-qualified URL, current-path-relative URL or false if
 544  *    no valid URL can be constructed
 545  */
 546 function wfExpandUrl( $url, $defaultProto = PROTO_CURRENT ) {
 547         global $wgServer, $wgCanonicalServer, $wgInternalServer, $wgRequest,
 548                 $wgHttpsPort;
 549         if ( $defaultProto === PROTO_CANONICAL ) {
 550                 $serverUrl = $wgCanonicalServer;
 551         } elseif ( $defaultProto === PROTO_INTERNAL && $wgInternalServer !== false ) {
 552                 // Make $wgInternalServer fall back to $wgServer if not set
 553                 $serverUrl = $wgInternalServer;
 554         } else {
 555                 $serverUrl = $wgServer;
 556                 if ( $defaultProto === PROTO_CURRENT ) {
 557                         $defaultProto = $wgRequest->getProtocol() . '://';
 558                 }
 559         }
 560
 561         // Analyze $serverUrl to obtain its protocol
 562         $bits = wfParseUrl( $serverUrl );
 563         $serverHasProto = $bits && $bits['scheme'] != '';
 564
 565         if ( $defaultProto === PROTO_CANONICAL || $defaultProto === PROTO_INTERNAL ) {
 566                 if ( $serverHasProto ) {
 567                         $defaultProto = $bits['scheme'] . '://';
 568                 } else {
 569                         // $wgCanonicalServer or $wgInternalServer doesn't have a protocol.
 570                         // This really isn't supposed to happen. Fall back to HTTP in this
 571                         // ridiculous case.
 572                         $defaultProto = PROTO_HTTP;
 573                 }
 574         }
 575
 576         $defaultProtoWithoutSlashes = substr( $defaultProto, 0, -2 );
 577
 578         if ( substr( $url, 0, 2 ) == '//' ) {
 579                 $url = $defaultProtoWithoutSlashes . $url;
 580         } elseif ( substr( $url, 0, 1 ) == '/' ) {
 581                 // If $serverUrl is protocol-relative, prepend $defaultProtoWithoutSlashes,
 582                 // otherwise leave it alone.
 583                 $url = ( $serverHasProto ? '' : $defaultProtoWithoutSlashes ) . $serverUrl . $url;
 584         }
 585
 586         $bits = wfParseUrl( $url );
 587
 588         // ensure proper port for HTTPS arrives in URL
 589         // https://bugzilla.wikimedia.org/show_bug.cgi?id=65184
 590         if ( $defaultProto === PROTO_HTTPS && $wgHttpsPort != 443 ) {
 591                 $bits['port'] = $wgHttpsPort;
 592         }
 593
 594         if ( $bits && isset( $bits['path'] ) ) {
 595                 $bits['path'] = wfRemoveDotSegments( $bits['path'] );
 596                 return wfAssembleUrl( $bits );
 597         } elseif ( $bits ) {
 598                 # No path to expand
 599                 return $url;
 600         } elseif ( substr( $url, 0, 1 ) != '/' ) {
 601                 # URL is a relative path
 602                 return wfRemoveDotSegments( $url );
 603         }
 604
 605         # Expanded URL is not valid.
 606         return false;
 607 }
 608
 609 /**
 610  * This function will reassemble a URL parsed with wfParseURL.  This is useful
 611  * if you need to edit part of a URL and put it back together.
 612  *
 613  * This is the basic structure used (brackets contain keys for $urlParts):
 614  * [scheme][delimiter][user]:[pass]@[host]:[port][path]?[query]#[fragment]
 615  *
 616  * @todo Need to integrate this into wfExpandUrl (bug 32168)
 617  *
 618  * @since 1.19
 619  * @param array $urlParts URL parts, as output from wfParseUrl
 620  * @return string URL assembled from its component parts
 621  */
 622 function wfAssembleUrl( $urlParts ) {
 623         $result = '';
 624
 625         if ( isset( $urlParts['delimiter'] ) ) {
 626                 if ( isset( $urlParts['scheme'] ) ) {
 627                         $result .= $urlParts['scheme'];
 628                 }
 629
 630                 $result .= $urlParts['delimiter'];
 631         }
 632
 633         if ( isset( $urlParts['host'] ) ) {
 634                 if ( isset( $urlParts['user'] ) ) {
 635                         $result .= $urlParts['user'];
 636                         if ( isset( $urlParts['pass'] ) ) {
 637                                 $result .= ':' . $urlParts['pass'];
 638                         }
 639                         $result .= '@';
 640                 }
 641
 642                 $result .= $urlParts['host'];
 643
 644                 if ( isset( $urlParts['port'] ) ) {
 645                         $result .= ':' . $urlParts['port'];
 646                 }
 647         }
 648
 649         if ( isset( $urlParts['path'] ) ) {
 650                 $result .= $urlParts['path'];
 651         }
 652
 653         if ( isset( $urlParts['query'] ) ) {
 654                 $result .= '?' . $urlParts['query'];
 655         }
 656
 657         if ( isset( $urlParts['fragment'] ) ) {
 658                 $result .= '#' . $urlParts['fragment'];
 659         }
 660
 661         return $result;
 662 }
 663
 664 /**
 665  * Remove all dot-segments in the provided URL path.  For example,
 666  * '/a/./b/../c/' becomes '/a/c/'.  For details on the algorithm, please see
 667  * RFC3986 section 5.2.4.
 668  *
 669  * @todo Need to integrate this into wfExpandUrl (bug 32168)
 670  *
 671  * @param string $urlPath URL path, potentially containing dot-segments
 672  * @return string URL path with all dot-segments removed
 673  */
 674 function wfRemoveDotSegments( $urlPath ) {
 675         $output = '';
 676         $inputOffset = 0;
 677         $inputLength = strlen( $urlPath );
 678
 679         while ( $inputOffset < $inputLength ) {
 680                 $prefixLengthOne = substr( $urlPath, $inputOffset, 1 );
 681                 $prefixLengthTwo = substr( $urlPath, $inputOffset, 2 );
 682                 $prefixLengthThree = substr( $urlPath, $inputOffset, 3 );
 683                 $prefixLengthFour = substr( $urlPath, $inputOffset, 4 );
 684                 $trimOutput = false;
 685
 686                 if ( $prefixLengthTwo == './' ) {
 687                         # Step A, remove leading "./"
 688                         $inputOffset += 2;
 689                 } elseif ( $prefixLengthThree == '../' ) {
 690                         # Step A, remove leading "../"
 691                         $inputOffset += 3;
 692                 } elseif ( ( $prefixLengthTwo == '/.' ) && ( $inputOffset + 2 == $inputLength ) ) {
 693                         # Step B, replace leading "/.$" with "/"
 694                         $inputOffset += 1;
 695                         $urlPath[$inputOffset] = '/';
 696                 } elseif ( $prefixLengthThree == '/./' ) {
 697                         # Step B, replace leading "/./" with "/"
 698                         $inputOffset += 2;
 699                 } elseif ( $prefixLengthThree == '/..' && ( $inputOffset + 3 == $inputLength ) ) {
 700                         # Step C, replace leading "/..$" with "/" and
 701                         # remove last path component in output
 702                         $inputOffset += 2;
 703                         $urlPath[$inputOffset] = '/';
 704                         $trimOutput = true;
 705                 } elseif ( $prefixLengthFour == '/../' ) {
 706                         # Step C, replace leading "/../" with "/" and
 707                         # remove last path component in output
 708                         $inputOffset += 3;
 709                         $trimOutput = true;
 710                 } elseif ( ( $prefixLengthOne == '.' ) && ( $inputOffset + 1 == $inputLength ) ) {
 711                         # Step D, remove "^.$"
 712                         $inputOffset += 1;
 713                 } elseif ( ( $prefixLengthTwo == '..' ) && ( $inputOffset + 2 == $inputLength ) ) {
 714                         # Step D, remove "^..$"
 715                         $inputOffset += 2;
 716                 } else {
 717                         # Step E, move leading path segment to output
 718                         if ( $prefixLengthOne == '/' ) {
 719                                 $slashPos = strpos( $urlPath, '/', $inputOffset + 1 );
 720                         } else {
 721                                 $slashPos = strpos( $urlPath, '/', $inputOffset );
 722                         }
 723                         if ( $slashPos === false ) {
 724                                 $output .= substr( $urlPath, $inputOffset );
 725                                 $inputOffset = $inputLength;
 726                         } else {
 727                                 $output .= substr( $urlPath, $inputOffset, $slashPos - $inputOffset );
 728                                 $inputOffset += $slashPos - $inputOffset;
 729                         }
 730                 }
 731
 732                 if ( $trimOutput ) {
 733                         $slashPos = strrpos( $output, '/' );
 734                         if ( $slashPos === false ) {
 735                                 $output = '';
 736                         } else {
 737                                 $output = substr( $output, 0, $slashPos );
 738                         }
 739                 }
 740         }
 741
 742         return $output;
 743 }
 744
 745 /**
 746  * Returns a regular expression of url protocols
 747  *
 748  * @param bool $includeProtocolRelative If false, remove '//' from the returned protocol list.
 749  *        DO NOT USE this directly, use wfUrlProtocolsWithoutProtRel() instead
 750  * @return string
 751  */
 752 function wfUrlProtocols( $includeProtocolRelative = true ) {
 753         global $wgUrlProtocols;
 754
 755         // Cache return values separately based on $includeProtocolRelative
 756         static $withProtRel = null, $withoutProtRel = null;
 757         $cachedValue = $includeProtocolRelative ? $withProtRel : $withoutProtRel;
 758         if ( !is_null( $cachedValue ) ) {
 759                 return $cachedValue;
 760         }
 761
 762         // Support old-style $wgUrlProtocols strings, for backwards compatibility
 763         // with LocalSettings files from 1.5
 764         if ( is_array( $wgUrlProtocols ) ) {
 765                 $protocols = array();
 766                 foreach ( $wgUrlProtocols as $protocol ) {
 767                         // Filter out '//' if !$includeProtocolRelative
 768                         if ( $includeProtocolRelative || $protocol !== '//' ) {
 769                                 $protocols[] = preg_quote( $protocol, '/' );
 770                         }
 771                 }
 772
 773                 $retval = implode( '|', $protocols );
 774         } else {
 775                 // Ignore $includeProtocolRelative in this case
 776                 // This case exists for pre-1.6 compatibility, and we can safely assume
 777                 // that '//' won't appear in a pre-1.6 config because protocol-relative
 778                 // URLs weren't supported until 1.18
 779                 $retval = $wgUrlProtocols;
 780         }
 781
 782         // Cache return value
 783         if ( $includeProtocolRelative ) {
 784                 $withProtRel = $retval;
 785         } else {
 786                 $withoutProtRel = $retval;
 787         }
 788         return $retval;
 789 }
 790
 791 /**
 792  * Like wfUrlProtocols(), but excludes '//' from the protocol list. Use this if
 793  * you need a regex that matches all URL protocols but does not match protocol-
 794  * relative URLs
 795  * @return string
 796  */
 797 function wfUrlProtocolsWithoutProtRel() {
 798         return wfUrlProtocols( false );
 799 }
 800
 801 /**
 802  * parse_url() work-alike, but non-broken.  Differences:
 803  *
 804  * 1) Does not raise warnings on bad URLs (just returns false).
 805  * 2) Handles protocols that don't use :// (e.g., mailto: and news:, as well as
 806  *    protocol-relative URLs) correctly.
 807  * 3) Adds a "delimiter" element to the array, either '://', ':' or '//' (see (2)).
 808  *
 809  * @param string $url A URL to parse
 810  * @return string[] Bits of the URL in an associative array, per PHP docs
 811  */
 812 function wfParseUrl( $url ) {
 813         global $wgUrlProtocols; // Allow all protocols defined in DefaultSettings/LocalSettings.php
 814
 815         // Protocol-relative URLs are handled really badly by parse_url(). It's so
 816         // bad that the easiest way to handle them is to just prepend 'http:' and
 817         // strip the protocol out later.
 818         $wasRelative = substr( $url, 0, 2 ) == '//';
 819         if ( $wasRelative ) {
 820                 $url = "http:$url";
 821         }
 822         wfSuppressWarnings();
 823         $bits = parse_url( $url );
 824         wfRestoreWarnings();
 825         // parse_url() returns an array without scheme for some invalid URLs, e.g.
 826         // parse_url("%0Ahttp://example.com") == array( 'host' => '%0Ahttp', 'path' => 'example.com' )
 827         if ( !$bits || !isset( $bits['scheme'] ) ) {
 828                 return false;
 829         }
 830
 831         // parse_url() incorrectly handles schemes case-sensitively. Convert it to lowercase.
 832         $bits['scheme'] = strtolower( $bits['scheme'] );
 833
 834         // most of the protocols are followed by ://, but mailto: and sometimes news: not, check for it
 835         if ( in_array( $bits['scheme'] . '://', $wgUrlProtocols ) ) {
 836                 $bits['delimiter'] = '://';
 837         } elseif ( in_array( $bits['scheme'] . ':', $wgUrlProtocols ) ) {
 838                 $bits['delimiter'] = ':';
 839                 // parse_url detects for news: and mailto: the host part of an url as path
 840                 // We have to correct this wrong detection
 841                 if ( isset( $bits['path'] ) ) {
 842                         $bits['host'] = $bits['path'];
 843                         $bits['path'] = '';
 844                 }
 845         } else {
 846                 return false;
 847         }
 848
 849         /* Provide an empty host for eg. file:/// urls (see bug 28627) */
 850         if ( !isset( $bits['host'] ) ) {
 851                 $bits['host'] = '';
 852
 853                 // bug 45069
 854                 if ( isset( $bits['path'] ) ) {
 855                         /* parse_url loses the third / for file:///c:/ urls (but not on variants) */
 856                         if ( substr( $bits['path'], 0, 1 ) !== '/' ) {
 857                                 $bits['path'] = '/' . $bits['path'];
 858                         }
 859                 } else {
 860                         $bits['path'] = '';
 861                 }
 862         }
 863
 864         // If the URL was protocol-relative, fix scheme and delimiter
 865         if ( $wasRelative ) {
 866                 $bits['scheme'] = '';
 867                 $bits['delimiter'] = '//';
 868         }
 869         return $bits;
 870 }
 871
 872 /**
 873  * Take a URL, make sure it's expanded to fully qualified, and replace any
 874  * encoded non-ASCII Unicode characters with their UTF-8 original forms
 875  * for more compact display and legibility for local audiences.
 876  *
 877  * @todo handle punycode domains too
 878  *
 879  * @param string $url
 880  * @return string
 881  */
 882 function wfExpandIRI( $url ) {
 883         return preg_replace_callback(
 884                 '/((?:%[89A-F][0-9A-F])+)/i',
 885                 'wfExpandIRI_callback',
 886                 wfExpandUrl( $url )
 887         );
 888 }
 889
 890 /**
 891  * Private callback for wfExpandIRI
 892  * @param array $matches
 893  * @return string
 894  */
 895 function wfExpandIRI_callback( $matches ) {
 896         return urldecode( $matches[1] );
 897 }
 898
 899 /**
 900  * Make URL indexes, appropriate for the el_index field of externallinks.
 901  *
 902  * @param string $url
 903  * @return array
 904  */
 905 function wfMakeUrlIndexes( $url ) {
 906         $bits = wfParseUrl( $url );
 907
 908         // Reverse the labels in the hostname, convert to lower case
 909         // For emails reverse domainpart only
 910         if ( $bits['scheme'] == 'mailto' ) {
 911                 $mailparts = explode( '@', $bits['host'], 2 );
 912                 if ( count( $mailparts ) === 2 ) {
 913                         $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
 914                 } else {
 915                         // No domain specified, don't mangle it
 916                         $domainpart = '';
 917                 }
 918                 $reversedHost = $domainpart . '@' . $mailparts[0];
 919         } else {
 920                 $reversedHost = strtolower( implode( '.', array_reverse( explode( '.', $bits['host'] ) ) ) );
 921         }
 922         // Add an extra dot to the end
 923         // Why? Is it in wrong place in mailto links?
 924         if ( substr( $reversedHost, -1, 1 ) !== '.' ) {
 925                 $reversedHost .= '.';
 926         }
 927         // Reconstruct the pseudo-URL
 928         $prot = $bits['scheme'];
 929         $index = $prot . $bits['delimiter'] . $reversedHost;
 930         // Leave out user and password. Add the port, path, query and fragment
 931         if ( isset( $bits['port'] ) ) {
 932                 $index .= ':' . $bits['port'];
 933         }
 934         if ( isset( $bits['path'] ) ) {
 935                 $index .= $bits['path'];
 936         } else {
 937                 $index .= '/';
 938         }
 939         if ( isset( $bits['query'] ) ) {
 940                 $index .= '?' . $bits['query'];
 941         }
 942         if ( isset( $bits['fragment'] ) ) {
 943                 $index .= '#' . $bits['fragment'];
 944         }
 945
 946         if ( $prot == '' ) {
 947                 return array( "http:$index", "https:$index" );
 948         } else {
 949                 return array( $index );
 950         }
 951 }
 952
 953 /**
 954  * Check whether a given URL has a domain that occurs in a given set of domains
 955  * @param string $url URL
 956  * @param array $domains Array of domains (strings)
 957  * @return bool True if the host part of $url ends in one of the strings in $domains
 958  */
 959 function wfMatchesDomainList( $url, $domains ) {
 960         $bits = wfParseUrl( $url );
 961         if ( is_array( $bits ) && isset( $bits['host'] ) ) {
 962                 $host = '.' . $bits['host'];
 963                 foreach ( (array)$domains as $domain ) {
 964                         $domain = '.' . $domain;
 965                         if ( substr( $host, -strlen( $domain ) ) === $domain ) {
 966                                 return true;
 967                         }
 968                 }
 969         }
 970         return false;
 971 }
 972
 973 /**
 974  * Sends a line to the debug log if enabled or, optionally, to a comment in output.
 975  * In normal operation this is a NOP.
 976  *
 977  * Controlling globals:
 978  * $wgDebugLogFile - points to the log file
 979  * $wgDebugRawPage - if false, 'action=raw' hits will not result in debug output.
 980  * $wgDebugComments - if on, some debug items may appear in comments in the HTML output.
 981  *
 982  * @param string $text
 983  * @param string|bool $dest Destination of the message:
 984  *     - 'all': both to the log and HTML (debug toolbar or HTML comments)
 985  *     - 'log': only to the log and not in HTML
 986  *   For backward compatibility, it can also take a boolean:
 987  *     - true: same as 'all'
 988  *     - false: same as 'log'
 989  */
 990 function wfDebug( $text, $dest = 'all' ) {
 991         global $wgDebugLogFile, $wgDebugRawPage, $wgDebugLogPrefix;
 992
 993         if ( !$wgDebugRawPage && wfIsDebugRawPage() ) {
 994                 return;
 995         }
 996
 997         // Turn $dest into a string if it's a boolean (for b/c)
 998         if ( $dest === true ) {
 999                 $dest = 'all';
1000         } elseif ( $dest === false ) {
1001                 $dest = 'log';
1002         }
1003
1004         $timer = wfDebugTimer();
1005         if ( $timer !== '' ) {
1006                 $text = preg_replace( '/[^\n]/', $timer . '\0', $text, 1 );
1007         }
1008
1009         if ( $dest === 'all' ) {
1010                 MWDebug::debugMsg( $text );
1011         }
1012
1013         if ( $wgDebugLogFile != '' ) {
1014                 # Strip unprintables; they can switch terminal modes when binary data
1015                 # gets dumped, which is pretty annoying.
1016                 $text = preg_replace( '![\x00-\x08\x0b\x0c\x0e-\x1f]!', ' ', $text );
1017                 $text = $wgDebugLogPrefix . $text;
1018                 wfErrorLog( $text, $wgDebugLogFile );
1019         }
1020 }
1021
1022 /**
1023  * Returns true if debug logging should be suppressed if $wgDebugRawPage = false
1024  * @return bool
1025  */
1026 function wfIsDebugRawPage() {
1027         static $cache;
1028         if ( $cache !== null ) {
1029                 return $cache;
1030         }
1031         # Check for raw action using $_GET not $wgRequest, since the latter might not be initialised yet
1032         if ( ( isset( $_GET['action'] ) && $_GET['action'] == 'raw' )
1033                 || (
1034                         isset( $_SERVER['SCRIPT_NAME'] )
1035                         && substr( $_SERVER['SCRIPT_NAME'], -8 ) == 'load.php'
1036                 )
1037         ) {
1038                 $cache = true;
1039         } else {
1040                 $cache = false;
1041         }
1042         return $cache;
1043 }
1044
1045 /**
1046  * Get microsecond timestamps for debug logs
1047  *
1048  * @return string
1049  */
1050 function wfDebugTimer() {
1051         global $wgDebugTimestamps, $wgRequestTime;
1052
1053         if ( !$wgDebugTimestamps ) {
1054                 return '';
1055         }
1056
1057         $prefix = sprintf( "%6.4f", microtime( true ) - $wgRequestTime );
1058         $mem = sprintf( "%5.1fM", ( memory_get_usage( true ) / ( 1024 * 1024 ) ) );
1059         return "$prefix $mem  ";
1060 }
1061
1062 /**
1063  * Send a line giving PHP memory usage.
1064  *
1065  * @param bool $exact Print exact byte values instead of kibibytes (default: false)
1066  */
1067 function wfDebugMem( $exact = false ) {
1068         $mem = memory_get_usage();
1069         if ( !$exact ) {
1070                 $mem = floor( $mem / 1024 ) . ' KiB';
1071         } else {
1072                 $mem .= ' B';
1073         }
1074         wfDebug( "Memory usage: $mem\n" );
1075 }
1076
1077 /**
1078  * Send a line to a supplementary debug log file, if configured, or main debug log if not.
1079  * To configure a supplementary log file, set $wgDebugLogGroups[$logGroup] to a string
1080  * filename or an associative array mapping 'destination' to the desired filename. The
1081  * associative array may also contain a 'sample' key with an integer value, specifying
1082  * a sampling factor.
1083  *
1084  * @since 1.23 support for sampling log messages via $wgDebugLogGroups.
1085  *
1086  * @param string $logGroup
1087  * @param string $text
1088  * @param string|bool $dest Destination of the message:
1089  *     - 'all': both to the log and HTML (debug toolbar or HTML comments)
1090  *     - 'log': only to the log and not in HTML
1091  *     - 'private': only to the specifc log if set in $wgDebugLogGroups and
1092  *       discarded otherwise
1093  *   For backward compatibility, it can also take a boolean:
1094  *     - true: same as 'all'
1095  *     - false: same as 'private'
1096  */
1097 function wfDebugLog( $logGroup, $text, $dest = 'all' ) {
1098         global $wgDebugLogGroups;
1099
1100         $text = trim( $text ) . "\n";
1101
1102         // Turn $dest into a string if it's a boolean (for b/c)
1103         if ( $dest === true ) {
1104                 $dest = 'all';
1105         } elseif ( $dest === false ) {
1106                 $dest = 'private';
1107         }
1108
1109         if ( !isset( $wgDebugLogGroups[$logGroup] ) ) {
1110                 if ( $dest !== 'private' ) {
1111                         wfDebug( "[$logGroup] $text", $dest );
1112                 }
1113                 return;
1114         }
1115
1116         if ( $dest === 'all' ) {
1117                 MWDebug::debugMsg( "[$logGroup] $text" );
1118         }
1119
1120         $logConfig = $wgDebugLogGroups[$logGroup];
1121         if ( $logConfig === false ) {
1122                 return;
1123         }
1124         if ( is_array( $logConfig ) ) {
1125                 if ( isset( $logConfig['sample'] ) && mt_rand( 1, $logConfig['sample'] ) !== 1 ) {
1126                         return;
1127                 }
1128                 $destination = $logConfig['destination'];
1129         } else {
1130                 $destination = strval( $logConfig );
1131         }
1132
1133         $time = wfTimestamp( TS_DB );
1134         $wiki = wfWikiID();
1135         $host = wfHostname();
1136         wfErrorLog( "$time $host $wiki: $text", $destination );
1137 }
1138
1139 /**
1140  * Log for database errors
1141  *
1142  * @param string $text Database error message.
1143  */
1144 function wfLogDBError( $text ) {
1145         global $wgDBerrorLog, $wgDBerrorLogTZ;
1146         static $logDBErrorTimeZoneObject = null;
1147
1148         if ( $wgDBerrorLog ) {
1149                 $host = wfHostname();
1150                 $wiki = wfWikiID();
1151
1152                 if ( $wgDBerrorLogTZ && !$logDBErrorTimeZoneObject ) {
1153                         $logDBErrorTimeZoneObject = new DateTimeZone( $wgDBerrorLogTZ );
1154                 }
1155
1156                 // Workaround for https://bugs.php.net/bug.php?id=52063
1157                 // Can be removed when min PHP > 5.3.2
1158                 if ( $logDBErrorTimeZoneObject === null ) {
1159                         $d = date_create( "now" );
1160                 } else {
1161                         $d = date_create( "now", $logDBErrorTimeZoneObject );
1162                 }
1163
1164                 $date = $d->format( 'D M j G:i:s T Y' );
1165
1166                 $text = "$date\t$host\t$wiki\t" . trim( $text ) . "\n";
1167                 wfErrorLog( $text, $wgDBerrorLog );
1168         }
1169 }
1170
1171 /**
1172  * Throws a warning that $function is deprecated
1173  *
1174  * @param string $function
1175  * @param string|bool $version Version of MediaWiki that the function
1176  *    was deprecated in (Added in 1.19).
1177  * @param string|bool $component Added in 1.19.
1178  * @param int $callerOffset How far up the call stack is the original
1179  *    caller. 2 = function that called the function that called
1180  *    wfDeprecated (Added in 1.20)
1181  *
1182  * @return null
1183  */
1184 function wfDeprecated( $function, $version = false, $component = false, $callerOffset = 2 ) {
1185         MWDebug::deprecated( $function, $version, $component, $callerOffset + 1 );
1186 }
1187
1188 /**
1189  * Send a warning either to the debug log or in a PHP error depending on
1190  * $wgDevelopmentWarnings. To log warnings in production, use wfLogWarning() instead.
1191  *
1192  * @param string $msg message to send
1193  * @param int $callerOffset Number of items to go back in the backtrace to
1194  *        find the correct caller (1 = function calling wfWarn, ...)
1195  * @param int $level PHP error level; defaults to E_USER_NOTICE;
1196  *        only used when $wgDevelopmentWarnings is true
1197  */
1198 function wfWarn( $msg, $callerOffset = 1, $level = E_USER_NOTICE ) {
1199         MWDebug::warning( $msg, $callerOffset + 1, $level, 'auto' );
1200 }
1201
1202 /**
1203  * Send a warning as a PHP error and the debug log. This is intended for logging
1204  * warnings in production. For logging development warnings, use WfWarn instead.
1205  *
1206  * @param string $msg Message to send
1207  * @param int $callerOffset Number of items to go back in the backtrace to
1208  *        find the correct caller (1 = function calling wfLogWarning, ...)
1209  * @param int $level PHP error level; defaults to E_USER_WARNING
1210  */
1211 function wfLogWarning( $msg, $callerOffset = 1, $level = E_USER_WARNING ) {
1212         MWDebug::warning( $msg, $callerOffset + 1, $level, 'production' );
1213 }
1214
1215 /**
1216  * Log to a file without getting "file size exceeded" signals.
1217  *
1218  * Can also log to TCP or UDP with the syntax udp://host:port/prefix. This will
1219  * send lines to the specified port, prefixed by the specified prefix and a space.
1220  *
1221  * @param string $text
1222  * @param string $file Filename
1223  * @throws MWException
1224  */
1225 function wfErrorLog( $text, $file ) {
1226         if ( substr( $file, 0, 4 ) == 'udp:' ) {
1227                 # Needs the sockets extension
1228                 if ( preg_match( '!^(tcp|udp):(?://)?\[([0-9a-fA-F:]+)\]:(\d+)(?:/(.*))?$!', $file, $m ) ) {
1229                         // IPv6 bracketed host
1230                         $host = $m[2];
1231                         $port = intval( $m[3] );
1232                         $prefix = isset( $m[4] ) ? $m[4] : false;
1233                         $domain = AF_INET6;
1234                 } elseif ( preg_match( '!^(tcp|udp):(?://)?([a-zA-Z0-9.-]+):(\d+)(?:/(.*))?$!', $file, $m ) ) {
1235                         $host = $m[2];
1236                         if ( !IP::isIPv4( $host ) ) {
1237                                 $host = gethostbyname( $host );
1238                         }
1239                         $port = intval( $m[3] );
1240                         $prefix = isset( $m[4] ) ? $m[4] : false;
1241                         $domain = AF_INET;
1242                 } else {
1243                         throw new MWException( __METHOD__ . ': Invalid UDP specification' );
1244                 }
1245
1246                 // Clean it up for the multiplexer
1247                 if ( strval( $prefix ) !== '' ) {
1248                         $text = preg_replace( '/^/m', $prefix . ' ', $text );
1249
1250                         // Limit to 64KB
1251                         if ( strlen( $text ) > 65506 ) {
1252                                 $text = substr( $text, 0, 65506 );
1253                         }
1254
1255                         if ( substr( $text, -1 ) != "\n" ) {
1256                                 $text .= "\n";
1257                         }
1258                 } elseif ( strlen( $text ) > 65507 ) {
1259                         $text = substr( $text, 0, 65507 );
1260                 }
1261
1262                 $sock = socket_create( $domain, SOCK_DGRAM, SOL_UDP );
1263                 if ( !$sock ) {
1264                         return;
1265                 }
1266
1267                 socket_sendto( $sock, $text, strlen( $text ), 0, $host, $port );
1268                 socket_close( $sock );
1269         } else {
1270                 wfSuppressWarnings();
1271                 $exists = file_exists( $file );
1272                 $size = $exists ? filesize( $file ) : false;
1273                 if ( !$exists || ( $size !== false && $size + strlen( $text ) < 0x7fffffff ) ) {
1274                         file_put_contents( $file, $text, FILE_APPEND );
1275                 }
1276                 wfRestoreWarnings();
1277         }
1278 }
1279
1280 /**
1281  * @todo document
1282  */
1283 function wfLogProfilingData() {
1284         global $wgRequestTime, $wgDebugLogFile, $wgDebugLogGroups, $wgDebugRawPage;
1285         global $wgProfileLimit, $wgUser, $wgRequest;
1286
1287         StatCounter::singleton()->flush();
1288
1289         $profiler = Profiler::instance();
1290
1291         # Profiling must actually be enabled...
1292         if ( $profiler->isStub() ) {
1293                 return;
1294         }
1295
1296         // Get total page request time and only show pages that longer than
1297         // $wgProfileLimit time (default is 0)
1298         $elapsed = microtime( true ) - $wgRequestTime;
1299         if ( $elapsed <= $wgProfileLimit ) {
1300                 return;
1301         }
1302
1303         $profiler->logData();
1304
1305         // Check whether this should be logged in the debug file.
1306         if ( isset( $wgDebugLogGroups['profileoutput'] )
1307                 && $wgDebugLogGroups['profileoutput'] === false
1308         ) {
1309                 // Explicitely disabled
1310                 return;
1311         }
1312         if ( !isset( $wgDebugLogGroups['profileoutput'] ) && $wgDebugLogFile == '' ) {
1313                 // Logging not enabled; no point going further
1314                 return;
1315         }
1316         if ( !$wgDebugRawPage && wfIsDebugRawPage() ) {
1317                 return;
1318         }
1319
1320         $forward = '';
1321         if ( !empty( $_SERVER['HTTP_X_FORWARDED_FOR'] ) ) {
1322                 $forward = ' forwarded for ' . $_SERVER['HTTP_X_FORWARDED_FOR'];
1323         }
1324         if ( !empty( $_SERVER['HTTP_CLIENT_IP'] ) ) {
1325                 $forward .= ' client IP ' . $_SERVER['HTTP_CLIENT_IP'];
1326         }
1327         if ( !empty( $_SERVER['HTTP_FROM'] ) ) {
1328                 $forward .= ' from ' . $_SERVER['HTTP_FROM'];
1329         }
1330         if ( $forward ) {
1331                 $forward = "\t(proxied via {$_SERVER['REMOTE_ADDR']}{$forward})";
1332         }
1333         // Don't load $wgUser at this late stage just for statistics purposes
1334         // @todo FIXME: We can detect some anons even if it is not loaded. See User::getId()
1335         if ( $wgUser->isItemLoaded( 'id' ) && $wgUser->isAnon() ) {
1336                 $forward .= ' anon';
1337         }
1338
1339         // Command line script uses a FauxRequest object which does not have
1340         // any knowledge about an URL and throw an exception instead.
1341         try {
1342                 $requestUrl = $wgRequest->getRequestURL();
1343         } catch ( MWException $e ) {
1344                 $requestUrl = 'n/a';
1345         }
1346
1347         $log = sprintf( "%s\t%04.3f\t%s\n",
1348                 gmdate( 'YmdHis' ), $elapsed,
1349                 urldecode( $requestUrl . $forward ) );
1350
1351         wfDebugLog( 'profileoutput', $log . $profiler->getOutput() );
1352 }
1353
1354 /**
1355  * Increment a statistics counter
1356  *
1357  * @param string $key
1358  * @param int $count
1359  * @return void
1360  */
1361 function wfIncrStats( $key, $count = 1 ) {
1362         StatCounter::singleton()->incr( $key, $count );
1363 }
1364
1365 /**
1366  * Check whether the wiki is in read-only mode.
1367  *
1368  * @return bool
1369  */
1370 function wfReadOnly() {
1371         return wfReadOnlyReason() !== false;
1372 }
1373
1374 /**
1375  * Get the value of $wgReadOnly or the contents of $wgReadOnlyFile.
1376  *
1377  * @return string|bool String when in read-only mode; false otherwise
1378  */
1379 function wfReadOnlyReason() {
1380         global $wgReadOnly, $wgReadOnlyFile;
1381
1382         if ( $wgReadOnly === null ) {
1383                 // Set $wgReadOnly for faster access next time
1384                 if ( is_file( $wgReadOnlyFile ) && filesize( $wgReadOnlyFile ) > 0 ) {
1385                         $wgReadOnly = file_get_contents( $wgReadOnlyFile );
1386                 } else {
1387                         $wgReadOnly = false;
1388                 }
1389         }
1390
1391         return $wgReadOnly;
1392 }
1393
1394 /**
1395  * Return a Language object from $langcode
1396  *
1397  * @param Language|string|bool $langcode Either:
1398  *                  - a Language object
1399  *                  - code of the language to get the message for, if it is
1400  *                    a valid code create a language for that language, if
1401  *                    it is a string but not a valid code then make a basic
1402  *                    language object
1403  *                  - a boolean: if it's false then use the global object for
1404  *                    the current user's language (as a fallback for the old parameter
1405  *                    functionality), or if it is true then use global object
1406  *                    for the wiki's content language.
1407  * @return Language
1408  */
1409 function wfGetLangObj( $langcode = false ) {
1410         # Identify which language to get or create a language object for.
1411         # Using is_object here due to Stub objects.
1412         if ( is_object( $langcode ) ) {
1413                 # Great, we already have the object (hopefully)!
1414                 return $langcode;
1415         }
1416
1417         global $wgContLang, $wgLanguageCode;
1418         if ( $langcode === true || $langcode === $wgLanguageCode ) {
1419                 # $langcode is the language code of the wikis content language object.
1420                 # or it is a boolean and value is true
1421                 return $wgContLang;
1422         }
1423
1424         global $wgLang;
1425         if ( $langcode === false || $langcode === $wgLang->getCode() ) {
1426                 # $langcode is the language code of user language object.
1427                 # or it was a boolean and value is false
1428                 return $wgLang;
1429         }
1430
1431         $validCodes = array_keys( Language::fetchLanguageNames() );
1432         if ( in_array( $langcode, $validCodes ) ) {
1433                 # $langcode corresponds to a valid language.
1434                 return Language::factory( $langcode );
1435         }
1436
1437         # $langcode is a string, but not a valid language code; use content language.
1438         wfDebug( "Invalid language code passed to wfGetLangObj, falling back to content language.\n" );
1439         return $wgContLang;
1440 }
1441
1442 /**
1443  * This is the function for getting translated interface messages.
1444  *
1445  * @see Message class for documentation how to use them.
1446  * @see https://www.mediawiki.org/wiki/Manual:Messages_API
1447  *
1448  * This function replaces all old wfMsg* functions.
1449  *
1450  * @param string $key Message key
1451  * @param mixed [$params,...] Normal message parameters
1452  * @return Message
1453  *
1454  * @since 1.17
1455  *
1456  * @see Message::__construct
1457  */
1458 function wfMessage( $key /*...*/ ) {
1459         $params = func_get_args();
1460         array_shift( $params );
1461         if ( isset( $params[0] ) && is_array( $params[0] ) ) {
1462                 $params = $params[0];
1463         }
1464         return new Message( $key, $params );
1465 }
1466
1467 /**
1468  * This function accepts multiple message keys and returns a message instance
1469  * for the first message which is non-empty. If all messages are empty then an
1470  * instance of the first message key is returned.
1471  *
1472  * @param string|string[] [$keys,...] Message keys
1473  * @return Message
1474  *
1475  * @since 1.18
1476  *
1477  * @see Message::newFallbackSequence
1478  */
1479 function wfMessageFallback( /*...*/ ) {
1480         $args = func_get_args();
1481         return call_user_func_array( 'Message::newFallbackSequence', $args );
1482 }
1483
1484 /**
1485  * Get a message from anywhere, for the current user language.
1486  *
1487  * Use wfMsgForContent() instead if the message should NOT
1488  * change depending on the user preferences.
1489  *
1490  * @deprecated since 1.18
1491  *
1492  * @param string $key lookup key for the message, usually
1493  *    defined in languages/Language.php
1494  *
1495  * Parameters to the message, which can be used to insert variable text into
1496  * it, can be passed to this function in the following formats:
1497  * - One per argument, starting at the second parameter
1498  * - As an array in the second parameter
1499  * These are not shown in the function definition.
1500  *
1501  * @return string
1502  */
1503 function wfMsg( $key ) {
1504         wfDeprecated( __METHOD__, '1.21' );
1505
1506         $args = func_get_args();
1507         array_shift( $args );
1508         return wfMsgReal( $key, $args );
1509 }
1510
1511 /**
1512  * Same as above except doesn't transform the message
1513  *
1514  * @deprecated since 1.18
1515  *
1516  * @param string $key
1517  * @return string
1518  */
1519 function wfMsgNoTrans( $key ) {
1520         wfDeprecated( __METHOD__, '1.21' );
1521
1522         $args = func_get_args();
1523         array_shift( $args );
1524         return wfMsgReal( $key, $args, true, false, false );
1525 }
1526
1527 /**
1528  * Get a message from anywhere, for the current global language
1529  * set with $wgLanguageCode.
1530  *
1531  * Use this if the message should NOT change dependent on the
1532  * language set in the user's preferences. This is the case for
1533  * most text written into logs, as well as link targets (such as
1534  * the name of the copyright policy page). Link titles, on the
1535  * other hand, should be shown in the UI language.
1536  *
1537  * Note that MediaWiki allows users to change the user interface
1538  * language in their preferences, but a single installation
1539  * typically only contains content in one language.
1540  *
1541  * Be wary of this distinction: If you use wfMsg() where you should
1542  * use wfMsgForContent(), a user of the software may have to
1543  * customize potentially hundreds of messages in
1544  * order to, e.g., fix a link in every possible language.
1545  *
1546  * @deprecated since 1.18
1547  *
1548  * @param string $key Lookup key for the message, usually
1549  *     defined in languages/Language.php
1550  * @return string
1551  */
1552 function wfMsgForContent( $key ) {
1553         wfDeprecated( __METHOD__, '1.21' );
1554
1555         global $wgForceUIMsgAsContentMsg;
1556         $args = func_get_args();
1557         array_shift( $args );
1558         $forcontent = true;
1559         if ( is_array( $wgForceUIMsgAsContentMsg )
1560                 && in_array( $key, $wgForceUIMsgAsContentMsg )
1561         ) {
1562                 $forcontent = false;
1563         }
1564         return wfMsgReal( $key, $args, true, $forcontent );
1565 }
1566
1567 /**
1568  * Same as above except doesn't transform the message
1569  *
1570  * @deprecated since 1.18
1571  *
1572  * @param string $key
1573  * @return string
1574  */
1575 function wfMsgForContentNoTrans( $key ) {
1576         wfDeprecated( __METHOD__, '1.21' );
1577
1578         global $wgForceUIMsgAsContentMsg;
1579         $args = func_get_args();
1580         array_shift( $args );
1581         $forcontent = true;
1582         if ( is_array( $wgForceUIMsgAsContentMsg )
1583                 && in_array( $key, $wgForceUIMsgAsContentMsg )
1584         ) {
1585                 $forcontent = false;
1586         }
1587         return wfMsgReal( $key, $args, true, $forcontent, false );
1588 }
1589
1590 /**
1591  * Really get a message
1592  *
1593  * @deprecated since 1.18
1594  *
1595  * @param string $key Key to get.
1596  * @param array $args
1597  * @param bool $useDB
1598  * @param string|bool $forContent Language code, or false for user lang, true for content lang.
1599  * @param bool $transform Whether or not to transform the message.
1600  * @return string The requested message.
1601  */
1602 function wfMsgReal( $key, $args, $useDB = true, $forContent = false, $transform = true ) {
1603         wfDeprecated( __METHOD__, '1.21' );
1604
1605         wfProfileIn( __METHOD__ );
1606         $message = wfMsgGetKey( $key, $useDB, $forContent, $transform );
1607         $message = wfMsgReplaceArgs( $message, $args );
1608         wfProfileOut( __METHOD__ );
1609         return $message;
1610 }
1611
1612 /**
1613  * Fetch a message string value, but don't replace any keys yet.
1614  *
1615  * @deprecated since 1.18
1616  *
1617  * @param string $key
1618  * @param bool $useDB
1619  * @param string|bool $langCode Code of the language to get the message for, or
1620  *   behaves as a content language switch if it is a boolean.
1621  * @param bool $transform Whether to parse magic words, etc.
1622  * @return string
1623  */
1624 function wfMsgGetKey( $key, $useDB = true, $langCode = false, $transform = true ) {
1625         wfDeprecated( __METHOD__, '1.21' );
1626
1627         wfRunHooks( 'NormalizeMessageKey', array( &$key, &$useDB, &$langCode, &$transform ) );
1628
1629         $cache = MessageCache::singleton();
1630         $message = $cache->get( $key, $useDB, $langCode );
1631         if ( $message === false ) {
1632                 $message = '&lt;' . htmlspecialchars( $key ) . '&gt;';
1633         } elseif ( $transform ) {
1634                 $message = $cache->transform( $message );
1635         }
1636         return $message;
1637 }
1638
1639 /**
1640  * Replace message parameter keys on the given formatted output.
1641  *
1642  * @param string $message
1643  * @param array $args
1644  * @return string
1645  * @private
1646  */
1647 function wfMsgReplaceArgs( $message, $args ) {
1648         # Fix windows line-endings
1649         # Some messages are split with explode("\n", $msg)
1650         $message = str_replace( "\r", '', $message );
1651
1652         // Replace arguments
1653         if ( count( $args ) ) {
1654                 if ( is_array( $args[0] ) ) {
1655                         $args = array_values( $args[0] );
1656                 }
1657                 $replacementKeys = array();
1658                 foreach ( $args as $n => $param ) {
1659                         $replacementKeys['$' . ( $n + 1 )] = $param;
1660                 }
1661                 $message = strtr( $message, $replacementKeys );
1662         }
1663
1664         return $message;
1665 }
1666
1667 /**
1668  * Return an HTML-escaped version of a message.
1669  * Parameter replacements, if any, are done *after* the HTML-escaping,
1670  * so parameters may contain HTML (eg links or form controls). Be sure
1671  * to pre-escape them if you really do want plaintext, or just wrap
1672  * the whole thing in htmlspecialchars().
1673  *
1674  * @deprecated since 1.18
1675  *
1676  * @param string $key
1677  * @param string [$args,...] Parameters
1678  * @return string
1679  */
1680 function wfMsgHtml( $key ) {
1681         wfDeprecated( __METHOD__, '1.21' );
1682
1683         $args = func_get_args();
1684         array_shift( $args );
1685         return wfMsgReplaceArgs( htmlspecialchars( wfMsgGetKey( $key ) ), $args );
1686 }
1687
1688 /**
1689  * Return an HTML version of message
1690  * Parameter replacements, if any, are done *after* parsing the wiki-text message,
1691  * so parameters may contain HTML (eg links or form controls). Be sure
1692  * to pre-escape them if you really do want plaintext, or just wrap
1693  * the whole thing in htmlspecialchars().
1694  *
1695  * @deprecated since 1.18
1696  *
1697  * @param string $key
1698  * @param string [$args,...] Parameters
1699  * @return string
1700  */
1701 function wfMsgWikiHtml( $key ) {
1702         wfDeprecated( __METHOD__, '1.21' );
1703
1704         $args = func_get_args();
1705         array_shift( $args );
1706         return wfMsgReplaceArgs(
1707                 MessageCache::singleton()->parse( wfMsgGetKey( $key ), null,
1708                 /* can't be set to false */ true, /* interface */ true )->getText(),
1709                 $args );
1710 }
1711
1712 /**
1713  * Returns message in the requested format
1714  *
1715  * @deprecated since 1.18
1716  *
1717  * @param string $key Key of the message
1718  * @param array $options Processing rules.
1719  *   Can take the following options:
1720  *     parse: parses wikitext to HTML
1721  *     parseinline: parses wikitext to HTML and removes the surrounding
1722  *       p's added by parser or tidy
1723  *     escape: filters message through htmlspecialchars
1724  *     escapenoentities: same, but allows entity references like &#160; through
1725  *     replaceafter: parameters are substituted after parsing or escaping
1726  *     parsemag: transform the message using magic phrases
1727  *     content: fetch message for content language instead of interface
1728  *   Also can accept a single associative argument, of the form 'language' => 'xx':
1729  *     language: Language object or language code to fetch message for
1730  *       (overridden by content).
1731  * Behavior for conflicting options (e.g., parse+parseinline) is undefined.
1732  *
1733  * @return string
1734  */
1735 function wfMsgExt( $key, $options ) {
1736         wfDeprecated( __METHOD__, '1.21' );
1737
1738         $args = func_get_args();
1739         array_shift( $args );
1740         array_shift( $args );
1741         $options = (array)$options;
1742
1743         foreach ( $options as $arrayKey => $option ) {
1744                 if ( !preg_match( '/^[0-9]+|language$/', $arrayKey ) ) {
1745                         # An unknown index, neither numeric nor "language"
1746                         wfWarn( "wfMsgExt called with incorrect parameter key $arrayKey", 1, E_USER_WARNING );
1747                 } elseif ( preg_match( '/^[0-9]+$/', $arrayKey ) && !in_array( $option,
1748                 array( 'parse', 'parseinline', 'escape', 'escapenoentities',
1749                 'replaceafter', 'parsemag', 'content' ) ) ) {
1750                         # A numeric index with unknown value
1751                         wfWarn( "wfMsgExt called with incorrect parameter $option", 1, E_USER_WARNING );
1752                 }
1753         }
1754
1755         if ( in_array( 'content', $options, true ) ) {
1756                 $forContent = true;
1757                 $langCode = true;
1758                 $langCodeObj = null;
1759         } elseif ( array_key_exists( 'language', $options ) ) {
1760                 $forContent = false;
1761                 $langCode = wfGetLangObj( $options['language'] );
1762                 $langCodeObj = $langCode;
1763         } else {
1764                 $forContent = false;
1765                 $langCode = false;
1766                 $langCodeObj = null;
1767         }
1768
1769         $string = wfMsgGetKey( $key, /*DB*/true, $langCode, /*Transform*/false );
1770
1771         if ( !in_array( 'replaceafter', $options, true ) ) {
1772                 $string = wfMsgReplaceArgs( $string, $args );
1773         }
1774
1775         $messageCache = MessageCache::singleton();
1776         $parseInline = in_array( 'parseinline', $options, true );
1777         if ( in_array( 'parse', $options, true ) || $parseInline ) {
1778                 $string = $messageCache->parse( $string, null, true, !$forContent, $langCodeObj );
1779                 if ( $string instanceof ParserOutput ) {
1780                         $string = $string->getText();
1781                 }
1782
1783                 if ( $parseInline ) {
1784                         $string = Parser::stripOuterParagraph( $string );
1785                 }
1786         } elseif ( in_array( 'parsemag', $options, true ) ) {
1787                 $string = $messageCache->transform( $string,
1788                                 !$forContent, $langCodeObj );
1789         }
1790
1791         if ( in_array( 'escape', $options, true ) ) {
1792                 $string = htmlspecialchars ( $string );
1793         } elseif ( in_array( 'escapenoentities', $options, true ) ) {
1794                 $string = Sanitizer::escapeHtmlAllowEntities( $string );
1795         }
1796
1797         if ( in_array( 'replaceafter', $options, true ) ) {
1798                 $string = wfMsgReplaceArgs( $string, $args );
1799         }
1800
1801         return $string;
1802 }
1803
1804 /**
1805  * Since wfMsg() and co suck, they don't return false if the message key they
1806  * looked up didn't exist but instead the key wrapped in <>'s, this function checks for the
1807  * nonexistence of messages by checking the MessageCache::get() result directly.
1808  *
1809  * @deprecated since 1.18. Use Message::isDisabled().
1810  *
1811  * @param string $key The message key looked up
1812  * @return bool True if the message *doesn't* exist.
1813  */
1814 function wfEmptyMsg( $key ) {
1815         wfDeprecated( __METHOD__, '1.21' );
1816
1817         return MessageCache::singleton()->get( $key, /*useDB*/true, /*content*/false ) === false;
1818 }
1819
1820 /**
1821  * Throw a debugging exception. This function previously once exited the process,
1822  * but now throws an exception instead, with similar results.
1823  *
1824  * @deprecated since 1.22; just throw an MWException yourself
1825  * @param string $msg Message shown when dying.
1826  * @throws MWException
1827  */
1828 function wfDebugDieBacktrace( $msg = '' ) {
1829         wfDeprecated( __FUNCTION__, '1.22' );
1830         throw new MWException( $msg );
1831 }
1832
1833 /**
1834  * Fetch server name for use in error reporting etc.
1835  * Use real server name if available, so we know which machine
1836  * in a server farm generated the current page.
1837  *
1838  * @return string
1839  */
1840 function wfHostname() {
1841         static $host;
1842         if ( is_null( $host ) ) {
1843
1844                 # Hostname overriding
1845                 global $wgOverrideHostname;
1846                 if ( $wgOverrideHostname !== false ) {
1847                         # Set static and skip any detection
1848                         $host = $wgOverrideHostname;
1849                         return $host;
1850                 }
1851
1852                 if ( function_exists( 'posix_uname' ) ) {
1853                         // This function not present on Windows
1854                         $uname = posix_uname();
1855                 } else {
1856                         $uname = false;
1857                 }
1858                 if ( is_array( $uname ) && isset( $uname['nodename'] ) ) {
1859                         $host = $uname['nodename'];
1860                 } elseif ( getenv( 'COMPUTERNAME' ) ) {
1861                         # Windows computer name
1862                         $host = getenv( 'COMPUTERNAME' );
1863                 } else {
1864                         # This may be a virtual server.
1865                         $host = $_SERVER['SERVER_NAME'];
1866                 }
1867         }
1868         return $host;
1869 }
1870
1871 /**
1872  * Returns a script tag that stores the amount of time it took MediaWiki to
1873  * handle the request in milliseconds as 'wgBackendResponseTime'.
1874  *
1875  * If $wgShowHostnames is true, the script will also set 'wgHostname' to the
1876  * hostname of the server handling the request.
1877  *
1878  * @return string
1879  */
1880 function wfReportTime() {
1881         global $wgRequestTime, $wgShowHostnames;
1882
1883         $responseTime = round( ( microtime( true ) - $wgRequestTime ) * 1000 );
1884         $reportVars = array( 'wgBackendResponseTime' => $responseTime );
1885         if ( $wgShowHostnames ) {
1886                 $reportVars[ 'wgHostname' ] = wfHostname();
1887         }
1888         return Skin::makeVariablesScript( $reportVars );
1889 }
1890
1891 /**
1892  * Safety wrapper for debug_backtrace().
1893  *
1894  * With Zend Optimizer 3.2.0 loaded, this causes segfaults under somewhat
1895  * murky circumstances, which may be triggered in part by stub objects
1896  * or other fancy talking'.
1897  *
1898  * Will return an empty array if Zend Optimizer is detected or if
1899  * debug_backtrace is disabled, otherwise the output from
1900  * debug_backtrace() (trimmed).
1901  *
1902  * @param int $limit This parameter can be used to limit the number of stack frames returned
1903  *
1904  * @return array Array of backtrace information
1905  */
1906 function wfDebugBacktrace( $limit = 0 ) {
1907         static $disabled = null;
1908
1909         if ( extension_loaded( 'Zend Optimizer' ) ) {
1910                 wfDebug( "Zend Optimizer detected; skipping debug_backtrace for safety.\n" );
1911                 return array();
1912         }
1913
1914         if ( is_null( $disabled ) ) {
1915                 $disabled = false;
1916                 $functions = explode( ',', ini_get( 'disable_functions' ) );
1917                 $functions = array_map( 'trim', $functions );
1918                 $functions = array_map( 'strtolower', $functions );
1919                 if ( in_array( 'debug_backtrace', $functions ) ) {
1920                         wfDebug( "debug_backtrace is in disabled_functions\n" );
1921                         $disabled = true;
1922                 }
1923         }
1924         if ( $disabled ) {
1925                 return array();
1926         }
1927
1928         if ( $limit && version_compare( PHP_VERSION, '5.4.0', '>=' ) ) {
1929                 return array_slice( debug_backtrace( DEBUG_BACKTRACE_PROVIDE_OBJECT, $limit + 1 ), 1 );
1930         } else {
1931                 return array_slice( debug_backtrace(), 1 );
1932         }
1933 }
1934
1935 /**
1936  * Get a debug backtrace as a string
1937  *
1938  * @return string
1939  */
1940 function wfBacktrace() {
1941         global $wgCommandLineMode;
1942
1943         if ( $wgCommandLineMode ) {
1944                 $msg = '';
1945         } else {
1946                 $msg = "<ul>\n";
1947         }
1948         $backtrace = wfDebugBacktrace();
1949         foreach ( $backtrace as $call ) {
1950                 if ( isset( $call['file'] ) ) {
1951                         $f = explode( DIRECTORY_SEPARATOR, $call['file'] );
1952                         $file = $f[count( $f ) - 1];
1953                 } else {
1954                         $file = '-';
1955                 }
1956                 if ( isset( $call['line'] ) ) {
1957                         $line = $call['line'];
1958                 } else {
1959                         $line = '-';
1960                 }
1961                 if ( $wgCommandLineMode ) {
1962                         $msg .= "$file line $line calls ";
1963                 } else {
1964                         $msg .= '<li>' . $file . ' line ' . $line . ' calls ';
1965                 }
1966                 if ( !empty( $call['class'] ) ) {
1967                         $msg .= $call['class'] . $call['type'];
1968                 }
1969                 $msg .= $call['function'] . '()';
1970
1971                 if ( $wgCommandLineMode ) {
1972                         $msg .= "\n";
1973                 } else {
1974                         $msg .= "</li>\n";
1975                 }
1976         }
1977         if ( $wgCommandLineMode ) {
1978                 $msg .= "\n";
1979         } else {
1980                 $msg .= "</ul>\n";
1981         }
1982
1983         return $msg;
1984 }
1985
1986 /**
1987  * Get the name of the function which called this function
1988  * wfGetCaller( 1 ) is the function with the wfGetCaller() call (ie. __FUNCTION__)
1989  * wfGetCaller( 2 ) [default] is the caller of the function running wfGetCaller()
1990  * wfGetCaller( 3 ) is the parent of that.
1991  *
1992  * @param int $level
1993  * @return string
1994  */
1995 function wfGetCaller( $level = 2 ) {
1996         $backtrace = wfDebugBacktrace( $level + 1 );
1997         if ( isset( $backtrace[$level] ) ) {
1998                 return wfFormatStackFrame( $backtrace[$level] );
1999         } else {
2000                 return 'unknown';
2001         }
2002 }
2003
2004 /**
2005  * Return a string consisting of callers in the stack. Useful sometimes
2006  * for profiling specific points.
2007  *
2008  * @param int $limit The maximum depth of the stack frame to return, or false for the entire stack.
2009  * @return string
2010  */
2011 function wfGetAllCallers( $limit = 3 ) {
2012         $trace = array_reverse( wfDebugBacktrace() );
2013         if ( !$limit || $limit > count( $trace ) - 1 ) {
2014                 $limit = count( $trace ) - 1;
2015         }
2016         $trace = array_slice( $trace, -$limit - 1, $limit );
2017         return implode( '/', array_map( 'wfFormatStackFrame', $trace ) );
2018 }
2019
2020 /**
2021  * Return a string representation of frame
2022  *
2023  * @param array $frame
2024  * @return string
2025  */
2026 function wfFormatStackFrame( $frame ) {
2027         return isset( $frame['class'] ) ?
2028                 $frame['class'] . '::' . $frame['function'] :
2029                 $frame['function'];
2030 }
2031
2032 /* Some generic result counters, pulled out of SearchEngine */
2033
2034 /**
2035  * @todo document
2036  *
2037  * @param int $offset
2038  * @param int $limit
2039  * @return string
2040  */
2041 function wfShowingResults( $offset, $limit ) {
2042         return wfMessage( 'showingresults' )->numParams( $limit, $offset + 1 )->parse();
2043 }
2044
2045 /**
2046  * Generate (prev x| next x) (20|50|100...) type links for paging
2047  *
2048  * @param string $offset
2049  * @param int $limit
2050  * @param string $link
2051  * @param string $query Optional URL query parameter string
2052  * @param bool $atend Optional param for specified if this is the last page
2053  * @return string
2054  * @deprecated since 1.19; use Language::viewPrevNext() instead
2055  */
2056 function wfViewPrevNext( $offset, $limit, $link, $query = '', $atend = false ) {
2057         wfDeprecated( __METHOD__, '1.19' );
2058
2059         global $wgLang;
2060
2061         $query = wfCgiToArray( $query );
2062
2063         if ( is_object( $link ) ) {
2064                 $title = $link;
2065         } else {
2066                 $title = Title::newFromText( $link );
2067                 if ( is_null( $title ) ) {
2068                         return false;
2069                 }
2070         }
2071
2072         return $wgLang->viewPrevNext( $title, $offset, $limit, $query, $atend );
2073 }
2074
2075 /**
2076  * @todo document
2077  * @todo FIXME: We may want to blacklist some broken browsers
2078  *
2079  * @param bool $force
2080  * @return bool Whereas client accept gzip compression
2081  */
2082 function wfClientAcceptsGzip( $force = false ) {
2083         static $result = null;
2084         if ( $result === null || $force ) {
2085                 $result = false;
2086                 if ( isset( $_SERVER['HTTP_ACCEPT_ENCODING'] ) ) {
2087                         # @todo FIXME: We may want to blacklist some broken browsers
2088                         $m = array();
2089                         if ( preg_match(
2090                                         '/\bgzip(?:;(q)=([0-9]+(?:\.[0-9]+)))?\b/',
2091                                         $_SERVER['HTTP_ACCEPT_ENCODING'],
2092                                         $m
2093                                 )
2094                         ) {
2095                                 if ( isset( $m[2] ) && ( $m[1] == 'q' ) && ( $m[2] == 0 ) ) {
2096                                         $result = false;
2097                                         return $result;
2098                                 }
2099                                 wfDebug( "wfClientAcceptsGzip: client accepts gzip.\n" );
2100                                 $result = true;
2101                         }
2102                 }
2103         }
2104         return $result;
2105 }
2106
2107 /**
2108  * Obtain the offset and limit values from the request string;
2109  * used in special pages
2110  *
2111  * @param int $deflimit Default limit if none supplied
2112  * @param string $optionname Name of a user preference to check against
2113  * @return array
2114  */
2115 function wfCheckLimits( $deflimit = 50, $optionname = 'rclimit' ) {
2116         global $wgRequest;
2117         return $wgRequest->getLimitOffset( $deflimit, $optionname );
2118 }
2119
2120 /**
2121  * Escapes the given text so that it may be output using addWikiText()
2122  * without any linking, formatting, etc. making its way through. This
2123  * is achieved by substituting certain characters with HTML entities.
2124  * As required by the callers, "<nowiki>" is not used.
2125  *
2126  * @param string $text Text to be escaped
2127  * @return string
2128  */
2129 function wfEscapeWikiText( $text ) {
2130         static $repl = null, $repl2 = null;
2131         if ( $repl === null ) {
2132                 $repl = array(
2133                         '"' => '&#34;', '&' => '&#38;', "'" => '&#39;', '<' => '&#60;',
2134                         '=' => '&#61;', '>' => '&#62;', '[' => '&#91;', ']' => '&#93;',
2135                         '{' => '&#123;', '|' => '&#124;', '}' => '&#125;', ';' => '&#59;',
2136                         "\n#" => "\n&#35;", "\r#" => "\r&#35;",
2137                         "\n*" => "\n&#42;", "\r*" => "\r&#42;",
2138                         "\n:" => "\n&#58;", "\r:" => "\r&#58;",
2139                         "\n " => "\n&#32;", "\r " => "\r&#32;",
2140                         "\n\n" => "\n&#10;", "\r\n" => "&#13;\n",
2141                         "\n\r" => "\n&#13;", "\r\r" => "\r&#13;",
2142                         "\n\t" => "\n&#9;", "\r\t" => "\r&#9;", // "\n\t\n" is treated like "\n\n"
2143                         "\n----" => "\n&#45;---", "\r----" => "\r&#45;---",
2144                         '__' => '_&#95;', '://' => '&#58;//',
2145                 );
2146
2147                 // We have to catch everything "\s" matches in PCRE
2148                 foreach ( array( 'ISBN', 'RFC', 'PMID' ) as $magic ) {
2149                         $repl["$magic "] = "$magic&#32;";
2150                         $repl["$magic\t"] = "$magic&#9;";
2151                         $repl["$magic\r"] = "$magic&#13;";
2152                         $repl["$magic\n"] = "$magic&#10;";
2153                         $repl["$magic\f"] = "$magic&#12;";
2154                 }
2155
2156                 // And handle protocols that don't use "://"
2157                 global $wgUrlProtocols;
2158                 $repl2 = array();
2159                 foreach ( $wgUrlProtocols as $prot ) {
2160                         if ( substr( $prot, -1 ) === ':' ) {
2161                                 $repl2[] = preg_quote( substr( $prot, 0, -1 ), '/' );
2162                         }
2163                 }
2164                 $repl2 = $repl2 ? '/\b(' . join( '|', $repl2 ) . '):/i' : '/^(?!)/';
2165         }
2166         $text = substr( strtr( "\n$text", $repl ), 1 );
2167         $text = preg_replace( $repl2, '$1&#58;', $text );
2168         return $text;
2169 }
2170
2171 /**
2172  * Get the current unix timestamp with microseconds.  Useful for profiling
2173  * @deprecated since 1.22; call microtime() directly
2174  * @return float
2175  */
2176 function wfTime() {
2177         wfDeprecated( __FUNCTION__, '1.22' );
2178         return microtime( true );
2179 }
2180
2181 /**
2182  * Sets dest to source and returns the original value of dest
2183  * If source is NULL, it just returns the value, it doesn't set the variable
2184  * If force is true, it will set the value even if source is NULL
2185  *
2186  * @param mixed $dest
2187  * @param mixed $source
2188  * @param bool $force
2189  * @return mixed
2190  */
2191 function wfSetVar( &$dest, $source, $force = false ) {
2192         $temp = $dest;
2193         if ( !is_null( $source ) || $force ) {
2194                 $dest = $source;
2195         }
2196         return $temp;
2197 }
2198
2199 /**
2200  * As for wfSetVar except setting a bit
2201  *
2202  * @param int $dest
2203  * @param int $bit
2204  * @param bool $state
2205  *
2206  * @return bool
2207  */
2208 function wfSetBit( &$dest, $bit, $state = true ) {
2209         $temp = (bool)( $dest & $bit );
2210         if ( !is_null( $state ) ) {
2211                 if ( $state ) {
2212                         $dest |= $bit;
2213                 } else {
2214                         $dest &= ~$bit;
2215                 }
2216         }
2217         return $temp;
2218 }
2219
2220 /**
2221  * A wrapper around the PHP function var_export().
2222  * Either print it or add it to the regular output ($wgOut).
2223  *
2224  * @param mixed $var A PHP variable to dump.
2225  */
2226 function wfVarDump( $var ) {
2227         global $wgOut;
2228         $s = str_replace( "\n", "<br />\n", var_export( $var, true ) . "\n" );
2229         if ( headers_sent() || !isset( $wgOut ) || !is_object( $wgOut ) ) {
2230                 print $s;
2231         } else {
2232                 $wgOut->addHTML( $s );
2233         }
2234 }
2235
2236 /**
2237  * Provide a simple HTTP error.
2238  *
2239  * @param int|string $code
2240  * @param string $label
2241  * @param string $desc
2242  */
2243 function wfHttpError( $code, $label, $desc ) {
2244         global $wgOut;
2245         $wgOut->disable();
2246         header( "HTTP/1.0 $code $label" );
2247         header( "Status: $code $label" );
2248         $wgOut->sendCacheControl();
2249
2250         header( 'Content-type: text/html; charset=utf-8' );
2251         print "<!doctype html>" .
2252                 '<html><head><title>' .
2253                 htmlspecialchars( $label ) .
2254                 '</title></head><body><h1>' .
2255                 htmlspecialchars( $label ) .
2256                 '</h1><p>' .
2257                 nl2br( htmlspecialchars( $desc ) ) .
2258                 "</p></body></html>\n";
2259 }
2260
2261 /**
2262  * Clear away any user-level output buffers, discarding contents.
2263  *
2264  * Suitable for 'starting afresh', for instance when streaming
2265  * relatively large amounts of data without buffering, or wanting to
2266  * output image files without ob_gzhandler's compression.
2267  *
2268  * The optional $resetGzipEncoding parameter controls suppression of
2269  * the Content-Encoding header sent by ob_gzhandler; by default it
2270  * is left. See comments for wfClearOutputBuffers() for why it would
2271  * be used.
2272  *
2273  * Note that some PHP configuration options may add output buffer
2274  * layers which cannot be removed; these are left in place.
2275  *
2276  * @param bool $resetGzipEncoding
2277  */
2278 function wfResetOutputBuffers( $resetGzipEncoding = true ) {
2279         if ( $resetGzipEncoding ) {
2280                 // Suppress Content-Encoding and Content-Length
2281                 // headers from 1.10+s wfOutputHandler
2282                 global $wgDisableOutputCompression;
2283                 $wgDisableOutputCompression = true;
2284         }
2285         while ( $status = ob_get_status() ) {
2286                 if ( $status['type'] == 0 /* PHP_OUTPUT_HANDLER_INTERNAL */ ) {
2287                         // Probably from zlib.output_compression or other
2288                         // PHP-internal setting which can't be removed.
2289                         //
2290                         // Give up, and hope the result doesn't break
2291                         // output behavior.
2292                         break;
2293                 }
2294                 if ( !ob_end_clean() ) {
2295                         // Could not remove output buffer handler; abort now
2296                         // to avoid getting in some kind of infinite loop.
2297                         break;
2298                 }
2299                 if ( $resetGzipEncoding ) {
2300                         if ( $status['name'] == 'ob_gzhandler' ) {
2301                                 // Reset the 'Content-Encoding' field set by this handler
2302                                 // so we can start fresh.
2303                                 header_remove( 'Content-Encoding' );
2304                                 break;
2305                         }
2306                 }
2307         }
2308 }
2309
2310 /**
2311  * More legible than passing a 'false' parameter to wfResetOutputBuffers():
2312  *
2313  * Clear away output buffers, but keep the Content-Encoding header
2314  * produced by ob_gzhandler, if any.
2315  *
2316  * This should be used for HTTP 304 responses, where you need to
2317  * preserve the Content-Encoding header of the real result, but
2318  * also need to suppress the output of ob_gzhandler to keep to spec
2319  * and avoid breaking Firefox in rare cases where the headers and
2320  * body are broken over two packets.
2321  */
2322 function wfClearOutputBuffers() {
2323         wfResetOutputBuffers( false );
2324 }
2325
2326 /**
2327  * Converts an Accept-* header into an array mapping string values to quality
2328  * factors
2329  *
2330  * @param string $accept
2331  * @param string $def default
2332  * @return float[] Associative array of string => float pairs
2333  */
2334 function wfAcceptToPrefs( $accept, $def = '*/*' ) {
2335         # No arg means accept anything (per HTTP spec)
2336         if ( !$accept ) {
2337                 return array( $def => 1.0 );
2338         }
2339
2340         $prefs = array();
2341
2342         $parts = explode( ',', $accept );
2343
2344         foreach ( $parts as $part ) {
2345                 # @todo FIXME: Doesn't deal with params like 'text/html; level=1'
2346                 $values = explode( ';', trim( $part ) );
2347                 $match = array();
2348                 if ( count( $values ) == 1 ) {
2349                         $prefs[$values[0]] = 1.0;
2350                 } elseif ( preg_match( '/q\s*=\s*(\d*\.\d+)/', $values[1], $match ) ) {
2351                         $prefs[$values[0]] = floatval( $match[1] );
2352                 }
2353         }
2354
2355         return $prefs;
2356 }
2357
2358 /**
2359  * Checks if a given MIME type matches any of the keys in the given
2360  * array. Basic wildcards are accepted in the array keys.
2361  *
2362  * Returns the matching MIME type (or wildcard) if a match, otherwise
2363  * NULL if no match.
2364  *
2365  * @param string $type
2366  * @param array $avail
2367  * @return string
2368  * @private
2369  */
2370 function mimeTypeMatch( $type, $avail ) {
2371         if ( array_key_exists( $type, $avail ) ) {
2372                 return $type;
2373         } else {
2374                 $parts = explode( '/', $type );
2375                 if ( array_key_exists( $parts[0] . '/*', $avail ) ) {
2376                         return $parts[0] . '/*';
2377                 } elseif ( array_key_exists( '*/*', $avail ) ) {
2378                         return '*/*';
2379                 } else {
2380                         return null;
2381                 }
2382         }
2383 }
2384
2385 /**
2386  * Returns the 'best' match between a client's requested internet media types
2387  * and the server's list of available types. Each list should be an associative
2388  * array of type to preference (preference is a float between 0.0 and 1.0).
2389  * Wildcards in the types are acceptable.
2390  *
2391  * @param array $cprefs Client's acceptable type list
2392  * @param array $sprefs Server's offered types
2393  * @return string
2394  *
2395  * @todo FIXME: Doesn't handle params like 'text/plain; charset=UTF-8'
2396  * XXX: generalize to negotiate other stuff
2397  */
2398 function wfNegotiateType( $cprefs, $sprefs ) {
2399         $combine = array();
2400
2401         foreach ( array_keys( $sprefs ) as $type ) {
2402                 $parts = explode( '/', $type );
2403                 if ( $parts[1] != '*' ) {
2404                         $ckey = mimeTypeMatch( $type, $cprefs );
2405                         if ( $ckey ) {
2406                                 $combine[$type] = $sprefs[$type] * $cprefs[$ckey];
2407                         }
2408                 }
2409         }
2410
2411         foreach ( array_keys( $cprefs ) as $type ) {
2412                 $parts = explode( '/', $type );
2413                 if ( $parts[1] != '*' && !array_key_exists( $type, $sprefs ) ) {
2414                         $skey = mimeTypeMatch( $type, $sprefs );
2415                         if ( $skey ) {
2416                                 $combine[$type] = $sprefs[$skey] * $cprefs[$type];
2417                         }
2418                 }
2419         }
2420
2421         $bestq = 0;
2422         $besttype = null;
2423
2424         foreach ( array_keys( $combine ) as $type ) {
2425                 if ( $combine[$type] > $bestq ) {
2426                         $besttype = $type;
2427                         $bestq = $combine[$type];
2428                 }
2429         }
2430
2431         return $besttype;
2432 }
2433
2434 /**
2435  * Reference-counted warning suppression
2436  *
2437  * @param bool $end
2438  */
2439 function wfSuppressWarnings( $end = false ) {
2440         static $suppressCount = 0;
2441         static $originalLevel = false;
2442
2443         if ( $end ) {
2444                 if ( $suppressCount ) {
2445                         --$suppressCount;
2446                         if ( !$suppressCount ) {
2447                                 error_reporting( $originalLevel );
2448                         }
2449                 }
2450         } else {
2451                 if ( !$suppressCount ) {
2452                         $originalLevel = error_reporting( E_ALL & ~(
2453                                 E_WARNING |
2454                                 E_NOTICE |
2455                                 E_USER_WARNING |
2456                                 E_USER_NOTICE |
2457                                 E_DEPRECATED |
2458                                 E_USER_DEPRECATED |
2459                                 E_STRICT
2460                         ) );
2461                 }
2462                 ++$suppressCount;
2463         }
2464 }
2465
2466 /**
2467  * Restore error level to previous value
2468  */
2469 function wfRestoreWarnings() {
2470         wfSuppressWarnings( true );
2471 }
2472
2473 # Autodetect, convert and provide timestamps of various types
2474
2475 /**
2476  * Unix time - the number of seconds since 1970-01-01 00:00:00 UTC
2477  */
2478 define( 'TS_UNIX', 0 );
2479
2480 /**
2481  * MediaWiki concatenated string timestamp (YYYYMMDDHHMMSS)
2482  */
2483 define( 'TS_MW', 1 );
2484
2485 /**
2486  * MySQL DATETIME (YYYY-MM-DD HH:MM:SS)
2487  */
2488 define( 'TS_DB', 2 );
2489
2490 /**
2491  * RFC 2822 format, for E-mail and HTTP headers
2492  */
2493 define( 'TS_RFC2822', 3 );
2494
2495 /**
2496  * ISO 8601 format with no timezone: 1986-02-09T20:00:00Z
2497  *
2498  * This is used by Special:Export
2499  */
2500 define( 'TS_ISO_8601', 4 );
2501
2502 /**
2503  * An Exif timestamp (YYYY:MM:DD HH:MM:SS)
2504  *
2505  * @see http://exif.org/Exif2-2.PDF The Exif 2.2 spec, see page 28 for the
2506  *       DateTime tag and page 36 for the DateTimeOriginal and
2507  *       DateTimeDigitized tags.
2508  */
2509 define( 'TS_EXIF', 5 );
2510
2511 /**
2512  * Oracle format time.
2513  */
2514 define( 'TS_ORACLE', 6 );
2515
2516 /**
2517  * Postgres format time.
2518  */
2519 define( 'TS_POSTGRES', 7 );
2520
2521 /**
2522  * ISO 8601 basic format with no timezone: 19860209T200000Z.  This is used by ResourceLoader
2523  */
2524 define( 'TS_ISO_8601_BASIC', 9 );
2525
2526 /**
2527  * Get a timestamp string in one of various formats
2528  *
2529  * @param mixed $outputtype A timestamp in one of the supported formats, the
2530  *   function will autodetect which format is supplied and act accordingly.
2531  * @param mixed $ts Optional timestamp to convert, default 0 for the current time
2532  * @return string|bool String / false The same date in the format specified in $outputtype or false
2533  */
2534 function wfTimestamp( $outputtype = TS_UNIX, $ts = 0 ) {
2535         try {
2536                 $timestamp = new MWTimestamp( $ts );
2537                 return $timestamp->getTimestamp( $outputtype );
2538         } catch ( TimestampException $e ) {
2539                 wfDebug( "wfTimestamp() fed bogus time value: TYPE=$outputtype; VALUE=$ts\n" );
2540                 return false;
2541         }
2542 }
2543
2544 /**
2545  * Return a formatted timestamp, or null if input is null.
2546  * For dealing with nullable timestamp columns in the database.
2547  *
2548  * @param int $outputtype
2549  * @param string $ts
2550  * @return string
2551  */
2552 function wfTimestampOrNull( $outputtype = TS_UNIX, $ts = null ) {
2553         if ( is_null( $ts ) ) {
2554                 return null;
2555         } else {
2556                 return wfTimestamp( $outputtype, $ts );
2557         }
2558 }
2559
2560 /**
2561  * Convenience function; returns MediaWiki timestamp for the present time.
2562  *
2563  * @return string
2564  */
2565 function wfTimestampNow() {
2566         # return NOW
2567         return wfTimestamp( TS_MW, time() );
2568 }
2569
2570 /**
2571  * Check if the operating system is Windows
2572  *
2573  * @return bool True if it's Windows, false otherwise.
2574  */
2575 function wfIsWindows() {
2576         static $isWindows = null;
2577         if ( $isWindows === null ) {
2578                 $isWindows = substr( php_uname(), 0, 7 ) == 'Windows';
2579         }
2580         return $isWindows;
2581 }
2582
2583 /**
2584  * Check if we are running under HHVM
2585  *
2586  * @return bool
2587  */
2588 function wfIsHHVM() {
2589         return defined( 'HHVM_VERSION' );
2590 }
2591
2592 /**
2593  * Swap two variables
2594  *
2595  * @param mixed $x
2596  * @param mixed $y
2597  */
2598 function swap( &$x, &$y ) {
2599         $z = $x;
2600         $x = $y;
2601         $y = $z;
2602 }
2603
2604 /**
2605  * Tries to get the system directory for temporary files. First
2606  * $wgTmpDirectory is checked, and then the TMPDIR, TMP, and TEMP
2607  * environment variables are then checked in sequence, and if none are
2608  * set try sys_get_temp_dir().
2609  *
2610  * NOTE: When possible, use instead the tmpfile() function to create
2611  * temporary files to avoid race conditions on file creation, etc.
2612  *
2613  * @return string
2614  */
2615 function wfTempDir() {
2616         global $wgTmpDirectory;
2617
2618         if ( $wgTmpDirectory !== false ) {
2619                 return $wgTmpDirectory;
2620         }
2621
2622         $tmpDir = array_map( "getenv", array( 'TMPDIR', 'TMP', 'TEMP' ) );
2623
2624         foreach ( $tmpDir as $tmp ) {
2625                 if ( $tmp && file_exists( $tmp ) && is_dir( $tmp ) && is_writable( $tmp ) ) {
2626                         return $tmp;
2627                 }
2628         }
2629         return sys_get_temp_dir();
2630 }
2631
2632 /**
2633  * Make directory, and make all parent directories if they don't exist
2634  *
2635  * @param string $dir Full path to directory to create
2636  * @param int $mode Chmod value to use, default is $wgDirectoryMode
2637  * @param string $caller Optional caller param for debugging.
2638  * @throws MWException
2639  * @return bool
2640  */
2641 function wfMkdirParents( $dir, $mode = null, $caller = null ) {
2642         global $wgDirectoryMode;
2643
2644         if ( FileBackend::isStoragePath( $dir ) ) { // sanity
2645                 throw new MWException( __FUNCTION__ . " given storage path '$dir'." );
2646         }
2647
2648         if ( !is_null( $caller ) ) {
2649                 wfDebug( "$caller: called wfMkdirParents($dir)\n" );
2650         }
2651
2652         if ( strval( $dir ) === '' || ( file_exists( $dir ) && is_dir( $dir ) ) ) {
2653                 return true;
2654         }
2655
2656         $dir = str_replace( array( '\\', '/' ), DIRECTORY_SEPARATOR, $dir );
2657
2658         if ( is_null( $mode ) ) {
2659                 $mode = $wgDirectoryMode;
2660         }
2661
2662         // Turn off the normal warning, we're doing our own below
2663         wfSuppressWarnings();
2664         $ok = mkdir( $dir, $mode, true ); // PHP5 <3
2665         wfRestoreWarnings();
2666
2667         if ( !$ok ) {
2668                 //directory may have been created on another request since we last checked
2669                 if ( is_dir( $dir ) ) {
2670                         return true;
2671                 }
2672
2673                 // PHP doesn't report the path in its warning message, so add our own to aid in diagnosis.
2674                 wfLogWarning( sprintf( "failed to mkdir \"%s\" mode 0%o", $dir, $mode ) );
2675         }
2676         return $ok;
2677 }
2678
2679 /**
2680  * Remove a directory and all its content.
2681  * Does not hide error.
2682  * @param string $dir
2683  */
2684 function wfRecursiveRemoveDir( $dir ) {
2685         wfDebug( __FUNCTION__ . "( $dir )\n" );
2686         // taken from http://de3.php.net/manual/en/function.rmdir.php#98622
2687         if ( is_dir( $dir ) ) {
2688                 $objects = scandir( $dir );
2689                 foreach ( $objects as $object ) {
2690                         if ( $object != "." && $object != ".." ) {
2691                                 if ( filetype( $dir . '/' . $object ) == "dir" ) {
2692                                         wfRecursiveRemoveDir( $dir . '/' . $object );
2693                                 } else {
2694                                         unlink( $dir . '/' . $object );
2695                                 }
2696                         }
2697                 }
2698                 reset( $objects );
2699                 rmdir( $dir );
2700         }
2701 }
2702
2703 /**
2704  * @param int $nr The number to format
2705  * @param int $acc The number of digits after the decimal point, default 2
2706  * @param bool $round Whether or not to round the value, default true
2707  * @return string
2708  */
2709 function wfPercent( $nr, $acc = 2, $round = true ) {
2710         $ret = sprintf( "%.${acc}f", $nr );
2711         return $round ? round( $ret, $acc ) . '%' : "$ret%";
2712 }
2713
2714 /**
2715  * Safety wrapper around ini_get() for boolean settings.
2716  * The values returned from ini_get() are pre-normalized for settings
2717  * set via php.ini or php_flag/php_admin_flag... but *not*
2718  * for those set via php_value/php_admin_value.
2719  *
2720  * It's fairly common for people to use php_value instead of php_flag,
2721  * which can leave you with an 'off' setting giving a false positive
2722  * for code that just takes the ini_get() return value as a boolean.
2723  *
2724  * To make things extra interesting, setting via php_value accepts
2725  * "true" and "yes" as true, but php.ini and php_flag consider them false. :)
2726  * Unrecognized values go false... again opposite PHP's own coercion
2727  * from string to bool.
2728  *
2729  * Luckily, 'properly' set settings will always come back as '0' or '1',
2730  * so we only have to worry about them and the 'improper' settings.
2731  *
2732  * I frickin' hate PHP... :P
2733  *
2734  * @param string $setting
2735  * @return bool
2736  */
2737 function wfIniGetBool( $setting ) {
2738         $val = strtolower( ini_get( $setting ) );
2739         // 'on' and 'true' can't have whitespace around them, but '1' can.
2740         return $val == 'on'
2741                 || $val == 'true'
2742                 || $val == 'yes'
2743                 || preg_match( "/^\s*[+-]?0*[1-9]/", $val ); // approx C atoi() function
2744 }
2745
2746 /**
2747  * Windows-compatible version of escapeshellarg()
2748  * Windows doesn't recognise single-quotes in the shell, but the escapeshellarg()
2749  * function puts single quotes in regardless of OS.
2750  *
2751  * Also fixes the locale problems on Linux in PHP 5.2.6+ (bug backported to
2752  * earlier distro releases of PHP)
2753  *
2754  * @param string [$args,...]
2755  * @return string
2756  */
2757 function wfEscapeShellArg( /*...*/ ) {
2758         wfInitShellLocale();
2759
2760         $args = func_get_args();
2761         $first = true;
2762         $retVal = '';
2763         foreach ( $args as $arg ) {
2764                 if ( !$first ) {
2765                         $retVal .= ' ';
2766                 } else {
2767                         $first = false;
2768                 }
2769
2770                 if ( wfIsWindows() ) {
2771                         // Escaping for an MSVC-style command line parser and CMD.EXE
2772                         // @codingStandardsIgnoreStart For long URLs
2773                         // Refs:
2774                         //  * http://web.archive.org/web/20020708081031/http://mailman.lyra.org/pipermail/scite-interest/2002-March/000436.html
2775                         //  * http://technet.microsoft.com/en-us/library/cc723564.aspx
2776                         //  * Bug #13518
2777                         //  * CR r63214
2778                         // Double the backslashes before any double quotes. Escape the double quotes.
2779                         // @codingStandardsIgnoreEnd
2780                         $tokens = preg_split( '/(\\\\*")/', $arg, -1, PREG_SPLIT_DELIM_CAPTURE );
2781                         $arg = '';
2782                         $iteration = 0;
2783                         foreach ( $tokens as $token ) {
2784                                 if ( $iteration % 2 == 1 ) {
2785                                         // Delimiter, a double quote preceded by zero or more slashes
2786                                         $arg .= str_replace( '\\', '\\\\', substr( $token, 0, -1 ) ) . '\\"';
2787                                 } elseif ( $iteration % 4 == 2 ) {
2788                                         // ^ in $token will be outside quotes, need to be escaped
2789                                         $arg .= str_replace( '^', '^^', $token );
2790                                 } else { // $iteration % 4 == 0
2791                                         // ^ in $token will appear inside double quotes, so leave as is
2792                                         $arg .= $token;
2793                                 }
2794                                 $iteration++;
2795                         }
2796                         // Double the backslashes before the end of the string, because
2797                         // we will soon add a quote
2798                         $m = array();
2799                         if ( preg_match( '/^(.*?)(\\\\+)$/', $arg, $m ) ) {
2800                                 $arg = $m[1] . str_replace( '\\', '\\\\', $m[2] );
2801                         }
2802
2803                         // Add surrounding quotes
2804                         $retVal .= '"' . $arg . '"';
2805                 } else {
2806                         $retVal .= escapeshellarg( $arg );
2807                 }
2808         }
2809         return $retVal;
2810 }
2811
2812 /**
2813  * Check if wfShellExec() is effectively disabled via php.ini config
2814  *
2815  * @return bool|string False or one of (safemode,disabled)
2816  * @since 1.22
2817  */
2818 function wfShellExecDisabled() {
2819         static $disabled = null;
2820         if ( is_null( $disabled ) ) {
2821                 $disabled = false;
2822                 if ( wfIniGetBool( 'safe_mode' ) ) {
2823                         wfDebug( "wfShellExec can't run in safe_mode, PHP's exec functions are too broken.\n" );
2824                         $disabled = 'safemode';
2825                 } else {
2826                         $functions = explode( ',', ini_get( 'disable_functions' ) );
2827                         $functions = array_map( 'trim', $functions );
2828                         $functions = array_map( 'strtolower', $functions );
2829                         if ( in_array( 'proc_open', $functions ) ) {
2830                                 wfDebug( "proc_open is in disabled_functions\n" );
2831                                 $disabled = 'disabled';
2832                         }
2833                 }
2834         }
2835         return $disabled;
2836 }
2837
2838 /**
2839  * Execute a shell command, with time and memory limits mirrored from the PHP
2840  * configuration if supported.
2841  *
2842  * @param string $cmd Command line, properly escaped for shell.
2843  * @param null|mixed &$retval Optional, will receive the program's exit code.
2844  *   (non-zero is usually failure). If there is an error from
2845  *   read, select, or proc_open(), this will be set to -1.
2846  * @param array $environ Optional environment variables which should be
2847  *   added to the executed command environment.
2848  * @param array $limits Optional array with limits(filesize, memory, time, walltime)
2849  *   this overwrites the global wgMaxShell* limits.
2850  * @param array $options Array of options:
2851  *   - duplicateStderr: Set this to true to duplicate stderr to stdout,
2852  *     including errors from limit.sh
2853  *
2854  * @return string Collected stdout as a string
2855  */
2856 function wfShellExec( $cmd, &$retval = null, $environ = array(),
2857         $limits = array(), $options = array()
2858 ) {
2859         global $IP, $wgMaxShellMemory, $wgMaxShellFileSize, $wgMaxShellTime,
2860                 $wgMaxShellWallClockTime, $wgShellCgroup;
2861
2862         $disabled = wfShellExecDisabled();
2863         if ( $disabled ) {
2864                 $retval = 1;
2865                 return $disabled == 'safemode' ?
2866                         'Unable to run external programs in safe mode.' :
2867                         'Unable to run external programs, proc_open() is disabled.';
2868         }
2869
2870         $includeStderr = isset( $options['duplicateStderr'] ) && $options['duplicateStderr'];
2871
2872         wfInitShellLocale();
2873
2874         $envcmd = '';
2875         foreach ( $environ as $k => $v ) {
2876                 if ( wfIsWindows() ) {
2877                         /* Surrounding a set in quotes (method used by wfEscapeShellArg) makes the quotes themselves
2878                          * appear in the environment variable, so we must use carat escaping as documented in
2879                          * http://technet.microsoft.com/en-us/library/cc723564.aspx
2880                          * Note however that the quote isn't listed there, but is needed, and the parentheses
2881                          * are listed there but doesn't appear to need it.
2882                          */
2883                         $envcmd .= "set $k=" . preg_replace( '/([&|()<>^"])/', '^\\1', $v ) . '&& ';
2884                 } else {
2885                         /* Assume this is a POSIX shell, thus required to accept variable assignments before the command
2886                          * http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_09_01
2887                          */
2888                         $envcmd .= "$k=" . escapeshellarg( $v ) . ' ';
2889                 }
2890         }
2891         $cmd = $envcmd . $cmd;
2892
2893         $useLogPipe = false;
2894         if ( php_uname( 's' ) == 'Linux' ) {
2895                 $time = intval ( isset( $limits['time'] ) ? $limits['time'] : $wgMaxShellTime );
2896                 if ( isset( $limits['walltime'] ) ) {
2897                         $wallTime = intval( $limits['walltime'] );
2898                 } elseif ( isset( $limits['time'] ) ) {
2899                         $wallTime = $time;
2900                 } else {
2901                         $wallTime = intval( $wgMaxShellWallClockTime );
2902                 }
2903                 $mem = intval ( isset( $limits['memory'] ) ? $limits['memory'] : $wgMaxShellMemory );
2904                 $filesize = intval ( isset( $limits['filesize'] ) ? $limits['filesize'] : $wgMaxShellFileSize );
2905
2906                 if ( $time > 0 || $mem > 0 || $filesize > 0 || $wallTime > 0 ) {
2907                         $cmd = '/bin/bash ' . escapeshellarg( "$IP/includes/limit.sh" ) . ' ' .
2908                                 escapeshellarg( $cmd ) . ' ' .
2909                                 escapeshellarg(
2910                                         "MW_INCLUDE_STDERR=" . ( $includeStderr ? '1' : '' ) . ';' .
2911                                         "MW_CPU_LIMIT=$time; " .
2912                                         'MW_CGROUP=' . escapeshellarg( $wgShellCgroup ) . '; ' .
2913                                         "MW_MEM_LIMIT=$mem; " .
2914                                         "MW_FILE_SIZE_LIMIT=$filesize; " .
2915                                         "MW_WALL_CLOCK_LIMIT=$wallTime; " .
2916                                         "MW_USE_LOG_PIPE=yes"
2917                                 );
2918                         $useLogPipe = true;
2919                 } elseif ( $includeStderr ) {
2920                         $cmd .= ' 2>&1';
2921                 }
2922         } elseif ( $includeStderr ) {
2923                 $cmd .= ' 2>&1';
2924         }
2925         wfDebug( "wfShellExec: $cmd\n" );
2926
2927         $desc = array(
2928                 0 => array( 'file', 'php://stdin', 'r' ),
2929                 1 => array( 'pipe', 'w' ),
2930                 2 => array( 'file', 'php://stderr', 'w' ) );
2931         if ( $useLogPipe ) {
2932                 $desc[3] = array( 'pipe', 'w' );
2933         }
2934         $pipes = null;
2935         $proc = proc_open( $cmd, $desc, $pipes );
2936         if ( !$proc ) {
2937                 wfDebugLog( 'exec', "proc_open() failed: $cmd" );
2938                 $retval = -1;
2939                 return '';
2940         }
2941         $outBuffer = $logBuffer = '';
2942         $emptyArray = array();
2943         $status = false;
2944         $logMsg = false;
2945
2946         // According to the documentation, it is possible for stream_select()
2947         // to fail due to EINTR. I haven't managed to induce this in testing
2948         // despite sending various signals. If it did happen, the error
2949         // message would take the form:
2950         //
2951         // stream_select(): unable to select [4]: Interrupted system call (max_fd=5)
2952         //
2953         // where [4] is the value of the macro EINTR and "Interrupted system
2954         // call" is string which according to the Linux manual is "possibly"
2955         // localised according to LC_MESSAGES.
2956         $eintr = defined( 'SOCKET_EINTR' ) ? SOCKET_EINTR : 4;
2957         $eintrMessage = "stream_select(): unable to select [$eintr]";
2958
2959         // Build a table mapping resource IDs to pipe FDs to work around a
2960         // PHP 5.3 issue in which stream_select() does not preserve array keys
2961         // <https://bugs.php.net/bug.php?id=53427>.
2962         $fds = array();
2963         foreach ( $pipes as $fd => $pipe ) {
2964                 $fds[(int)$pipe] = $fd;
2965         }
2966
2967         while ( true ) {
2968                 $status = proc_get_status( $proc );
2969                 if ( !$status['running'] ) {
2970                         break;
2971                 }
2972                 $status = false;
2973
2974                 $readyPipes = $pipes;
2975
2976                 // Clear last error
2977                 // @codingStandardsIgnoreStart Generic.PHP.NoSilencedErrors.Discouraged
2978                 @trigger_error( '' );
2979                 if ( @stream_select( $readyPipes, $emptyArray, $emptyArray, null ) === false ) {
2980                         // @codingStandardsIgnoreEnd
2981                         $error = error_get_last();
2982                         if ( strncmp( $error['message'], $eintrMessage, strlen( $eintrMessage ) ) == 0 ) {
2983                                 continue;
2984                         } else {
2985                                 trigger_error( $error['message'], E_USER_WARNING );
2986                                 $logMsg = $error['message'];
2987                                 break;
2988                         }
2989                 }
2990                 foreach ( $readyPipes as $pipe ) {
2991                         $block = fread( $pipe, 65536 );
2992                         $fd = $fds[(int)$pipe];
2993                         if ( $block === '' ) {
2994                                 // End of file
2995                                 fclose( $pipes[$fd] );
2996                                 unset( $pipes[$fd] );
2997                                 if ( !$pipes ) {
2998                                         break 2;
2999                                 }
3000                         } elseif ( $block === false ) {
3001                                 // Read error
3002                                 $logMsg = "Error reading from pipe";
3003                                 break 2;
3004                         } elseif ( $fd == 1 ) {
3005                                 // From stdout
3006                                 $outBuffer .= $block;
3007                         } elseif ( $fd == 3 ) {
3008                                 // From log FD
3009                                 $logBuffer .= $block;
3010                                 if ( strpos( $block, "\n" ) !== false ) {
3011                                         $lines = explode( "\n", $logBuffer );
3012                                         $logBuffer = array_pop( $lines );
3013                                         foreach ( $lines as $line ) {
3014                                                 wfDebugLog( 'exec', $line );
3015                                         }
3016                                 }
3017                         }
3018                 }
3019         }
3020
3021         foreach ( $pipes as $pipe ) {
3022                 fclose( $pipe );
3023         }
3024
3025         // Use the status previously collected if possible, since proc_get_status()
3026         // just calls waitpid() which will not return anything useful the second time.
3027         if ( $status === false ) {
3028                 $status = proc_get_status( $proc );
3029         }
3030
3031         if ( $logMsg !== false ) {
3032                 // Read/select error
3033                 $retval = -1;
3034                 proc_close( $proc );
3035         } elseif ( $status['signaled'] ) {
3036                 $logMsg = "Exited with signal {$status['termsig']}";
3037                 $retval = 128 + $status['termsig'];
3038                 proc_close( $proc );
3039         } else {
3040                 if ( $status['running'] ) {
3041                         $retval = proc_close( $proc );
3042                 } else {
3043                         $retval = $status['exitcode'];
3044                         proc_close( $proc );
3045                 }
3046                 if ( $retval == 127 ) {
3047                         $logMsg = "Possibly missing executable file";
3048                 } elseif ( $retval >= 129 && $retval <= 192 ) {
3049                         $logMsg = "Probably exited with signal " . ( $retval - 128 );
3050                 }
3051         }
3052
3053         if ( $logMsg !== false ) {
3054                 wfDebugLog( 'exec', "$logMsg: $cmd" );
3055         }
3056
3057         return $outBuffer;
3058 }
3059
3060 /**
3061  * Execute a shell command, returning both stdout and stderr. Convenience
3062  * function, as all the arguments to wfShellExec can become unwieldy.
3063  *
3064  * @note This also includes errors from limit.sh, e.g. if $wgMaxShellFileSize is exceeded.
3065  * @param string $cmd Command line, properly escaped for shell.
3066  * @param null|mixed &$retval Optional, will receive the program's exit code.
3067  *   (non-zero is usually failure)
3068  * @param array $environ optional environment variables which should be
3069  *   added to the executed command environment.
3070  * @param array $limits Optional array with limits(filesize, memory, time, walltime)
3071  *   this overwrites the global wgShellMax* limits.
3072  * @return string Collected stdout and stderr as a string
3073  */
3074 function wfShellExecWithStderr( $cmd, &$retval = null, $environ = array(), $limits = array() ) {
3075         return wfShellExec( $cmd, $retval, $environ, $limits, array( 'duplicateStderr' => true ) );
3076 }
3077
3078 /**
3079  * Workaround for http://bugs.php.net/bug.php?id=45132
3080  * escapeshellarg() destroys non-ASCII characters if LANG is not a UTF-8 locale
3081  */
3082 function wfInitShellLocale() {
3083         static $done = false;
3084         if ( $done ) {
3085                 return;
3086         }
3087         $done = true;
3088         global $wgShellLocale;
3089         if ( !wfIniGetBool( 'safe_mode' ) ) {
3090                 putenv( "LC_CTYPE=$wgShellLocale" );
3091                 setlocale( LC_CTYPE, $wgShellLocale );
3092         }
3093 }
3094
3095 /**
3096  * Alias to wfShellWikiCmd()
3097  *
3098  * @see wfShellWikiCmd()
3099  */
3100 function wfShellMaintenanceCmd( $script, array $parameters = array(), array $options = array() ) {
3101         return wfShellWikiCmd( $script, $parameters, $options );
3102 }
3103
3104 /**
3105  * Generate a shell-escaped command line string to run a MediaWiki cli script.
3106  * Note that $parameters should be a flat array and an option with an argument
3107  * should consist of two consecutive items in the array (do not use "--option value").
3108  *
3109  * @param string $script MediaWiki cli script path
3110  * @param array $parameters Arguments and options to the script
3111  * @param array $options Associative array of options:
3112  *              'php': The path to the php executable
3113  *              'wrapper': Path to a PHP wrapper to handle the maintenance script
3114  * @return string
3115  */
3116 function wfShellWikiCmd( $script, array $parameters = array(), array $options = array() ) {
3117         global $wgPhpCli;
3118         // Give site config file a chance to run the script in a wrapper.
3119         // The caller may likely want to call wfBasename() on $script.
3120         wfRunHooks( 'wfShellWikiCmd', array( &$script, &$parameters, &$options ) );
3121         $cmd = isset( $options['php'] ) ? array( $options['php'] ) : array( $wgPhpCli );
3122         if ( isset( $options['wrapper'] ) ) {
3123                 $cmd[] = $options['wrapper'];
3124         }
3125         $cmd[] = $script;
3126         // Escape each parameter for shell
3127         return implode( " ", array_map( 'wfEscapeShellArg', array_merge( $cmd, $parameters ) ) );
3128 }
3129
3130 /**
3131  * wfMerge attempts to merge differences between three texts.
3132  * Returns true for a clean merge and false for failure or a conflict.
3133  *
3134  * @param string $old
3135  * @param string $mine
3136  * @param string $yours
3137  * @param string $result
3138  * @return bool
3139  */
3140 function wfMerge( $old, $mine, $yours, &$result ) {
3141         global $wgDiff3;
3142
3143         # This check may also protect against code injection in
3144         # case of broken installations.
3145         wfSuppressWarnings();
3146         $haveDiff3 = $wgDiff3 && file_exists( $wgDiff3 );
3147         wfRestoreWarnings();
3148
3149         if ( !$haveDiff3 ) {
3150                 wfDebug( "diff3 not found\n" );
3151                 return false;
3152         }
3153
3154         # Make temporary files
3155         $td = wfTempDir();
3156         $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
3157         $mytextFile = fopen( $mytextName = tempnam( $td, 'merge-mine-' ), 'w' );
3158         $yourtextFile = fopen( $yourtextName = tempnam( $td, 'merge-your-' ), 'w' );
3159
3160         # NOTE: diff3 issues a warning to stderr if any of the files does not end with
3161         #       a newline character. To avoid this, we normalize the trailing whitespace before
3162         #       creating the diff.
3163
3164         fwrite( $oldtextFile, rtrim( $old ) . "\n" );
3165         fclose( $oldtextFile );
3166         fwrite( $mytextFile, rtrim( $mine ) . "\n" );
3167         fclose( $mytextFile );
3168         fwrite( $yourtextFile, rtrim( $yours ) . "\n" );
3169         fclose( $yourtextFile );
3170
3171         # Check for a conflict
3172         $cmd = wfEscapeShellArg( $wgDiff3 ) . ' -a --overlap-only ' .
3173                 wfEscapeShellArg( $mytextName ) . ' ' .
3174                 wfEscapeShellArg( $oldtextName ) . ' ' .
3175                 wfEscapeShellArg( $yourtextName );
3176         $handle = popen( $cmd, 'r' );
3177
3178         if ( fgets( $handle, 1024 ) ) {
3179                 $conflict = true;
3180         } else {
3181                 $conflict = false;
3182         }
3183         pclose( $handle );
3184
3185         # Merge differences
3186         $cmd = wfEscapeShellArg( $wgDiff3 ) . ' -a -e --merge ' .
3187                 wfEscapeShellArg( $mytextName, $oldtextName, $yourtextName );
3188         $handle = popen( $cmd, 'r' );
3189         $result = '';
3190         do {
3191                 $data = fread( $handle, 8192 );
3192                 if ( strlen( $data ) == 0 ) {
3193                         break;
3194                 }
3195                 $result .= $data;
3196         } while ( true );
3197         pclose( $handle );
3198         unlink( $mytextName );
3199         unlink( $oldtextName );
3200         unlink( $yourtextName );
3201
3202         if ( $result === '' && $old !== '' && !$conflict ) {
3203                 wfDebug( "Unexpected null result from diff3. Command: $cmd\n" );
3204                 $conflict = true;
3205         }
3206         return !$conflict;
3207 }
3208
3209 /**
3210  * Returns unified plain-text diff of two texts.
3211  * Useful for machine processing of diffs.
3212  *
3213  * @param string $before The text before the changes.
3214  * @param string $after The text after the changes.
3215  * @param string $params Command-line options for the diff command.
3216  * @return string Unified diff of $before and $after
3217  */
3218 function wfDiff( $before, $after, $params = '-u' ) {
3219         if ( $before == $after ) {
3220                 return '';
3221         }
3222
3223         global $wgDiff;
3224         wfSuppressWarnings();
3225         $haveDiff = $wgDiff && file_exists( $wgDiff );
3226         wfRestoreWarnings();
3227
3228         # This check may also protect against code injection in
3229         # case of broken installations.
3230         if ( !$haveDiff ) {
3231                 wfDebug( "diff executable not found\n" );
3232                 $diffs = new Diff( explode( "\n", $before ), explode( "\n", $after ) );
3233                 $format = new UnifiedDiffFormatter();
3234                 return $format->format( $diffs );
3235         }
3236
3237         # Make temporary files
3238         $td = wfTempDir();
3239         $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
3240         $newtextFile = fopen( $newtextName = tempnam( $td, 'merge-your-' ), 'w' );
3241
3242         fwrite( $oldtextFile, $before );
3243         fclose( $oldtextFile );
3244         fwrite( $newtextFile, $after );
3245         fclose( $newtextFile );
3246
3247         // Get the diff of the two files
3248         $cmd = "$wgDiff " . $params . ' ' . wfEscapeShellArg( $oldtextName, $newtextName );
3249
3250         $h = popen( $cmd, 'r' );
3251
3252         $diff = '';
3253
3254         do {
3255                 $data = fread( $h, 8192 );
3256                 if ( strlen( $data ) == 0 ) {
3257                         break;
3258                 }
3259                 $diff .= $data;
3260         } while ( true );
3261
3262         // Clean up
3263         pclose( $h );
3264         unlink( $oldtextName );
3265         unlink( $newtextName );
3266
3267         // Kill the --- and +++ lines. They're not useful.
3268         $diff_lines = explode( "\n", $diff );
3269         if ( strpos( $diff_lines[0], '---' ) === 0 ) {
3270                 unset( $diff_lines[0] );
3271         }
3272         if ( strpos( $diff_lines[1], '+++' ) === 0 ) {
3273                 unset( $diff_lines[1] );
3274         }
3275
3276         $diff = implode( "\n", $diff_lines );
3277
3278         return $diff;
3279 }
3280
3281 /**
3282  * This function works like "use VERSION" in Perl, the program will die with a
3283  * backtrace if the current version of PHP is less than the version provided
3284  *
3285  * This is useful for extensions which due to their nature are not kept in sync
3286  * with releases, and might depend on other versions of PHP than the main code
3287  *
3288  * Note: PHP might die due to parsing errors in some cases before it ever
3289  *       manages to call this function, such is life
3290  *
3291  * @see perldoc -f use
3292  *
3293  * @param string|int|float $req_ver The version to check, can be a string, an integer, or a float
3294  * @throws MWException
3295  */
3296 function wfUsePHP( $req_ver ) {
3297         $php_ver = PHP_VERSION;
3298
3299         if ( version_compare( $php_ver, (string)$req_ver, '<' ) ) {
3300                 throw new MWException( "PHP $req_ver required--this is only $php_ver" );
3301         }
3302 }
3303
3304 /**
3305  * This function works like "use VERSION" in Perl except it checks the version
3306  * of MediaWiki, the program will die with a backtrace if the current version
3307  * of MediaWiki is less than the version provided.
3308  *
3309  * This is useful for extensions which due to their nature are not kept in sync
3310  * with releases
3311  *
3312  * Note: Due to the behavior of PHP's version_compare() which is used in this
3313  * function, if you want to allow the 'wmf' development versions add a 'c' (or
3314  * any single letter other than 'a', 'b' or 'p') as a post-fix to your
3315  * targeted version number. For example if you wanted to allow any variation
3316  * of 1.22 use `wfUseMW( '1.22c' )`. Using an 'a' or 'b' instead of 'c' will
3317  * not result in the same comparison due to the internal logic of
3318  * version_compare().
3319  *
3320  * @see perldoc -f use
3321  *
3322  * @param string|int|float $req_ver The version to check, can be a string, an integer, or a float
3323  * @throws MWException
3324  */
3325 function wfUseMW( $req_ver ) {
3326         global $wgVersion;
3327
3328         if ( version_compare( $wgVersion, (string)$req_ver, '<' ) ) {
3329                 throw new MWException( "MediaWiki $req_ver required--this is only $wgVersion" );
3330         }
3331 }
3332
3333 /**
3334  * Return the final portion of a pathname.
3335  * Reimplemented because PHP5's "basename()" is buggy with multibyte text.
3336  * http://bugs.php.net/bug.php?id=33898
3337  *
3338  * PHP's basename() only considers '\' a pathchar on Windows and Netware.
3339  * We'll consider it so always, as we don't want '\s' in our Unix paths either.
3340  *
3341  * @param string $path
3342  * @param string $suffix String to remove if present
3343  * @return string
3344  */
3345 function wfBaseName( $path, $suffix = '' ) {
3346         if ( $suffix == '' ) {
3347                 $encSuffix = '';
3348         } else {
3349                 $encSuffix = '(?:' . preg_quote( $suffix, '#' ) . ')?';
3350         }
3351
3352         $matches = array();
3353         if ( preg_match( "#([^/\\\\]*?){$encSuffix}[/\\\\]*$#", $path, $matches ) ) {
3354                 return $matches[1];
3355         } else {
3356                 return '';
3357         }
3358 }
3359
3360 /**
3361  * Generate a relative path name to the given file.
3362  * May explode on non-matching case-insensitive paths,
3363  * funky symlinks, etc.
3364  *
3365  * @param string $path Absolute destination path including target filename
3366  * @param string $from Absolute source path, directory only
3367  * @return string
3368  */
3369 function wfRelativePath( $path, $from ) {
3370         // Normalize mixed input on Windows...
3371         $path = str_replace( '/', DIRECTORY_SEPARATOR, $path );
3372         $from = str_replace( '/', DIRECTORY_SEPARATOR, $from );
3373
3374         // Trim trailing slashes -- fix for drive root
3375         $path = rtrim( $path, DIRECTORY_SEPARATOR );
3376         $from = rtrim( $from, DIRECTORY_SEPARATOR );
3377
3378         $pieces = explode( DIRECTORY_SEPARATOR, dirname( $path ) );
3379         $against = explode( DIRECTORY_SEPARATOR, $from );
3380
3381         if ( $pieces[0] !== $against[0] ) {
3382                 // Non-matching Windows drive letters?
3383                 // Return a full path.
3384                 return $path;
3385         }
3386
3387         // Trim off common prefix
3388         while ( count( $pieces ) && count( $against )
3389                 && $pieces[0] == $against[0] ) {
3390                 array_shift( $pieces );
3391                 array_shift( $against );
3392         }
3393
3394         // relative dots to bump us to the parent
3395         while ( count( $against ) ) {
3396                 array_unshift( $pieces, '..' );
3397                 array_shift( $against );
3398         }
3399
3400         array_push( $pieces, wfBaseName( $path ) );
3401
3402         return implode( DIRECTORY_SEPARATOR, $pieces );
3403 }
3404
3405 /**
3406  * Convert an arbitrarily-long digit string from one numeric base
3407  * to another, optionally zero-padding to a minimum column width.
3408  *
3409  * Supports base 2 through 36; digit values 10-36 are represented
3410  * as lowercase letters a-z. Input is case-insensitive.
3411  *
3412  * @param string $input Input number
3413  * @param int $sourceBase Base of the input number
3414  * @param int $destBase Desired base of the output
3415  * @param int $pad Minimum number of digits in the output (pad with zeroes)
3416  * @param bool $lowercase Whether to output in lowercase or uppercase
3417  * @param string $engine Either "gmp", "bcmath", or "php"
3418  * @return string|bool The output number as a string, or false on error
3419  */
3420 function wfBaseConvert( $input, $sourceBase, $destBase, $pad = 1,
3421         $lowercase = true, $engine = 'auto'
3422 ) {
3423         $input = (string)$input;
3424         if (
3425                 $sourceBase < 2 ||
3426                 $sourceBase > 36 ||
3427                 $destBase < 2 ||
3428                 $destBase > 36 ||
3429                 $sourceBase != (int)$sourceBase ||
3430                 $destBase != (int)$destBase ||
3431                 $pad != (int)$pad ||
3432                 !preg_match(
3433                         "/^[" . substr( '0123456789abcdefghijklmnopqrstuvwxyz', 0, $sourceBase ) . "]+$/i",
3434                         $input
3435                 )
3436         ) {
3437                 return false;
3438         }
3439
3440         static $baseChars = array(
3441                 10 => 'a', 11 => 'b', 12 => 'c', 13 => 'd', 14 => 'e', 15 => 'f',
3442                 16 => 'g', 17 => 'h', 18 => 'i', 19 => 'j', 20 => 'k', 21 => 'l',
3443                 22 => 'm', 23 => 'n', 24 => 'o', 25 => 'p', 26 => 'q', 27 => 'r',
3444                 28 => 's', 29 => 't', 30 => 'u', 31 => 'v', 32 => 'w', 33 => 'x',
3445                 34 => 'y', 35 => 'z',
3446
3447                 '0' => 0, '1' => 1, '2' => 2, '3' => 3, '4' => 4, '5' => 5,
3448                 '6' => 6, '7' => 7, '8' => 8, '9' => 9, 'a' => 10, 'b' => 11,
3449                 'c' => 12, 'd' => 13, 'e' => 14, 'f' => 15, 'g' => 16, 'h' => 17,
3450                 'i' => 18, 'j' => 19, 'k' => 20, 'l' => 21, 'm' => 22, 'n' => 23,
3451                 'o' => 24, 'p' => 25, 'q' => 26, 'r' => 27, 's' => 28, 't' => 29,
3452                 'u' => 30, 'v' => 31, 'w' => 32, 'x' => 33, 'y' => 34, 'z' => 35
3453         );
3454
3455         if ( extension_loaded( 'gmp' ) && ( $engine == 'auto' || $engine == 'gmp' ) ) {
3456                 $result = gmp_strval( gmp_init( $input, $sourceBase ), $destBase );
3457         } elseif ( extension_loaded( 'bcmath' ) && ( $engine == 'auto' || $engine == 'bcmath' ) ) {
3458                 $decimal = '0';
3459                 foreach ( str_split( strtolower( $input ) ) as $char ) {
3460                         $decimal = bcmul( $decimal, $sourceBase );
3461                         $decimal = bcadd( $decimal, $baseChars[$char] );
3462                 }
3463
3464                 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3465                 for ( $result = ''; bccomp( $decimal, 0 ); $decimal = bcdiv( $decimal, $destBase, 0 ) ) {
3466                         $result .= $baseChars[bcmod( $decimal, $destBase )];
3467                 }
3468                 // @codingStandardsIgnoreEnd
3469
3470                 $result = strrev( $result );
3471         } else {
3472                 $inDigits = array();
3473                 foreach ( str_split( strtolower( $input ) ) as $char ) {
3474                         $inDigits[] = $baseChars[$char];
3475                 }
3476
3477                 // Iterate over the input, modulo-ing out an output digit
3478                 // at a time until input is gone.
3479                 $result = '';
3480                 while ( $inDigits ) {
3481                         $work = 0;
3482                         $workDigits = array();
3483
3484                         // Long division...
3485                         foreach ( $inDigits as $digit ) {
3486                                 $work *= $sourceBase;
3487                                 $work += $digit;
3488
3489                                 if ( $workDigits || $work >= $destBase ) {
3490                                         $workDigits[] = (int)( $work / $destBase );
3491                                 }
3492                                 $work %= $destBase;
3493                         }
3494
3495                         // All that division leaves us with a remainder,
3496                         // which is conveniently our next output digit.
3497                         $result .= $baseChars[$work];
3498
3499                         // And we continue!
3500                         $inDigits = $workDigits;
3501                 }
3502
3503                 $result = strrev( $result );
3504         }
3505
3506         if ( !$lowercase ) {
3507                 $result = strtoupper( $result );
3508         }
3509
3510         return str_pad( $result, $pad, '0', STR_PAD_LEFT );
3511 }
3512
3513 /**
3514  * Check if there is sufficient entropy in php's built-in session generation
3515  *
3516  * @return bool true = there is sufficient entropy
3517  */
3518 function wfCheckEntropy() {
3519         return (
3520                         ( wfIsWindows() && version_compare( PHP_VERSION, '5.3.3', '>=' ) )
3521                         || ini_get( 'session.entropy_file' )
3522                 )
3523                 && intval( ini_get( 'session.entropy_length' ) ) >= 32;
3524 }
3525
3526 /**
3527  * Override session_id before session startup if php's built-in
3528  * session generation code is not secure.
3529  */
3530 function wfFixSessionID() {
3531         // If the cookie or session id is already set we already have a session and should abort
3532         if ( isset( $_COOKIE[session_name()] ) || session_id() ) {
3533                 return;
3534         }
3535
3536         // PHP's built-in session entropy is enabled if:
3537         // - entropy_file is set or you're on Windows with php 5.3.3+
3538         // - AND entropy_length is > 0
3539         // We treat it as disabled if it doesn't have an entropy length of at least 32
3540         $entropyEnabled = wfCheckEntropy();
3541
3542         // If built-in entropy is not enabled or not sufficient override PHP's
3543         // built in session id generation code
3544         if ( !$entropyEnabled ) {
3545                 wfDebug( __METHOD__ . ": PHP's built in entropy is disabled or not sufficient, " .
3546                         "overriding session id generation using our cryptrand source.\n" );
3547                 session_id( MWCryptRand::generateHex( 32 ) );
3548         }
3549 }
3550
3551 /**
3552  * Reset the session_id
3553  *
3554  * @since 1.22
3555  */
3556 function wfResetSessionID() {
3557         global $wgCookieSecure;
3558         $oldSessionId = session_id();
3559         $cookieParams = session_get_cookie_params();
3560         if ( wfCheckEntropy() && $wgCookieSecure == $cookieParams['secure'] ) {
3561                 session_regenerate_id( false );
3562         } else {
3563                 $tmp = $_SESSION;
3564                 session_destroy();
3565                 wfSetupSession( MWCryptRand::generateHex( 32 ) );
3566                 $_SESSION = $tmp;
3567         }
3568         $newSessionId = session_id();
3569         wfRunHooks( 'ResetSessionID', array( $oldSessionId, $newSessionId ) );
3570 }
3571
3572 /**
3573  * Initialise php session
3574  *
3575  * @param bool $sessionId
3576  */
3577 function wfSetupSession( $sessionId = false ) {
3578         global $wgSessionsInMemcached, $wgSessionsInObjectCache, $wgCookiePath, $wgCookieDomain,
3579                         $wgCookieSecure, $wgCookieHttpOnly, $wgSessionHandler;
3580         if ( $wgSessionsInObjectCache || $wgSessionsInMemcached ) {
3581                 ObjectCacheSessionHandler::install();
3582         } elseif ( $wgSessionHandler && $wgSessionHandler != ini_get( 'session.save_handler' ) ) {
3583                 # Only set this if $wgSessionHandler isn't null and session.save_handler
3584                 # hasn't already been set to the desired value (that causes errors)
3585                 ini_set( 'session.save_handler', $wgSessionHandler );
3586         }
3587         session_set_cookie_params(
3588                 0, $wgCookiePath, $wgCookieDomain, $wgCookieSecure, $wgCookieHttpOnly );
3589         session_cache_limiter( 'private, must-revalidate' );
3590         if ( $sessionId ) {
3591                 session_id( $sessionId );
3592         } else {
3593                 wfFixSessionID();
3594         }
3595         wfSuppressWarnings();
3596         session_start();
3597         wfRestoreWarnings();
3598 }
3599
3600 /**
3601  * Get an object from the precompiled serialized directory
3602  *
3603  * @param string $name
3604  * @return mixed The variable on success, false on failure
3605  */
3606 function wfGetPrecompiledData( $name ) {
3607         global $IP;
3608
3609         $file = "$IP/serialized/$name";
3610         if ( file_exists( $file ) ) {
3611                 $blob = file_get_contents( $file );
3612                 if ( $blob ) {
3613                         return unserialize( $blob );
3614                 }
3615         }
3616         return false;
3617 }
3618
3619 /**
3620  * Get a cache key
3621  *
3622  * @param string [$args,...]
3623  * @return string
3624  */
3625 function wfMemcKey( /*...*/ ) {
3626         global $wgCachePrefix;
3627         $prefix = $wgCachePrefix === false ? wfWikiID() : $wgCachePrefix;
3628         $args = func_get_args();
3629         $key = $prefix . ':' . implode( ':', $args );
3630         $key = str_replace( ' ', '_', $key );
3631         return $key;
3632 }
3633
3634 /**
3635  * Get a cache key for a foreign DB
3636  *
3637  * @param string $db
3638  * @param string $prefix
3639  * @param string [$args,...]
3640  * @return string
3641  */
3642 function wfForeignMemcKey( $db, $prefix /*...*/ ) {
3643         $args = array_slice( func_get_args(), 2 );
3644         if ( $prefix ) {
3645                 $key = "$db-$prefix:" . implode( ':', $args );
3646         } else {
3647                 $key = $db . ':' . implode( ':', $args );
3648         }
3649         return str_replace( ' ', '_', $key );
3650 }
3651
3652 /**
3653  * Get an ASCII string identifying this wiki
3654  * This is used as a prefix in memcached keys
3655  *
3656  * @return string
3657  */
3658 function wfWikiID() {
3659         global $wgDBprefix, $wgDBname;
3660         if ( $wgDBprefix ) {
3661                 return "$wgDBname-$wgDBprefix";
3662         } else {
3663                 return $wgDBname;
3664         }
3665 }
3666
3667 /**
3668  * Split a wiki ID into DB name and table prefix
3669  *
3670  * @param string $wiki
3671  *
3672  * @return array
3673  */
3674 function wfSplitWikiID( $wiki ) {
3675         $bits = explode( '-', $wiki, 2 );
3676         if ( count( $bits ) < 2 ) {
3677                 $bits[] = '';
3678         }
3679         return $bits;
3680 }
3681
3682 /**
3683  * Get a Database object.
3684  *
3685  * @param int $db Index of the connection to get. May be DB_MASTER for the
3686  *            master (for write queries), DB_SLAVE for potentially lagged read
3687  *            queries, or an integer >= 0 for a particular server.
3688  *
3689  * @param string|string[] $groups Query groups. An array of group names that this query
3690  *                belongs to. May contain a single string if the query is only
3691  *                in one group.
3692  *
3693  * @param string|bool $wiki The wiki ID, or false for the current wiki
3694  *
3695  * Note: multiple calls to wfGetDB(DB_SLAVE) during the course of one request
3696  * will always return the same object, unless the underlying connection or load
3697  * balancer is manually destroyed.
3698  *
3699  * Note 2: use $this->getDB() in maintenance scripts that may be invoked by
3700  * updater to ensure that a proper database is being updated.
3701  *
3702  * @return DatabaseBase
3703  */
3704 function &wfGetDB( $db, $groups = array(), $wiki = false ) {
3705         return wfGetLB( $wiki )->getConnection( $db, $groups, $wiki );
3706 }
3707
3708 /**
3709  * Get a load balancer object.
3710  *
3711  * @param string|bool $wiki wiki ID, or false for the current wiki
3712  * @return LoadBalancer
3713  */
3714 function wfGetLB( $wiki = false ) {
3715         return wfGetLBFactory()->getMainLB( $wiki );
3716 }
3717
3718 /**
3719  * Get the load balancer factory object
3720  *
3721  * @return LBFactory
3722  */
3723 function &wfGetLBFactory() {
3724         return LBFactory::singleton();
3725 }
3726
3727 /**
3728  * Find a file.
3729  * Shortcut for RepoGroup::singleton()->findFile()
3730  *
3731  * @param string $title or Title object
3732  * @param array $options Associative array of options:
3733  *     time:           requested time for an archived image, or false for the
3734  *                     current version. An image object will be returned which was
3735  *                     created at the specified time.
3736  *
3737  *     ignoreRedirect: If true, do not follow file redirects
3738  *
3739  *     private:        If true, return restricted (deleted) files if the current
3740  *                     user is allowed to view them. Otherwise, such files will not
3741  *                     be found.
3742  *
3743  *     bypassCache:    If true, do not use the process-local cache of File objects
3744  *
3745  * @return File|bool File, or false if the file does not exist
3746  */
3747 function wfFindFile( $title, $options = array() ) {
3748         return RepoGroup::singleton()->findFile( $title, $options );
3749 }
3750
3751 /**
3752  * Get an object referring to a locally registered file.
3753  * Returns a valid placeholder object if the file does not exist.
3754  *
3755  * @param Title|string $title
3756  * @return LocalFile|null A File, or null if passed an invalid Title
3757  */
3758 function wfLocalFile( $title ) {
3759         return RepoGroup::singleton()->getLocalRepo()->newFile( $title );
3760 }
3761
3762 /**
3763  * Should low-performance queries be disabled?
3764  *
3765  * @return bool
3766  * @codeCoverageIgnore
3767  */
3768 function wfQueriesMustScale() {
3769         global $wgMiserMode;
3770         return $wgMiserMode
3771                 || ( SiteStats::pages() > 100000
3772                 && SiteStats::edits() > 1000000
3773                 && SiteStats::users() > 10000 );
3774 }
3775
3776 /**
3777  * Get the path to a specified script file, respecting file
3778  * extensions; this is a wrapper around $wgScriptExtension etc.
3779  * except for 'index' and 'load' which use $wgScript/$wgLoadScript
3780  *
3781  * @param string $script Script filename, sans extension
3782  * @return string
3783  */
3784 function wfScript( $script = 'index' ) {
3785         global $wgScriptPath, $wgScriptExtension, $wgScript, $wgLoadScript;
3786         if ( $script === 'index' ) {
3787                 return $wgScript;
3788         } elseif ( $script === 'load' ) {
3789                 return $wgLoadScript;
3790         } else {
3791                 return "{$wgScriptPath}/{$script}{$wgScriptExtension}";
3792         }
3793 }
3794
3795 /**
3796  * Get the script URL.
3797  *
3798  * @return string script URL
3799  */
3800 function wfGetScriptUrl() {
3801         if ( isset( $_SERVER['SCRIPT_NAME'] ) ) {
3802                 #
3803                 # as it was called, minus the query string.
3804                 #
3805                 # Some sites use Apache rewrite rules to handle subdomains,
3806                 # and have PHP set up in a weird way that causes PHP_SELF
3807                 # to contain the rewritten URL instead of the one that the
3808                 # outside world sees.
3809                 #
3810                 # If in this mode, use SCRIPT_URL instead, which mod_rewrite
3811                 # provides containing the "before" URL.
3812                 return $_SERVER['SCRIPT_NAME'];
3813         } else {
3814                 return $_SERVER['URL'];
3815         }
3816 }
3817
3818 /**
3819  * Convenience function converts boolean values into "true"
3820  * or "false" (string) values
3821  *
3822  * @param bool $value
3823  * @return string
3824  */
3825 function wfBoolToStr( $value ) {
3826         return $value ? 'true' : 'false';
3827 }
3828
3829 /**
3830  * Get a platform-independent path to the null file, e.g. /dev/null
3831  *
3832  * @return string
3833  */
3834 function wfGetNull() {
3835         return wfIsWindows() ? 'NUL' : '/dev/null';
3836 }
3837
3838 /**
3839  * Modern version of wfWaitForSlaves(). Instead of looking at replication lag
3840  * and waiting for it to go down, this waits for the slaves to catch up to the
3841  * master position. Use this when updating very large numbers of rows, as
3842  * in maintenance scripts, to avoid causing too much lag.  Of course, this is
3843  * a no-op if there are no slaves.
3844  *
3845  * @param int|bool $maxLag (deprecated)
3846  * @param string|bool $wiki Wiki identifier accepted by wfGetLB
3847  * @param string|bool $cluster Cluster name accepted by LBFactory. Default: false.
3848  */
3849 function wfWaitForSlaves( $maxLag = false, $wiki = false, $cluster = false ) {
3850         if ( $cluster !== false ) {
3851                 $lb = wfGetLBFactory()->getExternalLB( $cluster );
3852         } else {
3853                 $lb = wfGetLB( $wiki );
3854         }
3855
3856         // bug 27975 - Don't try to wait for slaves if there are none
3857         // Prevents permission error when getting master position
3858         if ( $lb->getServerCount() > 1 ) {
3859                 $dbw = $lb->getConnection( DB_MASTER, array(), $wiki );
3860                 $pos = $dbw->getMasterPos();
3861                 // The DBMS may not support getMasterPos() or the whole
3862                 // load balancer might be fake (e.g. $wgAllDBsAreLocalhost).
3863                 if ( $pos !== false ) {
3864                         $lb->waitForAll( $pos );
3865                 }
3866         }
3867 }
3868
3869 /**
3870  * Count down from $seconds to zero on the terminal, with a one-second pause
3871  * between showing each number. For use in command-line scripts.
3872  *
3873  * @codeCoverageIgnore
3874  * @param int $seconds
3875  */
3876 function wfCountDown( $seconds ) {
3877         for ( $i = $seconds; $i >= 0; $i-- ) {
3878                 if ( $i != $seconds ) {
3879                         echo str_repeat( "\x08", strlen( $i + 1 ) );
3880                 }
3881                 echo $i;
3882                 flush();
3883                 if ( $i ) {
3884                         sleep( 1 );
3885                 }
3886         }
3887         echo "\n";
3888 }
3889
3890 /**
3891  * Replace all invalid characters with -
3892  * Additional characters can be defined in $wgIllegalFileChars (see bug 20489)
3893  * By default, $wgIllegalFileChars = ':'
3894  *
3895  * @param string $name Filename to process
3896  * @return string
3897  */
3898 function wfStripIllegalFilenameChars( $name ) {
3899         global $wgIllegalFileChars;
3900         $illegalFileChars = $wgIllegalFileChars ? "|[" . $wgIllegalFileChars . "]" : '';
3901         $name = wfBaseName( $name );
3902         $name = preg_replace(
3903                 "/[^" . Title::legalChars() . "]" . $illegalFileChars . "/",
3904                 '-',
3905                 $name
3906         );
3907         return $name;
3908 }
3909
3910 /**
3911  * Set PHP's memory limit to the larger of php.ini or $wgMemoryLimit;
3912  *
3913  * @return int Value the memory limit was set to.
3914  */
3915 function wfMemoryLimit() {
3916         global $wgMemoryLimit;
3917         $memlimit = wfShorthandToInteger( ini_get( 'memory_limit' ) );
3918         if ( $memlimit != -1 ) {
3919                 $conflimit = wfShorthandToInteger( $wgMemoryLimit );
3920                 if ( $conflimit == -1 ) {
3921                         wfDebug( "Removing PHP's memory limit\n" );
3922                         wfSuppressWarnings();
3923                         ini_set( 'memory_limit', $conflimit );
3924                         wfRestoreWarnings();
3925                         return $conflimit;
3926                 } elseif ( $conflimit > $memlimit ) {
3927                         wfDebug( "Raising PHP's memory limit to $conflimit bytes\n" );
3928                         wfSuppressWarnings();
3929                         ini_set( 'memory_limit', $conflimit );
3930                         wfRestoreWarnings();
3931                         return $conflimit;
3932                 }
3933         }
3934         return $memlimit;
3935 }
3936
3937 /**
3938  * Converts shorthand byte notation to integer form
3939  *
3940  * @param string $string
3941  * @return int
3942  */
3943 function wfShorthandToInteger( $string = '' ) {
3944         $string = trim( $string );
3945         if ( $string === '' ) {
3946                 return -1;
3947         }
3948         $last = $string[strlen( $string ) - 1];
3949         $val = intval( $string );
3950         switch ( $last ) {
3951                 case 'g':
3952                 case 'G':
3953                         $val *= 1024;
3954                         // break intentionally missing
3955                 case 'm':
3956                 case 'M':
3957                         $val *= 1024;
3958                         // break intentionally missing
3959                 case 'k':
3960                 case 'K':
3961                         $val *= 1024;
3962         }
3963
3964         return $val;
3965 }
3966
3967 /**
3968  * Get the normalised IETF language tag
3969  * See unit test for examples.
3970  *
3971  * @param string $code The language code.
3972  * @return string The language code which complying with BCP 47 standards.
3973  */
3974 function wfBCP47( $code ) {
3975         $codeSegment = explode( '-', $code );
3976         $codeBCP = array();
3977         foreach ( $codeSegment as $segNo => $seg ) {
3978                 // when previous segment is x, it is a private segment and should be lc
3979                 if ( $segNo > 0 && strtolower( $codeSegment[( $segNo - 1 )] ) == 'x' ) {
3980                         $codeBCP[$segNo] = strtolower( $seg );
3981                 // ISO 3166 country code
3982                 } elseif ( ( strlen( $seg ) == 2 ) && ( $segNo > 0 ) ) {
3983                         $codeBCP[$segNo] = strtoupper( $seg );
3984                 // ISO 15924 script code
3985                 } elseif ( ( strlen( $seg ) == 4 ) && ( $segNo > 0 ) ) {
3986                         $codeBCP[$segNo] = ucfirst( strtolower( $seg ) );
3987                 // Use lowercase for other cases
3988                 } else {
3989                         $codeBCP[$segNo] = strtolower( $seg );
3990                 }
3991         }
3992         $langCode = implode( '-', $codeBCP );
3993         return $langCode;
3994 }
3995
3996 /**
3997  * Get a cache object.
3998  *
3999  * @param int $inputType Cache type, one the the CACHE_* constants.
4000  * @return BagOStuff
4001  */
4002 function wfGetCache( $inputType ) {
4003         return ObjectCache::getInstance( $inputType );
4004 }
4005
4006 /**
4007  * Get the main cache object
4008  *
4009  * @return BagOStuff
4010  */
4011 function wfGetMainCache() {
4012         global $wgMainCacheType;
4013         return ObjectCache::getInstance( $wgMainCacheType );
4014 }
4015
4016 /**
4017  * Get the cache object used by the message cache
4018  *
4019  * @return BagOStuff
4020  */
4021 function wfGetMessageCacheStorage() {
4022         global $wgMessageCacheType;
4023         return ObjectCache::getInstance( $wgMessageCacheType );
4024 }
4025
4026 /**
4027  * Get the cache object used by the parser cache
4028  *
4029  * @return BagOStuff
4030  */
4031 function wfGetParserCacheStorage() {
4032         global $wgParserCacheType;
4033         return ObjectCache::getInstance( $wgParserCacheType );
4034 }
4035
4036 /**
4037  * Get the cache object used by the language converter
4038  *
4039  * @return BagOStuff
4040  */
4041 function wfGetLangConverterCacheStorage() {
4042         global $wgLanguageConverterCacheType;
4043         return ObjectCache::getInstance( $wgLanguageConverterCacheType );
4044 }
4045
4046 /**
4047  * Call hook functions defined in $wgHooks
4048  *
4049  * @param string $event Event name
4050  * @param array $args Parameters passed to hook functions
4051  * @param string|null $deprecatedVersion Optionally mark hook as deprecated with version number
4052  *
4053  * @return bool True if no handler aborted the hook
4054  */
4055 function wfRunHooks( $event, array $args = array(), $deprecatedVersion = null ) {
4056         return Hooks::run( $event, $args, $deprecatedVersion );
4057 }
4058
4059 /**
4060  * Wrapper around php's unpack.
4061  *
4062  * @param string $format The format string (See php's docs)
4063  * @param string $data A binary string of binary data
4064  * @param int|bool $length The minimum length of $data or false. This is to
4065  *      prevent reading beyond the end of $data. false to disable the check.
4066  *
4067  * Also be careful when using this function to read unsigned 32 bit integer
4068  * because php might make it negative.
4069  *
4070  * @throws MWException if $data not long enough, or if unpack fails
4071  * @return array Associative array of the extracted data
4072  */
4073 function wfUnpack( $format, $data, $length = false ) {
4074         if ( $length !== false ) {
4075                 $realLen = strlen( $data );
4076                 if ( $realLen < $length ) {
4077                         throw new MWException( "Tried to use wfUnpack on a "
4078                                 . "string of length $realLen, but needed one "
4079                                 . "of at least length $length."
4080                         );
4081                 }
4082         }
4083
4084         wfSuppressWarnings();
4085         $result = unpack( $format, $data );
4086         wfRestoreWarnings();
4087
4088         if ( $result === false ) {
4089                 // If it cannot extract the packed data.
4090                 throw new MWException( "unpack could not unpack binary data" );
4091         }
4092         return $result;
4093 }
4094
4095 /**
4096  * Determine if an image exists on the 'bad image list'.
4097  *
4098  * The format of MediaWiki:Bad_image_list is as follows:
4099  *    * Only list items (lines starting with "*") are considered
4100  *    * The first link on a line must be a link to a bad image
4101  *    * Any subsequent links on the same line are considered to be exceptions,
4102  *      i.e. articles where the image may occur inline.
4103  *
4104  * @param string $name The image name to check
4105  * @param Title|bool $contextTitle The page on which the image occurs, if known
4106  * @param string $blacklist Wikitext of a file blacklist
4107  * @return bool
4108  */
4109 function wfIsBadImage( $name, $contextTitle = false, $blacklist = null ) {
4110         static $badImageCache = null; // based on bad_image_list msg
4111         wfProfileIn( __METHOD__ );
4112
4113         # Handle redirects
4114         $redirectTitle = RepoGroup::singleton()->checkRedirect( Title::makeTitle( NS_FILE, $name ) );
4115         if ( $redirectTitle ) {
4116                 $name = $redirectTitle->getDBkey();
4117         }
4118
4119         # Run the extension hook
4120         $bad = false;
4121         if ( !wfRunHooks( 'BadImage', array( $name, &$bad ) ) ) {
4122                 wfProfileOut( __METHOD__ );
4123                 return $bad;
4124         }
4125
4126         $cacheable = ( $blacklist === null );
4127         if ( $cacheable && $badImageCache !== null ) {
4128                 $badImages = $badImageCache;
4129         } else { // cache miss
4130                 if ( $blacklist === null ) {
4131                         $blacklist = wfMessage( 'bad_image_list' )->inContentLanguage()->plain(); // site list
4132                 }
4133                 # Build the list now
4134                 $badImages = array();
4135                 $lines = explode( "\n", $blacklist );
4136                 foreach ( $lines as $line ) {
4137                         # List items only
4138                         if ( substr( $line, 0, 1 ) !== '*' ) {
4139                                 continue;
4140                         }
4141
4142                         # Find all links
4143                         $m = array();
4144                         if ( !preg_match_all( '/\[\[:?(.*?)\]\]/', $line, $m ) ) {
4145                                 continue;
4146                         }
4147
4148                         $exceptions = array();
4149                         $imageDBkey = false;
4150                         foreach ( $m[1] as $i => $titleText ) {
4151                                 $title = Title::newFromText( $titleText );
4152                                 if ( !is_null( $title ) ) {
4153                                         if ( $i == 0 ) {
4154                                                 $imageDBkey = $title->getDBkey();
4155                                         } else {
4156                                                 $exceptions[$title->getPrefixedDBkey()] = true;
4157                                         }
4158                                 }
4159                         }
4160
4161                         if ( $imageDBkey !== false ) {
4162                                 $badImages[$imageDBkey] = $exceptions;
4163                         }
4164                 }
4165                 if ( $cacheable ) {
4166                         $badImageCache = $badImages;
4167                 }
4168         }
4169
4170         $contextKey = $contextTitle ? $contextTitle->getPrefixedDBkey() : false;
4171         $bad = isset( $badImages[$name] ) && !isset( $badImages[$name][$contextKey] );
4172         wfProfileOut( __METHOD__ );
4173         return $bad;
4174 }
4175
4176 /**
4177  * Determine whether the client at a given source IP is likely to be able to
4178  * access the wiki via HTTPS.
4179  *
4180  * @param string $ip The IPv4/6 address in the normal human-readable form
4181  * @return bool
4182  */
4183 function wfCanIPUseHTTPS( $ip ) {
4184         $canDo = true;
4185         wfRunHooks( 'CanIPUseHTTPS', array( $ip, &$canDo ) );
4186         return !!$canDo;
4187 }
4188
4189 /**
4190  * Work out the IP address based on various globals
4191  * For trusted proxies, use the XFF client IP (first of the chain)
4192  *
4193  * @deprecated since 1.19; call $wgRequest->getIP() directly.
4194  * @return string
4195  */
4196 function wfGetIP() {
4197         wfDeprecated( __METHOD__, '1.19' );
4198         global $wgRequest;
4199         return $wgRequest->getIP();
4200 }
4201
4202 /**
4203  * Checks if an IP is a trusted proxy provider.
4204  * Useful to tell if X-Forwarded-For data is possibly bogus.
4205  * Squid cache servers for the site are whitelisted.
4206  * @deprecated Since 1.24, use IP::isTrustedProxy()
4207  *
4208  * @param string $ip
4209  * @return bool
4210  */
4211 function wfIsTrustedProxy( $ip ) {
4212         return IP::isTrustedProxy( $ip );
4213 }
4214
4215 /**
4216  * Checks if an IP matches a proxy we've configured.
4217  * @deprecated Since 1.24, use IP::isConfiguredProxy()
4218  *
4219  * @param string $ip
4220  * @return bool
4221  * @since 1.23 Supports CIDR ranges in $wgSquidServersNoPurge
4222  */
4223 function wfIsConfiguredProxy( $ip ) {
4224         return IP::isTrustedProxy( $ip );
4225 }