includes/GlobalFunctions.php

   1 <?php
   2 /**
   3  * Global functions used everywhere
   4  * @file
   5  */
   6
   7 if ( !defined( 'MEDIAWIKI' ) ) {
   8         die( "This file is part of MediaWiki, it is not a valid entry point" );
   9 }
  10
  11 // Hide compatibility functions from Doxygen
  12 /// @cond
  13
  14 /**
  15  * Compatibility functions
  16  *
  17  * We support PHP 5.2.3 and up.
  18  * Re-implementations of newer functions or functions in non-standard
  19  * PHP extensions may be included here.
  20  */
  21
  22 if( !function_exists( 'iconv' ) ) {
  23         /**
  24          * @codeCoverageIgnore
  25          * @return string
  26          */
  27         function iconv( $from, $to, $string ) {
  28                 return Fallback::iconv( $from, $to, $string );
  29         }
  30 }
  31
  32 if ( !function_exists( 'mb_substr' ) ) {
  33         /**
  34          * @codeCoverageIgnore
  35          * @return string
  36          */
  37         function mb_substr( $str, $start, $count='end' ) {
  38                 return Fallback::mb_substr( $str, $start, $count );
  39         }
  40
  41         /**
  42          * @codeCoverageIgnore
  43          * @return int
  44          */
  45         function mb_substr_split_unicode( $str, $splitPos ) {
  46                 return Fallback::mb_substr_split_unicode( $str, $splitPos );
  47         }
  48 }
  49
  50 if ( !function_exists( 'mb_strlen' ) ) {
  51         /**
  52          * @codeCoverageIgnore
  53          * @return int
  54          */
  55         function mb_strlen( $str, $enc = '' ) {
  56                 return Fallback::mb_strlen( $str, $enc );
  57         }
  58 }
  59
  60 if( !function_exists( 'mb_strpos' ) ) {
  61         /**
  62          * @codeCoverageIgnore
  63          * @return int
  64          */
  65         function mb_strpos( $haystack, $needle, $offset = 0, $encoding = '' ) {
  66                 return Fallback::mb_strpos( $haystack, $needle, $offset, $encoding );
  67         }
  68
  69 }
  70
  71 if( !function_exists( 'mb_strrpos' ) ) {
  72         /**
  73          * @codeCoverageIgnore
  74          * @return int
  75          */
  76         function mb_strrpos( $haystack, $needle, $offset = 0, $encoding = '' ) {
  77                 return Fallback::mb_strrpos( $haystack, $needle, $offset, $encoding );
  78         }
  79 }
  80
  81
  82 // Support for Wietse Venema's taint feature
  83 if ( !function_exists( 'istainted' ) ) {
  84         /**
  85          * @codeCoverageIgnore
  86          * @return int
  87          */
  88         function istainted( $var ) {
  89                 return 0;
  90         }
  91         /** @codeCoverageIgnore */
  92         function taint( $var, $level = 0 ) {}
  93         /** @codeCoverageIgnore */
  94         function untaint( $var, $level = 0 ) {}
  95         define( 'TC_HTML', 1 );
  96         define( 'TC_SHELL', 1 );
  97         define( 'TC_MYSQL', 1 );
  98         define( 'TC_PCRE', 1 );
  99         define( 'TC_SELF', 1 );
 100 }
 101 /// @endcond
 102
 103 /**
 104  * Like array_diff( $a, $b ) except that it works with two-dimensional arrays.
 105  * @param $a array
 106  * @param $b array
 107  * @return array
 108  */
 109 function wfArrayDiff2( $a, $b ) {
 110         return array_udiff( $a, $b, 'wfArrayDiff2_cmp' );
 111 }
 112
 113 /**
 114  * @param $a
 115  * @param $b
 116  * @return int
 117  */
 118 function wfArrayDiff2_cmp( $a, $b ) {
 119         if ( !is_array( $a ) ) {
 120                 return strcmp( $a, $b );
 121         } elseif ( count( $a ) !== count( $b ) ) {
 122                 return count( $a ) < count( $b ) ? -1 : 1;
 123         } else {
 124                 reset( $a );
 125                 reset( $b );
 126                 while( ( list( , $valueA ) = each( $a ) ) && ( list( , $valueB ) = each( $b ) ) ) {
 127                         $cmp = strcmp( $valueA, $valueB );
 128                         if ( $cmp !== 0 ) {
 129                                 return $cmp;
 130                         }
 131                 }
 132                 return 0;
 133         }
 134 }
 135
 136 /**
 137  * Array lookup
 138  * Returns an array where the values in the first array are replaced by the
 139  * values in the second array with the corresponding keys
 140  *
 141  * @param $a Array
 142  * @param $b Array
 143  * @return array
 144  */
 145 function wfArrayLookup( $a, $b ) {
 146         return array_flip( array_intersect( array_flip( $a ), array_keys( $b ) ) );
 147 }
 148
 149 /**
 150  * Appends to second array if $value differs from that in $default
 151  *
 152  * @param $key String|Int
 153  * @param $value Mixed
 154  * @param $default Mixed
 155  * @param $changed Array to alter
 156  */
 157 function wfAppendToArrayIfNotDefault( $key, $value, $default, &$changed ) {
 158         if ( is_null( $changed ) ) {
 159                 throw new MWException( 'GlobalFunctions::wfAppendToArrayIfNotDefault got null' );
 160         }
 161         if ( $default[$key] !== $value ) {
 162                 $changed[$key] = $value;
 163         }
 164 }
 165
 166 /**
 167  * Backwards array plus for people who haven't bothered to read the PHP manual
 168  * XXX: will not darn your socks for you.
 169  *
 170  * @param $array1 Array
 171  * @param [$array2, [...]] Arrays
 172  * @return Array
 173  */
 174 function wfArrayMerge( $array1/* ... */ ) {
 175         $args = func_get_args();
 176         $args = array_reverse( $args, true );
 177         $out = array();
 178         foreach ( $args as $arg ) {
 179                 $out += $arg;
 180         }
 181         return $out;
 182 }
 183
 184 /**
 185  * Merge arrays in the style of getUserPermissionsErrors, with duplicate removal
 186  * e.g.
 187  *      wfMergeErrorArrays(
 188  *              array( array( 'x' ) ),
 189  *              array( array( 'x', '2' ) ),
 190  *              array( array( 'x' ) ),
 191  *              array( array( 'y' ) )
 192  *      );
 193  * returns:
 194  *              array(
 195  *              array( 'x', '2' ),
 196  *              array( 'x' ),
 197  *              array( 'y' )
 198  *      )
 199  * @param varargs
 200  * @return Array
 201  */
 202 function wfMergeErrorArrays( /*...*/ ) {
 203         $args = func_get_args();
 204         $out = array();
 205         foreach ( $args as $errors ) {
 206                 foreach ( $errors as $params ) {
 207                         # @todo FIXME: Sometimes get nested arrays for $params,
 208                         # which leads to E_NOTICEs
 209                         $spec = implode( "\t", $params );
 210                         $out[$spec] = $params;
 211                 }
 212         }
 213         return array_values( $out );
 214 }
 215
 216 /**
 217  * Insert array into another array after the specified *KEY*
 218  *
 219  * @param $array Array: The array.
 220  * @param $insert Array: The array to insert.
 221  * @param $after Mixed: The key to insert after
 222  * @return Array
 223  */
 224 function wfArrayInsertAfter( $array, $insert, $after ) {
 225         // Find the offset of the element to insert after.
 226         $keys = array_keys( $array );
 227         $offsetByKey = array_flip( $keys );
 228
 229         $offset = $offsetByKey[$after];
 230
 231         // Insert at the specified offset
 232         $before = array_slice( $array, 0, $offset + 1, true );
 233         $after = array_slice( $array, $offset + 1, count( $array ) - $offset, true );
 234
 235         $output = $before + $insert + $after;
 236
 237         return $output;
 238 }
 239
 240 /**
 241  * Recursively converts the parameter (an object) to an array with the same data
 242  *
 243  * @param $objOrArray Object|Array
 244  * @param $recursive Bool
 245  * @return Array
 246  */
 247 function wfObjectToArray( $objOrArray, $recursive = true ) {
 248         $array = array();
 249         if( is_object( $objOrArray ) ) {
 250                 $objOrArray = get_object_vars( $objOrArray );
 251         }
 252         foreach ( $objOrArray as $key => $value ) {
 253                 if ( $recursive && ( is_object( $value ) || is_array( $value ) ) ) {
 254                         $value = wfObjectToArray( $value );
 255                 }
 256
 257                 $array[$key] = $value;
 258         }
 259
 260         return $array;
 261 }
 262
 263 /**
 264  * Wrapper around array_map() which also taints variables
 265  *
 266  * @param  $function Callback
 267  * @param  $input Array
 268  * @return Array
 269  */
 270 function wfArrayMap( $function, $input ) {
 271         $ret = array_map( $function, $input );
 272         foreach ( $ret as $key => $value ) {
 273                 $taint = istainted( $input[$key] );
 274                 if ( $taint ) {
 275                         taint( $ret[$key], $taint );
 276                 }
 277         }
 278         return $ret;
 279 }
 280
 281 /**
 282  * Get a random decimal value between 0 and 1, in a way
 283  * not likely to give duplicate values for any realistic
 284  * number of articles.
 285  *
 286  * @return string
 287  */
 288 function wfRandom() {
 289         # The maximum random value is "only" 2^31-1, so get two random
 290         # values to reduce the chance of dupes
 291         $max = mt_getrandmax() + 1;
 292         $rand = number_format( ( mt_rand() * $max + mt_rand() )
 293                 / $max / $max, 12, '.', '' );
 294         return $rand;
 295 }
 296
 297 /**
 298  * We want some things to be included as literal characters in our title URLs
 299  * for prettiness, which urlencode encodes by default.  According to RFC 1738,
 300  * all of the following should be safe:
 301  *
 302  * ;:@&=$-_.+!*'(),
 303  *
 304  * But + is not safe because it's used to indicate a space; &= are only safe in
 305  * paths and not in queries (and we don't distinguish here); ' seems kind of
 306  * scary; and urlencode() doesn't touch -_. to begin with.  Plus, although /
 307  * is reserved, we don't care.  So the list we unescape is:
 308  *
 309  * ;:@$!*(),/
 310  *
 311  * However, IIS7 redirects fail when the url contains a colon (Bug 22709),
 312  * so no fancy : for IIS7.
 313  *
 314  * %2F in the page titles seems to fatally break for some reason.
 315  *
 316  * @param $s String:
 317  * @return string
 318 */
 319 function wfUrlencode( $s ) {
 320         static $needle;
 321         if ( is_null( $s ) ) {
 322                 $needle = null;
 323                 return '';
 324         }
 325
 326         if ( is_null( $needle ) ) {
 327                 $needle = array( '%3B', '%40', '%24', '%21', '%2A', '%28', '%29', '%2C', '%2F' );
 328                 if ( !isset( $_SERVER['SERVER_SOFTWARE'] ) || ( strpos( $_SERVER['SERVER_SOFTWARE'], 'Microsoft-IIS/7' ) === false ) ) {
 329                         $needle[] = '%3A';
 330                 }
 331         }
 332
 333         $s = urlencode( $s );
 334         $s = str_ireplace(
 335                 $needle,
 336                 array( ';', '@', '$', '!', '*', '(', ')', ',', '/', ':' ),
 337                 $s
 338         );
 339
 340         return $s;
 341 }
 342
 343 /**
 344  * This function takes two arrays as input, and returns a CGI-style string, e.g.
 345  * "days=7&limit=100". Options in the first array override options in the second.
 346  * Options set to null or false will not be output.
 347  *
 348  * @param $array1 Array ( String|Array )
 349  * @param $array2 Array ( String|Array )
 350  * @param $prefix String
 351  * @return String
 352  */
 353 function wfArrayToCGI( $array1, $array2 = null, $prefix = '' ) {
 354         if ( !is_null( $array2 ) ) {
 355                 $array1 = $array1 + $array2;
 356         }
 357
 358         $cgi = '';
 359         foreach ( $array1 as $key => $value ) {
 360                 if ( !is_null($value) && $value !== false ) {
 361                         if ( $cgi != '' ) {
 362                                 $cgi .= '&';
 363                         }
 364                         if ( $prefix !== '' ) {
 365                                 $key = $prefix . "[$key]";
 366                         }
 367                         if ( is_array( $value ) ) {
 368                                 $firstTime = true;
 369                                 foreach ( $value as $k => $v ) {
 370                                         $cgi .= $firstTime ? '' : '&';
 371                                         if ( is_array( $v ) ) {
 372                                                 $cgi .= wfArrayToCGI( $v, null, $key . "[$k]" );
 373                                         } else {
 374                                                 $cgi .= urlencode( $key . "[$k]" ) . '=' . urlencode( $v );
 375                                         }
 376                                         $firstTime = false;
 377                                 }
 378                         } else {
 379                                 if ( is_object( $value ) ) {
 380                                         $value = $value->__toString();
 381                                 }
 382                                 $cgi .= urlencode( $key ) . '=' . urlencode( $value );
 383                         }
 384                 }
 385         }
 386         return $cgi;
 387 }
 388
 389 /**
 390  * This is the logical opposite of wfArrayToCGI(): it accepts a query string as
 391  * its argument and returns the same string in array form.  This allows compa-
 392  * tibility with legacy functions that accept raw query strings instead of nice
 393  * arrays.  Of course, keys and values are urldecode()d.
 394  *
 395  * @param $query String: query string
 396  * @return array Array version of input
 397  */
 398 function wfCgiToArray( $query ) {
 399         if ( isset( $query[0] ) && $query[0] == '?' ) {
 400                 $query = substr( $query, 1 );
 401         }
 402         $bits = explode( '&', $query );
 403         $ret = array();
 404         foreach ( $bits as $bit ) {
 405                 if ( $bit === '' ) {
 406                         continue;
 407                 }
 408                 if ( strpos( $bit, '=' ) === false ) {
 409                         // Pieces like &qwerty become 'qwerty' => '' (at least this is what php does)
 410                         $key = $bit;
 411                         $value = '';
 412                 } else {
 413                         list( $key, $value ) = explode( '=', $bit );
 414                 }
 415                 $key = urldecode( $key );
 416                 $value = urldecode( $value );
 417                 if ( strpos( $key, '[' ) !== false ) {
 418                         $keys = array_reverse( explode( '[', $key ) );
 419                         $key = array_pop( $keys );
 420                         $temp = $value;
 421                         foreach ( $keys as $k ) {
 422                                 $k = substr( $k, 0, -1 );
 423                                 $temp = array( $k => $temp );
 424                         }
 425                         if ( isset( $ret[$key] ) ) {
 426                                 $ret[$key] = array_merge( $ret[$key], $temp );
 427                         } else {
 428                                 $ret[$key] = $temp;
 429                         }
 430                 } else {
 431                         $ret[$key] = $value;
 432                 }
 433         }
 434         return $ret;
 435 }
 436
 437 /**
 438  * Append a query string to an existing URL, which may or may not already
 439  * have query string parameters already. If so, they will be combined.
 440  *
 441  * @param $url String
 442  * @param $query Mixed: string or associative array
 443  * @return string
 444  */
 445 function wfAppendQuery( $url, $query ) {
 446         if ( is_array( $query ) ) {
 447                 $query = wfArrayToCGI( $query );
 448         }
 449         if( $query != '' ) {
 450                 if( false === strpos( $url, '?' ) ) {
 451                         $url .= '?';
 452                 } else {
 453                         $url .= '&';
 454                 }
 455                 $url .= $query;
 456         }
 457         return $url;
 458 }
 459
 460 /**
 461  * Expand a potentially local URL to a fully-qualified URL.  Assumes $wgServer
 462  * is correct.
 463  *
 464  * The meaning of the PROTO_* constants is as follows:
 465  * PROTO_HTTP: Output a URL starting with http://
 466  * PROTO_HTTPS: Output a URL starting with https://
 467  * PROTO_RELATIVE: Output a URL starting with // (protocol-relative URL)
 468  * PROTO_CURRENT: Output a URL starting with either http:// or https:// , depending on which protocol was used for the current incoming request
 469  * PROTO_CANONICAL: For URLs without a domain, like /w/index.php , use $wgCanonicalServer. For protocol-relative URLs, use the protocol of $wgCanonicalServer
 470  * PROTO_INTERNAL: Like PROTO_CANONICAL, but uses $wgInternalServer instead of $wgCanonicalServer
 471  *
 472  * @todo this won't work with current-path-relative URLs
 473  * like "subdir/foo.html", etc.
 474  *
 475  * @param $url String: either fully-qualified or a local path + query
 476  * @param $defaultProto Mixed: one of the PROTO_* constants. Determines the
 477  *                             protocol to use if $url or $wgServer is
 478  *                             protocol-relative
 479  * @return string Fully-qualified URL, current-path-relative URL or false if
 480  *                no valid URL can be constructed
 481  */
 482 function wfExpandUrl( $url, $defaultProto = PROTO_CURRENT ) {
 483         global $wgServer, $wgCanonicalServer, $wgInternalServer;
 484         $serverUrl = $wgServer;
 485         if ( $defaultProto === PROTO_CANONICAL ) {
 486                 $serverUrl = $wgCanonicalServer;
 487         }
 488         // Make $wgInternalServer fall back to $wgServer if not set
 489         if ( $defaultProto === PROTO_INTERNAL && $wgInternalServer !== false ) {
 490                 $serverUrl = $wgInternalServer;
 491         }
 492         if ( $defaultProto === PROTO_CURRENT ) {
 493                 $defaultProto = WebRequest::detectProtocol() . '://';
 494         }
 495
 496         // Analyze $serverUrl to obtain its protocol
 497         $bits = wfParseUrl( $serverUrl );
 498         $serverHasProto = $bits && $bits['scheme'] != '';
 499
 500         if ( $defaultProto === PROTO_CANONICAL || $defaultProto === PROTO_INTERNAL ) {
 501                 if ( $serverHasProto ) {
 502                         $defaultProto = $bits['scheme'] . '://';
 503                 } else {
 504                         // $wgCanonicalServer or $wgInternalServer doesn't have a protocol. This really isn't supposed to happen
 505                         // Fall back to HTTP in this ridiculous case
 506                         $defaultProto = PROTO_HTTP;
 507                 }
 508         }
 509
 510         $defaultProtoWithoutSlashes = substr( $defaultProto, 0, -2 );
 511
 512         if ( substr( $url, 0, 2 ) == '//' ) {
 513                 $url = $defaultProtoWithoutSlashes . $url;
 514         } elseif ( substr( $url, 0, 1 ) == '/' ) {
 515                 // If $serverUrl is protocol-relative, prepend $defaultProtoWithoutSlashes, otherwise leave it alone
 516                 $url = ( $serverHasProto ? '' : $defaultProtoWithoutSlashes ) . $serverUrl . $url;
 517         }
 518
 519         $bits = wfParseUrl( $url );
 520         if ( $bits && isset( $bits['path'] ) ) {
 521                 $bits['path'] = wfRemoveDotSegments( $bits['path'] );
 522                 return wfAssembleUrl( $bits );
 523         } elseif ( $bits ) {
 524                 # No path to expand
 525                 return $url;
 526         } elseif ( substr( $url, 0, 1 ) != '/' ) {
 527                 # URL is a relative path
 528                 return wfRemoveDotSegments( $url );
 529         }
 530
 531         # Expanded URL is not valid.
 532         return false;
 533 }
 534
 535 /**
 536  * This function will reassemble a URL parsed with wfParseURL.  This is useful
 537  * if you need to edit part of a URL and put it back together.
 538  *
 539  * This is the basic structure used (brackets contain keys for $urlParts):
 540  * [scheme][delimiter][user]:[pass]@[host]:[port][path]?[query]#[fragment]
 541  *
 542  * @todo Need to integrate this into wfExpandUrl (bug 32168)
 543  *
 544  * @param $urlParts Array URL parts, as output from wfParseUrl
 545  * @return string URL assembled from its component parts
 546  */
 547 function wfAssembleUrl( $urlParts ) {
 548         $result = '';
 549
 550         if ( isset( $urlParts['delimiter'] ) ) {
 551                 if ( isset( $urlParts['scheme'] ) ) {
 552                         $result .= $urlParts['scheme'];
 553                 }
 554
 555                 $result .= $urlParts['delimiter'];
 556         }
 557
 558         if ( isset( $urlParts['host'] ) ) {
 559                 if ( isset( $urlParts['user'] ) ) {
 560                         $result .= $urlParts['user'];
 561                         if ( isset( $urlParts['pass'] ) ) {
 562                                 $result .= ':' . $urlParts['pass'];
 563                         }
 564                         $result .= '@';
 565                 }
 566
 567                 $result .= $urlParts['host'];
 568
 569                 if ( isset( $urlParts['port'] ) ) {
 570                         $result .= ':' . $urlParts['port'];
 571                 }
 572         }
 573
 574         if ( isset( $urlParts['path'] ) ) {
 575                 $result .= $urlParts['path'];
 576         }
 577
 578         if ( isset( $urlParts['query'] ) ) {
 579                 $result .= '?' . $urlParts['query'];
 580         }
 581
 582         if ( isset( $urlParts['fragment'] ) ) {
 583                 $result .= '#' . $urlParts['fragment'];
 584         }
 585
 586         return $result;
 587 }
 588
 589 /**
 590  * Remove all dot-segments in the provided URL path.  For example,
 591  * '/a/./b/../c/' becomes '/a/c/'.  For details on the algorithm, please see
 592  * RFC3986 section 5.2.4.
 593  *
 594  * @todo Need to integrate this into wfExpandUrl (bug 32168)
 595  *
 596  * @param $urlPath String URL path, potentially containing dot-segments
 597  * @return string URL path with all dot-segments removed
 598  */
 599 function wfRemoveDotSegments( $urlPath ) {
 600         $output = '';
 601         $inputOffset = 0;
 602         $inputLength = strlen( $urlPath );
 603
 604         while ( $inputOffset < $inputLength ) {
 605                 $prefixLengthOne = substr( $urlPath, $inputOffset, 1 );
 606                 $prefixLengthTwo = substr( $urlPath, $inputOffset, 2 );
 607                 $prefixLengthThree = substr( $urlPath, $inputOffset, 3 );
 608                 $prefixLengthFour = substr( $urlPath, $inputOffset, 4 );
 609                 $trimOutput = false;
 610
 611                 if ( $prefixLengthTwo == './' ) {
 612                         # Step A, remove leading "./"
 613                         $inputOffset += 2;
 614                 } elseif ( $prefixLengthThree == '../' ) {
 615                         # Step A, remove leading "../"
 616                         $inputOffset += 3;
 617                 } elseif ( ( $prefixLengthTwo == '/.' ) && ( $inputOffset + 2 == $inputLength ) ) {
 618                         # Step B, replace leading "/.$" with "/"
 619                         $inputOffset += 1;
 620                         $urlPath[$inputOffset] = '/';
 621                 } elseif ( $prefixLengthThree == '/./' ) {
 622                         # Step B, replace leading "/./" with "/"
 623                         $inputOffset += 2;
 624                 } elseif ( $prefixLengthThree == '/..' && ( $inputOffset + 3 == $inputLength ) ) {
 625                         # Step C, replace leading "/..$" with "/" and
 626                         # remove last path component in output
 627                         $inputOffset += 2;
 628                         $urlPath[$inputOffset] = '/';
 629                         $trimOutput = true;
 630                 } elseif ( $prefixLengthFour == '/../' ) {
 631                         # Step C, replace leading "/../" with "/" and
 632                         # remove last path component in output
 633                         $inputOffset += 3;
 634                         $trimOutput = true;
 635                 } elseif ( ( $prefixLengthOne == '.' ) && ( $inputOffset + 1 == $inputLength ) ) {
 636                         # Step D, remove "^.$"
 637                         $inputOffset += 1;
 638                 } elseif ( ( $prefixLengthTwo == '..' ) && ( $inputOffset + 2 == $inputLength ) ) {
 639                         # Step D, remove "^..$"
 640                         $inputOffset += 2;
 641                 } else {
 642                         # Step E, move leading path segment to output
 643                         if ( $prefixLengthOne == '/' ) {
 644                                 $slashPos = strpos( $urlPath, '/', $inputOffset + 1 );
 645                         } else {
 646                                 $slashPos = strpos( $urlPath, '/', $inputOffset );
 647                         }
 648                         if ( $slashPos === false ) {
 649                                 $output .= substr( $urlPath, $inputOffset );
 650                                 $inputOffset = $inputLength;
 651                         } else {
 652                                 $output .= substr( $urlPath, $inputOffset, $slashPos - $inputOffset );
 653                                 $inputOffset += $slashPos - $inputOffset;
 654                         }
 655                 }
 656
 657                 if ( $trimOutput ) {
 658                         $slashPos = strrpos( $output, '/' );
 659                         if ( $slashPos === false ) {
 660                                 $output = '';
 661                         } else {
 662                                 $output = substr( $output, 0, $slashPos );
 663                         }
 664                 }
 665         }
 666
 667         return $output;
 668 }
 669
 670 /**
 671  * Returns a regular expression of url protocols
 672  *
 673  * @param $includeProtocolRelative bool If false, remove '//' from the returned protocol list.
 674  *        DO NOT USE this directly, use wfUrlProtocolsWithoutProtRel() instead
 675  * @return String
 676  */
 677 function wfUrlProtocols( $includeProtocolRelative = true ) {
 678         global $wgUrlProtocols;
 679
 680         // Cache return values separately based on $includeProtocolRelative
 681         static $withProtRel = null, $withoutProtRel = null;
 682         $cachedValue = $includeProtocolRelative ? $withProtRel : $withoutProtRel;
 683         if ( !is_null( $cachedValue ) ) {
 684                 return $cachedValue;
 685         }
 686
 687         // Support old-style $wgUrlProtocols strings, for backwards compatibility
 688         // with LocalSettings files from 1.5
 689         if ( is_array( $wgUrlProtocols ) ) {
 690                 $protocols = array();
 691                 foreach ( $wgUrlProtocols as $protocol ) {
 692                         // Filter out '//' if !$includeProtocolRelative
 693                         if ( $includeProtocolRelative || $protocol !== '//' ) {
 694                                 $protocols[] = preg_quote( $protocol, '/' );
 695                         }
 696                 }
 697
 698                 $retval = implode( '|', $protocols );
 699         } else {
 700                 // Ignore $includeProtocolRelative in this case
 701                 // This case exists for pre-1.6 compatibility, and we can safely assume
 702                 // that '//' won't appear in a pre-1.6 config because protocol-relative
 703                 // URLs weren't supported until 1.18
 704                 $retval = $wgUrlProtocols;
 705         }
 706
 707         // Cache return value
 708         if ( $includeProtocolRelative ) {
 709                 $withProtRel = $retval;
 710         } else {
 711                 $withoutProtRel = $retval;
 712         }
 713         return $retval;
 714 }
 715
 716 /**
 717  * Like wfUrlProtocols(), but excludes '//' from the protocol list. Use this if
 718  * you need a regex that matches all URL protocols but does not match protocol-
 719  * relative URLs
 720  * @return String
 721  */
 722 function wfUrlProtocolsWithoutProtRel() {
 723         return wfUrlProtocols( false );
 724 }
 725
 726 /**
 727  * parse_url() work-alike, but non-broken.  Differences:
 728  *
 729  * 1) Does not raise warnings on bad URLs (just returns false)
 730  * 2) Handles protocols that don't use :// (e.g., mailto: and news: , as well as protocol-relative URLs) correctly
 731  * 3) Adds a "delimiter" element to the array, either '://', ':' or '//' (see (2))
 732  *
 733  * @param $url String: a URL to parse
 734  * @return Array: bits of the URL in an associative array, per PHP docs
 735  */
 736 function wfParseUrl( $url ) {
 737         global $wgUrlProtocols; // Allow all protocols defined in DefaultSettings/LocalSettings.php
 738
 739         // Protocol-relative URLs are handled really badly by parse_url(). It's so bad that the easiest
 740         // way to handle them is to just prepend 'http:' and strip the protocol out later
 741         $wasRelative = substr( $url, 0, 2 ) == '//';
 742         if ( $wasRelative ) {
 743                 $url = "http:$url";
 744         }
 745         wfSuppressWarnings();
 746         $bits = parse_url( $url );
 747         wfRestoreWarnings();
 748         // parse_url() returns an array without scheme for some invalid URLs, e.g.
 749         // parse_url("%0Ahttp://example.com") == array( 'host' => '%0Ahttp', 'path' => 'example.com' )
 750         if ( !$bits || !isset( $bits['scheme'] ) ) {
 751                 return false;
 752         }
 753
 754         // most of the protocols are followed by ://, but mailto: and sometimes news: not, check for it
 755         if ( in_array( $bits['scheme'] . '://', $wgUrlProtocols ) ) {
 756                 $bits['delimiter'] = '://';
 757         } elseif ( in_array( $bits['scheme'] . ':', $wgUrlProtocols ) ) {
 758                 $bits['delimiter'] = ':';
 759                 // parse_url detects for news: and mailto: the host part of an url as path
 760                 // We have to correct this wrong detection
 761                 if ( isset( $bits['path'] ) ) {
 762                         $bits['host'] = $bits['path'];
 763                         $bits['path'] = '';
 764                 }
 765         } else {
 766                 return false;
 767         }
 768
 769         /* Provide an empty host for eg. file:/// urls (see bug 28627) */
 770         if ( !isset( $bits['host'] ) ) {
 771                 $bits['host'] = '';
 772
 773                 /* parse_url loses the third / for file:///c:/ urls (but not on variants) */
 774                 if ( substr( $bits['path'], 0, 1 ) !== '/' ) {
 775                         $bits['path'] = '/' . $bits['path'];
 776                 }
 777         }
 778
 779         // If the URL was protocol-relative, fix scheme and delimiter
 780         if ( $wasRelative ) {
 781                 $bits['scheme'] = '';
 782                 $bits['delimiter'] = '//';
 783         }
 784         return $bits;
 785 }
 786
 787 /**
 788  * Make URL indexes, appropriate for the el_index field of externallinks.
 789  *
 790  * @param $url String
 791  * @return array
 792  */
 793 function wfMakeUrlIndexes( $url ) {
 794         $bits = wfParseUrl( $url );
 795
 796         // Reverse the labels in the hostname, convert to lower case
 797         // For emails reverse domainpart only
 798         if ( $bits['scheme'] == 'mailto' ) {
 799                 $mailparts = explode( '@', $bits['host'], 2 );
 800                 if ( count( $mailparts ) === 2 ) {
 801                         $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
 802                 } else {
 803                         // No domain specified, don't mangle it
 804                         $domainpart = '';
 805                 }
 806                 $reversedHost = $domainpart . '@' . $mailparts[0];
 807         } else {
 808                 $reversedHost = strtolower( implode( '.', array_reverse( explode( '.', $bits['host'] ) ) ) );
 809         }
 810         // Add an extra dot to the end
 811         // Why? Is it in wrong place in mailto links?
 812         if ( substr( $reversedHost, -1, 1 ) !== '.' ) {
 813                 $reversedHost .= '.';
 814         }
 815         // Reconstruct the pseudo-URL
 816         $prot = $bits['scheme'];
 817         $index = $prot . $bits['delimiter'] . $reversedHost;
 818         // Leave out user and password. Add the port, path, query and fragment
 819         if ( isset( $bits['port'] ) ) {
 820                 $index .= ':' . $bits['port'];
 821         }
 822         if ( isset( $bits['path'] ) ) {
 823                 $index .= $bits['path'];
 824         } else {
 825                 $index .= '/';
 826         }
 827         if ( isset( $bits['query'] ) ) {
 828                 $index .= '?' . $bits['query'];
 829         }
 830         if ( isset( $bits['fragment'] ) ) {
 831                 $index .= '#' . $bits['fragment'];
 832         }
 833
 834         if ( $prot == '' ) {
 835                 return array( "http:$index", "https:$index" );
 836         } else {
 837                 return array( $index );
 838         }
 839 }
 840
 841 /**
 842  * Check whether a given URL has a domain that occurs in a given set of domains
 843  * @param $url string URL
 844  * @param $domains array Array of domains (strings)
 845  * @return bool True if the host part of $url ends in one of the strings in $domains
 846  */
 847 function wfMatchesDomainList( $url, $domains ) {
 848         $bits = wfParseUrl( $url );
 849         if ( is_array( $bits ) && isset( $bits['host'] ) ) {
 850                 foreach ( (array)$domains as $domain ) {
 851                         // FIXME: This gives false positives. http://nds-nl.wikipedia.org will match nl.wikipedia.org
 852                         // We should use something that interprets dots instead
 853                         if ( substr( $bits['host'], -strlen( $domain ) ) === $domain ) {
 854                                 return true;
 855                         }
 856                 }
 857         }
 858         return false;
 859 }
 860
 861 /**
 862  * Sends a line to the debug log if enabled or, optionally, to a comment in output.
 863  * In normal operation this is a NOP.
 864  *
 865  * Controlling globals:
 866  * $wgDebugLogFile - points to the log file
 867  * $wgProfileOnly - if set, normal debug messages will not be recorded.
 868  * $wgDebugRawPage - if false, 'action=raw' hits will not result in debug output.
 869  * $wgDebugComments - if on, some debug items may appear in comments in the HTML output.
 870  *
 871  * @param $text String
 872  * @param $logonly Bool: set true to avoid appearing in HTML when $wgDebugComments is set
 873  */
 874 function wfDebug( $text, $logonly = false ) {
 875         global $wgOut, $wgDebugLogFile, $wgDebugComments, $wgProfileOnly, $wgDebugRawPage;
 876         global $wgDebugLogPrefix, $wgShowDebug;
 877
 878         static $cache = array(); // Cache of unoutputted messages
 879         $text = wfDebugTimer() . $text;
 880
 881         if ( !$wgDebugRawPage && wfIsDebugRawPage() ) {
 882                 return;
 883         }
 884
 885         if ( ( $wgDebugComments || $wgShowDebug ) && !$logonly ) {
 886                 $cache[] = $text;
 887
 888                 if ( isset( $wgOut ) && is_object( $wgOut ) ) {
 889                         // add the message and any cached messages to the output
 890                         array_map( array( $wgOut, 'debug' ), $cache );
 891                         $cache = array();
 892                 }
 893         }
 894         if ( wfRunHooks( 'Debug', array( $text, null /* no log group */ ) ) ) {
 895                 if ( $wgDebugLogFile != '' && !$wgProfileOnly ) {
 896                         # Strip unprintables; they can switch terminal modes when binary data
 897                         # gets dumped, which is pretty annoying.
 898                         $text = preg_replace( '![\x00-\x08\x0b\x0c\x0e-\x1f]!', ' ', $text );
 899                         $text = $wgDebugLogPrefix . $text;
 900                         wfErrorLog( $text, $wgDebugLogFile );
 901                 }
 902         }
 903
 904         MWDebug::debugMsg( $text );
 905 }
 906
 907 /**
 908  * Returns true if debug logging should be suppressed if $wgDebugRawPage = false
 909  * @return bool
 910  */
 911 function wfIsDebugRawPage() {
 912         static $cache;
 913         if ( $cache !== null ) {
 914                 return $cache;
 915         }
 916         # Check for raw action using $_GET not $wgRequest, since the latter might not be initialised yet
 917         if ( ( isset( $_GET['action'] ) && $_GET['action'] == 'raw' )
 918                 || (
 919                         isset( $_SERVER['SCRIPT_NAME'] )
 920                         && substr( $_SERVER['SCRIPT_NAME'], -8 ) == 'load.php'
 921                 ) )
 922         {
 923                 $cache = true;
 924         } else {
 925                 $cache = false;
 926         }
 927         return $cache;
 928 }
 929
 930 /**
 931  * Get microsecond timestamps for debug logs
 932  *
 933  * @return string
 934  */
 935 function wfDebugTimer() {
 936         global $wgDebugTimestamps, $wgRequestTime;
 937
 938         if ( !$wgDebugTimestamps ) {
 939                 return '';
 940         }
 941
 942         $prefix = sprintf( "%6.4f", microtime( true ) - $wgRequestTime );
 943         $mem = sprintf( "%5.1fM", ( memory_get_usage( true ) / ( 1024 * 1024 ) ) );
 944         return "$prefix $mem  ";
 945 }
 946
 947 /**
 948  * Send a line giving PHP memory usage.
 949  *
 950  * @param $exact Bool: print exact values instead of kilobytes (default: false)
 951  */
 952 function wfDebugMem( $exact = false ) {
 953         $mem = memory_get_usage();
 954         if( !$exact ) {
 955                 $mem = floor( $mem / 1024 ) . ' kilobytes';
 956         } else {
 957                 $mem .= ' bytes';
 958         }
 959         wfDebug( "Memory usage: $mem\n" );
 960 }
 961
 962 /**
 963  * Send a line to a supplementary debug log file, if configured, or main debug log if not.
 964  * $wgDebugLogGroups[$logGroup] should be set to a filename to send to a separate log.
 965  *
 966  * @param $logGroup String
 967  * @param $text String
 968  * @param $public Bool: whether to log the event in the public log if no private
 969  *                     log file is specified, (default true)
 970  */
 971 function wfDebugLog( $logGroup, $text, $public = true ) {
 972         global $wgDebugLogGroups;
 973         $text = trim( $text ) . "\n";
 974         if( isset( $wgDebugLogGroups[$logGroup] ) ) {
 975                 $time = wfTimestamp( TS_DB );
 976                 $wiki = wfWikiID();
 977                 $host = wfHostname();
 978                 if ( wfRunHooks( 'Debug', array( $text, $logGroup ) ) ) {
 979                         wfErrorLog( "$time $host $wiki: $text", $wgDebugLogGroups[$logGroup] );
 980                 }
 981         } elseif ( $public === true ) {
 982                 wfDebug( $text, true );
 983         }
 984 }
 985
 986 /**
 987  * Log for database errors
 988  *
 989  * @param $text String: database error message.
 990  */
 991 function wfLogDBError( $text ) {
 992         global $wgDBerrorLog;
 993         if ( $wgDBerrorLog ) {
 994                 $host = wfHostname();
 995                 $wiki = wfWikiID();
 996                 $text = date( 'D M j G:i:s T Y' ) . "\t$host\t$wiki\t$text";
 997                 wfErrorLog( $text, $wgDBerrorLog );
 998         }
 999 }
1000
1001 /**
1002  * Throws a warning that $function is deprecated
1003  *
1004  * @param $function String
1005  * @param $version String|bool: Version of MediaWiki that the function was deprecated in (Added in 1.19).
1006  * @param $component String|bool: Added in 1.19.
1007  * @param $callerOffset integer: How far up the callstack is the original
1008  *    caller. 2 = function that called the function that called
1009  *    wfDeprecated (Added in 1.20)
1010  *
1011  * @return null
1012  */
1013 function wfDeprecated( $function, $version = false, $component = false, $callerOffset = 2 ) {
1014         static $functionsWarned = array();
1015
1016         MWDebug::deprecated( $function, $version, $component );
1017
1018         if ( !isset( $functionsWarned[$function] ) ) {
1019                 $functionsWarned[$function] = true;
1020
1021                 if ( $version ) {
1022                         global $wgDeprecationReleaseLimit;
1023
1024                         if ( $wgDeprecationReleaseLimit && $component === false ) {
1025                                 # Strip -* off the end of $version so that branches can use the
1026                                 # format #.##-branchname to avoid issues if the branch is merged into
1027                                 # a version of MediaWiki later than what it was branched from
1028                                 $comparableVersion = preg_replace( '/-.*$/', '', $version );
1029
1030                                 # If the comparableVersion is larger than our release limit then
1031                                 # skip the warning message for the deprecation
1032                                 if ( version_compare( $wgDeprecationReleaseLimit, $comparableVersion, '<' ) ) {
1033                                         return;
1034                                 }
1035                         }
1036
1037                         $component = $component === false ? 'MediaWiki' : $component;
1038                         wfWarn( "Use of $function was deprecated in $component $version.", $callerOffset );
1039                 } else {
1040                         wfWarn( "Use of $function is deprecated.", $callerOffset );
1041                 }
1042         }
1043 }
1044
1045 /**
1046  * Send a warning either to the debug log or in a PHP error depending on
1047  * $wgDevelopmentWarnings
1048  *
1049  * @param $msg String: message to send
1050  * @param $callerOffset Integer: number of items to go back in the backtrace to
1051  *        find the correct caller (1 = function calling wfWarn, ...)
1052  * @param $level Integer: PHP error level; only used when $wgDevelopmentWarnings
1053  *        is true
1054  */
1055 function wfWarn( $msg, $callerOffset = 1, $level = E_USER_NOTICE ) {
1056         global $wgDevelopmentWarnings;
1057
1058         MWDebug::warning( $msg, $callerOffset + 2 );
1059
1060         $callers = wfDebugBacktrace();
1061         if ( isset( $callers[$callerOffset + 1] ) ) {
1062                 $callerfunc = $callers[$callerOffset + 1];
1063                 $callerfile = $callers[$callerOffset];
1064                 if ( isset( $callerfile['file'] ) && isset( $callerfile['line'] ) ) {
1065                         $file = $callerfile['file'] . ' at line ' . $callerfile['line'];
1066                 } else {
1067                         $file = '(internal function)';
1068                 }
1069                 $func = '';
1070                 if ( isset( $callerfunc['class'] ) ) {
1071                         $func .= $callerfunc['class'] . '::';
1072                 }
1073                 if ( isset( $callerfunc['function'] ) ) {
1074                         $func .= $callerfunc['function'];
1075                 }
1076                 $msg .= " [Called from $func in $file]";
1077         }
1078
1079         if ( $wgDevelopmentWarnings ) {
1080                 trigger_error( $msg, $level );
1081         } else {
1082                 wfDebug( "$msg\n" );
1083         }
1084 }
1085
1086 /**
1087  * Log to a file without getting "file size exceeded" signals.
1088  *
1089  * Can also log to TCP or UDP with the syntax udp://host:port/prefix. This will
1090  * send lines to the specified port, prefixed by the specified prefix and a space.
1091  *
1092  * @param $text String
1093  * @param $file String filename
1094  */
1095 function wfErrorLog( $text, $file ) {
1096         if ( substr( $file, 0, 4 ) == 'udp:' ) {
1097                 # Needs the sockets extension
1098                 if ( preg_match( '!^(tcp|udp):(?://)?\[([0-9a-fA-F:]+)\]:(\d+)(?:/(.*))?$!', $file, $m ) ) {
1099                         // IPv6 bracketed host
1100                         $host = $m[2];
1101                         $port = intval( $m[3] );
1102                         $prefix = isset( $m[4] ) ? $m[4] : false;
1103                         $domain = AF_INET6;
1104                 } elseif ( preg_match( '!^(tcp|udp):(?://)?([a-zA-Z0-9.-]+):(\d+)(?:/(.*))?$!', $file, $m ) ) {
1105                         $host = $m[2];
1106                         if ( !IP::isIPv4( $host ) ) {
1107                                 $host = gethostbyname( $host );
1108                         }
1109                         $port = intval( $m[3] );
1110                         $prefix = isset( $m[4] ) ? $m[4] : false;
1111                         $domain = AF_INET;
1112                 } else {
1113                         throw new MWException( __METHOD__ . ': Invalid UDP specification' );
1114                 }
1115
1116                 // Clean it up for the multiplexer
1117                 if ( strval( $prefix ) !== '' ) {
1118                         $text = preg_replace( '/^/m', $prefix . ' ', $text );
1119
1120                         // Limit to 64KB
1121                         if ( strlen( $text ) > 65506 ) {
1122                                 $text = substr( $text, 0, 65506 );
1123                         }
1124
1125                         if ( substr( $text, -1 ) != "\n" ) {
1126                                 $text .= "\n";
1127                         }
1128                 } elseif ( strlen( $text ) > 65507 ) {
1129                         $text = substr( $text, 0, 65507 );
1130                 }
1131
1132                 $sock = socket_create( $domain, SOCK_DGRAM, SOL_UDP );
1133                 if ( !$sock ) {
1134                         return;
1135                 }
1136
1137                 socket_sendto( $sock, $text, strlen( $text ), 0, $host, $port );
1138                 socket_close( $sock );
1139         } else {
1140                 wfSuppressWarnings();
1141                 $exists = file_exists( $file );
1142                 $size = $exists ? filesize( $file ) : false;
1143                 if ( !$exists || ( $size !== false && $size + strlen( $text ) < 0x7fffffff ) ) {
1144                         file_put_contents( $file, $text, FILE_APPEND );
1145                 }
1146                 wfRestoreWarnings();
1147         }
1148 }
1149
1150 /**
1151  * @todo document
1152  */
1153 function wfLogProfilingData() {
1154         global $wgRequestTime, $wgDebugLogFile, $wgDebugRawPage, $wgRequest;
1155         global $wgProfileLimit, $wgUser;
1156
1157         $profiler = Profiler::instance();
1158
1159         # Profiling must actually be enabled...
1160         if ( $profiler->isStub() ) {
1161                 return;
1162         }
1163
1164         // Get total page request time and only show pages that longer than
1165         // $wgProfileLimit time (default is 0)
1166         $elapsed = microtime( true ) - $wgRequestTime;
1167         if ( $elapsed <= $wgProfileLimit ) {
1168                 return;
1169         }
1170
1171         $profiler->logData();
1172
1173         // Check whether this should be logged in the debug file.
1174         if ( $wgDebugLogFile == '' || ( !$wgDebugRawPage && wfIsDebugRawPage() ) ) {
1175                 return;
1176         }
1177
1178         $forward = '';
1179         if ( !empty( $_SERVER['HTTP_X_FORWARDED_FOR'] ) ) {
1180                 $forward = ' forwarded for ' . $_SERVER['HTTP_X_FORWARDED_FOR'];
1181         }
1182         if ( !empty( $_SERVER['HTTP_CLIENT_IP'] ) ) {
1183                 $forward .= ' client IP ' . $_SERVER['HTTP_CLIENT_IP'];
1184         }
1185         if ( !empty( $_SERVER['HTTP_FROM'] ) ) {
1186                 $forward .= ' from ' . $_SERVER['HTTP_FROM'];
1187         }
1188         if ( $forward ) {
1189                 $forward = "\t(proxied via {$_SERVER['REMOTE_ADDR']}{$forward})";
1190         }
1191         // Don't load $wgUser at this late stage just for statistics purposes
1192         // @todo FIXME: We can detect some anons even if it is not loaded. See User::getId()
1193         if ( $wgUser->isItemLoaded( 'id' ) && $wgUser->isAnon() ) {
1194                 $forward .= ' anon';
1195         }
1196         $log = sprintf( "%s\t%04.3f\t%s\n",
1197                 gmdate( 'YmdHis' ), $elapsed,
1198                 urldecode( $wgRequest->getRequestURL() . $forward ) );
1199
1200         wfErrorLog( $log . $profiler->getOutput(), $wgDebugLogFile );
1201 }
1202
1203 /**
1204  * Increment a statistics counter
1205  *
1206  * @param $key String
1207  * @param $count Int
1208  */
1209 function wfIncrStats( $key, $count = 1 ) {
1210         global $wgStatsMethod;
1211
1212         $count = intval( $count );
1213
1214         if( $wgStatsMethod == 'udp' ) {
1215                 global $wgUDPProfilerHost, $wgUDPProfilerPort, $wgDBname, $wgAggregateStatsID;
1216                 static $socket;
1217
1218                 $id = $wgAggregateStatsID !== false ? $wgAggregateStatsID : $wgDBname;
1219
1220                 if ( !$socket ) {
1221                         $socket = socket_create( AF_INET, SOCK_DGRAM, SOL_UDP );
1222                         $statline = "stats/{$id} - 1 1 1 1 1 -total\n";
1223                         socket_sendto(
1224                                 $socket,
1225                                 $statline,
1226                                 strlen( $statline ),
1227                                 0,
1228                                 $wgUDPProfilerHost,
1229                                 $wgUDPProfilerPort
1230                         );
1231                 }
1232                 $statline = "stats/{$id} - {$count} 1 1 1 1 {$key}\n";
1233                 wfSuppressWarnings();
1234                 socket_sendto(
1235                         $socket,
1236                         $statline,
1237                         strlen( $statline ),
1238                         0,
1239                         $wgUDPProfilerHost,
1240                         $wgUDPProfilerPort
1241                 );
1242                 wfRestoreWarnings();
1243         } elseif( $wgStatsMethod == 'cache' ) {
1244                 global $wgMemc;
1245                 $key = wfMemcKey( 'stats', $key );
1246                 if ( is_null( $wgMemc->incr( $key, $count ) ) ) {
1247                         $wgMemc->add( $key, $count );
1248                 }
1249         } else {
1250                 // Disabled
1251         }
1252 }
1253
1254 /**
1255  * Check if the wiki read-only lock file is present. This can be used to lock
1256  * off editing functions, but doesn't guarantee that the database will not be
1257  * modified.
1258  *
1259  * @return bool
1260  */
1261 function wfReadOnly() {
1262         global $wgReadOnlyFile, $wgReadOnly;
1263
1264         if ( !is_null( $wgReadOnly ) ) {
1265                 return (bool)$wgReadOnly;
1266         }
1267         if ( $wgReadOnlyFile == '' ) {
1268                 return false;
1269         }
1270         // Set $wgReadOnly for faster access next time
1271         if ( is_file( $wgReadOnlyFile ) ) {
1272                 $wgReadOnly = file_get_contents( $wgReadOnlyFile );
1273         } else {
1274                 $wgReadOnly = false;
1275         }
1276         return (bool)$wgReadOnly;
1277 }
1278
1279 /**
1280  * @return bool
1281  */
1282 function wfReadOnlyReason() {
1283         global $wgReadOnly;
1284         wfReadOnly();
1285         return $wgReadOnly;
1286 }
1287
1288 /**
1289  * Return a Language object from $langcode
1290  *
1291  * @param $langcode Mixed: either:
1292  *                  - a Language object
1293  *                  - code of the language to get the message for, if it is
1294  *                    a valid code create a language for that language, if
1295  *                    it is a string but not a valid code then make a basic
1296  *                    language object
1297  *                  - a boolean: if it's false then use the global object for
1298  *                    the current user's language (as a fallback for the old parameter
1299  *                    functionality), or if it is true then use global object
1300  *                    for the wiki's content language.
1301  * @return Language object
1302  */
1303 function wfGetLangObj( $langcode = false ) {
1304         # Identify which language to get or create a language object for.
1305         # Using is_object here due to Stub objects.
1306         if( is_object( $langcode ) ) {
1307                 # Great, we already have the object (hopefully)!
1308                 return $langcode;
1309         }
1310
1311         global $wgContLang, $wgLanguageCode;
1312         if( $langcode === true || $langcode === $wgLanguageCode ) {
1313                 # $langcode is the language code of the wikis content language object.
1314                 # or it is a boolean and value is true
1315                 return $wgContLang;
1316         }
1317
1318         global $wgLang;
1319         if( $langcode === false || $langcode === $wgLang->getCode() ) {
1320                 # $langcode is the language code of user language object.
1321                 # or it was a boolean and value is false
1322                 return $wgLang;
1323         }
1324
1325         $validCodes = array_keys( Language::fetchLanguageNames() );
1326         if( in_array( $langcode, $validCodes ) ) {
1327                 # $langcode corresponds to a valid language.
1328                 return Language::factory( $langcode );
1329         }
1330
1331         # $langcode is a string, but not a valid language code; use content language.
1332         wfDebug( "Invalid language code passed to wfGetLangObj, falling back to content language.\n" );
1333         return $wgContLang;
1334 }
1335
1336 /**
1337  * Old function when $wgBetterDirectionality existed
1338  * All usage removed, wfUILang can be removed in near future
1339  *
1340  * @deprecated since 1.18
1341  * @return Language
1342  */
1343 function wfUILang() {
1344         wfDeprecated( __METHOD__, '1.18' );
1345         global $wgLang;
1346         return $wgLang;
1347 }
1348
1349 /**
1350  * This is the new function for getting translated interface messages.
1351  * See the Message class for documentation how to use them.
1352  * The intention is that this function replaces all old wfMsg* functions.
1353  * @param $key \string Message key.
1354  * Varargs: normal message parameters.
1355  * @return Message
1356  * @since 1.17
1357  */
1358 function wfMessage( $key /*...*/) {
1359         $params = func_get_args();
1360         array_shift( $params );
1361         if ( isset( $params[0] ) && is_array( $params[0] ) ) {
1362                 $params = $params[0];
1363         }
1364         return new Message( $key, $params );
1365 }
1366
1367 /**
1368  * This function accepts multiple message keys and returns a message instance
1369  * for the first message which is non-empty. If all messages are empty then an
1370  * instance of the first message key is returned.
1371  * @param varargs: message keys
1372  * @return Message
1373  * @since 1.18
1374  */
1375 function wfMessageFallback( /*...*/ ) {
1376         $args = func_get_args();
1377         return MWFunction::callArray( 'Message::newFallbackSequence', $args );
1378 }
1379
1380 /**
1381  * Get a message from anywhere, for the current user language.
1382  *
1383  * Use wfMsgForContent() instead if the message should NOT
1384  * change depending on the user preferences.
1385  *
1386  * @param $key String: lookup key for the message, usually
1387  *    defined in languages/Language.php
1388  *
1389  * Parameters to the message, which can be used to insert variable text into
1390  * it, can be passed to this function in the following formats:
1391  * - One per argument, starting at the second parameter
1392  * - As an array in the second parameter
1393  * These are not shown in the function definition.
1394  *
1395  * @return String
1396  */
1397 function wfMsg( $key ) {
1398         $args = func_get_args();
1399         array_shift( $args );
1400         return wfMsgReal( $key, $args );
1401 }
1402
1403 /**
1404  * Same as above except doesn't transform the message
1405  *
1406  * @param $key String
1407  * @return String
1408  */
1409 function wfMsgNoTrans( $key ) {
1410         $args = func_get_args();
1411         array_shift( $args );
1412         return wfMsgReal( $key, $args, true, false, false );
1413 }
1414
1415 /**
1416  * Get a message from anywhere, for the current global language
1417  * set with $wgLanguageCode.
1418  *
1419  * Use this if the message should NOT change dependent on the
1420  * language set in the user's preferences. This is the case for
1421  * most text written into logs, as well as link targets (such as
1422  * the name of the copyright policy page). Link titles, on the
1423  * other hand, should be shown in the UI language.
1424  *
1425  * Note that MediaWiki allows users to change the user interface
1426  * language in their preferences, but a single installation
1427  * typically only contains content in one language.
1428  *
1429  * Be wary of this distinction: If you use wfMsg() where you should
1430  * use wfMsgForContent(), a user of the software may have to
1431  * customize potentially hundreds of messages in
1432  * order to, e.g., fix a link in every possible language.
1433  *
1434  * @param $key String: lookup key for the message, usually
1435  *     defined in languages/Language.php
1436  * @return String
1437  */
1438 function wfMsgForContent( $key ) {
1439         global $wgForceUIMsgAsContentMsg;
1440         $args = func_get_args();
1441         array_shift( $args );
1442         $forcontent = true;
1443         if( is_array( $wgForceUIMsgAsContentMsg ) &&
1444                 in_array( $key, $wgForceUIMsgAsContentMsg ) )
1445         {
1446                 $forcontent = false;
1447         }
1448         return wfMsgReal( $key, $args, true, $forcontent );
1449 }
1450
1451 /**
1452  * Same as above except doesn't transform the message
1453  *
1454  * @param $key String
1455  * @return String
1456  */
1457 function wfMsgForContentNoTrans( $key ) {
1458         global $wgForceUIMsgAsContentMsg;
1459         $args = func_get_args();
1460         array_shift( $args );
1461         $forcontent = true;
1462         if( is_array( $wgForceUIMsgAsContentMsg ) &&
1463                 in_array( $key, $wgForceUIMsgAsContentMsg ) )
1464         {
1465                 $forcontent = false;
1466         }
1467         return wfMsgReal( $key, $args, true, $forcontent, false );
1468 }
1469
1470 /**
1471  * Really get a message
1472  *
1473  * @param $key String: key to get.
1474  * @param $args
1475  * @param $useDB Boolean
1476  * @param $forContent Mixed: Language code, or false for user lang, true for content lang.
1477  * @param $transform Boolean: Whether or not to transform the message.
1478  * @return String: the requested message.
1479  */
1480 function wfMsgReal( $key, $args, $useDB = true, $forContent = false, $transform = true ) {
1481         wfProfileIn( __METHOD__ );
1482         $message = wfMsgGetKey( $key, $useDB, $forContent, $transform );
1483         $message = wfMsgReplaceArgs( $message, $args );
1484         wfProfileOut( __METHOD__ );
1485         return $message;
1486 }
1487
1488 /**
1489  * Fetch a message string value, but don't replace any keys yet.
1490  *
1491  * @param $key String
1492  * @param $useDB Bool
1493  * @param $langCode String: Code of the language to get the message for, or
1494  *                  behaves as a content language switch if it is a boolean.
1495  * @param $transform Boolean: whether to parse magic words, etc.
1496  * @return string
1497  */
1498 function wfMsgGetKey( $key, $useDB = true, $langCode = false, $transform = true ) {
1499         wfRunHooks( 'NormalizeMessageKey', array( &$key, &$useDB, &$langCode, &$transform ) );
1500
1501         $cache = MessageCache::singleton();
1502         $message = $cache->get( $key, $useDB, $langCode );
1503         if( $message === false ) {
1504                 $message = '&lt;' . htmlspecialchars( $key ) . '&gt;';
1505         } elseif ( $transform ) {
1506                 $message = $cache->transform( $message );
1507         }
1508         return $message;
1509 }
1510
1511 /**
1512  * Replace message parameter keys on the given formatted output.
1513  *
1514  * @param $message String
1515  * @param $args Array
1516  * @return string
1517  * @private
1518  */
1519 function wfMsgReplaceArgs( $message, $args ) {
1520         # Fix windows line-endings
1521         # Some messages are split with explode("\n", $msg)
1522         $message = str_replace( "\r", '', $message );
1523
1524         // Replace arguments
1525         if ( count( $args ) ) {
1526                 if ( is_array( $args[0] ) ) {
1527                         $args = array_values( $args[0] );
1528                 }
1529                 $replacementKeys = array();
1530                 foreach( $args as $n => $param ) {
1531                         $replacementKeys['$' . ( $n + 1 )] = $param;
1532                 }
1533                 $message = strtr( $message, $replacementKeys );
1534         }
1535
1536         return $message;
1537 }
1538
1539 /**
1540  * Return an HTML-escaped version of a message.
1541  * Parameter replacements, if any, are done *after* the HTML-escaping,
1542  * so parameters may contain HTML (eg links or form controls). Be sure
1543  * to pre-escape them if you really do want plaintext, or just wrap
1544  * the whole thing in htmlspecialchars().
1545  *
1546  * @param $key String
1547  * @param string ... parameters
1548  * @return string
1549  */
1550 function wfMsgHtml( $key ) {
1551         $args = func_get_args();
1552         array_shift( $args );
1553         return wfMsgReplaceArgs( htmlspecialchars( wfMsgGetKey( $key ) ), $args );
1554 }
1555
1556 /**
1557  * Return an HTML version of message
1558  * Parameter replacements, if any, are done *after* parsing the wiki-text message,
1559  * so parameters may contain HTML (eg links or form controls). Be sure
1560  * to pre-escape them if you really do want plaintext, or just wrap
1561  * the whole thing in htmlspecialchars().
1562  *
1563  * @param $key String
1564  * @param string ... parameters
1565  * @return string
1566  */
1567 function wfMsgWikiHtml( $key ) {
1568         $args = func_get_args();
1569         array_shift( $args );
1570         return wfMsgReplaceArgs(
1571                 MessageCache::singleton()->parse( wfMsgGetKey( $key ), null,
1572                 /* can't be set to false */ true, /* interface */ true )->getText(),
1573                 $args );
1574 }
1575
1576 /**
1577  * Returns message in the requested format
1578  * @param $key String: key of the message
1579  * @param $options Array: processing rules. Can take the following options:
1580  *   <i>parse</i>: parses wikitext to HTML
1581  *   <i>parseinline</i>: parses wikitext to HTML and removes the surrounding
1582  *       p's added by parser or tidy
1583  *   <i>escape</i>: filters message through htmlspecialchars
1584  *   <i>escapenoentities</i>: same, but allows entity references like &#160; through
1585  *   <i>replaceafter</i>: parameters are substituted after parsing or escaping
1586  *   <i>parsemag</i>: transform the message using magic phrases
1587  *   <i>content</i>: fetch message for content language instead of interface
1588  * Also can accept a single associative argument, of the form 'language' => 'xx':
1589  *   <i>language</i>: Language object or language code to fetch message for
1590  *       (overriden by <i>content</i>).
1591  * Behavior for conflicting options (e.g., parse+parseinline) is undefined.
1592  *
1593  * @return String
1594  */
1595 function wfMsgExt( $key, $options ) {
1596         $args = func_get_args();
1597         array_shift( $args );
1598         array_shift( $args );
1599         $options = (array)$options;
1600
1601         foreach( $options as $arrayKey => $option ) {
1602                 if( !preg_match( '/^[0-9]+|language$/', $arrayKey ) ) {
1603                         # An unknown index, neither numeric nor "language"
1604                         wfWarn( "wfMsgExt called with incorrect parameter key $arrayKey", 1, E_USER_WARNING );
1605                 } elseif( preg_match( '/^[0-9]+$/', $arrayKey ) && !in_array( $option,
1606                 array( 'parse', 'parseinline', 'escape', 'escapenoentities',
1607                 'replaceafter', 'parsemag', 'content' ) ) ) {
1608                         # A numeric index with unknown value
1609                         wfWarn( "wfMsgExt called with incorrect parameter $option", 1, E_USER_WARNING );
1610                 }
1611         }
1612
1613         if( in_array( 'content', $options, true ) ) {
1614                 $forContent = true;
1615                 $langCode = true;
1616                 $langCodeObj = null;
1617         } elseif( array_key_exists( 'language', $options ) ) {
1618                 $forContent = false;
1619                 $langCode = wfGetLangObj( $options['language'] );
1620                 $langCodeObj = $langCode;
1621         } else {
1622                 $forContent = false;
1623                 $langCode = false;
1624                 $langCodeObj = null;
1625         }
1626
1627         $string = wfMsgGetKey( $key, /*DB*/true, $langCode, /*Transform*/false );
1628
1629         if( !in_array( 'replaceafter', $options, true ) ) {
1630                 $string = wfMsgReplaceArgs( $string, $args );
1631         }
1632
1633         $messageCache = MessageCache::singleton();
1634         $parseInline = in_array( 'parseinline', $options, true );
1635         if( in_array( 'parse', $options, true ) || $parseInline ) {
1636                 $string = $messageCache->parse( $string, null, true, !$forContent, $langCodeObj );
1637                 if ( $string instanceof ParserOutput ) {
1638                         $string = $string->getText();
1639                 }
1640
1641                 if ( $parseInline ) {
1642                         $m = array();
1643                         if( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $string, $m ) ) {
1644                                 $string = $m[1];
1645                         }
1646                 }
1647         } elseif ( in_array( 'parsemag', $options, true ) ) {
1648                 $string = $messageCache->transform( $string,
1649                                 !$forContent, $langCodeObj );
1650         }
1651
1652         if ( in_array( 'escape', $options, true ) ) {
1653                 $string = htmlspecialchars ( $string );
1654         } elseif ( in_array( 'escapenoentities', $options, true ) ) {
1655                 $string = Sanitizer::escapeHtmlAllowEntities( $string );
1656         }
1657
1658         if( in_array( 'replaceafter', $options, true ) ) {
1659                 $string = wfMsgReplaceArgs( $string, $args );
1660         }
1661
1662         return $string;
1663 }
1664
1665 /**
1666  * Since wfMsg() and co suck, they don't return false if the message key they
1667  * looked up didn't exist but a XHTML string, this function checks for the
1668  * nonexistance of messages by checking the MessageCache::get() result directly.
1669  *
1670  * @param $key      String: the message key looked up
1671  * @return Boolean True if the message *doesn't* exist.
1672  */
1673 function wfEmptyMsg( $key ) {
1674         return MessageCache::singleton()->get( $key, /*useDB*/true, /*content*/false ) === false;
1675 }
1676
1677 /**
1678  * Throw a debugging exception. This function previously once exited the process,
1679  * but now throws an exception instead, with similar results.
1680  *
1681  * @param $msg String: message shown when dying.
1682  */
1683 function wfDebugDieBacktrace( $msg = '' ) {
1684         throw new MWException( $msg );
1685 }
1686
1687 /**
1688  * Fetch server name for use in error reporting etc.
1689  * Use real server name if available, so we know which machine
1690  * in a server farm generated the current page.
1691  *
1692  * @return string
1693  */
1694 function wfHostname() {
1695         static $host;
1696         if ( is_null( $host ) ) {
1697                 if ( function_exists( 'posix_uname' ) ) {
1698                         // This function not present on Windows
1699                         $uname = posix_uname();
1700                 } else {
1701                         $uname = false;
1702                 }
1703                 if( is_array( $uname ) && isset( $uname['nodename'] ) ) {
1704                         $host = $uname['nodename'];
1705                 } elseif ( getenv( 'COMPUTERNAME' ) ) {
1706                         # Windows computer name
1707                         $host = getenv( 'COMPUTERNAME' );
1708                 } else {
1709                         # This may be a virtual server.
1710                         $host = $_SERVER['SERVER_NAME'];
1711                 }
1712         }
1713         return $host;
1714 }
1715
1716 /**
1717  * Returns a HTML comment with the elapsed time since request.
1718  * This method has no side effects.
1719  *
1720  * @return string
1721  */
1722 function wfReportTime() {
1723         global $wgRequestTime, $wgShowHostnames;
1724
1725         $elapsed = microtime( true ) - $wgRequestTime;
1726
1727         return $wgShowHostnames
1728                 ? sprintf( '<!-- Served by %s in %01.3f secs. -->', wfHostname(), $elapsed )
1729                 : sprintf( '<!-- Served in %01.3f secs. -->', $elapsed );
1730 }
1731
1732 /**
1733  * Safety wrapper for debug_backtrace().
1734  *
1735  * With Zend Optimizer 3.2.0 loaded, this causes segfaults under somewhat
1736  * murky circumstances, which may be triggered in part by stub objects
1737  * or other fancy talkin'.
1738  *
1739  * Will return an empty array if Zend Optimizer is detected or if
1740  * debug_backtrace is disabled, otherwise the output from
1741  * debug_backtrace() (trimmed).
1742  *
1743  * @param $limit int This parameter can be used to limit the number of stack frames returned
1744  *
1745  * @return array of backtrace information
1746  */
1747 function wfDebugBacktrace( $limit = 0 ) {
1748         static $disabled = null;
1749
1750         if( extension_loaded( 'Zend Optimizer' ) ) {
1751                 wfDebug( "Zend Optimizer detected; skipping debug_backtrace for safety.\n" );
1752                 return array();
1753         }
1754
1755         if ( is_null( $disabled ) ) {
1756                 $disabled = false;
1757                 $functions = explode( ',', ini_get( 'disable_functions' ) );
1758                 $functions = array_map( 'trim', $functions );
1759                 $functions = array_map( 'strtolower', $functions );
1760                 if ( in_array( 'debug_backtrace', $functions ) ) {
1761                         wfDebug( "debug_backtrace is in disabled_functions\n" );
1762                         $disabled = true;
1763                 }
1764         }
1765         if ( $disabled ) {
1766                 return array();
1767         }
1768
1769         if ( $limit && version_compare( PHP_VERSION, '5.4.0', '>=' ) ) {
1770                 return array_slice( debug_backtrace( DEBUG_BACKTRACE_PROVIDE_OBJECT, $limit ), 1 );
1771         } else {
1772                 return array_slice( debug_backtrace(), 1 );
1773         }
1774 }
1775
1776 /**
1777  * Get a debug backtrace as a string
1778  *
1779  * @return string
1780  */
1781 function wfBacktrace() {
1782         global $wgCommandLineMode;
1783
1784         if ( $wgCommandLineMode ) {
1785                 $msg = '';
1786         } else {
1787                 $msg = "<ul>\n";
1788         }
1789         $backtrace = wfDebugBacktrace();
1790         foreach( $backtrace as $call ) {
1791                 if( isset( $call['file'] ) ) {
1792                         $f = explode( DIRECTORY_SEPARATOR, $call['file'] );
1793                         $file = $f[count( $f ) - 1];
1794                 } else {
1795                         $file = '-';
1796                 }
1797                 if( isset( $call['line'] ) ) {
1798                         $line = $call['line'];
1799                 } else {
1800                         $line = '-';
1801                 }
1802                 if ( $wgCommandLineMode ) {
1803                         $msg .= "$file line $line calls ";
1804                 } else {
1805                         $msg .= '<li>' . $file . ' line ' . $line . ' calls ';
1806                 }
1807                 if( !empty( $call['class'] ) ) {
1808                         $msg .= $call['class'] . $call['type'];
1809                 }
1810                 $msg .= $call['function'] . '()';
1811
1812                 if ( $wgCommandLineMode ) {
1813                         $msg .= "\n";
1814                 } else {
1815                         $msg .= "</li>\n";
1816                 }
1817         }
1818         if ( $wgCommandLineMode ) {
1819                 $msg .= "\n";
1820         } else {
1821                 $msg .= "</ul>\n";
1822         }
1823
1824         return $msg;
1825 }
1826
1827 /**
1828  * Get the name of the function which called this function
1829  *
1830  * @param $level Int
1831  * @return Bool|string
1832  */
1833 function wfGetCaller( $level = 2 ) {
1834         $backtrace = wfDebugBacktrace( $level );
1835         if ( isset( $backtrace[$level] ) ) {
1836                 return wfFormatStackFrame( $backtrace[$level] );
1837         } else {
1838                 $caller = 'unknown';
1839         }
1840         return $caller;
1841 }
1842
1843 /**
1844  * Return a string consisting of callers in the stack. Useful sometimes
1845  * for profiling specific points.
1846  *
1847  * @param $limit int The maximum depth of the stack frame to return, or false for
1848  *               the entire stack.
1849  * @return String
1850  */
1851 function wfGetAllCallers( $limit = 3 ) {
1852         $trace = array_reverse( wfDebugBacktrace() );
1853         if ( !$limit || $limit > count( $trace ) - 1 ) {
1854                 $limit = count( $trace ) - 1;
1855         }
1856         $trace = array_slice( $trace, -$limit - 1, $limit );
1857         return implode( '/', array_map( 'wfFormatStackFrame', $trace ) );
1858 }
1859
1860 /**
1861  * Return a string representation of frame
1862  *
1863  * @param $frame Array
1864  * @return Bool
1865  */
1866 function wfFormatStackFrame( $frame ) {
1867         return isset( $frame['class'] ) ?
1868                 $frame['class'] . '::' . $frame['function'] :
1869                 $frame['function'];
1870 }
1871
1872
1873 /* Some generic result counters, pulled out of SearchEngine */
1874
1875
1876 /**
1877  * @todo document
1878  *
1879  * @param $offset Int
1880  * @param $limit Int
1881  * @return String
1882  */
1883 function wfShowingResults( $offset, $limit ) {
1884         global $wgLang;
1885         return wfMsgExt(
1886                 'showingresults',
1887                 array( 'parseinline' ),
1888                 $wgLang->formatNum( $limit ),
1889                 $wgLang->formatNum( $offset + 1 )
1890         );
1891 }
1892
1893 /**
1894  * Generate (prev x| next x) (20|50|100...) type links for paging
1895  *
1896  * @param $offset String
1897  * @param $limit Integer
1898  * @param $link String
1899  * @param $query String: optional URL query parameter string
1900  * @param $atend Bool: optional param for specified if this is the last page
1901  * @return String
1902  * @deprecated in 1.19; use Language::viewPrevNext() instead
1903  */
1904 function wfViewPrevNext( $offset, $limit, $link, $query = '', $atend = false ) {
1905         wfDeprecated( __METHOD__, '1.19' );
1906
1907         global $wgLang;
1908
1909         $query = wfCgiToArray( $query );
1910
1911         if( is_object( $link ) ) {
1912                 $title = $link;
1913         } else {
1914                 $title = Title::newFromText( $link );
1915                 if( is_null( $title ) ) {
1916                         return false;
1917                 }
1918         }
1919
1920         return $wgLang->viewPrevNext( $title, $offset, $limit, $query, $atend );
1921 }
1922
1923 /**
1924  * Make a list item, used by various special pages
1925  *
1926  * @param $page String Page link
1927  * @param $details String Text between brackets
1928  * @param $oppositedm Boolean   Add the direction mark opposite to your
1929  *                                                              language, to display text properly
1930  * @return String
1931  * @deprecated since 1.19; use Language::specialList() instead
1932  */
1933 function wfSpecialList( $page, $details, $oppositedm = true ) {
1934         wfDeprecated( __METHOD__, '1.19' );
1935
1936         global $wgLang;
1937         return $wgLang->specialList( $page, $details, $oppositedm );
1938 }
1939
1940 /**
1941  * @todo document
1942  * @todo FIXME: We may want to blacklist some broken browsers
1943  *
1944  * @param $force Bool
1945  * @return bool Whereas client accept gzip compression
1946  */
1947 function wfClientAcceptsGzip( $force = false ) {
1948         static $result = null;
1949         if ( $result === null || $force ) {
1950                 $result = false;
1951                 if( isset( $_SERVER['HTTP_ACCEPT_ENCODING'] ) ) {
1952                         # @todo FIXME: We may want to blacklist some broken browsers
1953                         $m = array();
1954                         if( preg_match(
1955                                 '/\bgzip(?:;(q)=([0-9]+(?:\.[0-9]+)))?\b/',
1956                                 $_SERVER['HTTP_ACCEPT_ENCODING'],
1957                                 $m )
1958                         )
1959                         {
1960                                 if( isset( $m[2] ) && ( $m[1] == 'q' ) && ( $m[2] == 0 ) ) {
1961                                         $result = false;
1962                                         return $result;
1963                                 }
1964                                 wfDebug( "wfClientAcceptsGzip: client accepts gzip.\n" );
1965                                 $result = true;
1966                         }
1967                 }
1968         }
1969         return $result;
1970 }
1971
1972 /**
1973  * Obtain the offset and limit values from the request string;
1974  * used in special pages
1975  *
1976  * @param $deflimit Int default limit if none supplied
1977  * @param $optionname String Name of a user preference to check against
1978  * @return array
1979  *
1980  */
1981 function wfCheckLimits( $deflimit = 50, $optionname = 'rclimit' ) {
1982         global $wgRequest;
1983         return $wgRequest->getLimitOffset( $deflimit, $optionname );
1984 }
1985
1986 /**
1987  * Escapes the given text so that it may be output using addWikiText()
1988  * without any linking, formatting, etc. making its way through. This
1989  * is achieved by substituting certain characters with HTML entities.
1990  * As required by the callers, <nowiki> is not used.
1991  *
1992  * @param $text String: text to be escaped
1993  * @return String
1994  */
1995 function wfEscapeWikiText( $text ) {
1996         $text = strtr( "\n$text", array(
1997                 '"' => '&#34;', '&' => '&#38;', "'" => '&#39;', '<' => '&#60;',
1998                 '=' => '&#61;', '>' => '&#62;', '[' => '&#91;', ']' => '&#93;',
1999                 '{' => '&#123;', '|' => '&#124;', '}' => '&#125;',
2000                 "\n#" => "\n&#35;", "\n*" => "\n&#42;",
2001                 "\n:" => "\n&#58;", "\n;" => "\n&#59;",
2002                 '://' => '&#58;//', 'ISBN ' => 'ISBN&#32;', 'RFC ' => 'RFC&#32;',
2003         ) );
2004         return substr( $text, 1 );
2005 }
2006
2007 /**
2008  * Get the current unix timetstamp with microseconds.  Useful for profiling
2009  * @return Float
2010  */
2011 function wfTime() {
2012         return microtime( true );
2013 }
2014
2015 /**
2016  * Sets dest to source and returns the original value of dest
2017  * If source is NULL, it just returns the value, it doesn't set the variable
2018  * If force is true, it will set the value even if source is NULL
2019  *
2020  * @param $dest Mixed
2021  * @param $source Mixed
2022  * @param $force Bool
2023  * @return Mixed
2024  */
2025 function wfSetVar( &$dest, $source, $force = false ) {
2026         $temp = $dest;
2027         if ( !is_null( $source ) || $force ) {
2028                 $dest = $source;
2029         }
2030         return $temp;
2031 }
2032
2033 /**
2034  * As for wfSetVar except setting a bit
2035  *
2036  * @param $dest Int
2037  * @param $bit Int
2038  * @param $state Bool
2039  *
2040  * @return bool
2041  */
2042 function wfSetBit( &$dest, $bit, $state = true ) {
2043         $temp = (bool)( $dest & $bit );
2044         if ( !is_null( $state ) ) {
2045                 if ( $state ) {
2046                         $dest |= $bit;
2047                 } else {
2048                         $dest &= ~$bit;
2049                 }
2050         }
2051         return $temp;
2052 }
2053
2054 /**
2055  * A wrapper around the PHP function var_export().
2056  * Either print it or add it to the regular output ($wgOut).
2057  *
2058  * @param $var mixed A PHP variable to dump.
2059  */
2060 function wfVarDump( $var ) {
2061         global $wgOut;
2062         $s = str_replace( "\n", "<br />\n", var_export( $var, true ) . "\n" );
2063         if ( headers_sent() || !isset( $wgOut ) || !is_object( $wgOut ) ) {
2064                 print $s;
2065         } else {
2066                 $wgOut->addHTML( $s );
2067         }
2068 }
2069
2070 /**
2071  * Provide a simple HTTP error.
2072  *
2073  * @param $code Int|String
2074  * @param $label String
2075  * @param $desc String
2076  */
2077 function wfHttpError( $code, $label, $desc ) {
2078         global $wgOut;
2079         $wgOut->disable();
2080         header( "HTTP/1.0 $code $label" );
2081         header( "Status: $code $label" );
2082         $wgOut->sendCacheControl();
2083
2084         header( 'Content-type: text/html; charset=utf-8' );
2085         print "<!doctype html>" .
2086                 '<html><head><title>' .
2087                 htmlspecialchars( $label ) .
2088                 '</title></head><body><h1>' .
2089                 htmlspecialchars( $label ) .
2090                 '</h1><p>' .
2091                 nl2br( htmlspecialchars( $desc ) ) .
2092                 "</p></body></html>\n";
2093 }
2094
2095 /**
2096  * Clear away any user-level output buffers, discarding contents.
2097  *
2098  * Suitable for 'starting afresh', for instance when streaming
2099  * relatively large amounts of data without buffering, or wanting to
2100  * output image files without ob_gzhandler's compression.
2101  *
2102  * The optional $resetGzipEncoding parameter controls suppression of
2103  * the Content-Encoding header sent by ob_gzhandler; by default it
2104  * is left. See comments for wfClearOutputBuffers() for why it would
2105  * be used.
2106  *
2107  * Note that some PHP configuration options may add output buffer
2108  * layers which cannot be removed; these are left in place.
2109  *
2110  * @param $resetGzipEncoding Bool
2111  */
2112 function wfResetOutputBuffers( $resetGzipEncoding = true ) {
2113         if( $resetGzipEncoding ) {
2114                 // Suppress Content-Encoding and Content-Length
2115                 // headers from 1.10+s wfOutputHandler
2116                 global $wgDisableOutputCompression;
2117                 $wgDisableOutputCompression = true;
2118         }
2119         while( $status = ob_get_status() ) {
2120                 if( $status['type'] == 0 /* PHP_OUTPUT_HANDLER_INTERNAL */ ) {
2121                         // Probably from zlib.output_compression or other
2122                         // PHP-internal setting which can't be removed.
2123                         //
2124                         // Give up, and hope the result doesn't break
2125                         // output behavior.
2126                         break;
2127                 }
2128                 if( !ob_end_clean() ) {
2129                         // Could not remove output buffer handler; abort now
2130                         // to avoid getting in some kind of infinite loop.
2131                         break;
2132                 }
2133                 if( $resetGzipEncoding ) {
2134                         if( $status['name'] == 'ob_gzhandler' ) {
2135                                 // Reset the 'Content-Encoding' field set by this handler
2136                                 // so we can start fresh.
2137                                 if ( function_exists( 'header_remove' ) ) {
2138                                         // Available since PHP 5.3.0
2139                                         header_remove( 'Content-Encoding' );
2140                                 } else {
2141                                         // We need to provide a valid content-coding. See bug 28069
2142                                         header( 'Content-Encoding: identity' );
2143                                 }
2144                                 break;
2145                         }
2146                 }
2147         }
2148 }
2149
2150 /**
2151  * More legible than passing a 'false' parameter to wfResetOutputBuffers():
2152  *
2153  * Clear away output buffers, but keep the Content-Encoding header
2154  * produced by ob_gzhandler, if any.
2155  *
2156  * This should be used for HTTP 304 responses, where you need to
2157  * preserve the Content-Encoding header of the real result, but
2158  * also need to suppress the output of ob_gzhandler to keep to spec
2159  * and avoid breaking Firefox in rare cases where the headers and
2160  * body are broken over two packets.
2161  */
2162 function wfClearOutputBuffers() {
2163         wfResetOutputBuffers( false );
2164 }
2165
2166 /**
2167  * Converts an Accept-* header into an array mapping string values to quality
2168  * factors
2169  *
2170  * @param $accept String
2171  * @param $def String default
2172  * @return Array
2173  */
2174 function wfAcceptToPrefs( $accept, $def = '*/*' ) {
2175         # No arg means accept anything (per HTTP spec)
2176         if( !$accept ) {
2177                 return array( $def => 1.0 );
2178         }
2179
2180         $prefs = array();
2181
2182         $parts = explode( ',', $accept );
2183
2184         foreach( $parts as $part ) {
2185                 # @todo FIXME: Doesn't deal with params like 'text/html; level=1'
2186                 $values = explode( ';', trim( $part ) );
2187                 $match = array();
2188                 if ( count( $values ) == 1 ) {
2189                         $prefs[$values[0]] = 1.0;
2190                 } elseif ( preg_match( '/q\s*=\s*(\d*\.\d+)/', $values[1], $match ) ) {
2191                         $prefs[$values[0]] = floatval( $match[1] );
2192                 }
2193         }
2194
2195         return $prefs;
2196 }
2197
2198 /**
2199  * Checks if a given MIME type matches any of the keys in the given
2200  * array. Basic wildcards are accepted in the array keys.
2201  *
2202  * Returns the matching MIME type (or wildcard) if a match, otherwise
2203  * NULL if no match.
2204  *
2205  * @param $type String
2206  * @param $avail Array
2207  * @return string
2208  * @private
2209  */
2210 function mimeTypeMatch( $type, $avail ) {
2211         if( array_key_exists( $type, $avail ) ) {
2212                 return $type;
2213         } else {
2214                 $parts = explode( '/', $type );
2215                 if( array_key_exists( $parts[0] . '/*', $avail ) ) {
2216                         return $parts[0] . '/*';
2217                 } elseif( array_key_exists( '*/*', $avail ) ) {
2218                         return '*/*';
2219                 } else {
2220                         return null;
2221                 }
2222         }
2223 }
2224
2225 /**
2226  * Returns the 'best' match between a client's requested internet media types
2227  * and the server's list of available types. Each list should be an associative
2228  * array of type to preference (preference is a float between 0.0 and 1.0).
2229  * Wildcards in the types are acceptable.
2230  *
2231  * @param $cprefs Array: client's acceptable type list
2232  * @param $sprefs Array: server's offered types
2233  * @return string
2234  *
2235  * @todo FIXME: Doesn't handle params like 'text/plain; charset=UTF-8'
2236  * XXX: generalize to negotiate other stuff
2237  */
2238 function wfNegotiateType( $cprefs, $sprefs ) {
2239         $combine = array();
2240
2241         foreach( array_keys( $sprefs ) as $type ) {
2242                 $parts = explode( '/', $type );
2243                 if( $parts[1] != '*' ) {
2244                         $ckey = mimeTypeMatch( $type, $cprefs );
2245                         if( $ckey ) {
2246                                 $combine[$type] = $sprefs[$type] * $cprefs[$ckey];
2247                         }
2248                 }
2249         }
2250
2251         foreach( array_keys( $cprefs ) as $type ) {
2252                 $parts = explode( '/', $type );
2253                 if( $parts[1] != '*' && !array_key_exists( $type, $sprefs ) ) {
2254                         $skey = mimeTypeMatch( $type, $sprefs );
2255                         if( $skey ) {
2256                                 $combine[$type] = $sprefs[$skey] * $cprefs[$type];
2257                         }
2258                 }
2259         }
2260
2261         $bestq = 0;
2262         $besttype = null;
2263
2264         foreach( array_keys( $combine ) as $type ) {
2265                 if( $combine[$type] > $bestq ) {
2266                         $besttype = $type;
2267                         $bestq = $combine[$type];
2268                 }
2269         }
2270
2271         return $besttype;
2272 }
2273
2274 /**
2275  * Reference-counted warning suppression
2276  *
2277  * @param $end Bool
2278  */
2279 function wfSuppressWarnings( $end = false ) {
2280         static $suppressCount = 0;
2281         static $originalLevel = false;
2282
2283         if ( $end ) {
2284                 if ( $suppressCount ) {
2285                         --$suppressCount;
2286                         if ( !$suppressCount ) {
2287                                 error_reporting( $originalLevel );
2288                         }
2289                 }
2290         } else {
2291                 if ( !$suppressCount ) {
2292                         // E_DEPRECATED is undefined in PHP 5.2
2293                         if( !defined( 'E_DEPRECATED' ) ) {
2294                                 define( 'E_DEPRECATED', 8192 );
2295                         }
2296                         $originalLevel = error_reporting( E_ALL & ~( E_WARNING | E_NOTICE | E_USER_WARNING | E_USER_NOTICE | E_DEPRECATED ) );
2297                 }
2298                 ++$suppressCount;
2299         }
2300 }
2301
2302 /**
2303  * Restore error level to previous value
2304  */
2305 function wfRestoreWarnings() {
2306         wfSuppressWarnings( true );
2307 }
2308
2309 # Autodetect, convert and provide timestamps of various types
2310
2311 /**
2312  * Unix time - the number of seconds since 1970-01-01 00:00:00 UTC
2313  */
2314 define( 'TS_UNIX', 0 );
2315
2316 /**
2317  * MediaWiki concatenated string timestamp (YYYYMMDDHHMMSS)
2318  */
2319 define( 'TS_MW', 1 );
2320
2321 /**
2322  * MySQL DATETIME (YYYY-MM-DD HH:MM:SS)
2323  */
2324 define( 'TS_DB', 2 );
2325
2326 /**
2327  * RFC 2822 format, for E-mail and HTTP headers
2328  */
2329 define( 'TS_RFC2822', 3 );
2330
2331 /**
2332  * ISO 8601 format with no timezone: 1986-02-09T20:00:00Z
2333  *
2334  * This is used by Special:Export
2335  */
2336 define( 'TS_ISO_8601', 4 );
2337
2338 /**
2339  * An Exif timestamp (YYYY:MM:DD HH:MM:SS)
2340  *
2341  * @see http://exif.org/Exif2-2.PDF The Exif 2.2 spec, see page 28 for the
2342  *       DateTime tag and page 36 for the DateTimeOriginal and
2343  *       DateTimeDigitized tags.
2344  */
2345 define( 'TS_EXIF', 5 );
2346
2347 /**
2348  * Oracle format time.
2349  */
2350 define( 'TS_ORACLE', 6 );
2351
2352 /**
2353  * Postgres format time.
2354  */
2355 define( 'TS_POSTGRES', 7 );
2356
2357 /**
2358  * DB2 format time
2359  */
2360 define( 'TS_DB2', 8 );
2361
2362 /**
2363  * ISO 8601 basic format with no timezone: 19860209T200000Z.  This is used by ResourceLoader
2364  */
2365 define( 'TS_ISO_8601_BASIC', 9 );
2366
2367 /**
2368  * Get a timestamp string in one of various formats
2369  *
2370  * @param $outputtype Mixed: A timestamp in one of the supported formats, the
2371  *                    function will autodetect which format is supplied and act
2372  *                    accordingly.
2373  * @param $ts Mixed: the timestamp to convert or 0 for the current timestamp
2374  * @return Mixed: String / false The same date in the format specified in $outputtype or false
2375  */
2376 function wfTimestamp( $outputtype = TS_UNIX, $ts = 0 ) {
2377         $uts = 0;
2378         $da = array();
2379         $strtime = '';
2380
2381         if ( !$ts ) { // We want to catch 0, '', null... but not date strings starting with a letter.
2382                 $uts = time();
2383                 $strtime = "@$uts";
2384         } elseif ( preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)$/D', $ts, $da ) ) {
2385                 # TS_DB
2386         } elseif ( preg_match( '/^(\d{4}):(\d\d):(\d\d) (\d\d):(\d\d):(\d\d)$/D', $ts, $da ) ) {
2387                 # TS_EXIF
2388         } elseif ( preg_match( '/^(\d{4})(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$/D', $ts, $da ) ) {
2389                 # TS_MW
2390         } elseif ( preg_match( '/^-?\d{1,13}$/D', $ts ) ) {
2391                 # TS_UNIX
2392                 $uts = $ts;
2393                 $strtime = "@$ts"; // http://php.net/manual/en/datetime.formats.compound.php
2394         } elseif ( preg_match( '/^\d{2}-\d{2}-\d{4} \d{2}:\d{2}:\d{2}.\d{6}$/', $ts ) ) {
2395                 # TS_ORACLE // session altered to DD-MM-YYYY HH24:MI:SS.FF6
2396                 $strtime = preg_replace( '/(\d\d)\.(\d\d)\.(\d\d)(\.(\d+))?/', "$1:$2:$3",
2397                                 str_replace( '+00:00', 'UTC', $ts ) );
2398         } elseif ( preg_match( '/^(\d{4})-(\d{2})-(\d{2})T(\d{2}):(\d{2}):(\d{2})(?:\.*\d*)?Z$/', $ts, $da ) ) {
2399                 # TS_ISO_8601
2400         } elseif ( preg_match( '/^(\d{4})(\d{2})(\d{2})T(\d{2})(\d{2})(\d{2})(?:\.*\d*)?Z$/', $ts, $da ) ) {
2401                 #TS_ISO_8601_BASIC
2402         } elseif ( preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)\.*\d*[\+\- ](\d\d)$/', $ts, $da ) ) {
2403                 # TS_POSTGRES
2404         } elseif ( preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)\.*\d* GMT$/', $ts, $da ) ) {
2405                 # TS_POSTGRES
2406         } elseif (preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)\.\d\d\d$/', $ts, $da ) ) {
2407                 # TS_DB2
2408         } elseif ( preg_match( '/^[ \t\r\n]*([A-Z][a-z]{2},[ \t\r\n]*)?' . # Day of week
2409                                                         '\d\d?[ \t\r\n]*[A-Z][a-z]{2}[ \t\r\n]*\d{2}(?:\d{2})?' .  # dd Mon yyyy
2410                                                         '[ \t\r\n]*\d\d[ \t\r\n]*:[ \t\r\n]*\d\d[ \t\r\n]*:[ \t\r\n]*\d\d/S', $ts ) ) { # hh:mm:ss
2411                 # TS_RFC2822, accepting a trailing comment. See http://www.squid-cache.org/mail-archive/squid-users/200307/0122.html / r77171
2412                 # The regex is a superset of rfc2822 for readability
2413                 $strtime = strtok( $ts, ';' );
2414         } elseif ( preg_match( '/^[A-Z][a-z]{5,8}, \d\d-[A-Z][a-z]{2}-\d{2} \d\d:\d\d:\d\d/', $ts ) ) {
2415                 # TS_RFC850
2416                 $strtime = $ts;
2417         } elseif ( preg_match( '/^[A-Z][a-z]{2} [A-Z][a-z]{2} +\d{1,2} \d\d:\d\d:\d\d \d{4}/', $ts ) ) {
2418                 # asctime
2419                 $strtime = $ts;
2420         } else {
2421                 # Bogus value...
2422                 wfDebug("wfTimestamp() fed bogus time value: TYPE=$outputtype; VALUE=$ts\n");
2423
2424                 return false;
2425         }
2426
2427         static $formats = array(
2428                 TS_UNIX => 'U',
2429                 TS_MW => 'YmdHis',
2430                 TS_DB => 'Y-m-d H:i:s',
2431                 TS_ISO_8601 => 'Y-m-d\TH:i:s\Z',
2432                 TS_ISO_8601_BASIC => 'Ymd\THis\Z',
2433                 TS_EXIF => 'Y:m:d H:i:s', // This shouldn't ever be used, but is included for completeness
2434                 TS_RFC2822 => 'D, d M Y H:i:s',
2435                 TS_ORACLE => 'd-m-Y H:i:s.000000', // Was 'd-M-y h.i.s A' . ' +00:00' before r51500
2436                 TS_POSTGRES => 'Y-m-d H:i:s',
2437                 TS_DB2 => 'Y-m-d H:i:s',
2438         );
2439
2440         if ( !isset( $formats[$outputtype] ) ) {
2441                 throw new MWException( 'wfTimestamp() called with illegal output type.' );
2442         }
2443
2444         if ( function_exists( "date_create" ) ) {
2445                 if ( count( $da ) ) {
2446                         $ds = sprintf("%04d-%02d-%02dT%02d:%02d:%02d.00+00:00",
2447                                 (int)$da[1], (int)$da[2], (int)$da[3],
2448                                 (int)$da[4], (int)$da[5], (int)$da[6]);
2449
2450                         $d = date_create( $ds, new DateTimeZone( 'GMT' ) );
2451                 } elseif ( $strtime ) {
2452                         $d = date_create( $strtime, new DateTimeZone( 'GMT' ) );
2453                 } else {
2454                         return false;
2455                 }
2456
2457                 if ( !$d ) {
2458                         wfDebug("wfTimestamp() fed bogus time value: $outputtype; $ts\n");
2459                         return false;
2460                 }
2461
2462                 $output = $d->format( $formats[$outputtype] );
2463         } else {
2464                 if ( count( $da ) ) {
2465                         // Warning! gmmktime() acts oddly if the month or day is set to 0
2466                         // We may want to handle that explicitly at some point
2467                         $uts = gmmktime( (int)$da[4], (int)$da[5], (int)$da[6],
2468                                 (int)$da[2], (int)$da[3], (int)$da[1] );
2469                 } elseif ( $strtime ) {
2470                         $uts = strtotime( $strtime );
2471                 }
2472
2473                 if ( $uts === false ) {
2474                         wfDebug("wfTimestamp() can't parse the timestamp (non 32-bit time? Update php): $outputtype; $ts\n");
2475                         return false;
2476                 }
2477
2478                 if ( TS_UNIX == $outputtype ) {
2479                         return $uts;
2480                 }
2481                 $output = gmdate( $formats[$outputtype], $uts );
2482         }
2483
2484         if ( ( $outputtype == TS_RFC2822 ) || ( $outputtype == TS_POSTGRES ) ) {
2485                 $output .= ' GMT';
2486         }
2487
2488         return $output;
2489 }
2490
2491 /**
2492  * Return a formatted timestamp, or null if input is null.
2493  * For dealing with nullable timestamp columns in the database.
2494  *
2495  * @param $outputtype Integer
2496  * @param $ts String
2497  * @return String
2498  */
2499 function wfTimestampOrNull( $outputtype = TS_UNIX, $ts = null ) {
2500         if( is_null( $ts ) ) {
2501                 return null;
2502         } else {
2503                 return wfTimestamp( $outputtype, $ts );
2504         }
2505 }
2506
2507 /**
2508  * Convenience function; returns MediaWiki timestamp for the present time.
2509  *
2510  * @return string
2511  */
2512 function wfTimestampNow() {
2513         # return NOW
2514         return wfTimestamp( TS_MW, time() );
2515 }
2516
2517 /**
2518  * Check if the operating system is Windows
2519  *
2520  * @return Bool: true if it's Windows, False otherwise.
2521  */
2522 function wfIsWindows() {
2523         static $isWindows = null;
2524         if ( $isWindows === null ) {
2525                 $isWindows = substr( php_uname(), 0, 7 ) == 'Windows';
2526         }
2527         return $isWindows;
2528 }
2529
2530 /**
2531  * Check if we are running under HipHop
2532  *
2533  * @return Bool
2534  */
2535 function wfIsHipHop() {
2536         return function_exists( 'hphp_thread_set_warmup_enabled' );
2537 }
2538
2539 /**
2540  * Swap two variables
2541  *
2542  * @param $x Mixed
2543  * @param $y Mixed
2544  */
2545 function swap( &$x, &$y ) {
2546         $z = $x;
2547         $x = $y;
2548         $y = $z;
2549 }
2550
2551 /**
2552  * Tries to get the system directory for temporary files. The TMPDIR, TMP, and
2553  * TEMP environment variables are then checked in sequence, and if none are set
2554  * try sys_get_temp_dir() for PHP >= 5.2.1. All else fails, return /tmp for Unix
2555  * or C:\Windows\Temp for Windows and hope for the best.
2556  * It is common to call it with tempnam().
2557  *
2558  * NOTE: When possible, use instead the tmpfile() function to create
2559  * temporary files to avoid race conditions on file creation, etc.
2560  *
2561  * @return String
2562  */
2563 function wfTempDir() {
2564         foreach( array( 'TMPDIR', 'TMP', 'TEMP' ) as $var ) {
2565                 $tmp = getenv( $var );
2566                 if( $tmp && file_exists( $tmp ) && is_dir( $tmp ) && is_writable( $tmp ) ) {
2567                         return $tmp;
2568                 }
2569         }
2570         if( function_exists( 'sys_get_temp_dir' ) ) {
2571                 return sys_get_temp_dir();
2572         }
2573         # Usual defaults
2574         return wfIsWindows() ? 'C:\Windows\Temp' : '/tmp';
2575 }
2576
2577 /**
2578  * Make directory, and make all parent directories if they don't exist
2579  *
2580  * @param $dir String: full path to directory to create
2581  * @param $mode Integer: chmod value to use, default is $wgDirectoryMode
2582  * @param $caller String: optional caller param for debugging.
2583  * @return bool
2584  */
2585 function wfMkdirParents( $dir, $mode = null, $caller = null ) {
2586         global $wgDirectoryMode;
2587
2588         if ( FileBackend::isStoragePath( $dir ) ) { // sanity
2589                 throw new MWException( __FUNCTION__ . " given storage path `$dir`.");
2590         }
2591
2592         if ( !is_null( $caller ) ) {
2593                 wfDebug( "$caller: called wfMkdirParents($dir)\n" );
2594         }
2595
2596         if( strval( $dir ) === '' || file_exists( $dir ) ) {
2597                 return true;
2598         }
2599
2600         $dir = str_replace( array( '\\', '/' ), DIRECTORY_SEPARATOR, $dir );
2601
2602         if ( is_null( $mode ) ) {
2603                 $mode = $wgDirectoryMode;
2604         }
2605
2606         // Turn off the normal warning, we're doing our own below
2607         wfSuppressWarnings();
2608         $ok = mkdir( $dir, $mode, true ); // PHP5 <3
2609         wfRestoreWarnings();
2610
2611         if( !$ok ) {
2612                 // PHP doesn't report the path in its warning message, so add our own to aid in diagnosis.
2613                 trigger_error( __FUNCTION__ . ": failed to mkdir \"$dir\" mode $mode", E_USER_WARNING );
2614         }
2615         return $ok;
2616 }
2617
2618 /**
2619  * Remove a directory and all its content.
2620  * Does not hide error.
2621  */
2622 function wfRecursiveRemoveDir( $dir ) {
2623         wfDebug( __FUNCTION__ . "( $dir )\n" );
2624         // taken from http://de3.php.net/manual/en/function.rmdir.php#98622
2625         if ( is_dir( $dir ) ) {
2626                 $objects = scandir( $dir );
2627                 foreach ( $objects as $object ) {
2628                         if ( $object != "." && $object != ".." ) {
2629                                 if ( filetype( $dir . '/' . $object ) == "dir" ) {
2630                                         wfRecursiveRemoveDir( $dir . '/' . $object );
2631                                 } else {
2632                                         unlink( $dir . '/' . $object );
2633                                 }
2634                         }
2635                 }
2636                 reset( $objects );
2637                 rmdir( $dir );
2638         }
2639 }
2640
2641 /**
2642  * @param $nr Mixed: the number to format
2643  * @param $acc Integer: the number of digits after the decimal point, default 2
2644  * @param $round Boolean: whether or not to round the value, default true
2645  * @return float
2646  */
2647 function wfPercent( $nr, $acc = 2, $round = true ) {
2648         $ret = sprintf( "%.${acc}f", $nr );
2649         return $round ? round( $ret, $acc ) . '%' : "$ret%";
2650 }
2651
2652 /**
2653  * Find out whether or not a mixed variable exists in a string
2654  *
2655  * @param $needle String
2656  * @param $str String
2657  * @param $insensitive Boolean
2658  * @return Boolean
2659  */
2660 function in_string( $needle, $str, $insensitive = false ) {
2661         $func = 'strpos';
2662         if( $insensitive ) $func = 'stripos';
2663
2664         return $func( $str, $needle ) !== false;
2665 }
2666
2667 /**
2668  * Safety wrapper around ini_get() for boolean settings.
2669  * The values returned from ini_get() are pre-normalized for settings
2670  * set via php.ini or php_flag/php_admin_flag... but *not*
2671  * for those set via php_value/php_admin_value.
2672  *
2673  * It's fairly common for people to use php_value instead of php_flag,
2674  * which can leave you with an 'off' setting giving a false positive
2675  * for code that just takes the ini_get() return value as a boolean.
2676  *
2677  * To make things extra interesting, setting via php_value accepts
2678  * "true" and "yes" as true, but php.ini and php_flag consider them false. :)
2679  * Unrecognized values go false... again opposite PHP's own coercion
2680  * from string to bool.
2681  *
2682  * Luckily, 'properly' set settings will always come back as '0' or '1',
2683  * so we only have to worry about them and the 'improper' settings.
2684  *
2685  * I frickin' hate PHP... :P
2686  *
2687  * @param $setting String
2688  * @return Bool
2689  */
2690 function wfIniGetBool( $setting ) {
2691         $val = ini_get( $setting );
2692         // 'on' and 'true' can't have whitespace around them, but '1' can.
2693         return strtolower( $val ) == 'on'
2694                 || strtolower( $val ) == 'true'
2695                 || strtolower( $val ) == 'yes'
2696                 || preg_match( "/^\s*[+-]?0*[1-9]/", $val ); // approx C atoi() function
2697 }
2698
2699 /**
2700  * Wrapper function for PHP's dl(). This doesn't work in most situations from
2701  * PHP 5.3 onward, and is usually disabled in shared environments anyway.
2702  *
2703  * @param $extension String A PHP extension. The file suffix (.so or .dll)
2704  *                          should be omitted
2705  * @param $fileName String Name of the library, if not $extension.suffix
2706  * @return Bool - Whether or not the extension is loaded
2707  */
2708 function wfDl( $extension, $fileName = null ) {
2709         if( extension_loaded( $extension ) ) {
2710                 return true;
2711         }
2712
2713         $canDl = false;
2714         $sapi = php_sapi_name();
2715         if( version_compare( PHP_VERSION, '5.3.0', '<' ) ||
2716                 $sapi == 'cli' || $sapi == 'cgi' || $sapi == 'embed' )
2717         {
2718                 $canDl = ( function_exists( 'dl' ) && is_callable( 'dl' )
2719                 && wfIniGetBool( 'enable_dl' ) && !wfIniGetBool( 'safe_mode' ) );
2720         }
2721
2722         if( $canDl ) {
2723                 $fileName = $fileName ? $fileName : $extension;
2724                 if( wfIsWindows() ) {
2725                         $fileName = 'php_' . $fileName;
2726                 }
2727                 wfSuppressWarnings();
2728                 dl( $fileName . '.' . PHP_SHLIB_SUFFIX );
2729                 wfRestoreWarnings();
2730         }
2731         return extension_loaded( $extension );
2732 }
2733
2734 /**
2735  * Windows-compatible version of escapeshellarg()
2736  * Windows doesn't recognise single-quotes in the shell, but the escapeshellarg()
2737  * function puts single quotes in regardless of OS.
2738  *
2739  * Also fixes the locale problems on Linux in PHP 5.2.6+ (bug backported to
2740  * earlier distro releases of PHP)
2741  *
2742  * @param varargs
2743  * @return String
2744  */
2745 function wfEscapeShellArg( ) {
2746         wfInitShellLocale();
2747
2748         $args = func_get_args();
2749         $first = true;
2750         $retVal = '';
2751         foreach ( $args as $arg ) {
2752                 if ( !$first ) {
2753                         $retVal .= ' ';
2754                 } else {
2755                         $first = false;
2756                 }
2757
2758                 if ( wfIsWindows() ) {
2759                         // Escaping for an MSVC-style command line parser and CMD.EXE
2760                         // Refs:
2761                         //  * http://web.archive.org/web/20020708081031/http://mailman.lyra.org/pipermail/scite-interest/2002-March/000436.html
2762                         //  * http://technet.microsoft.com/en-us/library/cc723564.aspx
2763                         //  * Bug #13518
2764                         //  * CR r63214
2765                         // Double the backslashes before any double quotes. Escape the double quotes.
2766                         $tokens = preg_split( '/(\\\\*")/', $arg, -1, PREG_SPLIT_DELIM_CAPTURE );
2767                         $arg = '';
2768                         $iteration = 0;
2769                         foreach ( $tokens as $token ) {
2770                                 if ( $iteration % 2 == 1 ) {
2771                                         // Delimiter, a double quote preceded by zero or more slashes
2772                                         $arg .= str_replace( '\\', '\\\\', substr( $token, 0, -1 ) ) . '\\"';
2773                                 } elseif ( $iteration % 4 == 2 ) {
2774                                         // ^ in $token will be outside quotes, need to be escaped
2775                                         $arg .= str_replace( '^', '^^', $token );
2776                                 } else { // $iteration % 4 == 0
2777                                         // ^ in $token will appear inside double quotes, so leave as is
2778                                         $arg .= $token;
2779                                 }
2780                                 $iteration++;
2781                         }
2782                         // Double the backslashes before the end of the string, because
2783                         // we will soon add a quote
2784                         $m = array();
2785                         if ( preg_match( '/^(.*?)(\\\\+)$/', $arg, $m ) ) {
2786                                 $arg = $m[1] . str_replace( '\\', '\\\\', $m[2] );
2787                         }
2788
2789                         // Add surrounding quotes
2790                         $retVal .= '"' . $arg . '"';
2791                 } else {
2792                         $retVal .= escapeshellarg( $arg );
2793                 }
2794         }
2795         return $retVal;
2796 }
2797
2798 /**
2799  * Execute a shell command, with time and memory limits mirrored from the PHP
2800  * configuration if supported.
2801  * @param $cmd String Command line, properly escaped for shell.
2802  * @param &$retval null|Mixed optional, will receive the program's exit code.
2803  *                 (non-zero is usually failure)
2804  * @param $environ Array optional environment variables which should be
2805  *                 added to the executed command environment.
2806  * @return string collected stdout as a string (trailing newlines stripped)
2807  */
2808 function wfShellExec( $cmd, &$retval = null, $environ = array() ) {
2809         global $IP, $wgMaxShellMemory, $wgMaxShellFileSize, $wgMaxShellTime;
2810
2811         static $disabled;
2812         if ( is_null( $disabled ) ) {
2813                 $disabled = false;
2814                 if( wfIniGetBool( 'safe_mode' ) ) {
2815                         wfDebug( "wfShellExec can't run in safe_mode, PHP's exec functions are too broken.\n" );
2816                         $disabled = 'safemode';
2817                 } else {
2818                         $functions = explode( ',', ini_get( 'disable_functions' ) );
2819                         $functions = array_map( 'trim', $functions );
2820                         $functions = array_map( 'strtolower', $functions );
2821                         if ( in_array( 'passthru', $functions ) ) {
2822                                 wfDebug( "passthru is in disabled_functions\n" );
2823                                 $disabled = 'passthru';
2824                         }
2825                 }
2826         }
2827         if ( $disabled ) {
2828                 $retval = 1;
2829                 return $disabled == 'safemode' ?
2830                         'Unable to run external programs in safe mode.' :
2831                         'Unable to run external programs, passthru() is disabled.';
2832         }
2833
2834         wfInitShellLocale();
2835
2836         $envcmd = '';
2837         foreach( $environ as $k => $v ) {
2838                 if ( wfIsWindows() ) {
2839                         /* Surrounding a set in quotes (method used by wfEscapeShellArg) makes the quotes themselves
2840                          * appear in the environment variable, so we must use carat escaping as documented in
2841                          * http://technet.microsoft.com/en-us/library/cc723564.aspx
2842                          * Note however that the quote isn't listed there, but is needed, and the parentheses
2843                          * are listed there but doesn't appear to need it.
2844                          */
2845                         $envcmd .= "set $k=" . preg_replace( '/([&|()<>^"])/', '^\\1', $v ) . '&& ';
2846                 } else {
2847                         /* Assume this is a POSIX shell, thus required to accept variable assignments before the command
2848                          * http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_09_01
2849                          */
2850                         $envcmd .= "$k=" . escapeshellarg( $v ) . ' ';
2851                 }
2852         }
2853         $cmd = $envcmd . $cmd;
2854
2855         if ( wfIsWindows() ) {
2856                 if ( version_compare( PHP_VERSION, '5.3.0', '<' ) && /* Fixed in 5.3.0 :) */
2857                         ( version_compare( PHP_VERSION, '5.2.1', '>=' ) || php_uname( 's' ) == 'Windows NT' ) )
2858                 {
2859                         # Hack to work around PHP's flawed invocation of cmd.exe
2860                         # http://news.php.net/php.internals/21796
2861                         # Windows 9x doesn't accept any kind of quotes
2862                         $cmd = '"' . $cmd . '"';
2863                 }
2864         } elseif ( php_uname( 's' ) == 'Linux' ) {
2865                 $time = intval( $wgMaxShellTime );
2866                 $mem = intval( $wgMaxShellMemory );
2867                 $filesize = intval( $wgMaxShellFileSize );
2868
2869                 if ( $time > 0 && $mem > 0 ) {
2870                         $script = "$IP/bin/ulimit4.sh";
2871                         if ( is_executable( $script ) ) {
2872                                 $cmd = '/bin/bash ' . escapeshellarg( $script ) . " $time $mem $filesize " . escapeshellarg( $cmd );
2873                         }
2874                 }
2875         }
2876         wfDebug( "wfShellExec: $cmd\n" );
2877
2878         $retval = 1; // error by default?
2879         ob_start();
2880         passthru( $cmd, $retval );
2881         $output = ob_get_contents();
2882         ob_end_clean();
2883
2884         if ( $retval == 127 ) {
2885                 wfDebugLog( 'exec', "Possibly missing executable file: $cmd\n" );
2886         }
2887         return $output;
2888 }
2889
2890 /**
2891  * Workaround for http://bugs.php.net/bug.php?id=45132
2892  * escapeshellarg() destroys non-ASCII characters if LANG is not a UTF-8 locale
2893  */
2894 function wfInitShellLocale() {
2895         static $done = false;
2896         if ( $done ) {
2897                 return;
2898         }
2899         $done = true;
2900         global $wgShellLocale;
2901         if ( !wfIniGetBool( 'safe_mode' ) ) {
2902                 putenv( "LC_CTYPE=$wgShellLocale" );
2903                 setlocale( LC_CTYPE, $wgShellLocale );
2904         }
2905 }
2906
2907 /**
2908  * Generate a shell-escaped command line string to run a maintenance script.
2909  * Note that $parameters should be a flat array and an option with an argument
2910  * should consist of two consecutive items in the array (do not use "--option value").
2911  * @param $script string MediaWiki maintenance script path
2912  * @param $parameters Array Arguments and options to the script
2913  * @param $options Array Associative array of options:
2914  *              'php': The path to the php executable
2915  *              'wrapper': Path to a PHP wrapper to handle the maintenance script
2916  * @return Array
2917  */
2918 function wfShellMaintenanceCmd( $script, array $parameters = array(), array $options = array() ) {
2919         global $wgPhpCli;
2920         // Give site config file a chance to run the script in a wrapper.
2921         // The caller may likely want to call wfBasename() on $script.
2922         wfRunHooks( 'wfShellMaintenanceCmd', array( &$script, &$parameters, &$options ) );
2923         $cmd = isset( $options['php'] ) ? array( $options['php'] ) : array( $wgPhpCli );
2924         if ( isset( $options['wrapper'] ) ) {
2925                 $cmd[] = $options['wrapper'];
2926         }
2927         $cmd[] = $script;
2928         // Escape each parameter for shell
2929         return implode( " ", array_map( 'wfEscapeShellArg', array_merge( $cmd, $parameters ) ) );
2930 }
2931
2932 /**
2933  * wfMerge attempts to merge differences between three texts.
2934  * Returns true for a clean merge and false for failure or a conflict.
2935  *
2936  * @param $old String
2937  * @param $mine String
2938  * @param $yours String
2939  * @param $result String
2940  * @return Bool
2941  */
2942 function wfMerge( $old, $mine, $yours, &$result ) {
2943         global $wgDiff3;
2944
2945         # This check may also protect against code injection in
2946         # case of broken installations.
2947         wfSuppressWarnings();
2948         $haveDiff3 = $wgDiff3 && file_exists( $wgDiff3 );
2949         wfRestoreWarnings();
2950
2951         if( !$haveDiff3 ) {
2952                 wfDebug( "diff3 not found\n" );
2953                 return false;
2954         }
2955
2956         # Make temporary files
2957         $td = wfTempDir();
2958         $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
2959         $mytextFile = fopen( $mytextName = tempnam( $td, 'merge-mine-' ), 'w' );
2960         $yourtextFile = fopen( $yourtextName = tempnam( $td, 'merge-your-' ), 'w' );
2961
2962         fwrite( $oldtextFile, $old );
2963         fclose( $oldtextFile );
2964         fwrite( $mytextFile, $mine );
2965         fclose( $mytextFile );
2966         fwrite( $yourtextFile, $yours );
2967         fclose( $yourtextFile );
2968
2969         # Check for a conflict
2970         $cmd = $wgDiff3 . ' -a --overlap-only ' .
2971                 wfEscapeShellArg( $mytextName ) . ' ' .
2972                 wfEscapeShellArg( $oldtextName ) . ' ' .
2973                 wfEscapeShellArg( $yourtextName );
2974         $handle = popen( $cmd, 'r' );
2975
2976         if( fgets( $handle, 1024 ) ) {
2977                 $conflict = true;
2978         } else {
2979                 $conflict = false;
2980         }
2981         pclose( $handle );
2982
2983         # Merge differences
2984         $cmd = $wgDiff3 . ' -a -e --merge ' .
2985                 wfEscapeShellArg( $mytextName, $oldtextName, $yourtextName );
2986         $handle = popen( $cmd, 'r' );
2987         $result = '';
2988         do {
2989                 $data = fread( $handle, 8192 );
2990                 if ( strlen( $data ) == 0 ) {
2991                         break;
2992                 }
2993                 $result .= $data;
2994         } while ( true );
2995         pclose( $handle );
2996         unlink( $mytextName );
2997         unlink( $oldtextName );
2998         unlink( $yourtextName );
2999
3000         if ( $result === '' && $old !== '' && !$conflict ) {
3001                 wfDebug( "Unexpected null result from diff3. Command: $cmd\n" );
3002                 $conflict = true;
3003         }
3004         return !$conflict;
3005 }
3006
3007 /**
3008  * Returns unified plain-text diff of two texts.
3009  * Useful for machine processing of diffs.
3010  *
3011  * @param $before String: the text before the changes.
3012  * @param $after String: the text after the changes.
3013  * @param $params String: command-line options for the diff command.
3014  * @return String: unified diff of $before and $after
3015  */
3016 function wfDiff( $before, $after, $params = '-u' ) {
3017         if ( $before == $after ) {
3018                 return '';
3019         }
3020
3021         global $wgDiff;
3022         wfSuppressWarnings();
3023         $haveDiff = $wgDiff && file_exists( $wgDiff );
3024         wfRestoreWarnings();
3025
3026         # This check may also protect against code injection in
3027         # case of broken installations.
3028         if( !$haveDiff ) {
3029                 wfDebug( "diff executable not found\n" );
3030                 $diffs = new Diff( explode( "\n", $before ), explode( "\n", $after ) );
3031                 $format = new UnifiedDiffFormatter();
3032                 return $format->format( $diffs );
3033         }
3034
3035         # Make temporary files
3036         $td = wfTempDir();
3037         $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
3038         $newtextFile = fopen( $newtextName = tempnam( $td, 'merge-your-' ), 'w' );
3039
3040         fwrite( $oldtextFile, $before );
3041         fclose( $oldtextFile );
3042         fwrite( $newtextFile, $after );
3043         fclose( $newtextFile );
3044
3045         // Get the diff of the two files
3046         $cmd = "$wgDiff " . $params . ' ' . wfEscapeShellArg( $oldtextName, $newtextName );
3047
3048         $h = popen( $cmd, 'r' );
3049
3050         $diff = '';
3051
3052         do {
3053                 $data = fread( $h, 8192 );
3054                 if ( strlen( $data ) == 0 ) {
3055                         break;
3056                 }
3057                 $diff .= $data;
3058         } while ( true );
3059
3060         // Clean up
3061         pclose( $h );
3062         unlink( $oldtextName );
3063         unlink( $newtextName );
3064
3065         // Kill the --- and +++ lines. They're not useful.
3066         $diff_lines = explode( "\n", $diff );
3067         if ( strpos( $diff_lines[0], '---' ) === 0 ) {
3068                 unset( $diff_lines[0] );
3069         }
3070         if ( strpos( $diff_lines[1], '+++' ) === 0 ) {
3071                 unset( $diff_lines[1] );
3072         }
3073
3074         $diff = implode( "\n", $diff_lines );
3075
3076         return $diff;
3077 }
3078
3079 /**
3080  * This function works like "use VERSION" in Perl, the program will die with a
3081  * backtrace if the current version of PHP is less than the version provided
3082  *
3083  * This is useful for extensions which due to their nature are not kept in sync
3084  * with releases, and might depend on other versions of PHP than the main code
3085  *
3086  * Note: PHP might die due to parsing errors in some cases before it ever
3087  *       manages to call this function, such is life
3088  *
3089  * @see perldoc -f use
3090  *
3091  * @param $req_ver Mixed: the version to check, can be a string, an integer, or
3092  *                 a float
3093  */
3094 function wfUsePHP( $req_ver ) {
3095         $php_ver = PHP_VERSION;
3096
3097         if ( version_compare( $php_ver, (string)$req_ver, '<' ) ) {
3098                 throw new MWException( "PHP $req_ver required--this is only $php_ver" );
3099         }
3100 }
3101
3102 /**
3103  * This function works like "use VERSION" in Perl except it checks the version
3104  * of MediaWiki, the program will die with a backtrace if the current version
3105  * of MediaWiki is less than the version provided.
3106  *
3107  * This is useful for extensions which due to their nature are not kept in sync
3108  * with releases
3109  *
3110  * @see perldoc -f use
3111  *
3112  * @param $req_ver Mixed: the version to check, can be a string, an integer, or
3113  *                 a float
3114  */
3115 function wfUseMW( $req_ver ) {
3116         global $wgVersion;
3117
3118         if ( version_compare( $wgVersion, (string)$req_ver, '<' ) ) {
3119                 throw new MWException( "MediaWiki $req_ver required--this is only $wgVersion" );
3120         }
3121 }
3122
3123 /**
3124  * Return the final portion of a pathname.
3125  * Reimplemented because PHP5's basename() is buggy with multibyte text.
3126  * http://bugs.php.net/bug.php?id=33898
3127  *
3128  * PHP's basename() only considers '\' a pathchar on Windows and Netware.
3129  * We'll consider it so always, as we don't want \s in our Unix paths either.
3130  *
3131  * @param $path String
3132  * @param $suffix String: to remove if present
3133  * @return String
3134  */
3135 function wfBaseName( $path, $suffix = '' ) {
3136         $encSuffix = ( $suffix == '' )
3137                 ? ''
3138                 : ( '(?:' . preg_quote( $suffix, '#' ) . ')?' );
3139         $matches = array();
3140         if( preg_match( "#([^/\\\\]*?){$encSuffix}[/\\\\]*$#", $path, $matches ) ) {
3141                 return $matches[1];
3142         } else {
3143                 return '';
3144         }
3145 }
3146
3147 /**
3148  * Generate a relative path name to the given file.
3149  * May explode on non-matching case-insensitive paths,
3150  * funky symlinks, etc.
3151  *
3152  * @param $path String: absolute destination path including target filename
3153  * @param $from String: Absolute source path, directory only
3154  * @return String
3155  */
3156 function wfRelativePath( $path, $from ) {
3157         // Normalize mixed input on Windows...
3158         $path = str_replace( '/', DIRECTORY_SEPARATOR, $path );
3159         $from = str_replace( '/', DIRECTORY_SEPARATOR, $from );
3160
3161         // Trim trailing slashes -- fix for drive root
3162         $path = rtrim( $path, DIRECTORY_SEPARATOR );
3163         $from = rtrim( $from, DIRECTORY_SEPARATOR );
3164
3165         $pieces = explode( DIRECTORY_SEPARATOR, dirname( $path ) );
3166         $against = explode( DIRECTORY_SEPARATOR, $from );
3167
3168         if( $pieces[0] !== $against[0] ) {
3169                 // Non-matching Windows drive letters?
3170                 // Return a full path.
3171                 return $path;
3172         }
3173
3174         // Trim off common prefix
3175         while( count( $pieces ) && count( $against )
3176                 && $pieces[0] == $against[0] ) {
3177                 array_shift( $pieces );
3178                 array_shift( $against );
3179         }
3180
3181         // relative dots to bump us to the parent
3182         while( count( $against ) ) {
3183                 array_unshift( $pieces, '..' );
3184                 array_shift( $against );
3185         }
3186
3187         array_push( $pieces, wfBaseName( $path ) );
3188
3189         return implode( DIRECTORY_SEPARATOR, $pieces );
3190 }
3191
3192 /**
3193  * Do any deferred updates and clear the list
3194  *
3195  * @deprecated since 1.19
3196  * @see DeferredUpdates::doUpdate()
3197  * @param $commit string
3198  */
3199 function wfDoUpdates( $commit = '' ) {
3200         wfDeprecated( __METHOD__, '1.19' );
3201         DeferredUpdates::doUpdates( $commit );
3202 }
3203
3204 /**
3205  * Convert an arbitrarily-long digit string from one numeric base
3206  * to another, optionally zero-padding to a minimum column width.
3207  *
3208  * Supports base 2 through 36; digit values 10-36 are represented
3209  * as lowercase letters a-z. Input is case-insensitive.
3210  *
3211  * @param $input String: of digits
3212  * @param $sourceBase Integer: 2-36
3213  * @param $destBase Integer: 2-36
3214  * @param $pad Integer: 1 or greater
3215  * @param $lowercase Boolean
3216  * @return String or false on invalid input
3217  */
3218 function wfBaseConvert( $input, $sourceBase, $destBase, $pad = 1, $lowercase = true ) {
3219         $input = strval( $input );
3220         if( $sourceBase < 2 ||
3221                 $sourceBase > 36 ||
3222                 $destBase < 2 ||
3223                 $destBase > 36 ||
3224                 $pad < 1 ||
3225                 $sourceBase != intval( $sourceBase ) ||
3226                 $destBase != intval( $destBase ) ||
3227                 $pad != intval( $pad ) ||
3228                 !is_string( $input ) ||
3229                 $input == '' ) {
3230                 return false;
3231         }
3232         $digitChars = ( $lowercase ) ? '0123456789abcdefghijklmnopqrstuvwxyz' : '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ';
3233         $inDigits = array();
3234         $outChars = '';
3235
3236         // Decode and validate input string
3237         $input = strtolower( $input );
3238         for( $i = 0; $i < strlen( $input ); $i++ ) {
3239                 $n = strpos( $digitChars, $input[$i] );
3240                 if( $n === false || $n > $sourceBase ) {
3241                         return false;
3242                 }
3243                 $inDigits[] = $n;
3244         }
3245
3246         // Iterate over the input, modulo-ing out an output digit
3247         // at a time until input is gone.
3248         while( count( $inDigits ) ) {
3249                 $work = 0;
3250                 $workDigits = array();
3251
3252                 // Long division...
3253                 foreach( $inDigits as $digit ) {
3254                         $work *= $sourceBase;
3255                         $work += $digit;
3256
3257                         if( $work < $destBase ) {
3258                                 // Gonna need to pull another digit.
3259                                 if( count( $workDigits ) ) {
3260                                         // Avoid zero-padding; this lets us find
3261                                         // the end of the input very easily when
3262                                         // length drops to zero.
3263                                         $workDigits[] = 0;
3264                                 }
3265                         } else {
3266                                 // Finally! Actual division!
3267                                 $workDigits[] = intval( $work / $destBase );
3268
3269                                 // Isn't it annoying that most programming languages
3270                                 // don't have a single divide-and-remainder operator,
3271                                 // even though the CPU implements it that way?
3272                                 $work = $work % $destBase;
3273                         }
3274                 }
3275
3276                 // All that division leaves us with a remainder,
3277                 // which is conveniently our next output digit.
3278                 $outChars .= $digitChars[$work];
3279
3280                 // And we continue!
3281                 $inDigits = $workDigits;
3282         }
3283
3284         while( strlen( $outChars ) < $pad ) {
3285                 $outChars .= '0';
3286         }
3287
3288         return strrev( $outChars );
3289 }
3290
3291 /**
3292  * Create an object with a given name and an array of construct parameters
3293  *
3294  * @param $name String
3295  * @param $p Array: parameters
3296  * @return object
3297  * @deprecated since 1.18, warnings in 1.18, removal in 1.20
3298  */
3299 function wfCreateObject( $name, $p ) {
3300         wfDeprecated( __FUNCTION__, '1.18' );
3301         return MWFunction::newObj( $name, $p );
3302 }
3303
3304 /**
3305  * @return bool
3306  */
3307 function wfHttpOnlySafe() {
3308         global $wgHttpOnlyBlacklist;
3309
3310         if( isset( $_SERVER['HTTP_USER_AGENT'] ) ) {
3311                 foreach( $wgHttpOnlyBlacklist as $regex ) {
3312                         if( preg_match( $regex, $_SERVER['HTTP_USER_AGENT'] ) ) {
3313                                 return false;
3314                         }
3315                 }
3316         }
3317
3318         return true;
3319 }
3320
3321 /**
3322  * Initialise php session
3323  *
3324  * @param $sessionId Bool
3325  */
3326 function wfSetupSession( $sessionId = false ) {
3327         global $wgSessionsInMemcached, $wgCookiePath, $wgCookieDomain,
3328                         $wgCookieSecure, $wgCookieHttpOnly, $wgSessionHandler;
3329         if( $wgSessionsInMemcached ) {
3330                 if ( !defined( 'MW_COMPILED' ) ) {
3331                         global $IP;
3332                         require_once( "$IP/includes/cache/MemcachedSessions.php" );
3333                 }
3334                 session_set_save_handler( 'memsess_open', 'memsess_close', 'memsess_read',
3335                         'memsess_write', 'memsess_destroy', 'memsess_gc' );
3336
3337                 // It's necessary to register a shutdown function to call session_write_close(),
3338                 // because by the time the request shutdown function for the session module is
3339                 // called, $wgMemc has already been destroyed. Shutdown functions registered
3340                 // this way are called before object destruction.
3341                 register_shutdown_function( 'memsess_write_close' );
3342         } elseif( $wgSessionHandler && $wgSessionHandler != ini_get( 'session.save_handler' ) ) {
3343                 # Only set this if $wgSessionHandler isn't null and session.save_handler
3344                 # hasn't already been set to the desired value (that causes errors)
3345                 ini_set( 'session.save_handler', $wgSessionHandler );
3346         }
3347         $httpOnlySafe = wfHttpOnlySafe() && $wgCookieHttpOnly;
3348         wfDebugLog( 'cookie',
3349                 'session_set_cookie_params: "' . implode( '", "',
3350                         array(
3351                                 0,
3352                                 $wgCookiePath,
3353                                 $wgCookieDomain,
3354                                 $wgCookieSecure,
3355                                 $httpOnlySafe ) ) . '"' );
3356         session_set_cookie_params( 0, $wgCookiePath, $wgCookieDomain, $wgCookieSecure, $httpOnlySafe );
3357         session_cache_limiter( 'private, must-revalidate' );
3358         if ( $sessionId ) {
3359                 session_id( $sessionId );
3360         }
3361         wfSuppressWarnings();
3362         session_start();
3363         wfRestoreWarnings();
3364 }
3365
3366 /**
3367  * Get an object from the precompiled serialized directory
3368  *
3369  * @param $name String
3370  * @return Mixed: the variable on success, false on failure
3371  */
3372 function wfGetPrecompiledData( $name ) {
3373         global $IP;
3374
3375         $file = "$IP/serialized/$name";
3376         if ( file_exists( $file ) ) {
3377                 $blob = file_get_contents( $file );
3378                 if ( $blob ) {
3379                         return unserialize( $blob );
3380                 }
3381         }
3382         return false;
3383 }
3384
3385 /**
3386  * Get a cache key
3387  *
3388  * @param varargs
3389  * @return String
3390  */
3391 function wfMemcKey( /*... */ ) {
3392         global $wgCachePrefix;
3393         $prefix = $wgCachePrefix === false ? wfWikiID() : $wgCachePrefix;
3394         $args = func_get_args();
3395         $key = $prefix . ':' . implode( ':', $args );
3396         $key = str_replace( ' ', '_', $key );
3397         return $key;
3398 }
3399
3400 /**
3401  * Get a cache key for a foreign DB
3402  *
3403  * @param $db String
3404  * @param $prefix String
3405  * @param varargs String
3406  * @return String
3407  */
3408 function wfForeignMemcKey( $db, $prefix /*, ... */ ) {
3409         $args = array_slice( func_get_args(), 2 );
3410         if ( $prefix ) {
3411                 $key = "$db-$prefix:" . implode( ':', $args );
3412         } else {
3413                 $key = $db . ':' . implode( ':', $args );
3414         }
3415         return $key;
3416 }
3417
3418 /**
3419  * Get an ASCII string identifying this wiki
3420  * This is used as a prefix in memcached keys
3421  *
3422  * @return String
3423  */
3424 function wfWikiID() {
3425         global $wgDBprefix, $wgDBname;
3426         if ( $wgDBprefix ) {
3427                 return "$wgDBname-$wgDBprefix";
3428         } else {
3429                 return $wgDBname;
3430         }
3431 }
3432
3433 /**
3434  * Split a wiki ID into DB name and table prefix
3435  *
3436  * @param $wiki String
3437  *
3438  * @return array
3439  */
3440 function wfSplitWikiID( $wiki ) {
3441         $bits = explode( '-', $wiki, 2 );
3442         if ( count( $bits ) < 2 ) {
3443                 $bits[] = '';
3444         }
3445         return $bits;
3446 }
3447
3448 /**
3449  * Get a Database object.
3450  *
3451  * @param $db Integer: index of the connection to get. May be DB_MASTER for the
3452  *            master (for write queries), DB_SLAVE for potentially lagged read
3453  *            queries, or an integer >= 0 for a particular server.
3454  *
3455  * @param $groups Mixed: query groups. An array of group names that this query
3456  *                belongs to. May contain a single string if the query is only
3457  *                in one group.
3458  *
3459  * @param $wiki String: the wiki ID, or false for the current wiki
3460  *
3461  * Note: multiple calls to wfGetDB(DB_SLAVE) during the course of one request
3462  * will always return the same object, unless the underlying connection or load
3463  * balancer is manually destroyed.
3464  *
3465  * Note 2: use $this->getDB() in maintenance scripts that may be invoked by
3466  * updater to ensure that a proper database is being updated.
3467  *
3468  * @return DatabaseBase
3469  */
3470 function &wfGetDB( $db, $groups = array(), $wiki = false ) {
3471         return wfGetLB( $wiki )->getConnection( $db, $groups, $wiki );
3472 }
3473
3474 /**
3475  * Get a load balancer object.
3476  *
3477  * @param $wiki String: wiki ID, or false for the current wiki
3478  * @return LoadBalancer
3479  */
3480 function wfGetLB( $wiki = false ) {
3481         return wfGetLBFactory()->getMainLB( $wiki );
3482 }
3483
3484 /**
3485  * Get the load balancer factory object
3486  *
3487  * @return LBFactory
3488  */
3489 function &wfGetLBFactory() {
3490         return LBFactory::singleton();
3491 }
3492
3493 /**
3494  * Find a file.
3495  * Shortcut for RepoGroup::singleton()->findFile()
3496  *
3497  * @param $title String or Title object
3498  * @param $options array Associative array of options:
3499  *     time:           requested time for an archived image, or false for the
3500  *                     current version. An image object will be returned which was
3501  *                     created at the specified time.
3502  *
3503  *     ignoreRedirect: If true, do not follow file redirects
3504  *
3505  *     private:        If true, return restricted (deleted) files if the current
3506  *                     user is allowed to view them. Otherwise, such files will not
3507  *                     be found.
3508  *
3509  *     bypassCache:    If true, do not use the process-local cache of File objects
3510  *
3511  * @return File, or false if the file does not exist
3512  */
3513 function wfFindFile( $title, $options = array() ) {
3514         return RepoGroup::singleton()->findFile( $title, $options );
3515 }
3516
3517 /**
3518  * Get an object referring to a locally registered file.
3519  * Returns a valid placeholder object if the file does not exist.
3520  *
3521  * @param $title Title|String
3522  * @return File|null A File, or null if passed an invalid Title
3523  */
3524 function wfLocalFile( $title ) {
3525         return RepoGroup::singleton()->getLocalRepo()->newFile( $title );
3526 }
3527
3528 /**
3529  * Stream a file to the browser. Back-compat alias for StreamFile::stream()
3530  * @deprecated since 1.19
3531  */
3532 function wfStreamFile( $fname, $headers = array() ) {
3533         wfDeprecated( __FUNCTION__, '1.19' );
3534         StreamFile::stream( $fname, $headers );
3535 }
3536
3537 /**
3538  * Should low-performance queries be disabled?
3539  *
3540  * @return Boolean
3541  * @codeCoverageIgnore
3542  */
3543 function wfQueriesMustScale() {
3544         global $wgMiserMode;
3545         return $wgMiserMode
3546                 || ( SiteStats::pages() > 100000
3547                 && SiteStats::edits() > 1000000
3548                 && SiteStats::users() > 10000 );
3549 }
3550
3551 /**
3552  * Get the path to a specified script file, respecting file
3553  * extensions; this is a wrapper around $wgScriptExtension etc.
3554  *
3555  * @param $script String: script filename, sans extension
3556  * @return String
3557  */
3558 function wfScript( $script = 'index' ) {
3559         global $wgScriptPath, $wgScriptExtension;
3560         return "{$wgScriptPath}/{$script}{$wgScriptExtension}";
3561 }
3562
3563 /**
3564  * Get the script URL.
3565  *
3566  * @return string script URL
3567  */
3568 function wfGetScriptUrl() {
3569         if( isset( $_SERVER['SCRIPT_NAME'] ) ) {
3570                 #
3571                 # as it was called, minus the query string.
3572                 #
3573                 # Some sites use Apache rewrite rules to handle subdomains,
3574                 # and have PHP set up in a weird way that causes PHP_SELF
3575                 # to contain the rewritten URL instead of the one that the
3576                 # outside world sees.
3577                 #
3578                 # If in this mode, use SCRIPT_URL instead, which mod_rewrite
3579                 # provides containing the "before" URL.
3580                 return $_SERVER['SCRIPT_NAME'];
3581         } else {
3582                 return $_SERVER['URL'];
3583         }
3584 }
3585
3586 /**
3587  * Convenience function converts boolean values into "true"
3588  * or "false" (string) values
3589  *
3590  * @param $value Boolean
3591  * @return String
3592  */
3593 function wfBoolToStr( $value ) {
3594         return $value ? 'true' : 'false';
3595 }
3596
3597 /**
3598  * Load an extension messages file
3599  *
3600  * @deprecated since 1.16, warnings in 1.18, remove in 1.20
3601  * @codeCoverageIgnore
3602  */
3603 function wfLoadExtensionMessages() {
3604         wfDeprecated( __FUNCTION__, '1.16' );
3605 }
3606
3607 /**
3608  * Get a platform-independent path to the null file, e.g. /dev/null
3609  *
3610  * @return string
3611  */
3612 function wfGetNull() {
3613         return wfIsWindows()
3614                 ? 'NUL'
3615                 : '/dev/null';
3616 }
3617
3618 /**
3619  * Modern version of wfWaitForSlaves(). Instead of looking at replication lag
3620  * and waiting for it to go down, this waits for the slaves to catch up to the
3621  * master position. Use this when updating very large numbers of rows, as
3622  * in maintenance scripts, to avoid causing too much lag.  Of course, this is
3623  * a no-op if there are no slaves.
3624  *
3625  * @param $maxLag Integer (deprecated)
3626  * @param $wiki mixed Wiki identifier accepted by wfGetLB
3627  */
3628 function wfWaitForSlaves( $maxLag = false, $wiki = false ) {
3629         $lb = wfGetLB( $wiki );
3630         // bug 27975 - Don't try to wait for slaves if there are none
3631         // Prevents permission error when getting master position
3632         if ( $lb->getServerCount() > 1 ) {
3633                 $dbw = $lb->getConnection( DB_MASTER );
3634                 $pos = $dbw->getMasterPos();
3635                 $lb->waitForAll( $pos );
3636         }
3637 }
3638
3639 /**
3640  * Used to be used for outputting text in the installer/updater
3641  * @deprecated since 1.18, warnings in 1.18, remove in 1.20
3642  */
3643 function wfOut( $s ) {
3644         wfDeprecated( __FUNCTION__, '1.18' );
3645         global $wgCommandLineMode;
3646         if ( $wgCommandLineMode ) {
3647                 echo $s;
3648         } else {
3649                 echo htmlspecialchars( $s );
3650         }
3651         flush();
3652 }
3653
3654 /**
3655  * Count down from $n to zero on the terminal, with a one-second pause
3656  * between showing each number. For use in command-line scripts.
3657  * @codeCoverageIgnore
3658  * @param $n int
3659  */
3660 function wfCountDown( $n ) {
3661         for ( $i = $n; $i >= 0; $i-- ) {
3662                 if ( $i != $n ) {
3663                         echo str_repeat( "\x08", strlen( $i + 1 ) );
3664                 }
3665                 echo $i;
3666                 flush();
3667                 if ( $i ) {
3668                         sleep( 1 );
3669                 }
3670         }
3671         echo "\n";
3672 }
3673
3674 /**
3675  * Generate a random 32-character hexadecimal token.
3676  * @param $salt Mixed: some sort of salt, if necessary, to add to random
3677  *              characters before hashing.
3678  * @return string
3679  * @codeCoverageIgnore
3680  */
3681 function wfGenerateToken( $salt = '' ) {
3682         $salt = serialize( $salt );
3683         return md5( mt_rand( 0, 0x7fffffff ) . $salt );
3684 }
3685
3686 /**
3687  * Replace all invalid characters with -
3688  *
3689  * @param $name Mixed: filename to process
3690  * @return String
3691  */
3692 function wfStripIllegalFilenameChars( $name ) {
3693         global $wgIllegalFileChars;
3694         $name = wfBaseName( $name );
3695         $name = preg_replace(
3696                 "/[^" . Title::legalChars() . "]" .
3697                         ( $wgIllegalFileChars ? "|[" . $wgIllegalFileChars . "]" : '' ) .
3698                         "/",
3699                 '-',
3700                 $name
3701         );
3702         return $name;
3703 }
3704
3705 /**
3706  * Set PHP's memory limit to the larger of php.ini or $wgMemoryLimit;
3707  *
3708  * @return Integer value memory was set to.
3709  */
3710 function wfMemoryLimit() {
3711         global $wgMemoryLimit;
3712         $memlimit = wfShorthandToInteger( ini_get( 'memory_limit' ) );
3713         if( $memlimit != -1 ) {
3714                 $conflimit = wfShorthandToInteger( $wgMemoryLimit );
3715                 if( $conflimit == -1 ) {
3716                         wfDebug( "Removing PHP's memory limit\n" );
3717                         wfSuppressWarnings();
3718                         ini_set( 'memory_limit', $conflimit );
3719                         wfRestoreWarnings();
3720                         return $conflimit;
3721                 } elseif ( $conflimit > $memlimit ) {
3722                         wfDebug( "Raising PHP's memory limit to $conflimit bytes\n" );
3723                         wfSuppressWarnings();
3724                         ini_set( 'memory_limit', $conflimit );
3725                         wfRestoreWarnings();
3726                         return $conflimit;
3727                 }
3728         }
3729         return $memlimit;
3730 }
3731
3732 /**
3733  * Converts shorthand byte notation to integer form
3734  *
3735  * @param $string String
3736  * @return Integer
3737  */
3738 function wfShorthandToInteger( $string = '' ) {
3739         $string = trim( $string );
3740         if( $string === '' ) {
3741                 return -1;
3742         }
3743         $last = $string[strlen( $string ) - 1];
3744         $val = intval( $string );
3745         switch( $last ) {
3746                 case 'g':
3747                 case 'G':
3748                         $val *= 1024;
3749                         // break intentionally missing
3750                 case 'm':
3751                 case 'M':
3752                         $val *= 1024;
3753                         // break intentionally missing
3754                 case 'k':
3755                 case 'K':
3756                         $val *= 1024;
3757         }
3758
3759         return $val;
3760 }
3761
3762 /**
3763  * Get the normalised IETF language tag
3764  * See unit test for examples.
3765  *
3766  * @param $code String: The language code.
3767  * @return String: The language code which complying with BCP 47 standards.
3768  */
3769 function wfBCP47( $code ) {
3770         $codeSegment = explode( '-', $code );
3771         $codeBCP = array();
3772         foreach ( $codeSegment as $segNo => $seg ) {
3773                 if ( count( $codeSegment ) > 0 ) {
3774                         // when previous segment is x, it is a private segment and should be lc
3775                         if( $segNo > 0 && strtolower( $codeSegment[( $segNo - 1 )] ) == 'x' ) {
3776                                 $codeBCP[$segNo] = strtolower( $seg );
3777                         // ISO 3166 country code
3778                         } elseif ( ( strlen( $seg ) == 2 ) && ( $segNo > 0 ) ) {
3779                                 $codeBCP[$segNo] = strtoupper( $seg );
3780                         // ISO 15924 script code
3781                         } elseif ( ( strlen( $seg ) == 4 ) && ( $segNo > 0 ) ) {
3782                                 $codeBCP[$segNo] = ucfirst( strtolower( $seg ) );
3783                         // Use lowercase for other cases
3784                         } else {
3785                                 $codeBCP[$segNo] = strtolower( $seg );
3786                         }
3787                 } else {
3788                 // Use lowercase for single segment
3789                         $codeBCP[$segNo] = strtolower( $seg );
3790                 }
3791         }
3792         $langCode = implode( '-', $codeBCP );
3793         return $langCode;
3794 }
3795
3796 /**
3797  * Get a cache object.
3798  *
3799  * @param $inputType integer Cache type, one the the CACHE_* constants.
3800  * @return BagOStuff
3801  */
3802 function wfGetCache( $inputType ) {
3803         return ObjectCache::getInstance( $inputType );
3804 }
3805
3806 /**
3807  * Get the main cache object
3808  *
3809  * @return BagOStuff
3810  */
3811 function wfGetMainCache() {
3812         global $wgMainCacheType;
3813         return ObjectCache::getInstance( $wgMainCacheType );
3814 }
3815
3816 /**
3817  * Get the cache object used by the message cache
3818  *
3819  * @return BagOStuff
3820  */
3821 function wfGetMessageCacheStorage() {
3822         global $wgMessageCacheType;
3823         return ObjectCache::getInstance( $wgMessageCacheType );
3824 }
3825
3826 /**
3827  * Get the cache object used by the parser cache
3828  *
3829  * @return BagOStuff
3830  */
3831 function wfGetParserCacheStorage() {
3832         global $wgParserCacheType;
3833         return ObjectCache::getInstance( $wgParserCacheType );
3834 }
3835
3836 /**
3837  * Call hook functions defined in $wgHooks
3838  *
3839  * @param $event String: event name
3840  * @param $args Array: parameters passed to hook functions
3841  * @return Boolean True if no handler aborted the hook
3842  */
3843 function wfRunHooks( $event, $args = array() ) {
3844         return Hooks::run( $event, $args );
3845 }
3846
3847 /**
3848  * Wrapper around php's unpack.
3849  *
3850  * @param $format String: The format string (See php's docs)
3851  * @param $data: A binary string of binary data
3852  * @param $length integer or false: The minimun length of $data. This is to
3853  *      prevent reading beyond the end of $data. false to disable the check.
3854  *
3855  * Also be careful when using this function to read unsigned 32 bit integer
3856  * because php might make it negative.
3857  *
3858  * @throws MWException if $data not long enough, or if unpack fails
3859  * @return array Associative array of the extracted data
3860  */
3861 function wfUnpack( $format, $data, $length=false ) {
3862         if ( $length !== false ) {
3863                 $realLen = strlen( $data );
3864                 if ( $realLen < $length ) {
3865                         throw new MWException( "Tried to use wfUnpack on a "
3866                                 . "string of length $realLen, but needed one "
3867                                 . "of at least length $length."
3868                         );
3869                 }
3870         }
3871
3872         wfSuppressWarnings();
3873         $result = unpack( $format, $data );
3874         wfRestoreWarnings();
3875
3876         if ( $result === false ) {
3877                 // If it cannot extract the packed data.
3878                 throw new MWException( "unpack could not unpack binary data" );
3879         }
3880         return $result;
3881 }