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         /** @codeCoverageIgnore */
  24         function iconv( $from, $to, $string ) {
  25                 return Fallback::iconv( $from, $to, $string );
  26         }
  27 }
  28
  29 if ( !function_exists( 'mb_substr' ) ) {
  30         /** @codeCoverageIgnore */
  31         function mb_substr( $str, $start, $count='end' ) {
  32                 return Fallback::mb_substr( $str, $start, $count );
  33         }
  34
  35         /** @codeCoverageIgnore */
  36         function mb_substr_split_unicode( $str, $splitPos ) {
  37                 return Fallback::mb_substr_split_unicode( $str, $splitPos );
  38         }
  39 }
  40
  41 if ( !function_exists( 'mb_strlen' ) ) {
  42         /** @codeCoverageIgnore */
  43         function mb_strlen( $str, $enc = '' ) {
  44                 return Fallback::mb_strlen( $str, $enc );
  45         }
  46 }
  47
  48 if( !function_exists( 'mb_strpos' ) ) {
  49         /** @codeCoverageIgnore */
  50         function mb_strpos( $haystack, $needle, $offset = 0, $encoding = '' ) {
  51                 return Fallback::mb_strpos( $haystack, $needle, $offset, $encoding );
  52         }
  53
  54 }
  55
  56 if( !function_exists( 'mb_strrpos' ) ) {
  57         /** @codeCoverageIgnore */
  58         function mb_strrpos( $haystack, $needle, $offset = 0, $encoding = '' ) {
  59                 return Fallback::mb_strrpos( $haystack, $needle, $offset, $encoding );
  60         }
  61 }
  62
  63
  64 // Support for Wietse Venema's taint feature
  65 if ( !function_exists( 'istainted' ) ) {
  66         /** @codeCoverageIgnore */
  67         function istainted( $var ) {
  68                 return 0;
  69         }
  70         /** @codeCoverageIgnore */
  71         function taint( $var, $level = 0 ) {}
  72         /** @codeCoverageIgnore */
  73         function untaint( $var, $level = 0 ) {}
  74         define( 'TC_HTML', 1 );
  75         define( 'TC_SHELL', 1 );
  76         define( 'TC_MYSQL', 1 );
  77         define( 'TC_PCRE', 1 );
  78         define( 'TC_SELF', 1 );
  79 }
  80 /// @endcond
  81
  82 /**
  83  * Like array_diff( $a, $b ) except that it works with two-dimensional arrays.
  84  * @param $a array
  85  * @param $b array
  86  * @return array
  87  */
  88 function wfArrayDiff2( $a, $b ) {
  89         return array_udiff( $a, $b, 'wfArrayDiff2_cmp' );
  90 }
  91
  92 /**
  93  * @param $a
  94  * @param $b
  95  * @return int
  96  */
  97 function wfArrayDiff2_cmp( $a, $b ) {
  98         if ( !is_array( $a ) ) {
  99                 return strcmp( $a, $b );
 100         } elseif ( count( $a ) !== count( $b ) ) {
 101                 return count( $a ) < count( $b ) ? -1 : 1;
 102         } else {
 103                 reset( $a );
 104                 reset( $b );
 105                 while( ( list( , $valueA ) = each( $a ) ) && ( list( , $valueB ) = each( $b ) ) ) {
 106                         $cmp = strcmp( $valueA, $valueB );
 107                         if ( $cmp !== 0 ) {
 108                                 return $cmp;
 109                         }
 110                 }
 111                 return 0;
 112         }
 113 }
 114
 115 /**
 116  * Array lookup
 117  * Returns an array where the values in the first array are replaced by the
 118  * values in the second array with the corresponding keys
 119  *
 120  * @param $a Array
 121  * @param $b Array
 122  * @return array
 123  */
 124 function wfArrayLookup( $a, $b ) {
 125         return array_flip( array_intersect( array_flip( $a ), array_keys( $b ) ) );
 126 }
 127
 128 /**
 129  * Appends to second array if $value differs from that in $default
 130  *
 131  * @param $key String|Int
 132  * @param $value Mixed
 133  * @param $default Mixed
 134  * @param $changed Array to alter
 135  */
 136 function wfAppendToArrayIfNotDefault( $key, $value, $default, &$changed ) {
 137         if ( is_null( $changed ) ) {
 138                 throw new MWException( 'GlobalFunctions::wfAppendToArrayIfNotDefault got null' );
 139         }
 140         if ( $default[$key] !== $value ) {
 141                 $changed[$key] = $value;
 142         }
 143 }
 144
 145 /**
 146  * Backwards array plus for people who haven't bothered to read the PHP manual
 147  * XXX: will not darn your socks for you.
 148  *
 149  * @param $array1 Array
 150  * @param [$array2, [...]] Arrays
 151  * @return Array
 152  */
 153 function wfArrayMerge( $array1/* ... */ ) {
 154         $args = func_get_args();
 155         $args = array_reverse( $args, true );
 156         $out = array();
 157         foreach ( $args as $arg ) {
 158                 $out += $arg;
 159         }
 160         return $out;
 161 }
 162
 163 /**
 164  * Merge arrays in the style of getUserPermissionsErrors, with duplicate removal
 165  * e.g.
 166  *      wfMergeErrorArrays(
 167  *              array( array( 'x' ) ),
 168  *              array( array( 'x', '2' ) ),
 169  *              array( array( 'x' ) ),
 170  *              array( array( 'y' ) )
 171  *      );
 172  * returns:
 173  *              array(
 174  *              array( 'x', '2' ),
 175  *              array( 'x' ),
 176  *              array( 'y' )
 177  *      )
 178  * @param varargs
 179  * @return Array
 180  */
 181 function wfMergeErrorArrays( /*...*/ ) {
 182         $args = func_get_args();
 183         $out = array();
 184         foreach ( $args as $errors ) {
 185                 foreach ( $errors as $params ) {
 186                         # @todo FIXME: Sometimes get nested arrays for $params,
 187                         # which leads to E_NOTICEs
 188                         $spec = implode( "\t", $params );
 189                         $out[$spec] = $params;
 190                 }
 191         }
 192         return array_values( $out );
 193 }
 194
 195 /**
 196  * Insert array into another array after the specified *KEY*
 197  *
 198  * @param $array Array: The array.
 199  * @param $insert Array: The array to insert.
 200  * @param $after Mixed: The key to insert after
 201  * @return Array
 202  */
 203 function wfArrayInsertAfter( $array, $insert, $after ) {
 204         // Find the offset of the element to insert after.
 205         $keys = array_keys( $array );
 206         $offsetByKey = array_flip( $keys );
 207
 208         $offset = $offsetByKey[$after];
 209
 210         // Insert at the specified offset
 211         $before = array_slice( $array, 0, $offset + 1, true );
 212         $after = array_slice( $array, $offset + 1, count( $array ) - $offset, true );
 213
 214         $output = $before + $insert + $after;
 215
 216         return $output;
 217 }
 218
 219 /**
 220  * Recursively converts the parameter (an object) to an array with the same data
 221  *
 222  * @param $objOrArray Object|Array
 223  * @param $recursive Bool
 224  * @return Array
 225  */
 226 function wfObjectToArray( $objOrArray, $recursive = true ) {
 227         $array = array();
 228         if( is_object( $objOrArray ) ) {
 229                 $objOrArray = get_object_vars( $objOrArray );
 230         }
 231         foreach ( $objOrArray as $key => $value ) {
 232                 if ( $recursive && ( is_object( $value ) || is_array( $value ) ) ) {
 233                         $value = wfObjectToArray( $value );
 234                 }
 235
 236                 $array[$key] = $value;
 237         }
 238
 239         return $array;
 240 }
 241
 242 /**
 243  * Wrapper around array_map() which also taints variables
 244  *
 245  * @param  $function Callback
 246  * @param  $input Array
 247  * @return Array
 248  */
 249 function wfArrayMap( $function, $input ) {
 250         $ret = array_map( $function, $input );
 251         foreach ( $ret as $key => $value ) {
 252                 $taint = istainted( $input[$key] );
 253                 if ( $taint ) {
 254                         taint( $ret[$key], $taint );
 255                 }
 256         }
 257         return $ret;
 258 }
 259
 260 /**
 261  * Get a random decimal value between 0 and 1, in a way
 262  * not likely to give duplicate values for any realistic
 263  * number of articles.
 264  *
 265  * @return string
 266  */
 267 function wfRandom() {
 268         # The maximum random value is "only" 2^31-1, so get two random
 269         # values to reduce the chance of dupes
 270         $max = mt_getrandmax() + 1;
 271         $rand = number_format( ( mt_rand() * $max + mt_rand() )
 272                 / $max / $max, 12, '.', '' );
 273         return $rand;
 274 }
 275
 276 /**
 277  * We want some things to be included as literal characters in our title URLs
 278  * for prettiness, which urlencode encodes by default.  According to RFC 1738,
 279  * all of the following should be safe:
 280  *
 281  * ;:@&=$-_.+!*'(),
 282  *
 283  * But + is not safe because it's used to indicate a space; &= are only safe in
 284  * paths and not in queries (and we don't distinguish here); ' seems kind of
 285  * scary; and urlencode() doesn't touch -_. to begin with.  Plus, although /
 286  * is reserved, we don't care.  So the list we unescape is:
 287  *
 288  * ;:@$!*(),/
 289  *
 290  * However, IIS7 redirects fail when the url contains a colon (Bug 22709),
 291  * so no fancy : for IIS7.
 292  *
 293  * %2F in the page titles seems to fatally break for some reason.
 294  *
 295  * @param $s String:
 296  * @return string
 297 */
 298 function wfUrlencode( $s ) {
 299         static $needle;
 300         if ( is_null( $s ) ) {
 301                 $needle = null;
 302                 return '';
 303         }
 304
 305         if ( is_null( $needle ) ) {
 306                 $needle = array( '%3B', '%40', '%24', '%21', '%2A', '%28', '%29', '%2C', '%2F' );
 307                 if ( !isset( $_SERVER['SERVER_SOFTWARE'] ) || ( strpos( $_SERVER['SERVER_SOFTWARE'], 'Microsoft-IIS/7' ) === false ) ) {
 308                         $needle[] = '%3A';
 309                 }
 310         }
 311
 312         $s = urlencode( $s );
 313         $s = str_ireplace(
 314                 $needle,
 315                 array( ';', '@', '$', '!', '*', '(', ')', ',', '/', ':' ),
 316                 $s
 317         );
 318
 319         return $s;
 320 }
 321
 322 /**
 323  * This function takes two arrays as input, and returns a CGI-style string, e.g.
 324  * "days=7&limit=100". Options in the first array override options in the second.
 325  * Options set to "" will not be output.
 326  *
 327  * @param $array1 Array ( String|Array )
 328  * @param $array2 Array ( String|Array )
 329  * @param $prefix String
 330  * @return String
 331  */
 332 function wfArrayToCGI( $array1, $array2 = null, $prefix = '' ) {
 333         if ( !is_null( $array2 ) ) {
 334                 $array1 = $array1 + $array2;
 335         }
 336
 337         $cgi = '';
 338         foreach ( $array1 as $key => $value ) {
 339                 if ( $value !== '' ) {
 340                         if ( $cgi != '' ) {
 341                                 $cgi .= '&';
 342                         }
 343                         if ( $prefix !== '' ) {
 344                                 $key = $prefix . "[$key]";
 345                         }
 346                         if ( is_array( $value ) ) {
 347                                 $firstTime = true;
 348                                 foreach ( $value as $k => $v ) {
 349                                         $cgi .= $firstTime ? '' : '&';
 350                                         if ( is_array( $v ) ) {
 351                                                 $cgi .= wfArrayToCGI( $v, null, $key . "[$k]" );
 352                                         } else {
 353                                                 $cgi .= urlencode( $key . "[$k]" ) . '=' . urlencode( $v );
 354                                         }
 355                                         $firstTime = false;
 356                                 }
 357                         } else {
 358                                 if ( is_object( $value ) ) {
 359                                         $value = $value->__toString();
 360                                 }
 361                                 $cgi .= urlencode( $key ) . '=' . urlencode( $value );
 362                         }
 363                 }
 364         }
 365         return $cgi;
 366 }
 367
 368 /**
 369  * This is the logical opposite of wfArrayToCGI(): it accepts a query string as
 370  * its argument and returns the same string in array form.  This allows compa-
 371  * tibility with legacy functions that accept raw query strings instead of nice
 372  * arrays.  Of course, keys and values are urldecode()d.  Don't try passing in-
 373  * valid query strings, or it will explode.
 374  *
 375  * @param $query String: query string
 376  * @return array Array version of input
 377  */
 378 function wfCgiToArray( $query ) {
 379         if ( isset( $query[0] ) && $query[0] == '?' ) {
 380                 $query = substr( $query, 1 );
 381         }
 382         $bits = explode( '&', $query );
 383         $ret = array();
 384         foreach ( $bits as $bit ) {
 385                 if ( $bit === '' ) {
 386                         continue;
 387                 }
 388                 list( $key, $value ) = explode( '=', $bit );
 389                 $key = urldecode( $key );
 390                 $value = urldecode( $value );
 391                 if ( strpos( $key, '[' ) !== false ) {
 392                         $keys = array_reverse( explode( '[', $key ) );
 393                         $key = array_pop( $keys );
 394                         $temp = $value;
 395                         foreach ( $keys as $k ) {
 396                                 $k = substr( $k, 0, -1 );
 397                                 $temp = array( $k => $temp );
 398                         }
 399                         if ( isset( $ret[$key] ) ) {
 400                                 $ret[$key] = array_merge( $ret[$key], $temp );
 401                         } else {
 402                                 $ret[$key] = $temp;
 403                         }
 404                 } else {
 405                         $ret[$key] = $value;
 406                 }
 407         }
 408         return $ret;
 409 }
 410
 411 /**
 412  * Append a query string to an existing URL, which may or may not already
 413  * have query string parameters already. If so, they will be combined.
 414  *
 415  * @param $url String
 416  * @param $query Mixed: string or associative array
 417  * @return string
 418  */
 419 function wfAppendQuery( $url, $query ) {
 420         if ( is_array( $query ) ) {
 421                 $query = wfArrayToCGI( $query );
 422         }
 423         if( $query != '' ) {
 424                 if( false === strpos( $url, '?' ) ) {
 425                         $url .= '?';
 426                 } else {
 427                         $url .= '&';
 428                 }
 429                 $url .= $query;
 430         }
 431         return $url;
 432 }
 433
 434 /**
 435  * Expand a potentially local URL to a fully-qualified URL.  Assumes $wgServer
 436  * is correct.
 437  *
 438  * The meaning of the PROTO_* constants is as follows:
 439  * PROTO_HTTP: Output a URL starting with http://
 440  * PROTO_HTTPS: Output a URL starting with https://
 441  * PROTO_RELATIVE: Output a URL starting with // (protocol-relative URL)
 442  * PROTO_CURRENT: Output a URL starting with either http:// or https:// , depending on which protocol was used for the current incoming request
 443  * PROTO_CANONICAL: For URLs without a domain, like /w/index.php , use $wgCanonicalServer. For protocol-relative URLs, use the protocol of $wgCanonicalServer
 444  * PROTO_INTERNAL: Like PROTO_CANONICAL, but uses $wgInternalServer instead of $wgCanonicalServer
 445  *
 446  * @todo this won't work with current-path-relative URLs
 447  * like "subdir/foo.html", etc.
 448  *
 449  * @param $url String: either fully-qualified or a local path + query
 450  * @param $defaultProto Mixed: one of the PROTO_* constants. Determines the
 451  *                             protocol to use if $url or $wgServer is
 452  *                             protocol-relative
 453  * @return string Fully-qualified URL, current-path-relative URL or false if
 454  *                no valid URL can be constructed
 455  */
 456 function wfExpandUrl( $url, $defaultProto = PROTO_CURRENT ) {
 457         global $wgServer, $wgCanonicalServer, $wgInternalServer;
 458         $serverUrl = $wgServer;
 459         if ( $defaultProto === PROTO_CANONICAL ) {
 460                 $serverUrl = $wgCanonicalServer;
 461         }
 462         // Make $wgInternalServer fall back to $wgServer if not set
 463         if ( $defaultProto === PROTO_INTERNAL && $wgInternalServer !== false ) {
 464                 $serverUrl = $wgInternalServer;
 465         }
 466         if ( $defaultProto === PROTO_CURRENT ) {
 467                 $defaultProto = WebRequest::detectProtocol() . '://';
 468         }
 469
 470         // Analyze $serverUrl to obtain its protocol
 471         $bits = wfParseUrl( $serverUrl );
 472         $serverHasProto = $bits && $bits['scheme'] != '';
 473
 474         if ( $defaultProto === PROTO_CANONICAL || $defaultProto === PROTO_INTERNAL ) {
 475                 if ( $serverHasProto ) {
 476                         $defaultProto = $bits['scheme'] . '://';
 477                 } else {
 478                         // $wgCanonicalServer or $wgInternalServer doesn't have a protocol. This really isn't supposed to happen
 479                         // Fall back to HTTP in this ridiculous case
 480                         $defaultProto = PROTO_HTTP;
 481                 }
 482         }
 483
 484         $defaultProtoWithoutSlashes = substr( $defaultProto, 0, -2 );
 485
 486         if ( substr( $url, 0, 2 ) == '//' ) {
 487                 $url = $defaultProtoWithoutSlashes . $url;
 488         } elseif ( substr( $url, 0, 1 ) == '/' ) {
 489                 // If $serverUrl is protocol-relative, prepend $defaultProtoWithoutSlashes, otherwise leave it alone
 490                 $url = ( $serverHasProto ? '' : $defaultProtoWithoutSlashes ) . $serverUrl . $url;
 491         }
 492
 493         $bits = wfParseUrl( $url );
 494         if ( $bits && isset( $bits['path'] ) ) {
 495                 $bits['path'] = wfRemoveDotSegments( $bits['path'] );
 496                 return wfAssembleUrl( $bits );
 497         } elseif ( $bits ) {
 498                 # No path to expand
 499                 return $url;
 500         } elseif ( substr( $url, 0, 1 ) != '/' ) {
 501                 # URL is a relative path
 502                 return wfRemoveDotSegments( $url );
 503         }
 504
 505         # Expanded URL is not valid.
 506         return false;
 507 }
 508
 509 /**
 510  * This function will reassemble a URL parsed with wfParseURL.  This is useful
 511  * if you need to edit part of a URL and put it back together.
 512  *
 513  * This is the basic structure used (brackets contain keys for $urlParts):
 514  * [scheme][delimiter][user]:[pass]@[host]:[port][path]?[query]#[fragment]
 515  *
 516  * @todo Need to integrate this into wfExpandUrl (bug 32168)
 517  *
 518  * @param $urlParts Array URL parts, as output from wfParseUrl
 519  * @return string URL assembled from its component parts
 520  */
 521 function wfAssembleUrl( $urlParts ) {
 522         $result = '';
 523
 524         if ( isset( $urlParts['delimiter'] ) ) {
 525                 if ( isset( $urlParts['scheme'] ) ) {
 526                         $result .= $urlParts['scheme'];
 527                 }
 528
 529                 $result .= $urlParts['delimiter'];
 530         }
 531
 532         if ( isset( $urlParts['host'] ) ) {
 533                 if ( isset( $urlParts['user'] ) ) {
 534                         $result .= $urlParts['user'];
 535                         if ( isset( $urlParts['pass'] ) ) {
 536                                 $result .= ':' . $urlParts['pass'];
 537                         }
 538                         $result .= '@';
 539                 }
 540
 541                 $result .= $urlParts['host'];
 542
 543                 if ( isset( $urlParts['port'] ) ) {
 544                         $result .= ':' . $urlParts['port'];
 545                 }
 546         }
 547
 548         if ( isset( $urlParts['path'] ) ) {
 549                 $result .= $urlParts['path'];
 550         }
 551
 552         if ( isset( $urlParts['query'] ) ) {
 553                 $result .= '?' . $urlParts['query'];
 554         }
 555
 556         if ( isset( $urlParts['fragment'] ) ) {
 557                 $result .= '#' . $urlParts['fragment'];
 558         }
 559
 560         return $result;
 561 }
 562
 563 /**
 564  * Remove all dot-segments in the provided URL path.  For example,
 565  * '/a/./b/../c/' becomes '/a/c/'.  For details on the algorithm, please see
 566  * RFC3986 section 5.2.4.
 567  *
 568  * @todo Need to integrate this into wfExpandUrl (bug 32168)
 569  *
 570  * @param $urlPath String URL path, potentially containing dot-segments
 571  * @return string URL path with all dot-segments removed
 572  */
 573 function wfRemoveDotSegments( $urlPath ) {
 574         $output = '';
 575
 576         while ( $urlPath ) {
 577                 $matches = null;
 578                 if ( preg_match('%^\.\.?/%', $urlPath, $matches) ) {
 579                         # Step A, remove leading "../" or "./"
 580                         $urlPath = substr( $urlPath, strlen( $matches[0] ) );
 581                 } elseif ( preg_match( '%^/\.(/|$)%', $urlPath, $matches ) ) {
 582                         # Step B, replace leading "/.$" or "/./" with "/"
 583                         $start = strlen( $matches[0] );
 584                         $urlPath = '/' . substr( $urlPath, $start );
 585                 } elseif ( preg_match( '%^/\.\.(/|$)%', $urlPath, $matches ) ) {
 586                         # Step C, replace leading "/..$" or "/../" with "/" and
 587                         # remove last path component in output
 588                         $start = strlen( $matches[0] );
 589                         $urlPath = '/' . substr( $urlPath, $start );
 590                         $output = preg_replace('%(^|/)[^/]*$%', '', $output);
 591                 } elseif ( preg_match( '%^\.\.?$%', $urlPath, $matches ) ) {
 592                         # Step D, remove "^..$" or "^.$"
 593                         $urlPath = '';
 594                 } else {
 595                         # Step E, move leading path segment to output
 596                         preg_match( '%^/?[^/]*%', $urlPath, $matches );
 597                         $urlPath = substr( $urlPath, strlen( $matches[0] ) );
 598                         $output .= $matches[0];
 599                 }
 600         }
 601
 602         return $output;
 603 }
 604
 605 /**
 606  * Returns a regular expression of url protocols
 607  *
 608  * @param $includeProtocolRelative bool If false, remove '//' from the returned protocol list.
 609  *        DO NOT USE this directly, use wfUrlProtocolsWithoutProtRel() instead
 610  * @return String
 611  */
 612 function wfUrlProtocols( $includeProtocolRelative = true ) {
 613         global $wgUrlProtocols;
 614
 615         // Cache return values separately based on $includeProtocolRelative
 616         static $withProtRel = null, $withoutProtRel = null;
 617         $cachedValue = $includeProtocolRelative ? $withProtRel : $withoutProtRel;
 618         if ( !is_null( $cachedValue ) ) {
 619                 return $cachedValue;
 620         }
 621
 622         // Support old-style $wgUrlProtocols strings, for backwards compatibility
 623         // with LocalSettings files from 1.5
 624         if ( is_array( $wgUrlProtocols ) ) {
 625                 $protocols = array();
 626                 foreach ( $wgUrlProtocols as $protocol ) {
 627                         // Filter out '//' if !$includeProtocolRelative
 628                         if ( $includeProtocolRelative || $protocol !== '//' ) {
 629                                 $protocols[] = preg_quote( $protocol, '/' );
 630                         }
 631                 }
 632
 633                 $retval = implode( '|', $protocols );
 634         } else {
 635                 // Ignore $includeProtocolRelative in this case
 636                 // This case exists for pre-1.6 compatibility, and we can safely assume
 637                 // that '//' won't appear in a pre-1.6 config because protocol-relative
 638                 // URLs weren't supported until 1.18
 639                 $retval = $wgUrlProtocols;
 640         }
 641
 642         // Cache return value
 643         if ( $includeProtocolRelative ) {
 644                 $withProtRel = $retval;
 645         } else {
 646                 $withoutProtRel = $retval;
 647         }
 648         return $retval;
 649 }
 650
 651 /**
 652  * Like wfUrlProtocols(), but excludes '//' from the protocol list. Use this if
 653  * you need a regex that matches all URL protocols but does not match protocol-
 654  * relative URLs
 655  * @return String
 656  */
 657 function wfUrlProtocolsWithoutProtRel() {
 658         return wfUrlProtocols( false );
 659 }
 660
 661 /**
 662  * parse_url() work-alike, but non-broken.  Differences:
 663  *
 664  * 1) Does not raise warnings on bad URLs (just returns false)
 665  * 2) Handles protocols that don't use :// (e.g., mailto: and news: , as well as protocol-relative URLs) correctly
 666  * 3) Adds a "delimiter" element to the array, either '://', ':' or '//' (see (2))
 667  *
 668  * @param $url String: a URL to parse
 669  * @return Array: bits of the URL in an associative array, per PHP docs
 670  */
 671 function wfParseUrl( $url ) {
 672         global $wgUrlProtocols; // Allow all protocols defined in DefaultSettings/LocalSettings.php
 673
 674         // Protocol-relative URLs are handled really badly by parse_url(). It's so bad that the easiest
 675         // way to handle them is to just prepend 'http:' and strip the protocol out later
 676         $wasRelative = substr( $url, 0, 2 ) == '//';
 677         if ( $wasRelative ) {
 678                 $url = "http:$url";
 679         }
 680         wfSuppressWarnings();
 681         $bits = parse_url( $url );
 682         wfRestoreWarnings();
 683         // parse_url() returns an array without scheme for some invalid URLs, e.g.
 684         // parse_url("%0Ahttp://example.com") == array( 'host' => '%0Ahttp', 'path' => 'example.com' )
 685         if ( !$bits || !isset( $bits['scheme'] ) ) {
 686                 return false;
 687         }
 688
 689         // most of the protocols are followed by ://, but mailto: and sometimes news: not, check for it
 690         if ( in_array( $bits['scheme'] . '://', $wgUrlProtocols ) ) {
 691                 $bits['delimiter'] = '://';
 692         } elseif ( in_array( $bits['scheme'] . ':', $wgUrlProtocols ) ) {
 693                 $bits['delimiter'] = ':';
 694                 // parse_url detects for news: and mailto: the host part of an url as path
 695                 // We have to correct this wrong detection
 696                 if ( isset( $bits['path'] ) ) {
 697                         $bits['host'] = $bits['path'];
 698                         $bits['path'] = '';
 699                 }
 700         } else {
 701                 return false;
 702         }
 703
 704         /* Provide an empty host for eg. file:/// urls (see bug 28627) */
 705         if ( !isset( $bits['host'] ) ) {
 706                 $bits['host'] = '';
 707
 708                 /* parse_url loses the third / for file:///c:/ urls (but not on variants) */
 709                 if ( substr( $bits['path'], 0, 1 ) !== '/' ) {
 710                         $bits['path'] = '/' . $bits['path'];
 711                 }
 712         }
 713
 714         // If the URL was protocol-relative, fix scheme and delimiter
 715         if ( $wasRelative ) {
 716                 $bits['scheme'] = '';
 717                 $bits['delimiter'] = '//';
 718         }
 719         return $bits;
 720 }
 721
 722 /**
 723  * Make URL indexes, appropriate for the el_index field of externallinks.
 724  *
 725  * @param $url String
 726  * @return array
 727  */
 728 function wfMakeUrlIndexes( $url ) {
 729         $bits = wfParseUrl( $url );
 730
 731         // Reverse the labels in the hostname, convert to lower case
 732         // For emails reverse domainpart only
 733         if ( $bits['scheme'] == 'mailto' ) {
 734                 $mailparts = explode( '@', $bits['host'], 2 );
 735                 if ( count( $mailparts ) === 2 ) {
 736                         $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
 737                 } else {
 738                         // No domain specified, don't mangle it
 739                         $domainpart = '';
 740                 }
 741                 $reversedHost = $domainpart . '@' . $mailparts[0];
 742         } else {
 743                 $reversedHost = strtolower( implode( '.', array_reverse( explode( '.', $bits['host'] ) ) ) );
 744         }
 745         // Add an extra dot to the end
 746         // Why? Is it in wrong place in mailto links?
 747         if ( substr( $reversedHost, -1, 1 ) !== '.' ) {
 748                 $reversedHost .= '.';
 749         }
 750         // Reconstruct the pseudo-URL
 751         $prot = $bits['scheme'];
 752         $index = $prot . $bits['delimiter'] . $reversedHost;
 753         // Leave out user and password. Add the port, path, query and fragment
 754         if ( isset( $bits['port'] ) ) {
 755                 $index .= ':' . $bits['port'];
 756         }
 757         if ( isset( $bits['path'] ) ) {
 758                 $index .= $bits['path'];
 759         } else {
 760                 $index .= '/';
 761         }
 762         if ( isset( $bits['query'] ) ) {
 763                 $index .= '?' . $bits['query'];
 764         }
 765         if ( isset( $bits['fragment'] ) ) {
 766                 $index .= '#' . $bits['fragment'];
 767         }
 768
 769         if ( $prot == '' ) {
 770                 return array( "http:$index", "https:$index" );
 771         } else {
 772                 return array( $index );
 773         }
 774 }
 775
 776 /**
 777  * Check whether a given URL has a domain that occurs in a given set of domains
 778  * @param $url string URL
 779  * @param $domains array Array of domains (strings)
 780  * @return bool True if the host part of $url ends in one of the strings in $domains
 781  */
 782 function wfMatchesDomainList( $url, $domains ) {
 783         $bits = wfParseUrl( $url );
 784         if ( is_array( $bits ) && isset( $bits['host'] ) ) {
 785                 foreach ( (array)$domains as $domain ) {
 786                         // FIXME: This gives false positives. http://nds-nl.wikipedia.org will match nl.wikipedia.org
 787                         // We should use something that interprets dots instead
 788                         if ( substr( $bits['host'], -strlen( $domain ) ) === $domain ) {
 789                                 return true;
 790                         }
 791                 }
 792         }
 793         return false;
 794 }
 795
 796 /**
 797  * Sends a line to the debug log if enabled or, optionally, to a comment in output.
 798  * In normal operation this is a NOP.
 799  *
 800  * Controlling globals:
 801  * $wgDebugLogFile - points to the log file
 802  * $wgProfileOnly - if set, normal debug messages will not be recorded.
 803  * $wgDebugRawPage - if false, 'action=raw' hits will not result in debug output.
 804  * $wgDebugComments - if on, some debug items may appear in comments in the HTML output.
 805  *
 806  * @param $text String
 807  * @param $logonly Bool: set true to avoid appearing in HTML when $wgDebugComments is set
 808  */
 809 function wfDebug( $text, $logonly = false ) {
 810         global $wgOut, $wgDebugLogFile, $wgDebugComments, $wgProfileOnly, $wgDebugRawPage;
 811         global $wgDebugLogPrefix, $wgShowDebug;
 812
 813         static $cache = array(); // Cache of unoutputted messages
 814         $text = wfDebugTimer() . $text;
 815
 816         if ( !$wgDebugRawPage && wfIsDebugRawPage() ) {
 817                 return;
 818         }
 819
 820         if ( ( $wgDebugComments || $wgShowDebug ) && !$logonly ) {
 821                 $cache[] = $text;
 822
 823                 if ( isset( $wgOut ) && is_object( $wgOut ) ) {
 824                         // add the message and any cached messages to the output
 825                         array_map( array( $wgOut, 'debug' ), $cache );
 826                         $cache = array();
 827                 }
 828         }
 829         if ( wfRunHooks( 'Debug', array( $text, null /* no log group */ ) ) ) {
 830                 if ( $wgDebugLogFile != '' && !$wgProfileOnly ) {
 831                         # Strip unprintables; they can switch terminal modes when binary data
 832                         # gets dumped, which is pretty annoying.
 833                         $text = preg_replace( '![\x00-\x08\x0b\x0c\x0e-\x1a\x1c-\x1f]!', ' ', $text );
 834                         $text = $wgDebugLogPrefix . $text;
 835                         wfErrorLog( $text, $wgDebugLogFile );
 836                 }
 837         }
 838 }
 839
 840 /**
 841  * Returns true if debug logging should be suppressed if $wgDebugRawPage = false
 842  */
 843 function wfIsDebugRawPage() {
 844         static $cache;
 845         if ( $cache !== null ) {
 846                 return $cache;
 847         }
 848         # Check for raw action using $_GET not $wgRequest, since the latter might not be initialised yet
 849         if ( ( isset( $_GET['action'] ) && $_GET['action'] == 'raw' )
 850                 || (
 851                         isset( $_SERVER['SCRIPT_NAME'] )
 852                         && substr( $_SERVER['SCRIPT_NAME'], -8 ) == 'load.php'
 853                 ) )
 854         {
 855                 $cache = true;
 856         } else {
 857                 $cache = false;
 858         }
 859         return $cache;
 860 }
 861
 862 /**
 863  * Get microsecond timestamps for debug logs
 864  *
 865  * @return string
 866  */
 867 function wfDebugTimer() {
 868         global $wgDebugTimestamps;
 869         if ( !$wgDebugTimestamps ) {
 870                 return '';
 871         }
 872         static $start = null;
 873
 874         if ( $start === null ) {
 875                 $start = microtime( true );
 876                 $prefix = "\n$start";
 877         } else {
 878                 $prefix = sprintf( "%6.4f", microtime( true ) - $start );
 879         }
 880         $mem = sprintf( "%5.1fM", ( memory_get_usage( true ) / (1024*1024) ) );
 881         return "$prefix $mem  " ;
 882 }
 883
 884 /**
 885  * Send a line giving PHP memory usage.
 886  *
 887  * @param $exact Bool: print exact values instead of kilobytes (default: false)
 888  */
 889 function wfDebugMem( $exact = false ) {
 890         $mem = memory_get_usage();
 891         if( !$exact ) {
 892                 $mem = floor( $mem / 1024 ) . ' kilobytes';
 893         } else {
 894                 $mem .= ' bytes';
 895         }
 896         wfDebug( "Memory usage: $mem\n" );
 897 }
 898
 899 /**
 900  * Send a line to a supplementary debug log file, if configured, or main debug log if not.
 901  * $wgDebugLogGroups[$logGroup] should be set to a filename to send to a separate log.
 902  *
 903  * @param $logGroup String
 904  * @param $text String
 905  * @param $public Bool: whether to log the event in the public log if no private
 906  *                     log file is specified, (default true)
 907  */
 908 function wfDebugLog( $logGroup, $text, $public = true ) {
 909         global $wgDebugLogGroups, $wgShowHostnames;
 910         $text = trim( $text ) . "\n";
 911         if( isset( $wgDebugLogGroups[$logGroup] ) ) {
 912                 $time = wfTimestamp( TS_DB );
 913                 $wiki = wfWikiID();
 914                 if ( $wgShowHostnames ) {
 915                         $host = wfHostname();
 916                 } else {
 917                         $host = '';
 918                 }
 919                 if ( wfRunHooks( 'Debug', array( $text, $logGroup ) ) ) {
 920                         wfErrorLog( "$time $host $wiki: $text", $wgDebugLogGroups[$logGroup] );
 921                 }
 922         } elseif ( $public === true ) {
 923                 wfDebug( $text, true );
 924         }
 925 }
 926
 927 /**
 928  * Log for database errors
 929  *
 930  * @param $text String: database error message.
 931  */
 932 function wfLogDBError( $text ) {
 933         global $wgDBerrorLog, $wgDBname;
 934         if ( $wgDBerrorLog ) {
 935                 $host = trim(`hostname`);
 936                 $text = date( 'D M j G:i:s T Y' ) . "\t$host\t$wgDBname\t$text";
 937                 wfErrorLog( $text, $wgDBerrorLog );
 938         }
 939 }
 940
 941 /**
 942  * Log to a file without getting "file size exceeded" signals.
 943  *
 944  * Can also log to TCP or UDP with the syntax udp://host:port/prefix. This will
 945  * send lines to the specified port, prefixed by the specified prefix and a space.
 946  *
 947  * @param $text String
 948  * @param $file String filename
 949  */
 950 function wfErrorLog( $text, $file ) {
 951         if ( substr( $file, 0, 4 ) == 'udp:' ) {
 952                 # Needs the sockets extension
 953                 if ( preg_match( '!^(tcp|udp):(?://)?\[([0-9a-fA-F:]+)\]:(\d+)(?:/(.*))?$!', $file, $m ) ) {
 954                         // IPv6 bracketed host
 955                         $host = $m[2];
 956                         $port = intval( $m[3] );
 957                         $prefix = isset( $m[4] ) ? $m[4] : false;
 958                         $domain = AF_INET6;
 959                 } elseif ( preg_match( '!^(tcp|udp):(?://)?([a-zA-Z0-9.-]+):(\d+)(?:/(.*))?$!', $file, $m ) ) {
 960                         $host = $m[2];
 961                         if ( !IP::isIPv4( $host ) ) {
 962                                 $host = gethostbyname( $host );
 963                         }
 964                         $port = intval( $m[3] );
 965                         $prefix = isset( $m[4] ) ? $m[4] : false;
 966                         $domain = AF_INET;
 967                 } else {
 968                         throw new MWException( __METHOD__ . ': Invalid UDP specification' );
 969                 }
 970
 971                 // Clean it up for the multiplexer
 972                 if ( strval( $prefix ) !== '' ) {
 973                         $text = preg_replace( '/^/m', $prefix . ' ', $text );
 974
 975                         // Limit to 64KB
 976                         if ( strlen( $text ) > 65506 ) {
 977                                 $text = substr( $text, 0, 65506 );
 978                         }
 979
 980                         if ( substr( $text, -1 ) != "\n" ) {
 981                                 $text .= "\n";
 982                         }
 983                 } elseif ( strlen( $text ) > 65507 ) {
 984                         $text = substr( $text, 0, 65507 );
 985                 }
 986
 987                 $sock = socket_create( $domain, SOCK_DGRAM, SOL_UDP );
 988                 if ( !$sock ) {
 989                         return;
 990                 }
 991
 992                 socket_sendto( $sock, $text, strlen( $text ), 0, $host, $port );
 993                 socket_close( $sock );
 994         } else {
 995                 wfSuppressWarnings();
 996                 $exists = file_exists( $file );
 997                 $size = $exists ? filesize( $file ) : false;
 998                 if ( !$exists || ( $size !== false && $size + strlen( $text ) < 0x7fffffff ) ) {
 999                         file_put_contents( $file, $text, FILE_APPEND );
1000                 }
1001                 wfRestoreWarnings();
1002         }
1003 }
1004
1005 /**
1006  * @todo document
1007  */
1008 function wfLogProfilingData() {
1009         global $wgRequestTime, $wgDebugLogFile, $wgDebugRawPage, $wgRequest;
1010         global $wgProfileLimit, $wgUser;
1011
1012         $profiler = Profiler::instance();
1013
1014         # Profiling must actually be enabled...
1015         if ( $profiler->isStub() ) {
1016                 return;
1017         }
1018
1019         // Get total page request time and only show pages that longer than
1020         // $wgProfileLimit time (default is 0)
1021         $now = wfTime();
1022         $elapsed = $now - $wgRequestTime;
1023         if ( $elapsed <= $wgProfileLimit ) {
1024                 return;
1025         }
1026
1027         $profiler->logData();
1028
1029         // Check whether this should be logged in the debug file.
1030         if ( $wgDebugLogFile == '' || ( !$wgDebugRawPage && wfIsDebugRawPage() ) ) {
1031                 return;
1032         }
1033
1034         $forward = '';
1035         if ( !empty( $_SERVER['HTTP_X_FORWARDED_FOR'] ) ) {
1036                 $forward = ' forwarded for ' . $_SERVER['HTTP_X_FORWARDED_FOR'];
1037         }
1038         if ( !empty( $_SERVER['HTTP_CLIENT_IP'] ) ) {
1039                 $forward .= ' client IP ' . $_SERVER['HTTP_CLIENT_IP'];
1040         }
1041         if ( !empty( $_SERVER['HTTP_FROM'] ) ) {
1042                 $forward .= ' from ' . $_SERVER['HTTP_FROM'];
1043         }
1044         if ( $forward ) {
1045                 $forward = "\t(proxied via {$_SERVER['REMOTE_ADDR']}{$forward})";
1046         }
1047         // Don't load $wgUser at this late stage just for statistics purposes
1048         // @todo FIXME: We can detect some anons even if it is not loaded. See User::getId()
1049         if ( $wgUser->isItemLoaded( 'id' ) && $wgUser->isAnon() ) {
1050                 $forward .= ' anon';
1051         }
1052         $log = sprintf( "%s\t%04.3f\t%s\n",
1053                 gmdate( 'YmdHis' ), $elapsed,
1054                 urldecode( $wgRequest->getRequestURL() . $forward ) );
1055
1056         wfErrorLog( $log . $profiler->getOutput(), $wgDebugLogFile );
1057 }
1058
1059 /**
1060  * Check if the wiki read-only lock file is present. This can be used to lock
1061  * off editing functions, but doesn't guarantee that the database will not be
1062  * modified.
1063  *
1064  * @return bool
1065  */
1066 function wfReadOnly() {
1067         global $wgReadOnlyFile, $wgReadOnly;
1068
1069         if ( !is_null( $wgReadOnly ) ) {
1070                 return (bool)$wgReadOnly;
1071         }
1072         if ( $wgReadOnlyFile == '' ) {
1073                 return false;
1074         }
1075         // Set $wgReadOnly for faster access next time
1076         if ( is_file( $wgReadOnlyFile ) ) {
1077                 $wgReadOnly = file_get_contents( $wgReadOnlyFile );
1078         } else {
1079                 $wgReadOnly = false;
1080         }
1081         return (bool)$wgReadOnly;
1082 }
1083
1084 /**
1085  * @return bool
1086  */
1087 function wfReadOnlyReason() {
1088         global $wgReadOnly;
1089         wfReadOnly();
1090         return $wgReadOnly;
1091 }
1092
1093 /**
1094  * Return a Language object from $langcode
1095  *
1096  * @param $langcode Mixed: either:
1097  *                  - a Language object
1098  *                  - code of the language to get the message for, if it is
1099  *                    a valid code create a language for that language, if
1100  *                    it is a string but not a valid code then make a basic
1101  *                    language object
1102  *                  - a boolean: if it's false then use the global object for
1103  *                    the current user's language (as a fallback for the old parameter
1104  *                    functionality), or if it is true then use global object
1105  *                    for the wiki's content language.
1106  * @return Language object
1107  */
1108 function wfGetLangObj( $langcode = false ) {
1109         # Identify which language to get or create a language object for.
1110         # Using is_object here due to Stub objects.
1111         if( is_object( $langcode ) ) {
1112                 # Great, we already have the object (hopefully)!
1113                 return $langcode;
1114         }
1115
1116         global $wgContLang, $wgLanguageCode;
1117         if( $langcode === true || $langcode === $wgLanguageCode ) {
1118                 # $langcode is the language code of the wikis content language object.
1119                 # or it is a boolean and value is true
1120                 return $wgContLang;
1121         }
1122
1123         global $wgLang;
1124         if( $langcode === false || $langcode === $wgLang->getCode() ) {
1125                 # $langcode is the language code of user language object.
1126                 # or it was a boolean and value is false
1127                 return $wgLang;
1128         }
1129
1130         $validCodes = array_keys( Language::getLanguageNames() );
1131         if( in_array( $langcode, $validCodes ) ) {
1132                 # $langcode corresponds to a valid language.
1133                 return Language::factory( $langcode );
1134         }
1135
1136         # $langcode is a string, but not a valid language code; use content language.
1137         wfDebug( "Invalid language code passed to wfGetLangObj, falling back to content language.\n" );
1138         return $wgContLang;
1139 }
1140
1141 /**
1142  * Old function when $wgBetterDirectionality existed
1143  * Removed in core, kept in extensions for backwards compat.
1144  *
1145  * @deprecated since 1.18
1146  * @return Language
1147  */
1148 function wfUILang() {
1149         global $wgLang;
1150         return $wgLang;
1151 }
1152
1153 /**
1154  * This is the new function for getting translated interface messages.
1155  * See the Message class for documentation how to use them.
1156  * The intention is that this function replaces all old wfMsg* functions.
1157  * @param $key \string Message key.
1158  * Varargs: normal message parameters.
1159  * @return Message
1160  * @since 1.17
1161  */
1162 function wfMessage( $key /*...*/) {
1163         $params = func_get_args();
1164         array_shift( $params );
1165         if ( isset( $params[0] ) && is_array( $params[0] ) ) {
1166                 $params = $params[0];
1167         }
1168         return new Message( $key, $params );
1169 }
1170
1171 /**
1172  * This function accepts multiple message keys and returns a message instance
1173  * for the first message which is non-empty. If all messages are empty then an
1174  * instance of the first message key is returned.
1175  * @param varargs: message keys
1176  * @return Message
1177  * @since 1.18
1178  */
1179 function wfMessageFallback( /*...*/ ) {
1180         $args = func_get_args();
1181         return MWFunction::callArray( 'Message::newFallbackSequence', $args );
1182 }
1183
1184 /**
1185  * Get a message from anywhere, for the current user language.
1186  *
1187  * Use wfMsgForContent() instead if the message should NOT
1188  * change depending on the user preferences.
1189  *
1190  * @param $key String: lookup key for the message, usually
1191  *    defined in languages/Language.php
1192  *
1193  * Parameters to the message, which can be used to insert variable text into
1194  * it, can be passed to this function in the following formats:
1195  * - One per argument, starting at the second parameter
1196  * - As an array in the second parameter
1197  * These are not shown in the function definition.
1198  *
1199  * @return String
1200  */
1201 function wfMsg( $key ) {
1202         $args = func_get_args();
1203         array_shift( $args );
1204         return wfMsgReal( $key, $args );
1205 }
1206
1207 /**
1208  * Same as above except doesn't transform the message
1209  *
1210  * @param $key String
1211  * @return String
1212  */
1213 function wfMsgNoTrans( $key ) {
1214         $args = func_get_args();
1215         array_shift( $args );
1216         return wfMsgReal( $key, $args, true, false, false );
1217 }
1218
1219 /**
1220  * Get a message from anywhere, for the current global language
1221  * set with $wgLanguageCode.
1222  *
1223  * Use this if the message should NOT change dependent on the
1224  * language set in the user's preferences. This is the case for
1225  * most text written into logs, as well as link targets (such as
1226  * the name of the copyright policy page). Link titles, on the
1227  * other hand, should be shown in the UI language.
1228  *
1229  * Note that MediaWiki allows users to change the user interface
1230  * language in their preferences, but a single installation
1231  * typically only contains content in one language.
1232  *
1233  * Be wary of this distinction: If you use wfMsg() where you should
1234  * use wfMsgForContent(), a user of the software may have to
1235  * customize potentially hundreds of messages in
1236  * order to, e.g., fix a link in every possible language.
1237  *
1238  * @param $key String: lookup key for the message, usually
1239  *     defined in languages/Language.php
1240  * @return String
1241  */
1242 function wfMsgForContent( $key ) {
1243         global $wgForceUIMsgAsContentMsg;
1244         $args = func_get_args();
1245         array_shift( $args );
1246         $forcontent = true;
1247         if( is_array( $wgForceUIMsgAsContentMsg ) &&
1248                 in_array( $key, $wgForceUIMsgAsContentMsg ) )
1249         {
1250                 $forcontent = false;
1251         }
1252         return wfMsgReal( $key, $args, true, $forcontent );
1253 }
1254
1255 /**
1256  * Same as above except doesn't transform the message
1257  *
1258  * @param $key String
1259  * @return String
1260  */
1261 function wfMsgForContentNoTrans( $key ) {
1262         global $wgForceUIMsgAsContentMsg;
1263         $args = func_get_args();
1264         array_shift( $args );
1265         $forcontent = true;
1266         if( is_array( $wgForceUIMsgAsContentMsg ) &&
1267                 in_array( $key, $wgForceUIMsgAsContentMsg ) )
1268         {
1269                 $forcontent = false;
1270         }
1271         return wfMsgReal( $key, $args, true, $forcontent, false );
1272 }
1273
1274 /**
1275  * Really get a message
1276  *
1277  * @param $key String: key to get.
1278  * @param $args
1279  * @param $useDB Boolean
1280  * @param $forContent Mixed: Language code, or false for user lang, true for content lang.
1281  * @param $transform Boolean: Whether or not to transform the message.
1282  * @return String: the requested message.
1283  */
1284 function wfMsgReal( $key, $args, $useDB = true, $forContent = false, $transform = true ) {
1285         wfProfileIn( __METHOD__ );
1286         $message = wfMsgGetKey( $key, $useDB, $forContent, $transform );
1287         $message = wfMsgReplaceArgs( $message, $args );
1288         wfProfileOut( __METHOD__ );
1289         return $message;
1290 }
1291
1292 /**
1293  * Fetch a message string value, but don't replace any keys yet.
1294  *
1295  * @param $key String
1296  * @param $useDB Bool
1297  * @param $langCode String: Code of the language to get the message for, or
1298  *                  behaves as a content language switch if it is a boolean.
1299  * @param $transform Boolean: whether to parse magic words, etc.
1300  * @return string
1301  */
1302 function wfMsgGetKey( $key, $useDB = true, $langCode = false, $transform = true ) {
1303         wfRunHooks( 'NormalizeMessageKey', array( &$key, &$useDB, &$langCode, &$transform ) );
1304
1305         $cache = MessageCache::singleton();
1306         $message = $cache->get( $key, $useDB, $langCode );
1307         if( $message === false ) {
1308                 $message = '&lt;' . htmlspecialchars( $key ) . '&gt;';
1309         } elseif ( $transform ) {
1310                 $message = $cache->transform( $message );
1311         }
1312         return $message;
1313 }
1314
1315 /**
1316  * Replace message parameter keys on the given formatted output.
1317  *
1318  * @param $message String
1319  * @param $args Array
1320  * @return string
1321  * @private
1322  */
1323 function wfMsgReplaceArgs( $message, $args ) {
1324         # Fix windows line-endings
1325         # Some messages are split with explode("\n", $msg)
1326         $message = str_replace( "\r", '', $message );
1327
1328         // Replace arguments
1329         if ( count( $args ) ) {
1330                 if ( is_array( $args[0] ) ) {
1331                         $args = array_values( $args[0] );
1332                 }
1333                 $replacementKeys = array();
1334                 foreach( $args as $n => $param ) {
1335                         $replacementKeys['$' . ( $n + 1 )] = $param;
1336                 }
1337                 $message = strtr( $message, $replacementKeys );
1338         }
1339
1340         return $message;
1341 }
1342
1343 /**
1344  * Return an HTML-escaped version of a message.
1345  * Parameter replacements, if any, are done *after* the HTML-escaping,
1346  * so parameters may contain HTML (eg links or form controls). Be sure
1347  * to pre-escape them if you really do want plaintext, or just wrap
1348  * the whole thing in htmlspecialchars().
1349  *
1350  * @param $key String
1351  * @param string ... parameters
1352  * @return string
1353  */
1354 function wfMsgHtml( $key ) {
1355         $args = func_get_args();
1356         array_shift( $args );
1357         return wfMsgReplaceArgs( htmlspecialchars( wfMsgGetKey( $key ) ), $args );
1358 }
1359
1360 /**
1361  * Return an HTML version of message
1362  * Parameter replacements, if any, are done *after* parsing the wiki-text message,
1363  * so parameters may contain HTML (eg links or form controls). Be sure
1364  * to pre-escape them if you really do want plaintext, or just wrap
1365  * the whole thing in htmlspecialchars().
1366  *
1367  * @param $key String
1368  * @param string ... parameters
1369  * @return string
1370  */
1371 function wfMsgWikiHtml( $key ) {
1372         $args = func_get_args();
1373         array_shift( $args );
1374         return wfMsgReplaceArgs(
1375                 MessageCache::singleton()->parse( wfMsgGetKey( $key ), null,
1376                 /* can't be set to false */ true, /* interface */ true )->getText(),
1377                 $args );
1378 }
1379
1380 /**
1381  * Returns message in the requested format
1382  * @param $key String: key of the message
1383  * @param $options Array: processing rules. Can take the following options:
1384  *   <i>parse</i>: parses wikitext to HTML
1385  *   <i>parseinline</i>: parses wikitext to HTML and removes the surrounding
1386  *       p's added by parser or tidy
1387  *   <i>escape</i>: filters message through htmlspecialchars
1388  *   <i>escapenoentities</i>: same, but allows entity references like &#160; through
1389  *   <i>replaceafter</i>: parameters are substituted after parsing or escaping
1390  *   <i>parsemag</i>: transform the message using magic phrases
1391  *   <i>content</i>: fetch message for content language instead of interface
1392  * Also can accept a single associative argument, of the form 'language' => 'xx':
1393  *   <i>language</i>: Language object or language code to fetch message for
1394  *       (overriden by <i>content</i>).
1395  * Behavior for conflicting options (e.g., parse+parseinline) is undefined.
1396  *
1397  * @return String
1398  */
1399 function wfMsgExt( $key, $options ) {
1400         $args = func_get_args();
1401         array_shift( $args );
1402         array_shift( $args );
1403         $options = (array)$options;
1404
1405         foreach( $options as $arrayKey => $option ) {
1406                 if( !preg_match( '/^[0-9]+|language$/', $arrayKey ) ) {
1407                         # An unknown index, neither numeric nor "language"
1408                         wfWarn( "wfMsgExt called with incorrect parameter key $arrayKey", 1, E_USER_WARNING );
1409                 } elseif( preg_match( '/^[0-9]+$/', $arrayKey ) && !in_array( $option,
1410                 array( 'parse', 'parseinline', 'escape', 'escapenoentities',
1411                 'replaceafter', 'parsemag', 'content' ) ) ) {
1412                         # A numeric index with unknown value
1413                         wfWarn( "wfMsgExt called with incorrect parameter $option", 1, E_USER_WARNING );
1414                 }
1415         }
1416
1417         if( in_array( 'content', $options, true ) ) {
1418                 $forContent = true;
1419                 $langCode = true;
1420                 $langCodeObj = null;
1421         } elseif( array_key_exists( 'language', $options ) ) {
1422                 $forContent = false;
1423                 $langCode = wfGetLangObj( $options['language'] );
1424                 $langCodeObj = $langCode;
1425         } else {
1426                 $forContent = false;
1427                 $langCode = false;
1428                 $langCodeObj = null;
1429         }
1430
1431         $string = wfMsgGetKey( $key, /*DB*/true, $langCode, /*Transform*/false );
1432
1433         if( !in_array( 'replaceafter', $options, true ) ) {
1434                 $string = wfMsgReplaceArgs( $string, $args );
1435         }
1436
1437         $messageCache = MessageCache::singleton();
1438         if( in_array( 'parse', $options, true ) ) {
1439                 $string = $messageCache->parse( $string, null, true, !$forContent, $langCodeObj )->getText();
1440         } elseif ( in_array( 'parseinline', $options, true ) ) {
1441                 $string = $messageCache->parse( $string, null, true, !$forContent, $langCodeObj )->getText();
1442                 $m = array();
1443                 if( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $string, $m ) ) {
1444                         $string = $m[1];
1445                 }
1446         } elseif ( in_array( 'parsemag', $options, true ) ) {
1447                 $string = $messageCache->transform( $string,
1448                                 !$forContent, $langCodeObj );
1449         }
1450
1451         if ( in_array( 'escape', $options, true ) ) {
1452                 $string = htmlspecialchars ( $string );
1453         } elseif ( in_array( 'escapenoentities', $options, true ) ) {
1454                 $string = Sanitizer::escapeHtmlAllowEntities( $string );
1455         }
1456
1457         if( in_array( 'replaceafter', $options, true ) ) {
1458                 $string = wfMsgReplaceArgs( $string, $args );
1459         }
1460
1461         return $string;
1462 }
1463
1464 /**
1465  * Since wfMsg() and co suck, they don't return false if the message key they
1466  * looked up didn't exist but a XHTML string, this function checks for the
1467  * nonexistance of messages by checking the MessageCache::get() result directly.
1468  *
1469  * @param $key      String: the message key looked up
1470  * @return Boolean True if the message *doesn't* exist.
1471  */
1472 function wfEmptyMsg( $key ) {
1473         return MessageCache::singleton()->get( $key, /*useDB*/true, /*content*/false ) === false;
1474 }
1475
1476 /**
1477  * Throw a debugging exception. This function previously once exited the process,
1478  * but now throws an exception instead, with similar results.
1479  *
1480  * @param $msg String: message shown when dying.
1481  */
1482 function wfDebugDieBacktrace( $msg = '' ) {
1483         throw new MWException( $msg );
1484 }
1485
1486 /**
1487  * Fetch server name for use in error reporting etc.
1488  * Use real server name if available, so we know which machine
1489  * in a server farm generated the current page.
1490  *
1491  * @return string
1492  */
1493 function wfHostname() {
1494         static $host;
1495         if ( is_null( $host ) ) {
1496                 if ( function_exists( 'posix_uname' ) ) {
1497                         // This function not present on Windows
1498                         $uname = posix_uname();
1499                 } else {
1500                         $uname = false;
1501                 }
1502                 if( is_array( $uname ) && isset( $uname['nodename'] ) ) {
1503                         $host = $uname['nodename'];
1504                 } elseif ( getenv( 'COMPUTERNAME' ) ) {
1505                         # Windows computer name
1506                         $host = getenv( 'COMPUTERNAME' );
1507                 } else {
1508                         # This may be a virtual server.
1509                         $host = $_SERVER['SERVER_NAME'];
1510                 }
1511         }
1512         return $host;
1513 }
1514
1515 /**
1516  * Returns a HTML comment with the elapsed time since request.
1517  * This method has no side effects.
1518  *
1519  * @return string
1520  */
1521 function wfReportTime() {
1522         global $wgRequestTime, $wgShowHostnames;
1523
1524         $now = wfTime();
1525         $elapsed = $now - $wgRequestTime;
1526
1527         return $wgShowHostnames
1528                 ? sprintf( '<!-- Served by %s in %01.3f secs. -->', wfHostname(), $elapsed )
1529                 : sprintf( '<!-- Served in %01.3f secs. -->', $elapsed );
1530 }
1531
1532 /**
1533  * Safety wrapper for debug_backtrace().
1534  *
1535  * With Zend Optimizer 3.2.0 loaded, this causes segfaults under somewhat
1536  * murky circumstances, which may be triggered in part by stub objects
1537  * or other fancy talkin'.
1538  *
1539  * Will return an empty array if Zend Optimizer is detected or if
1540  * debug_backtrace is disabled, otherwise the output from
1541  * debug_backtrace() (trimmed).
1542  *
1543  * @param $limit int This parameter can be used to limit the number of stack frames returned
1544  *
1545  * @return array of backtrace information
1546  */
1547 function wfDebugBacktrace( $limit = 0 ) {
1548         static $disabled = null;
1549
1550         if( extension_loaded( 'Zend Optimizer' ) ) {
1551                 wfDebug( "Zend Optimizer detected; skipping debug_backtrace for safety.\n" );
1552                 return array();
1553         }
1554
1555         if ( is_null( $disabled ) ) {
1556                 $disabled = false;
1557                 $functions = explode( ',', ini_get( 'disable_functions' ) );
1558                 $functions = array_map( 'trim', $functions );
1559                 $functions = array_map( 'strtolower', $functions );
1560                 if ( in_array( 'debug_backtrace', $functions ) ) {
1561                         wfDebug( "debug_backtrace is in disabled_functions\n" );
1562                         $disabled = true;
1563                 }
1564         }
1565         if ( $disabled ) {
1566                 return array();
1567         }
1568
1569         if ( $limit && version_compare( PHP_VERSION, '5.4.0', '>=' ) ) {
1570                 return array_slice( debug_backtrace( DEBUG_BACKTRACE_PROVIDE_OBJECT, $limit ), 1 );
1571         } else {
1572                 return array_slice( debug_backtrace(), 1 );
1573         }
1574 }
1575
1576 /**
1577  * Get a debug backtrace as a string
1578  *
1579  * @return string
1580  */
1581 function wfBacktrace() {
1582         global $wgCommandLineMode;
1583
1584         if ( $wgCommandLineMode ) {
1585                 $msg = '';
1586         } else {
1587                 $msg = "<ul>\n";
1588         }
1589         $backtrace = wfDebugBacktrace();
1590         foreach( $backtrace as $call ) {
1591                 if( isset( $call['file'] ) ) {
1592                         $f = explode( DIRECTORY_SEPARATOR, $call['file'] );
1593                         $file = $f[count( $f ) - 1];
1594                 } else {
1595                         $file = '-';
1596                 }
1597                 if( isset( $call['line'] ) ) {
1598                         $line = $call['line'];
1599                 } else {
1600                         $line = '-';
1601                 }
1602                 if ( $wgCommandLineMode ) {
1603                         $msg .= "$file line $line calls ";
1604                 } else {
1605                         $msg .= '<li>' . $file . ' line ' . $line . ' calls ';
1606                 }
1607                 if( !empty( $call['class'] ) ) {
1608                         $msg .= $call['class'] . $call['type'];
1609                 }
1610                 $msg .= $call['function'] . '()';
1611
1612                 if ( $wgCommandLineMode ) {
1613                         $msg .= "\n";
1614                 } else {
1615                         $msg .= "</li>\n";
1616                 }
1617         }
1618         if ( $wgCommandLineMode ) {
1619                 $msg .= "\n";
1620         } else {
1621                 $msg .= "</ul>\n";
1622         }
1623
1624         return $msg;
1625 }
1626
1627 /**
1628  * Get the name of the function which called this function
1629  *
1630  * @param $level Int
1631  * @return Bool|string
1632  */
1633 function wfGetCaller( $level = 2 ) {
1634         $backtrace = wfDebugBacktrace( $level );
1635         if ( isset( $backtrace[$level] ) ) {
1636                 return wfFormatStackFrame( $backtrace[$level] );
1637         } else {
1638                 $caller = 'unknown';
1639         }
1640         return $caller;
1641 }
1642
1643 /**
1644  * Return a string consisting of callers in the stack. Useful sometimes
1645  * for profiling specific points.
1646  *
1647  * @param $limit The maximum depth of the stack frame to return, or false for
1648  *               the entire stack.
1649  * @return String
1650  */
1651 function wfGetAllCallers( $limit = 3 ) {
1652         $trace = array_reverse( wfDebugBacktrace() );
1653         if ( !$limit || $limit > count( $trace ) - 1 ) {
1654                 $limit = count( $trace ) - 1;
1655         }
1656         $trace = array_slice( $trace, -$limit - 1, $limit );
1657         return implode( '/', array_map( 'wfFormatStackFrame', $trace ) );
1658 }
1659
1660 /**
1661  * Return a string representation of frame
1662  *
1663  * @param $frame Array
1664  * @return Bool
1665  */
1666 function wfFormatStackFrame( $frame ) {
1667         return isset( $frame['class'] ) ?
1668                 $frame['class'] . '::' . $frame['function'] :
1669                 $frame['function'];
1670 }
1671
1672
1673 /* Some generic result counters, pulled out of SearchEngine */
1674
1675
1676 /**
1677  * @todo document
1678  *
1679  * @param $offset Int
1680  * @param $limit Int
1681  * @return String
1682  */
1683 function wfShowingResults( $offset, $limit ) {
1684         global $wgLang;
1685         return wfMsgExt(
1686                 'showingresults',
1687                 array( 'parseinline' ),
1688                 $wgLang->formatNum( $limit ),
1689                 $wgLang->formatNum( $offset + 1 )
1690         );
1691 }
1692
1693 /**
1694  * Generate (prev x| next x) (20|50|100...) type links for paging
1695  *
1696  * @param $offset String
1697  * @param $limit Integer
1698  * @param $link String
1699  * @param $query String: optional URL query parameter string
1700  * @param $atend Bool: optional param for specified if this is the last page
1701  * @return String
1702  * @deprecated in 1.19; use Language::viewPrevNext() instead
1703  */
1704 function wfViewPrevNext( $offset, $limit, $link, $query = '', $atend = false ) {
1705         global $wgLang;
1706
1707         $query = wfCgiToArray( $query );
1708
1709         if( is_object( $link ) ) {
1710                 $title = $link;
1711         } else {
1712                 $title = Title::newFromText( $link );
1713                 if( is_null( $title ) ) {
1714                         return false;
1715                 }
1716         }
1717
1718         return $wgLang->viewPrevNext( $title, $offset, $limit, $query, $atend );
1719 }
1720
1721 /**
1722  * Make a list item, used by various special pages
1723  *
1724  * @param $page String Page link
1725  * @param $details String Text between brackets
1726  * @param $oppositedm Boolean   Add the direction mark opposite to your
1727  *                                                              language, to display text properly
1728  * @return String
1729  * @deprecated since 1.19; use Language::specialList() instead
1730  */
1731 function wfSpecialList( $page, $details, $oppositedm = true ) {
1732         global $wgLang;
1733         return $wgLang->specialList( $page, $details, $oppositedm );
1734 }
1735
1736 /**
1737  * @todo document
1738  * @todo FIXME: We may want to blacklist some broken browsers
1739  *
1740  * @param $force Bool
1741  * @return bool Whereas client accept gzip compression
1742  */
1743 function wfClientAcceptsGzip( $force = false ) {
1744         static $result = null;
1745         if ( $result === null || $force ) {
1746                 $result = false;
1747                 if( isset( $_SERVER['HTTP_ACCEPT_ENCODING'] ) ) {
1748                         # @todo FIXME: We may want to blacklist some broken browsers
1749                         $m = array();
1750                         if( preg_match(
1751                                 '/\bgzip(?:;(q)=([0-9]+(?:\.[0-9]+)))?\b/',
1752                                 $_SERVER['HTTP_ACCEPT_ENCODING'],
1753                                 $m )
1754                         )
1755                         {
1756                                 if( isset( $m[2] ) && ( $m[1] == 'q' ) && ( $m[2] == 0 ) ) {
1757                                         $result = false;
1758                                         return $result;
1759                                 }
1760                                 wfDebug( "wfClientAcceptsGzip: client accepts gzip.\n" );
1761                                 $result = true;
1762                         }
1763                 }
1764         }
1765         return $result;
1766 }
1767
1768 /**
1769  * Obtain the offset and limit values from the request string;
1770  * used in special pages
1771  *
1772  * @param $deflimit Int default limit if none supplied
1773  * @param $optionname String Name of a user preference to check against
1774  * @return array
1775  *
1776  */
1777 function wfCheckLimits( $deflimit = 50, $optionname = 'rclimit' ) {
1778         global $wgRequest;
1779         return $wgRequest->getLimitOffset( $deflimit, $optionname );
1780 }
1781
1782 /**
1783  * Escapes the given text so that it may be output using addWikiText()
1784  * without any linking, formatting, etc. making its way through. This
1785  * is achieved by substituting certain characters with HTML entities.
1786  * As required by the callers, <nowiki> is not used.
1787  *
1788  * @param $text String: text to be escaped
1789  * @return String
1790  */
1791 function wfEscapeWikiText( $text ) {
1792         $text = strtr( "\n$text", array(
1793                 '"' => '&#34;', '&' => '&#38;', "'" => '&#39;', '<' => '&#60;',
1794                 '=' => '&#61;', '>' => '&#62;', '[' => '&#91;', ']' => '&#93;',
1795                 '{' => '&#123;', '|' => '&#124;', '}' => '&#125;',
1796                 "\n#" => "\n&#35;", "\n*" => "\n&#42;",
1797                 "\n:" => "\n&#58;", "\n;" => "\n&#59;",
1798                 '://' => '&#58;//', 'ISBN ' => 'ISBN&#32;', 'RFC ' => 'RFC&#32;',
1799         ) );
1800         return substr( $text, 1 );
1801 }
1802
1803 /**
1804  * Get the current unix timetstamp with microseconds.  Useful for profiling
1805  * @return Float
1806  */
1807 function wfTime() {
1808         return microtime( true );
1809 }
1810
1811 /**
1812  * Sets dest to source and returns the original value of dest
1813  * If source is NULL, it just returns the value, it doesn't set the variable
1814  * If force is true, it will set the value even if source is NULL
1815  *
1816  * @param $dest Mixed
1817  * @param $source Mixed
1818  * @param $force Bool
1819  * @return Mixed
1820  */
1821 function wfSetVar( &$dest, $source, $force = false ) {
1822         $temp = $dest;
1823         if ( !is_null( $source ) || $force ) {
1824                 $dest = $source;
1825         }
1826         return $temp;
1827 }
1828
1829 /**
1830  * As for wfSetVar except setting a bit
1831  *
1832  * @param $dest Int
1833  * @param $bit Int
1834  * @param $state Bool
1835  *
1836  * @return bool
1837  */
1838 function wfSetBit( &$dest, $bit, $state = true ) {
1839         $temp = (bool)( $dest & $bit );
1840         if ( !is_null( $state ) ) {
1841                 if ( $state ) {
1842                         $dest |= $bit;
1843                 } else {
1844                         $dest &= ~$bit;
1845                 }
1846         }
1847         return $temp;
1848 }
1849
1850 /**
1851  * A wrapper around the PHP function var_export().
1852  * Either print it or add it to the regular output ($wgOut).
1853  *
1854  * @param $var A PHP variable to dump.
1855  */
1856 function wfVarDump( $var ) {
1857         global $wgOut;
1858         $s = str_replace( "\n", "<br />\n", var_export( $var, true ) . "\n" );
1859         if ( headers_sent() || !isset( $wgOut ) || !is_object( $wgOut ) ) {
1860                 print $s;
1861         } else {
1862                 $wgOut->addHTML( $s );
1863         }
1864 }
1865
1866 /**
1867  * Provide a simple HTTP error.
1868  *
1869  * @param $code Int|String
1870  * @param $label String
1871  * @param $desc String
1872  */
1873 function wfHttpError( $code, $label, $desc ) {
1874         global $wgOut;
1875         $wgOut->disable();
1876         header( "HTTP/1.0 $code $label" );
1877         header( "Status: $code $label" );
1878         $wgOut->sendCacheControl();
1879
1880         header( 'Content-type: text/html; charset=utf-8' );
1881         print "<!DOCTYPE HTML PUBLIC \"-//IETF//DTD HTML 2.0//EN\">".
1882                 '<html><head><title>' .
1883                 htmlspecialchars( $label ) .
1884                 '</title></head><body><h1>' .
1885                 htmlspecialchars( $label ) .
1886                 '</h1><p>' .
1887                 nl2br( htmlspecialchars( $desc ) ) .
1888                 "</p></body></html>\n";
1889 }
1890
1891 /**
1892  * Clear away any user-level output buffers, discarding contents.
1893  *
1894  * Suitable for 'starting afresh', for instance when streaming
1895  * relatively large amounts of data without buffering, or wanting to
1896  * output image files without ob_gzhandler's compression.
1897  *
1898  * The optional $resetGzipEncoding parameter controls suppression of
1899  * the Content-Encoding header sent by ob_gzhandler; by default it
1900  * is left. See comments for wfClearOutputBuffers() for why it would
1901  * be used.
1902  *
1903  * Note that some PHP configuration options may add output buffer
1904  * layers which cannot be removed; these are left in place.
1905  *
1906  * @param $resetGzipEncoding Bool
1907  */
1908 function wfResetOutputBuffers( $resetGzipEncoding = true ) {
1909         if( $resetGzipEncoding ) {
1910                 // Suppress Content-Encoding and Content-Length
1911                 // headers from 1.10+s wfOutputHandler
1912                 global $wgDisableOutputCompression;
1913                 $wgDisableOutputCompression = true;
1914         }
1915         while( $status = ob_get_status() ) {
1916                 if( $status['type'] == 0 /* PHP_OUTPUT_HANDLER_INTERNAL */ ) {
1917                         // Probably from zlib.output_compression or other
1918                         // PHP-internal setting which can't be removed.
1919                         //
1920                         // Give up, and hope the result doesn't break
1921                         // output behavior.
1922                         break;
1923                 }
1924                 if( !ob_end_clean() ) {
1925                         // Could not remove output buffer handler; abort now
1926                         // to avoid getting in some kind of infinite loop.
1927                         break;
1928                 }
1929                 if( $resetGzipEncoding ) {
1930                         if( $status['name'] == 'ob_gzhandler' ) {
1931                                 // Reset the 'Content-Encoding' field set by this handler
1932                                 // so we can start fresh.
1933                                 if ( function_exists( 'header_remove' ) ) {
1934                                         // Available since PHP 5.3.0
1935                                         header_remove( 'Content-Encoding' );
1936                                 } else {
1937                                         // We need to provide a valid content-coding. See bug 28069
1938                                         header( 'Content-Encoding: identity' );
1939                                 }
1940                                 break;
1941                         }
1942                 }
1943         }
1944 }
1945
1946 /**
1947  * More legible than passing a 'false' parameter to wfResetOutputBuffers():
1948  *
1949  * Clear away output buffers, but keep the Content-Encoding header
1950  * produced by ob_gzhandler, if any.
1951  *
1952  * This should be used for HTTP 304 responses, where you need to
1953  * preserve the Content-Encoding header of the real result, but
1954  * also need to suppress the output of ob_gzhandler to keep to spec
1955  * and avoid breaking Firefox in rare cases where the headers and
1956  * body are broken over two packets.
1957  */
1958 function wfClearOutputBuffers() {
1959         wfResetOutputBuffers( false );
1960 }
1961
1962 /**
1963  * Converts an Accept-* header into an array mapping string values to quality
1964  * factors
1965  *
1966  * @param $accept String
1967  * @param $def String default
1968  * @return Array
1969  */
1970 function wfAcceptToPrefs( $accept, $def = '*/*' ) {
1971         # No arg means accept anything (per HTTP spec)
1972         if( !$accept ) {
1973                 return array( $def => 1.0 );
1974         }
1975
1976         $prefs = array();
1977
1978         $parts = explode( ',', $accept );
1979
1980         foreach( $parts as $part ) {
1981                 # @todo FIXME: Doesn't deal with params like 'text/html; level=1'
1982                 $values = explode( ';', trim( $part ) );
1983                 $match = array();
1984                 if ( count( $values ) == 1 ) {
1985                         $prefs[$values[0]] = 1.0;
1986                 } elseif ( preg_match( '/q\s*=\s*(\d*\.\d+)/', $values[1], $match ) ) {
1987                         $prefs[$values[0]] = floatval( $match[1] );
1988                 }
1989         }
1990
1991         return $prefs;
1992 }
1993
1994 /**
1995  * Checks if a given MIME type matches any of the keys in the given
1996  * array. Basic wildcards are accepted in the array keys.
1997  *
1998  * Returns the matching MIME type (or wildcard) if a match, otherwise
1999  * NULL if no match.
2000  *
2001  * @param $type String
2002  * @param $avail Array
2003  * @return string
2004  * @private
2005  */
2006 function mimeTypeMatch( $type, $avail ) {
2007         if( array_key_exists( $type, $avail ) ) {
2008                 return $type;
2009         } else {
2010                 $parts = explode( '/', $type );
2011                 if( array_key_exists( $parts[0] . '/*', $avail ) ) {
2012                         return $parts[0] . '/*';
2013                 } elseif( array_key_exists( '*/*', $avail ) ) {
2014                         return '*/*';
2015                 } else {
2016                         return null;
2017                 }
2018         }
2019 }
2020
2021 /**
2022  * Returns the 'best' match between a client's requested internet media types
2023  * and the server's list of available types. Each list should be an associative
2024  * array of type to preference (preference is a float between 0.0 and 1.0).
2025  * Wildcards in the types are acceptable.
2026  *
2027  * @param $cprefs Array: client's acceptable type list
2028  * @param $sprefs Array: server's offered types
2029  * @return string
2030  *
2031  * @todo FIXME: Doesn't handle params like 'text/plain; charset=UTF-8'
2032  * XXX: generalize to negotiate other stuff
2033  */
2034 function wfNegotiateType( $cprefs, $sprefs ) {
2035         $combine = array();
2036
2037         foreach( array_keys( $sprefs ) as $type ) {
2038                 $parts = explode( '/', $type );
2039                 if( $parts[1] != '*' ) {
2040                         $ckey = mimeTypeMatch( $type, $cprefs );
2041                         if( $ckey ) {
2042                                 $combine[$type] = $sprefs[$type] * $cprefs[$ckey];
2043                         }
2044                 }
2045         }
2046
2047         foreach( array_keys( $cprefs ) as $type ) {
2048                 $parts = explode( '/', $type );
2049                 if( $parts[1] != '*' && !array_key_exists( $type, $sprefs ) ) {
2050                         $skey = mimeTypeMatch( $type, $sprefs );
2051                         if( $skey ) {
2052                                 $combine[$type] = $sprefs[$skey] * $cprefs[$type];
2053                         }
2054                 }
2055         }
2056
2057         $bestq = 0;
2058         $besttype = null;
2059
2060         foreach( array_keys( $combine ) as $type ) {
2061                 if( $combine[$type] > $bestq ) {
2062                         $besttype = $type;
2063                         $bestq = $combine[$type];
2064                 }
2065         }
2066
2067         return $besttype;
2068 }
2069
2070 /**
2071  * Reference-counted warning suppression
2072  *
2073  * @param $end Bool
2074  */
2075 function wfSuppressWarnings( $end = false ) {
2076         static $suppressCount = 0;
2077         static $originalLevel = false;
2078
2079         if ( $end ) {
2080                 if ( $suppressCount ) {
2081                         --$suppressCount;
2082                         if ( !$suppressCount ) {
2083                                 error_reporting( $originalLevel );
2084                         }
2085                 }
2086         } else {
2087                 if ( !$suppressCount ) {
2088                         // E_DEPRECATED is undefined in PHP 5.2
2089                         if( !defined( 'E_DEPRECATED' ) ) {
2090                                 define( 'E_DEPRECATED', 8192 );
2091                         }
2092                         $originalLevel = error_reporting( E_ALL & ~( E_WARNING | E_NOTICE | E_USER_WARNING | E_USER_NOTICE | E_DEPRECATED ) );
2093                 }
2094                 ++$suppressCount;
2095         }
2096 }
2097
2098 /**
2099  * Restore error level to previous value
2100  */
2101 function wfRestoreWarnings() {
2102         wfSuppressWarnings( true );
2103 }
2104
2105 # Autodetect, convert and provide timestamps of various types
2106
2107 /**
2108  * Unix time - the number of seconds since 1970-01-01 00:00:00 UTC
2109  */
2110 define( 'TS_UNIX', 0 );
2111
2112 /**
2113  * MediaWiki concatenated string timestamp (YYYYMMDDHHMMSS)
2114  */
2115 define( 'TS_MW', 1 );
2116
2117 /**
2118  * MySQL DATETIME (YYYY-MM-DD HH:MM:SS)
2119  */
2120 define( 'TS_DB', 2 );
2121
2122 /**
2123  * RFC 2822 format, for E-mail and HTTP headers
2124  */
2125 define( 'TS_RFC2822', 3 );
2126
2127 /**
2128  * ISO 8601 format with no timezone: 1986-02-09T20:00:00Z
2129  *
2130  * This is used by Special:Export
2131  */
2132 define( 'TS_ISO_8601', 4 );
2133
2134 /**
2135  * An Exif timestamp (YYYY:MM:DD HH:MM:SS)
2136  *
2137  * @see http://exif.org/Exif2-2.PDF The Exif 2.2 spec, see page 28 for the
2138  *       DateTime tag and page 36 for the DateTimeOriginal and
2139  *       DateTimeDigitized tags.
2140  */
2141 define( 'TS_EXIF', 5 );
2142
2143 /**
2144  * Oracle format time.
2145  */
2146 define( 'TS_ORACLE', 6 );
2147
2148 /**
2149  * Postgres format time.
2150  */
2151 define( 'TS_POSTGRES', 7 );
2152
2153 /**
2154  * DB2 format time
2155  */
2156 define( 'TS_DB2', 8 );
2157
2158 /**
2159  * ISO 8601 basic format with no timezone: 19860209T200000Z.  This is used by ResourceLoader
2160  */
2161 define( 'TS_ISO_8601_BASIC', 9 );
2162
2163 /**
2164  * Get a timestamp string in one of various formats
2165  *
2166  * @param $outputtype Mixed: A timestamp in one of the supported formats, the
2167  *                    function will autodetect which format is supplied and act
2168  *                    accordingly.
2169  * @param $ts Mixed: the timestamp to convert or 0 for the current timestamp
2170  * @return Mixed: String / false The same date in the format specified in $outputtype or false
2171  */
2172 function wfTimestamp( $outputtype = TS_UNIX, $ts = 0 ) {
2173         $uts = 0;
2174         $da = array();
2175         $strtime = '';
2176
2177         if ( !$ts ) { // We want to catch 0, '', null... but not date strings starting with a letter.
2178                 $uts = time();
2179                 $strtime = "@$uts";
2180         } elseif ( preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)$/D', $ts, $da ) ) {
2181                 # TS_DB
2182         } elseif ( preg_match( '/^(\d{4}):(\d\d):(\d\d) (\d\d):(\d\d):(\d\d)$/D', $ts, $da ) ) {
2183                 # TS_EXIF
2184         } elseif ( preg_match( '/^(\d{4})(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$/D', $ts, $da ) ) {
2185                 # TS_MW
2186         } elseif ( preg_match( '/^-?\d{1,13}$/D', $ts ) ) {
2187                 # TS_UNIX
2188                 $uts = $ts;
2189                 $strtime = "@$ts"; // Undocumented?
2190         } elseif ( preg_match( '/^\d{2}-\d{2}-\d{4} \d{2}:\d{2}:\d{2}.\d{6}$/', $ts ) ) {
2191                 # TS_ORACLE // session altered to DD-MM-YYYY HH24:MI:SS.FF6
2192                 $strtime = preg_replace( '/(\d\d)\.(\d\d)\.(\d\d)(\.(\d+))?/', "$1:$2:$3",
2193                                 str_replace( '+00:00', 'UTC', $ts ) );
2194         } elseif ( preg_match( '/^(\d{4})-(\d{2})-(\d{2})T(\d{2}):(\d{2}):(\d{2})(?:\.*\d*)?Z$/', $ts, $da ) ) {
2195                 # TS_ISO_8601
2196         } elseif ( preg_match( '/^(\d{4})(\d{2})(\d{2})T(\d{2})(\d{2})(\d{2})(?:\.*\d*)?Z$/', $ts, $da ) ) {
2197                 #TS_ISO_8601_BASIC
2198         } elseif ( preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)\.*\d*[\+\- ](\d\d)$/', $ts, $da ) ) {
2199                 # TS_POSTGRES
2200         } elseif ( preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)\.*\d* GMT$/', $ts, $da ) ) {
2201                 # TS_POSTGRES
2202         } elseif (preg_match( '/^(\d{4})\-(\d\d)\-(\d\d) (\d\d):(\d\d):(\d\d)\.\d\d\d$/', $ts, $da ) ) {
2203                 # TS_DB2
2204         } elseif ( preg_match( '/^[ \t\r\n]*([A-Z][a-z]{2},[ \t\r\n]*)?' . # Day of week
2205                                                         '\d\d?[ \t\r\n]*[A-Z][a-z]{2}[ \t\r\n]*\d{2}(?:\d{2})?' .  # dd Mon yyyy
2206                                                         '[ \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
2207                 # TS_RFC2822, accepting a trailing comment. See http://www.squid-cache.org/mail-archive/squid-users/200307/0122.html / r77171
2208                 # The regex is a superset of rfc2822 for readability
2209                 $strtime = strtok( $ts, ';' );
2210         } 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 ) ) {
2211                 # TS_RFC850
2212                 $strtime = $ts;
2213         } 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 ) ) {
2214                 # asctime
2215                 $strtime = $ts;
2216         } else {
2217                 # Bogus value...
2218                 wfDebug("wfTimestamp() fed bogus time value: TYPE=$outputtype; VALUE=$ts\n");
2219
2220                 return false;
2221         }
2222
2223         static $formats = array(
2224                 TS_UNIX => 'U',
2225                 TS_MW => 'YmdHis',
2226                 TS_DB => 'Y-m-d H:i:s',
2227                 TS_ISO_8601 => 'Y-m-d\TH:i:s\Z',
2228                 TS_ISO_8601_BASIC => 'Ymd\THis\Z',
2229                 TS_EXIF => 'Y:m:d H:i:s', // This shouldn't ever be used, but is included for completeness
2230                 TS_RFC2822 => 'D, d M Y H:i:s',
2231                 TS_ORACLE => 'd-m-Y H:i:s.000000', // Was 'd-M-y h.i.s A' . ' +00:00' before r51500
2232                 TS_POSTGRES => 'Y-m-d H:i:s',
2233                 TS_DB2 => 'Y-m-d H:i:s',
2234         );
2235
2236         if ( !isset( $formats[$outputtype] ) ) {
2237                 throw new MWException( 'wfTimestamp() called with illegal output type.' );
2238         }
2239
2240         if ( function_exists( "date_create" ) ) {
2241                 if ( count( $da ) ) {
2242                         $ds = sprintf("%04d-%02d-%02dT%02d:%02d:%02d.00+00:00",
2243                                 (int)$da[1], (int)$da[2], (int)$da[3],
2244                                 (int)$da[4], (int)$da[5], (int)$da[6]);
2245
2246                         $d = date_create( $ds, new DateTimeZone( 'GMT' ) );
2247                 } elseif ( $strtime ) {
2248                         $d = date_create( $strtime, new DateTimeZone( 'GMT' ) );
2249                 } else {
2250                         return false;
2251                 }
2252
2253                 if ( !$d ) {
2254                         wfDebug("wfTimestamp() fed bogus time value: $outputtype; $ts\n");
2255                         return false;
2256                 }
2257
2258                 $output = $d->format( $formats[$outputtype] );
2259         } else {
2260                 if ( count( $da ) ) {
2261                         // Warning! gmmktime() acts oddly if the month or day is set to 0
2262                         // We may want to handle that explicitly at some point
2263                         $uts = gmmktime( (int)$da[4], (int)$da[5], (int)$da[6],
2264                                 (int)$da[2], (int)$da[3], (int)$da[1] );
2265                 } elseif ( $strtime ) {
2266                         $uts = strtotime( $strtime );
2267                 }
2268
2269                 if ( $uts === false ) {
2270                         wfDebug("wfTimestamp() can't parse the timestamp (non 32-bit time? Update php): $outputtype; $ts\n");
2271                         return false;
2272                 }
2273
2274                 if ( TS_UNIX == $outputtype ) {
2275                         return $uts;
2276                 }
2277                 $output = gmdate( $formats[$outputtype], $uts );
2278         }
2279
2280         if ( ( $outputtype == TS_RFC2822 ) || ( $outputtype == TS_POSTGRES ) ) {
2281                 $output .= ' GMT';
2282         }
2283
2284         return $output;
2285 }
2286
2287 /**
2288  * Return a formatted timestamp, or null if input is null.
2289  * For dealing with nullable timestamp columns in the database.
2290  *
2291  * @param $outputtype Integer
2292  * @param $ts String
2293  * @return String
2294  */
2295 function wfTimestampOrNull( $outputtype = TS_UNIX, $ts = null ) {
2296         if( is_null( $ts ) ) {
2297                 return null;
2298         } else {
2299                 return wfTimestamp( $outputtype, $ts );
2300         }
2301 }
2302
2303 /**
2304  * Convenience function; returns MediaWiki timestamp for the present time.
2305  *
2306  * @return string
2307  */
2308 function wfTimestampNow() {
2309         # return NOW
2310         return wfTimestamp( TS_MW, time() );
2311 }
2312
2313 /**
2314  * Check if the operating system is Windows
2315  *
2316  * @return Bool: true if it's Windows, False otherwise.
2317  */
2318 function wfIsWindows() {
2319         static $isWindows = null;
2320         if ( $isWindows === null ) {
2321                 $isWindows = substr( php_uname(), 0, 7 ) == 'Windows';
2322         }
2323         return $isWindows;
2324 }
2325
2326 /**
2327  * Check if we are running under HipHop
2328  *
2329  * @return Bool
2330  */
2331 function wfIsHipHop() {
2332         return function_exists( 'hphp_thread_set_warmup_enabled' );
2333 }
2334
2335 /**
2336  * Swap two variables
2337  *
2338  * @param $x Mixed
2339  * @param $y Mixed
2340  */
2341 function swap( &$x, &$y ) {
2342         $z = $x;
2343         $x = $y;
2344         $y = $z;
2345 }
2346
2347 /**
2348  * Tries to get the system directory for temporary files. The TMPDIR, TMP, and
2349  * TEMP environment variables are then checked in sequence, and if none are set
2350  * try sys_get_temp_dir() for PHP >= 5.2.1. All else fails, return /tmp for Unix
2351  * or C:\Windows\Temp for Windows and hope for the best.
2352  * It is common to call it with tempnam().
2353  *
2354  * NOTE: When possible, use instead the tmpfile() function to create
2355  * temporary files to avoid race conditions on file creation, etc.
2356  *
2357  * @return String
2358  */
2359 function wfTempDir() {
2360         foreach( array( 'TMPDIR', 'TMP', 'TEMP' ) as $var ) {
2361                 $tmp = getenv( $var );
2362                 if( $tmp && file_exists( $tmp ) && is_dir( $tmp ) && is_writable( $tmp ) ) {
2363                         return $tmp;
2364                 }
2365         }
2366         if( function_exists( 'sys_get_temp_dir' ) ) {
2367                 return sys_get_temp_dir();
2368         }
2369         # Usual defaults
2370         return wfIsWindows() ? 'C:\Windows\Temp' : '/tmp';
2371 }
2372
2373 /**
2374  * Make directory, and make all parent directories if they don't exist
2375  *
2376  * @param $dir String: full path to directory to create
2377  * @param $mode Integer: chmod value to use, default is $wgDirectoryMode
2378  * @param $caller String: optional caller param for debugging.
2379  * @return bool
2380  */
2381 function wfMkdirParents( $dir, $mode = null, $caller = null ) {
2382         global $wgDirectoryMode;
2383
2384         if ( !is_null( $caller ) ) {
2385                 wfDebug( "$caller: called wfMkdirParents($dir)\n" );
2386         }
2387
2388         if( strval( $dir ) === '' || file_exists( $dir ) ) {
2389                 return true;
2390         }
2391
2392         $dir = str_replace( array( '\\', '/' ), DIRECTORY_SEPARATOR, $dir );
2393
2394         if ( is_null( $mode ) ) {
2395                 $mode = $wgDirectoryMode;
2396         }
2397
2398         // Turn off the normal warning, we're doing our own below
2399         wfSuppressWarnings();
2400         $ok = mkdir( $dir, $mode, true ); // PHP5 <3
2401         wfRestoreWarnings();
2402
2403         if( !$ok ) {
2404                 // PHP doesn't report the path in its warning message, so add our own to aid in diagnosis.
2405                 trigger_error( __FUNCTION__ . ": failed to mkdir \"$dir\" mode $mode", E_USER_WARNING );
2406         }
2407         return $ok;
2408 }
2409
2410 /**
2411  * Increment a statistics counter
2412  *
2413  * @param $key String
2414  * @param $count Int
2415  */
2416 function wfIncrStats( $key, $count = 1 ) {
2417         global $wgStatsMethod;
2418
2419         $count = intval( $count );
2420
2421         if( $wgStatsMethod == 'udp' ) {
2422                 global $wgUDPProfilerHost, $wgUDPProfilerPort, $wgDBname, $wgAggregateStatsID;
2423                 static $socket;
2424
2425                 $id = $wgAggregateStatsID !== false ? $wgAggregateStatsID : $wgDBname;
2426
2427                 if ( !$socket ) {
2428                         $socket = socket_create( AF_INET, SOCK_DGRAM, SOL_UDP );
2429                         $statline = "stats/{$id} - {$count} 1 1 1 1 -total\n";
2430                         socket_sendto(
2431                                 $socket,
2432                                 $statline,
2433                                 strlen( $statline ),
2434                                 0,
2435                                 $wgUDPProfilerHost,
2436                                 $wgUDPProfilerPort
2437                         );
2438                 }
2439                 $statline = "stats/{$id} - {$count} 1 1 1 1 {$key}\n";
2440                 wfSuppressWarnings();
2441                 socket_sendto(
2442                         $socket,
2443                         $statline,
2444                         strlen( $statline ),
2445                         0,
2446                         $wgUDPProfilerHost,
2447                         $wgUDPProfilerPort
2448                 );
2449                 wfRestoreWarnings();
2450         } elseif( $wgStatsMethod == 'cache' ) {
2451                 global $wgMemc;
2452                 $key = wfMemcKey( 'stats', $key );
2453                 if ( is_null( $wgMemc->incr( $key, $count ) ) ) {
2454                         $wgMemc->add( $key, $count );
2455                 }
2456         } else {
2457                 // Disabled
2458         }
2459 }
2460
2461 /**
2462  * @param $nr Mixed: the number to format
2463  * @param $acc Integer: the number of digits after the decimal point, default 2
2464  * @param $round Boolean: whether or not to round the value, default true
2465  * @return float
2466  */
2467 function wfPercent( $nr, $acc = 2, $round = true ) {
2468         $ret = sprintf( "%.${acc}f", $nr );
2469         return $round ? round( $ret, $acc ) . '%' : "$ret%";
2470 }
2471
2472 /**
2473  * Find out whether or not a mixed variable exists in a string
2474  *
2475  * @param $needle String
2476  * @param $str String
2477  * @param $insensitive Boolean
2478  * @return Boolean
2479  */
2480 function in_string( $needle, $str, $insensitive = false ) {
2481         $func = 'strpos';
2482         if( $insensitive ) $func = 'stripos';
2483
2484         return $func( $str, $needle ) !== false;
2485 }
2486
2487 /**
2488  * Safety wrapper around ini_get() for boolean settings.
2489  * The values returned from ini_get() are pre-normalized for settings
2490  * set via php.ini or php_flag/php_admin_flag... but *not*
2491  * for those set via php_value/php_admin_value.
2492  *
2493  * It's fairly common for people to use php_value instead of php_flag,
2494  * which can leave you with an 'off' setting giving a false positive
2495  * for code that just takes the ini_get() return value as a boolean.
2496  *
2497  * To make things extra interesting, setting via php_value accepts
2498  * "true" and "yes" as true, but php.ini and php_flag consider them false. :)
2499  * Unrecognized values go false... again opposite PHP's own coercion
2500  * from string to bool.
2501  *
2502  * Luckily, 'properly' set settings will always come back as '0' or '1',
2503  * so we only have to worry about them and the 'improper' settings.
2504  *
2505  * I frickin' hate PHP... :P
2506  *
2507  * @param $setting String
2508  * @return Bool
2509  */
2510 function wfIniGetBool( $setting ) {
2511         $val = ini_get( $setting );
2512         // 'on' and 'true' can't have whitespace around them, but '1' can.
2513         return strtolower( $val ) == 'on'
2514                 || strtolower( $val ) == 'true'
2515                 || strtolower( $val ) == 'yes'
2516                 || preg_match( "/^\s*[+-]?0*[1-9]/", $val ); // approx C atoi() function
2517 }
2518
2519 /**
2520  * Wrapper function for PHP's dl(). This doesn't work in most situations from
2521  * PHP 5.3 onward, and is usually disabled in shared environments anyway.
2522  *
2523  * @param $extension String A PHP extension. The file suffix (.so or .dll)
2524  *                          should be omitted
2525  * @param $fileName String Name of the library, if not $extension.suffix
2526  * @return Bool - Whether or not the extension is loaded
2527  */
2528 function wfDl( $extension, $fileName = null ) {
2529         if( extension_loaded( $extension ) ) {
2530                 return true;
2531         }
2532
2533         $canDl = false;
2534         $sapi = php_sapi_name();
2535         if( version_compare( PHP_VERSION, '5.3.0', '<' ) ||
2536                 $sapi == 'cli' || $sapi == 'cgi' || $sapi == 'embed' )
2537         {
2538                 $canDl = ( function_exists( 'dl' ) && is_callable( 'dl' )
2539                 && wfIniGetBool( 'enable_dl' ) && !wfIniGetBool( 'safe_mode' ) );
2540         }
2541
2542         if( $canDl ) {
2543                 $fileName = $fileName ? $fileName : $extension;
2544                 if( wfIsWindows() ) {
2545                         $fileName = 'php_' . $fileName;
2546                 }
2547                 wfSuppressWarnings();
2548                 dl( $fileName . '.' . PHP_SHLIB_SUFFIX );
2549                 wfRestoreWarnings();
2550         }
2551         return extension_loaded( $extension );
2552 }
2553
2554 /**
2555  * Windows-compatible version of escapeshellarg()
2556  * Windows doesn't recognise single-quotes in the shell, but the escapeshellarg()
2557  * function puts single quotes in regardless of OS.
2558  *
2559  * Also fixes the locale problems on Linux in PHP 5.2.6+ (bug backported to
2560  * earlier distro releases of PHP)
2561  *
2562  * @param varargs
2563  * @return String
2564  */
2565 function wfEscapeShellArg( ) {
2566         wfInitShellLocale();
2567
2568         $args = func_get_args();
2569         $first = true;
2570         $retVal = '';
2571         foreach ( $args as $arg ) {
2572                 if ( !$first ) {
2573                         $retVal .= ' ';
2574                 } else {
2575                         $first = false;
2576                 }
2577
2578                 if ( wfIsWindows() ) {
2579                         // Escaping for an MSVC-style command line parser and CMD.EXE
2580                         // Refs:
2581                         //  * http://web.archive.org/web/20020708081031/http://mailman.lyra.org/pipermail/scite-interest/2002-March/000436.html
2582                         //  * http://technet.microsoft.com/en-us/library/cc723564.aspx
2583                         //  * Bug #13518
2584                         //  * CR r63214
2585                         // Double the backslashes before any double quotes. Escape the double quotes.
2586                         $tokens = preg_split( '/(\\\\*")/', $arg, -1, PREG_SPLIT_DELIM_CAPTURE );
2587                         $arg = '';
2588                         $iteration = 0;
2589                         foreach ( $tokens as $token ) {
2590                                 if ( $iteration % 2 == 1 ) {
2591                                         // Delimiter, a double quote preceded by zero or more slashes
2592                                         $arg .= str_replace( '\\', '\\\\', substr( $token, 0, -1 ) ) . '\\"';
2593                                 } elseif ( $iteration % 4 == 2 ) {
2594                                         // ^ in $token will be outside quotes, need to be escaped
2595                                         $arg .= str_replace( '^', '^^', $token );
2596                                 } else { // $iteration % 4 == 0
2597                                         // ^ in $token will appear inside double quotes, so leave as is
2598                                         $arg .= $token;
2599                                 }
2600                                 $iteration++;
2601                         }
2602                         // Double the backslashes before the end of the string, because
2603                         // we will soon add a quote
2604                         $m = array();
2605                         if ( preg_match( '/^(.*?)(\\\\+)$/', $arg, $m ) ) {
2606                                 $arg = $m[1] . str_replace( '\\', '\\\\', $m[2] );
2607                         }
2608
2609                         // Add surrounding quotes
2610                         $retVal .= '"' . $arg . '"';
2611                 } else {
2612                         $retVal .= escapeshellarg( $arg );
2613                 }
2614         }
2615         return $retVal;
2616 }
2617
2618 /**
2619  * Execute a shell command, with time and memory limits mirrored from the PHP
2620  * configuration if supported.
2621  * @param $cmd String Command line, properly escaped for shell.
2622  * @param &$retval optional, will receive the program's exit code.
2623  *                 (non-zero is usually failure)
2624  * @param $environ Array optional environment variables which should be
2625  *                 added to the executed command environment.
2626  * @return collected stdout as a string (trailing newlines stripped)
2627  */
2628 function wfShellExec( $cmd, &$retval = null, $environ = array() ) {
2629         global $IP, $wgMaxShellMemory, $wgMaxShellFileSize, $wgMaxShellTime;
2630
2631         static $disabled;
2632         if ( is_null( $disabled ) ) {
2633                 $disabled = false;
2634                 if( wfIniGetBool( 'safe_mode' ) ) {
2635                         wfDebug( "wfShellExec can't run in safe_mode, PHP's exec functions are too broken.\n" );
2636                         $disabled = 'safemode';
2637                 } else {
2638                         $functions = explode( ',', ini_get( 'disable_functions' ) );
2639                         $functions = array_map( 'trim', $functions );
2640                         $functions = array_map( 'strtolower', $functions );
2641                         if ( in_array( 'passthru', $functions ) ) {
2642                                 wfDebug( "passthru is in disabled_functions\n" );
2643                                 $disabled = 'passthru';
2644                         }
2645                 }
2646         }
2647         if ( $disabled ) {
2648                 $retval = 1;
2649                 return $disabled == 'safemode' ?
2650                         'Unable to run external programs in safe mode.' :
2651                         'Unable to run external programs, passthru() is disabled.';
2652         }
2653
2654         wfInitShellLocale();
2655
2656         $envcmd = '';
2657         foreach( $environ as $k => $v ) {
2658                 if ( wfIsWindows() ) {
2659                         /* Surrounding a set in quotes (method used by wfEscapeShellArg) makes the quotes themselves
2660                          * appear in the environment variable, so we must use carat escaping as documented in
2661                          * http://technet.microsoft.com/en-us/library/cc723564.aspx
2662                          * Note however that the quote isn't listed there, but is needed, and the parentheses
2663                          * are listed there but doesn't appear to need it.
2664                          */
2665                         $envcmd .= "set $k=" . preg_replace( '/([&|()<>^"])/', '^\\1', $v ) . '&& ';
2666                 } else {
2667                         /* Assume this is a POSIX shell, thus required to accept variable assignments before the command
2668                          * http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_09_01
2669                          */
2670                         $envcmd .= "$k=" . escapeshellarg( $v ) . ' ';
2671                 }
2672         }
2673         $cmd = $envcmd . $cmd;
2674
2675         if ( wfIsWindows() ) {
2676                 if ( version_compare( PHP_VERSION, '5.3.0', '<' ) && /* Fixed in 5.3.0 :) */
2677                         ( version_compare( PHP_VERSION, '5.2.1', '>=' ) || php_uname( 's' ) == 'Windows NT' ) )
2678                 {
2679                         # Hack to work around PHP's flawed invocation of cmd.exe
2680                         # http://news.php.net/php.internals/21796
2681                         # Windows 9x doesn't accept any kind of quotes
2682                         $cmd = '"' . $cmd . '"';
2683                 }
2684         } elseif ( php_uname( 's' ) == 'Linux' ) {
2685                 $time = intval( $wgMaxShellTime );
2686                 $mem = intval( $wgMaxShellMemory );
2687                 $filesize = intval( $wgMaxShellFileSize );
2688
2689                 if ( $time > 0 && $mem > 0 ) {
2690                         $script = "$IP/bin/ulimit4.sh";
2691                         if ( is_executable( $script ) ) {
2692                                 $cmd = '/bin/bash ' . escapeshellarg( $script ) . " $time $mem $filesize " . escapeshellarg( $cmd );
2693                         }
2694                 }
2695         }
2696         wfDebug( "wfShellExec: $cmd\n" );
2697
2698         $retval = 1; // error by default?
2699         ob_start();
2700         passthru( $cmd, $retval );
2701         $output = ob_get_contents();
2702         ob_end_clean();
2703
2704         if ( $retval == 127 ) {
2705                 wfDebugLog( 'exec', "Possibly missing executable file: $cmd\n" );
2706         }
2707         return $output;
2708 }
2709
2710 /**
2711  * Workaround for http://bugs.php.net/bug.php?id=45132
2712  * escapeshellarg() destroys non-ASCII characters if LANG is not a UTF-8 locale
2713  */
2714 function wfInitShellLocale() {
2715         static $done = false;
2716         if ( $done ) {
2717                 return;
2718         }
2719         $done = true;
2720         global $wgShellLocale;
2721         if ( !wfIniGetBool( 'safe_mode' ) ) {
2722                 putenv( "LC_CTYPE=$wgShellLocale" );
2723                 setlocale( LC_CTYPE, $wgShellLocale );
2724         }
2725 }
2726
2727 /**
2728  * Generate a shell-escaped command line string to run a maintenance script.
2729  * Note that $parameters should be a flat array and an option with an argument
2730  * should consist of two consecutive items in the array (do not use "--option value").
2731  * @param $script string MediaWiki maintenance script path
2732  * @param $parameters Array Arguments and options to the script
2733  * @param $options Array Associative array of options:
2734  *              'php': The path to the php executable
2735  *              'wrapper': Path to a PHP wrapper to handle the maintenance script
2736  * @return Array
2737  */
2738 function wfShellMaintenanceCmd( $script, array $parameters = array(), array $options = array() ) {
2739         global $wgPhpCli;
2740         // Give site config file a chance to run the script in a wrapper.
2741         // The caller may likely want to call wfBasename() on $script.
2742         wfRunHooks( 'wfShellMaintenanceCmd', array( &$script, &$parameters, &$options ) );
2743         $cmd = isset( $options['php'] ) ? array( $options['php'] ) : array( $wgPhpCli );
2744         if ( isset( $options['wrapper'] ) ) {
2745                 $cmd[] = $options['wrapper'];
2746         }
2747         $cmd[] = $script;
2748         // Escape each parameter for shell
2749         return implode( " ", array_map( 'wfEscapeShellArg', array_merge( $cmd, $parameters ) ) );
2750 }
2751
2752 /**
2753  * wfMerge attempts to merge differences between three texts.
2754  * Returns true for a clean merge and false for failure or a conflict.
2755  *
2756  * @param $old String
2757  * @param $mine String
2758  * @param $yours String
2759  * @param $result String
2760  * @return Bool
2761  */
2762 function wfMerge( $old, $mine, $yours, &$result ) {
2763         global $wgDiff3;
2764
2765         # This check may also protect against code injection in
2766         # case of broken installations.
2767         wfSuppressWarnings();
2768         $haveDiff3 = $wgDiff3 && file_exists( $wgDiff3 );
2769         wfRestoreWarnings();
2770
2771         if( !$haveDiff3 ) {
2772                 wfDebug( "diff3 not found\n" );
2773                 return false;
2774         }
2775
2776         # Make temporary files
2777         $td = wfTempDir();
2778         $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
2779         $mytextFile = fopen( $mytextName = tempnam( $td, 'merge-mine-' ), 'w' );
2780         $yourtextFile = fopen( $yourtextName = tempnam( $td, 'merge-your-' ), 'w' );
2781
2782         fwrite( $oldtextFile, $old );
2783         fclose( $oldtextFile );
2784         fwrite( $mytextFile, $mine );
2785         fclose( $mytextFile );
2786         fwrite( $yourtextFile, $yours );
2787         fclose( $yourtextFile );
2788
2789         # Check for a conflict
2790         $cmd = $wgDiff3 . ' -a --overlap-only ' .
2791                 wfEscapeShellArg( $mytextName ) . ' ' .
2792                 wfEscapeShellArg( $oldtextName ) . ' ' .
2793                 wfEscapeShellArg( $yourtextName );
2794         $handle = popen( $cmd, 'r' );
2795
2796         if( fgets( $handle, 1024 ) ) {
2797                 $conflict = true;
2798         } else {
2799                 $conflict = false;
2800         }
2801         pclose( $handle );
2802
2803         # Merge differences
2804         $cmd = $wgDiff3 . ' -a -e --merge ' .
2805                 wfEscapeShellArg( $mytextName, $oldtextName, $yourtextName );
2806         $handle = popen( $cmd, 'r' );
2807         $result = '';
2808         do {
2809                 $data = fread( $handle, 8192 );
2810                 if ( strlen( $data ) == 0 ) {
2811                         break;
2812                 }
2813                 $result .= $data;
2814         } while ( true );
2815         pclose( $handle );
2816         unlink( $mytextName );
2817         unlink( $oldtextName );
2818         unlink( $yourtextName );
2819
2820         if ( $result === '' && $old !== '' && !$conflict ) {
2821                 wfDebug( "Unexpected null result from diff3. Command: $cmd\n" );
2822                 $conflict = true;
2823         }
2824         return !$conflict;
2825 }
2826
2827 /**
2828  * Returns unified plain-text diff of two texts.
2829  * Useful for machine processing of diffs.
2830  *
2831  * @param $before String: the text before the changes.
2832  * @param $after String: the text after the changes.
2833  * @param $params String: command-line options for the diff command.
2834  * @return String: unified diff of $before and $after
2835  */
2836 function wfDiff( $before, $after, $params = '-u' ) {
2837         if ( $before == $after ) {
2838                 return '';
2839         }
2840
2841         global $wgDiff;
2842         wfSuppressWarnings();
2843         $haveDiff = $wgDiff && file_exists( $wgDiff );
2844         wfRestoreWarnings();
2845
2846         # This check may also protect against code injection in
2847         # case of broken installations.
2848         if( !$haveDiff ) {
2849                 wfDebug( "diff executable not found\n" );
2850                 $diffs = new Diff( explode( "\n", $before ), explode( "\n", $after ) );
2851                 $format = new UnifiedDiffFormatter();
2852                 return $format->format( $diffs );
2853         }
2854
2855         # Make temporary files
2856         $td = wfTempDir();
2857         $oldtextFile = fopen( $oldtextName = tempnam( $td, 'merge-old-' ), 'w' );
2858         $newtextFile = fopen( $newtextName = tempnam( $td, 'merge-your-' ), 'w' );
2859
2860         fwrite( $oldtextFile, $before );
2861         fclose( $oldtextFile );
2862         fwrite( $newtextFile, $after );
2863         fclose( $newtextFile );
2864
2865         // Get the diff of the two files
2866         $cmd = "$wgDiff " . $params . ' ' . wfEscapeShellArg( $oldtextName, $newtextName );
2867
2868         $h = popen( $cmd, 'r' );
2869
2870         $diff = '';
2871
2872         do {
2873                 $data = fread( $h, 8192 );
2874                 if ( strlen( $data ) == 0 ) {
2875                         break;
2876                 }
2877                 $diff .= $data;
2878         } while ( true );
2879
2880         // Clean up
2881         pclose( $h );
2882         unlink( $oldtextName );
2883         unlink( $newtextName );
2884
2885         // Kill the --- and +++ lines. They're not useful.
2886         $diff_lines = explode( "\n", $diff );
2887         if ( strpos( $diff_lines[0], '---' ) === 0 ) {
2888                 unset( $diff_lines[0] );
2889         }
2890         if ( strpos( $diff_lines[1], '+++' ) === 0 ) {
2891                 unset( $diff_lines[1] );
2892         }
2893
2894         $diff = implode( "\n", $diff_lines );
2895
2896         return $diff;
2897 }
2898
2899 /**
2900  * This function works like "use VERSION" in Perl, the program will die with a
2901  * backtrace if the current version of PHP is less than the version provided
2902  *
2903  * This is useful for extensions which due to their nature are not kept in sync
2904  * with releases, and might depend on other versions of PHP than the main code
2905  *
2906  * Note: PHP might die due to parsing errors in some cases before it ever
2907  *       manages to call this function, such is life
2908  *
2909  * @see perldoc -f use
2910  *
2911  * @param $req_ver Mixed: the version to check, can be a string, an integer, or
2912  *                 a float
2913  */
2914 function wfUsePHP( $req_ver ) {
2915         $php_ver = PHP_VERSION;
2916
2917         if ( version_compare( $php_ver, (string)$req_ver, '<' ) ) {
2918                 throw new MWException( "PHP $req_ver required--this is only $php_ver" );
2919         }
2920 }
2921
2922 /**
2923  * This function works like "use VERSION" in Perl except it checks the version
2924  * of MediaWiki, the program will die with a backtrace if the current version
2925  * of MediaWiki is less than the version provided.
2926  *
2927  * This is useful for extensions which due to their nature are not kept in sync
2928  * with releases
2929  *
2930  * @see perldoc -f use
2931  *
2932  * @param $req_ver Mixed: the version to check, can be a string, an integer, or
2933  *                 a float
2934  */
2935 function wfUseMW( $req_ver ) {
2936         global $wgVersion;
2937
2938         if ( version_compare( $wgVersion, (string)$req_ver, '<' ) ) {
2939                 throw new MWException( "MediaWiki $req_ver required--this is only $wgVersion" );
2940         }
2941 }
2942
2943 /**
2944  * Return the final portion of a pathname.
2945  * Reimplemented because PHP5's basename() is buggy with multibyte text.
2946  * http://bugs.php.net/bug.php?id=33898
2947  *
2948  * PHP's basename() only considers '\' a pathchar on Windows and Netware.
2949  * We'll consider it so always, as we don't want \s in our Unix paths either.
2950  *
2951  * @param $path String
2952  * @param $suffix String: to remove if present
2953  * @return String
2954  */
2955 function wfBaseName( $path, $suffix = '' ) {
2956         $encSuffix = ( $suffix == '' )
2957                 ? ''
2958                 : ( '(?:' . preg_quote( $suffix, '#' ) . ')?' );
2959         $matches = array();
2960         if( preg_match( "#([^/\\\\]*?){$encSuffix}[/\\\\]*$#", $path, $matches ) ) {
2961                 return $matches[1];
2962         } else {
2963                 return '';
2964         }
2965 }
2966
2967 /**
2968  * Generate a relative path name to the given file.
2969  * May explode on non-matching case-insensitive paths,
2970  * funky symlinks, etc.
2971  *
2972  * @param $path String: absolute destination path including target filename
2973  * @param $from String: Absolute source path, directory only
2974  * @return String
2975  */
2976 function wfRelativePath( $path, $from ) {
2977         // Normalize mixed input on Windows...
2978         $path = str_replace( '/', DIRECTORY_SEPARATOR, $path );
2979         $from = str_replace( '/', DIRECTORY_SEPARATOR, $from );
2980
2981         // Trim trailing slashes -- fix for drive root
2982         $path = rtrim( $path, DIRECTORY_SEPARATOR );
2983         $from = rtrim( $from, DIRECTORY_SEPARATOR );
2984
2985         $pieces = explode( DIRECTORY_SEPARATOR, dirname( $path ) );
2986         $against = explode( DIRECTORY_SEPARATOR, $from );
2987
2988         if( $pieces[0] !== $against[0] ) {
2989                 // Non-matching Windows drive letters?
2990                 // Return a full path.
2991                 return $path;
2992         }
2993
2994         // Trim off common prefix
2995         while( count( $pieces ) && count( $against )
2996                 && $pieces[0] == $against[0] ) {
2997                 array_shift( $pieces );
2998                 array_shift( $against );
2999         }
3000
3001         // relative dots to bump us to the parent
3002         while( count( $against ) ) {
3003                 array_unshift( $pieces, '..' );
3004                 array_shift( $against );
3005         }
3006
3007         array_push( $pieces, wfBaseName( $path ) );
3008
3009         return implode( DIRECTORY_SEPARATOR, $pieces );
3010 }
3011
3012 /**
3013  * Do any deferred updates and clear the list
3014  *
3015  * @deprecated since 1.19
3016  * @see DeferredUpdates::doUpdate()
3017  * @param $commit string
3018  */
3019 function wfDoUpdates( $commit = '' ) {
3020         DeferredUpdates::doUpdates( $commit );
3021 }
3022
3023 /**
3024  * Convert an arbitrarily-long digit string from one numeric base
3025  * to another, optionally zero-padding to a minimum column width.
3026  *
3027  * Supports base 2 through 36; digit values 10-36 are represented
3028  * as lowercase letters a-z. Input is case-insensitive.
3029  *
3030  * @param $input String: of digits
3031  * @param $sourceBase Integer: 2-36
3032  * @param $destBase Integer: 2-36
3033  * @param $pad Integer: 1 or greater
3034  * @param $lowercase Boolean
3035  * @return String or false on invalid input
3036  */
3037 function wfBaseConvert( $input, $sourceBase, $destBase, $pad = 1, $lowercase = true ) {
3038         $input = strval( $input );
3039         if( $sourceBase < 2 ||
3040                 $sourceBase > 36 ||
3041                 $destBase < 2 ||
3042                 $destBase > 36 ||
3043                 $pad < 1 ||
3044                 $sourceBase != intval( $sourceBase ) ||
3045                 $destBase != intval( $destBase ) ||
3046                 $pad != intval( $pad ) ||
3047                 !is_string( $input ) ||
3048                 $input == '' ) {
3049                 return false;
3050         }
3051         $digitChars = ( $lowercase ) ? '0123456789abcdefghijklmnopqrstuvwxyz' : '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ';
3052         $inDigits = array();
3053         $outChars = '';
3054
3055         // Decode and validate input string
3056         $input = strtolower( $input );
3057         for( $i = 0; $i < strlen( $input ); $i++ ) {
3058                 $n = strpos( $digitChars, $input[$i] );
3059                 if( $n === false || $n > $sourceBase ) {
3060                         return false;
3061                 }
3062                 $inDigits[] = $n;
3063         }
3064
3065         // Iterate over the input, modulo-ing out an output digit
3066         // at a time until input is gone.
3067         while( count( $inDigits ) ) {
3068                 $work = 0;
3069                 $workDigits = array();
3070
3071                 // Long division...
3072                 foreach( $inDigits as $digit ) {
3073                         $work *= $sourceBase;
3074                         $work += $digit;
3075
3076                         if( $work < $destBase ) {
3077                                 // Gonna need to pull another digit.
3078                                 if( count( $workDigits ) ) {
3079                                         // Avoid zero-padding; this lets us find
3080                                         // the end of the input very easily when
3081                                         // length drops to zero.
3082                                         $workDigits[] = 0;
3083                                 }
3084                         } else {
3085                                 // Finally! Actual division!
3086                                 $workDigits[] = intval( $work / $destBase );
3087
3088                                 // Isn't it annoying that most programming languages
3089                                 // don't have a single divide-and-remainder operator,
3090                                 // even though the CPU implements it that way?
3091                                 $work = $work % $destBase;
3092                         }
3093                 }
3094
3095                 // All that division leaves us with a remainder,
3096                 // which is conveniently our next output digit.
3097                 $outChars .= $digitChars[$work];
3098
3099                 // And we continue!
3100                 $inDigits = $workDigits;
3101         }
3102
3103         while( strlen( $outChars ) < $pad ) {
3104                 $outChars .= '0';
3105         }
3106
3107         return strrev( $outChars );
3108 }
3109
3110 /**
3111  * Create an object with a given name and an array of construct parameters
3112  *
3113  * @param $name String
3114  * @param $p Array: parameters
3115  * @return object
3116  * @deprecated since 1.18, warnings in 1.18, removal in 1.20
3117  */
3118 function wfCreateObject( $name, $p ) {
3119         wfDeprecated( __FUNCTION__ );
3120         return MWFunction::newObj( $name, $p );
3121 }
3122
3123 /**
3124  * @return bool
3125  */
3126 function wfHttpOnlySafe() {
3127         global $wgHttpOnlyBlacklist;
3128
3129         if( isset( $_SERVER['HTTP_USER_AGENT'] ) ) {
3130                 foreach( $wgHttpOnlyBlacklist as $regex ) {
3131                         if( preg_match( $regex, $_SERVER['HTTP_USER_AGENT'] ) ) {
3132                                 return false;
3133                         }
3134                 }
3135         }
3136
3137         return true;
3138 }
3139
3140 /**
3141  * Initialise php session
3142  *
3143  * @param $sessionId Bool
3144  */
3145 function wfSetupSession( $sessionId = false ) {
3146         global $wgSessionsInMemcached, $wgCookiePath, $wgCookieDomain,
3147                         $wgCookieSecure, $wgCookieHttpOnly, $wgSessionHandler;
3148         if( $wgSessionsInMemcached ) {
3149                 if ( !defined( 'MW_COMPILED' ) ) {
3150                         global $IP;
3151                         require_once( "$IP/includes/cache/MemcachedSessions.php" );
3152                 }
3153                 session_set_save_handler( 'memsess_open', 'memsess_close', 'memsess_read',
3154                         'memsess_write', 'memsess_destroy', 'memsess_gc' );
3155
3156                 // It's necessary to register a shutdown function to call session_write_close(),
3157                 // because by the time the request shutdown function for the session module is
3158                 // called, $wgMemc has already been destroyed. Shutdown functions registered
3159                 // this way are called before object destruction.
3160                 register_shutdown_function( 'memsess_write_close' );
3161         } elseif( $wgSessionHandler && $wgSessionHandler != ini_get( 'session.save_handler' ) ) {
3162                 # Only set this if $wgSessionHandler isn't null and session.save_handler
3163                 # hasn't already been set to the desired value (that causes errors)
3164                 ini_set( 'session.save_handler', $wgSessionHandler );
3165         }
3166         $httpOnlySafe = wfHttpOnlySafe() && $wgCookieHttpOnly;
3167         wfDebugLog( 'cookie',
3168                 'session_set_cookie_params: "' . implode( '", "',
3169                         array(
3170                                 0,
3171                                 $wgCookiePath,
3172                                 $wgCookieDomain,
3173                                 $wgCookieSecure,
3174                                 $httpOnlySafe ) ) . '"' );
3175         session_set_cookie_params( 0, $wgCookiePath, $wgCookieDomain, $wgCookieSecure, $httpOnlySafe );
3176         session_cache_limiter( 'private, must-revalidate' );
3177         if ( $sessionId ) {
3178                 session_id( $sessionId );
3179         }
3180         wfSuppressWarnings();
3181         session_start();
3182         wfRestoreWarnings();
3183 }
3184
3185 /**
3186  * Get an object from the precompiled serialized directory
3187  *
3188  * @param $name String
3189  * @return Mixed: the variable on success, false on failure
3190  */
3191 function wfGetPrecompiledData( $name ) {
3192         global $IP;
3193
3194         $file = "$IP/serialized/$name";
3195         if ( file_exists( $file ) ) {
3196                 $blob = file_get_contents( $file );
3197                 if ( $blob ) {
3198                         return unserialize( $blob );
3199                 }
3200         }
3201         return false;
3202 }
3203
3204 /**
3205  * Get a cache key
3206  *
3207  * @param varargs
3208  * @return String
3209  */
3210 function wfMemcKey( /*... */ ) {
3211         $args = func_get_args();
3212         $key = wfWikiID() . ':' . implode( ':', $args );
3213         $key = str_replace( ' ', '_', $key );
3214         return $key;
3215 }
3216
3217 /**
3218  * Get a cache key for a foreign DB
3219  *
3220  * @param $db String
3221  * @param $prefix String
3222  * @param varargs String
3223  * @return String
3224  */
3225 function wfForeignMemcKey( $db, $prefix /*, ... */ ) {
3226         $args = array_slice( func_get_args(), 2 );
3227         if ( $prefix ) {
3228                 $key = "$db-$prefix:" . implode( ':', $args );
3229         } else {
3230                 $key = $db . ':' . implode( ':', $args );
3231         }
3232         return $key;
3233 }
3234
3235 /**
3236  * Get an ASCII string identifying this wiki
3237  * This is used as a prefix in memcached keys
3238  *
3239  * @return String
3240  */
3241 function wfWikiID() {
3242         global $wgDBprefix, $wgDBname, $wgWikiID;
3243         if ( $wgWikiID !== false ) {
3244                 return $wgWikiID;
3245         } elseif ( $wgDBprefix ) {
3246                 return "$wgDBname-$wgDBprefix";
3247         } else {
3248                 return $wgDBname;
3249         }
3250 }
3251
3252 /**
3253  * Split a wiki ID into DB name and table prefix
3254  *
3255  * @param $wiki String
3256  *
3257  * @return array
3258  */
3259 function wfSplitWikiID( $wiki ) {
3260         $bits = explode( '-', $wiki, 2 );
3261         if ( count( $bits ) < 2 ) {
3262                 $bits[] = '';
3263         }
3264         return $bits;
3265 }
3266
3267 /**
3268  * Get a Database object.
3269  *
3270  * @param $db Integer: index of the connection to get. May be DB_MASTER for the
3271  *            master (for write queries), DB_SLAVE for potentially lagged read
3272  *            queries, or an integer >= 0 for a particular server.
3273  *
3274  * @param $groups Mixed: query groups. An array of group names that this query
3275  *                belongs to. May contain a single string if the query is only
3276  *                in one group.
3277  *
3278  * @param $wiki String: the wiki ID, or false for the current wiki
3279  *
3280  * Note: multiple calls to wfGetDB(DB_SLAVE) during the course of one request
3281  * will always return the same object, unless the underlying connection or load
3282  * balancer is manually destroyed.
3283  *
3284  * Note 2: use $this->getDB() in maintenance scripts that may be invoked by
3285  * updater to ensure that a proper database is being updated.
3286  *
3287  * @return DatabaseBase
3288  */
3289 function &wfGetDB( $db, $groups = array(), $wiki = false ) {
3290         return wfGetLB( $wiki )->getConnection( $db, $groups, $wiki );
3291 }
3292
3293 /**
3294  * Get a load balancer object.
3295  *
3296  * @param $wiki String: wiki ID, or false for the current wiki
3297  * @return LoadBalancer
3298  */
3299 function wfGetLB( $wiki = false ) {
3300         return wfGetLBFactory()->getMainLB( $wiki );
3301 }
3302
3303 /**
3304  * Get the load balancer factory object
3305  *
3306  * @return LBFactory
3307  */
3308 function &wfGetLBFactory() {
3309         return LBFactory::singleton();
3310 }
3311
3312 /**
3313  * Find a file.
3314  * Shortcut for RepoGroup::singleton()->findFile()
3315  *
3316  * @param $title String or Title object
3317  * @param $options Associative array of options:
3318  *     time:           requested time for an archived image, or false for the
3319  *                     current version. An image object will be returned which was
3320  *                     created at the specified time.
3321  *
3322  *     ignoreRedirect: If true, do not follow file redirects
3323  *
3324  *     private:        If true, return restricted (deleted) files if the current
3325  *                     user is allowed to view them. Otherwise, such files will not
3326  *                     be found.
3327  *
3328  *     bypassCache:    If true, do not use the process-local cache of File objects
3329  *
3330  * @return File, or false if the file does not exist
3331  */
3332 function wfFindFile( $title, $options = array() ) {
3333         return RepoGroup::singleton()->findFile( $title, $options );
3334 }
3335
3336 /**
3337  * Get an object referring to a locally registered file.
3338  * Returns a valid placeholder object if the file does not exist.
3339  *
3340  * @param $title Title or String
3341  * @return File|null A File, or null if passed an invalid Title
3342  */
3343 function wfLocalFile( $title ) {
3344         return RepoGroup::singleton()->getLocalRepo()->newFile( $title );
3345 }
3346
3347 /**
3348  * Stream a file to the browser. Back-compat alias for StreamFile::stream()
3349  * @deprecated since 1.19
3350  */
3351 function wfStreamFile( $fname, $headers = array() ) {
3352         StreamFile::stream( $fname, $headers );
3353 }
3354
3355 /**
3356  * Should low-performance queries be disabled?
3357  *
3358  * @return Boolean
3359  * @codeCoverageIgnore
3360  */
3361 function wfQueriesMustScale() {
3362         global $wgMiserMode;
3363         return $wgMiserMode
3364                 || ( SiteStats::pages() > 100000
3365                 && SiteStats::edits() > 1000000
3366                 && SiteStats::users() > 10000 );
3367 }
3368
3369 /**
3370  * Get the path to a specified script file, respecting file
3371  * extensions; this is a wrapper around $wgScriptExtension etc.
3372  *
3373  * @param $script String: script filename, sans extension
3374  * @return String
3375  */
3376 function wfScript( $script = 'index' ) {
3377         global $wgScriptPath, $wgScriptExtension;
3378         return "{$wgScriptPath}/{$script}{$wgScriptExtension}";
3379 }
3380
3381 /**
3382  * Get the script URL.
3383  *
3384  * @return script URL
3385  */
3386 function wfGetScriptUrl() {
3387         if( isset( $_SERVER['SCRIPT_NAME'] ) ) {
3388                 #
3389                 # as it was called, minus the query string.
3390                 #
3391                 # Some sites use Apache rewrite rules to handle subdomains,
3392                 # and have PHP set up in a weird way that causes PHP_SELF
3393                 # to contain the rewritten URL instead of the one that the
3394                 # outside world sees.
3395                 #
3396                 # If in this mode, use SCRIPT_URL instead, which mod_rewrite
3397                 # provides containing the "before" URL.
3398                 return $_SERVER['SCRIPT_NAME'];
3399         } else {
3400                 return $_SERVER['URL'];
3401         }
3402 }
3403
3404 /**
3405  * Convenience function converts boolean values into "true"
3406  * or "false" (string) values
3407  *
3408  * @param $value Boolean
3409  * @return String
3410  */
3411 function wfBoolToStr( $value ) {
3412         return $value ? 'true' : 'false';
3413 }
3414
3415 /**
3416  * Load an extension messages file
3417  *
3418  * @deprecated since 1.16, warnings in 1.18, remove in 1.20
3419  * @codeCoverageIgnore
3420  */
3421 function wfLoadExtensionMessages() {
3422         wfDeprecated( __FUNCTION__ );
3423 }
3424
3425 /**
3426  * Get a platform-independent path to the null file, e.g. /dev/null
3427  *
3428  * @return string
3429  */
3430 function wfGetNull() {
3431         return wfIsWindows()
3432                 ? 'NUL'
3433                 : '/dev/null';
3434 }
3435
3436 /**
3437  * Throws a warning that $function is deprecated
3438  *
3439  * @param $function String
3440  * @param $version String
3441  * @return null
3442  */
3443 function wfDeprecated( $function, $version = false ) {
3444         static $functionsWarned = array();
3445         if ( !isset( $functionsWarned[$function] ) ) {
3446                 $functionsWarned[$function] = true;
3447                 if ( $version ) {
3448                         global $wgDeprecationReleaseLimit;
3449                         if ( $wgDeprecationReleaseLimit ) {
3450                                 # Strip -* off the end of $version so that branches can use the
3451                                 # format #.##-branchname to avoid issues if the branch is merged into
3452                                 # a version of MediaWiki later than what it was branched from
3453                                 $comparableVersion = preg_replace( '/-.*$/', '', $version );
3454                                 # If the comparableVersion is larger than our release limit then
3455                                 # skip the warning message for the deprecation
3456                                 if ( version_compare( $wgDeprecationReleaseLimit, $comparableVersion, '<' ) ) {
3457                                         return;
3458                                 }
3459                         }
3460                         wfWarn( "Use of $function was deprecated in $version.", 2 );
3461                 } else {
3462                         wfWarn( "Use of $function is deprecated.", 2 );
3463                 }
3464         }
3465 }
3466
3467 /**
3468  * Send a warning either to the debug log or in a PHP error depending on
3469  * $wgDevelopmentWarnings
3470  *
3471  * @param $msg String: message to send
3472  * @param $callerOffset Integer: number of items to go back in the backtrace to
3473  *        find the correct caller (1 = function calling wfWarn, ...)
3474  * @param $level Integer: PHP error level; only used when $wgDevelopmentWarnings
3475  *        is true
3476  */
3477 function wfWarn( $msg, $callerOffset = 1, $level = E_USER_NOTICE ) {
3478         global $wgDevelopmentWarnings;
3479
3480         $callers = wfDebugBacktrace();
3481         if ( isset( $callers[$callerOffset + 1] ) ) {
3482                 $callerfunc = $callers[$callerOffset + 1];
3483                 $callerfile = $callers[$callerOffset];
3484                 if ( isset( $callerfile['file'] ) && isset( $callerfile['line'] ) ) {
3485                         $file = $callerfile['file'] . ' at line ' . $callerfile['line'];
3486                 } else {
3487                         $file = '(internal function)';
3488                 }
3489                 $func = '';
3490                 if ( isset( $callerfunc['class'] ) ) {
3491                         $func .= $callerfunc['class'] . '::';
3492                 }
3493                 if ( isset( $callerfunc['function'] ) ) {
3494                         $func .= $callerfunc['function'];
3495                 }
3496                 $msg .= " [Called from $func in $file]";
3497         }
3498
3499         if ( $wgDevelopmentWarnings ) {
3500                 trigger_error( $msg, $level );
3501         } else {
3502                 wfDebug( "$msg\n" );
3503         }
3504 }
3505
3506 /**
3507  * Modern version of wfWaitForSlaves(). Instead of looking at replication lag
3508  * and waiting for it to go down, this waits for the slaves to catch up to the
3509  * master position. Use this when updating very large numbers of rows, as
3510  * in maintenance scripts, to avoid causing too much lag.  Of course, this is
3511  * a no-op if there are no slaves.
3512  *
3513  * @param $maxLag Integer (deprecated)
3514  * @param $wiki mixed Wiki identifier accepted by wfGetLB
3515  */
3516 function wfWaitForSlaves( $maxLag = false, $wiki = false ) {
3517         $lb = wfGetLB( $wiki );
3518         // bug 27975 - Don't try to wait for slaves if there are none
3519         // Prevents permission error when getting master position
3520         if ( $lb->getServerCount() > 1 ) {
3521                 $dbw = $lb->getConnection( DB_MASTER );
3522                 $pos = $dbw->getMasterPos();
3523                 $lb->waitForAll( $pos );
3524         }
3525 }
3526
3527 /**
3528  * Used to be used for outputting text in the installer/updater
3529  * @deprecated since 1.18, warnings in 1.18, remove in 1.20
3530  */
3531 function wfOut( $s ) {
3532         wfDeprecated( __METHOD__ );
3533         global $wgCommandLineMode;
3534         if ( $wgCommandLineMode ) {
3535                 echo $s;
3536         } else {
3537                 echo htmlspecialchars( $s );
3538         }
3539         flush();
3540 }
3541
3542 /**
3543  * Count down from $n to zero on the terminal, with a one-second pause
3544  * between showing each number. For use in command-line scripts.
3545  * @codeCoverageIgnore
3546  * @param $n int
3547  */
3548 function wfCountDown( $n ) {
3549         for ( $i = $n; $i >= 0; $i-- ) {
3550                 if ( $i != $n ) {
3551                         echo str_repeat( "\x08", strlen( $i + 1 ) );
3552                 }
3553                 echo $i;
3554                 flush();
3555                 if ( $i ) {
3556                         sleep( 1 );
3557                 }
3558         }
3559         echo "\n";
3560 }
3561
3562 /**
3563  * Generate a random 32-character hexadecimal token.
3564  * @param $salt Mixed: some sort of salt, if necessary, to add to random
3565  *              characters before hashing.
3566  * @return string
3567  * @codeCoverageIgnore
3568  */
3569 function wfGenerateToken( $salt = '' ) {
3570         $salt = serialize( $salt );
3571         return md5( mt_rand( 0, 0x7fffffff ) . $salt );
3572 }
3573
3574 /**
3575  * Replace all invalid characters with -
3576  *
3577  * @param $name Mixed: filename to process
3578  * @return String
3579  */
3580 function wfStripIllegalFilenameChars( $name ) {
3581         global $wgIllegalFileChars;
3582         $name = wfBaseName( $name );
3583         $name = preg_replace(
3584                 "/[^" . Title::legalChars() . "]" .
3585                         ( $wgIllegalFileChars ? "|[" . $wgIllegalFileChars . "]" : '' ) .
3586                         "/",
3587                 '-',
3588                 $name
3589         );
3590         return $name;
3591 }
3592
3593 /**
3594  * Set PHP's memory limit to the larger of php.ini or $wgMemoryLimit;
3595  *
3596  * @return Integer value memory was set to.
3597  */
3598 function wfMemoryLimit() {
3599         global $wgMemoryLimit;
3600         $memlimit = wfShorthandToInteger( ini_get( 'memory_limit' ) );
3601         if( $memlimit != -1 ) {
3602                 $conflimit = wfShorthandToInteger( $wgMemoryLimit );
3603                 if( $conflimit == -1 ) {
3604                         wfDebug( "Removing PHP's memory limit\n" );
3605                         wfSuppressWarnings();
3606                         ini_set( 'memory_limit', $conflimit );
3607                         wfRestoreWarnings();
3608                         return $conflimit;
3609                 } elseif ( $conflimit > $memlimit ) {
3610                         wfDebug( "Raising PHP's memory limit to $conflimit bytes\n" );
3611                         wfSuppressWarnings();
3612                         ini_set( 'memory_limit', $conflimit );
3613                         wfRestoreWarnings();
3614                         return $conflimit;
3615                 }
3616         }
3617         return $memlimit;
3618 }
3619
3620 /**
3621  * Converts shorthand byte notation to integer form
3622  *
3623  * @param $string String
3624  * @return Integer
3625  */
3626 function wfShorthandToInteger( $string = '' ) {
3627         $string = trim( $string );
3628         if( $string === '' ) {
3629                 return -1;
3630         }
3631         $last = $string[strlen( $string ) - 1];
3632         $val = intval( $string );
3633         switch( $last ) {
3634                 case 'g':
3635                 case 'G':
3636                         $val *= 1024;
3637                         // break intentionally missing
3638                 case 'm':
3639                 case 'M':
3640                         $val *= 1024;
3641                         // break intentionally missing
3642                 case 'k':
3643                 case 'K':
3644                         $val *= 1024;
3645         }
3646
3647         return $val;
3648 }
3649
3650 /**
3651  * Get the normalised IETF language tag
3652  * See unit test for examples.
3653  *
3654  * @param $code String: The language code.
3655  * @return String: The language code which complying with BCP 47 standards.
3656  */
3657 function wfBCP47( $code ) {
3658         $codeSegment = explode( '-', $code );
3659         $codeBCP = array();
3660         foreach ( $codeSegment as $segNo => $seg ) {
3661                 if ( count( $codeSegment ) > 0 ) {
3662                         // when previous segment is x, it is a private segment and should be lc
3663                         if( $segNo > 0 && strtolower( $codeSegment[( $segNo - 1 )] ) == 'x' ) {
3664                                 $codeBCP[$segNo] = strtolower( $seg );
3665                         // ISO 3166 country code
3666                         } elseif ( ( strlen( $seg ) == 2 ) && ( $segNo > 0 ) ) {
3667                                 $codeBCP[$segNo] = strtoupper( $seg );
3668                         // ISO 15924 script code
3669                         } elseif ( ( strlen( $seg ) == 4 ) && ( $segNo > 0 ) ) {
3670                                 $codeBCP[$segNo] = ucfirst( strtolower( $seg ) );
3671                         // Use lowercase for other cases
3672                         } else {
3673                                 $codeBCP[$segNo] = strtolower( $seg );
3674                         }
3675                 } else {
3676                 // Use lowercase for single segment
3677                         $codeBCP[$segNo] = strtolower( $seg );
3678                 }
3679         }
3680         $langCode = implode( '-', $codeBCP );
3681         return $langCode;
3682 }
3683
3684 /**
3685  * Get a cache object.
3686  *
3687  * @param $inputType integer Cache type, one the the CACHE_* constants.
3688  * @return BagOStuff
3689  */
3690 function wfGetCache( $inputType ) {
3691         return ObjectCache::getInstance( $inputType );
3692 }
3693
3694 /**
3695  * Get the main cache object
3696  *
3697  * @return BagOStuff
3698  */
3699 function wfGetMainCache() {
3700         global $wgMainCacheType;
3701         return ObjectCache::getInstance( $wgMainCacheType );
3702 }
3703
3704 /**
3705  * Get the cache object used by the message cache
3706  *
3707  * @return BagOStuff
3708  */
3709 function wfGetMessageCacheStorage() {
3710         global $wgMessageCacheType;
3711         return ObjectCache::getInstance( $wgMessageCacheType );
3712 }
3713
3714 /**
3715  * Get the cache object used by the parser cache
3716  *
3717  * @return BagOStuff
3718  */
3719 function wfGetParserCacheStorage() {
3720         global $wgParserCacheType;
3721         return ObjectCache::getInstance( $wgParserCacheType );
3722 }
3723
3724 /**
3725  * Call hook functions defined in $wgHooks
3726  *
3727  * @param $event String: event name
3728  * @param $args Array: parameters passed to hook functions
3729  * @return Boolean
3730  */
3731 function wfRunHooks( $event, $args = array() ) {
3732         return Hooks::run( $event, $args );
3733 }
3734
3735 /**
3736  * Wrapper around php's unpack.
3737  *
3738  * @param $format String: The format string (See php's docs)
3739  * @param $data: A binary string of binary data
3740  * @param $length integer or false: The minimun length of $data. This is to
3741  *      prevent reading beyond the end of $data. false to disable the check.
3742  *
3743  * Also be careful when using this function to read unsigned 32 bit integer
3744  * because php might make it negative.
3745  *
3746  * @throws MWException if $data not long enough, or if unpack fails
3747  * @return Associative array of the extracted data
3748  */
3749 function wfUnpack( $format, $data, $length=false ) {
3750         if ( $length !== false ) {
3751                 $realLen = strlen( $data );
3752                 if ( $realLen < $length ) {
3753                         throw new MWException( "Tried to use wfUnpack on a "
3754                                 . "string of length $realLen, but needed one "
3755                                 . "of at least length $length."
3756                         );
3757                 }
3758         }
3759
3760         wfSuppressWarnings();
3761         $result = unpack( $format, $data );
3762         wfRestoreWarnings();
3763
3764         if ( $result === false ) {
3765                 // If it cannot extract the packed data.
3766                 throw new MWException( "unpack could not unpack binary data" );
3767         }
3768         return $result;
3769 }