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