= 5.4.0 // http://php.net/gzdecode if ( !function_exists( 'gzdecode' ) ) { /** * @codeCoverageIgnore * @param string $data * @return string */ function gzdecode( $data ) { return gzinflate( substr( $data, 10, -8 ) ); } } // hash_equals function only exists in PHP >= 5.6.0 // http://php.net/hash_equals if ( !function_exists( 'hash_equals' ) ) { /** * Check whether a user-provided string is equal to a fixed-length secret string * without revealing bytes of the secret string through timing differences. * * The usual way to compare strings (PHP's === operator or the underlying memcmp() * function in C) is to compare corresponding bytes and stop at the first difference, * which would take longer for a partial match than for a complete mismatch. This * is not secure when one of the strings (e.g. an HMAC or token) must remain secret * and the other may come from an attacker. Statistical analysis of timing measurements * over many requests may allow the attacker to guess the string's bytes one at a time * (and check his guesses) even if the timing differences are extremely small. * * When making such a security-sensitive comparison, it is essential that the sequence * in which instructions are executed and memory locations are accessed not depend on * the secret string's value. HOWEVER, for simplicity, we do not attempt to minimize * the inevitable leakage of the string's length. That is generally known anyway as * a chararacteristic of the hash function used to compute the secret value. * * Longer explanation: http://www.emerose.com/timing-attacks-explained * * @codeCoverageIgnore * @param string $known_string Fixed-length secret string to compare against * @param string $user_string User-provided string * @return bool True if the strings are the same, false otherwise */ function hash_equals( $known_string, $user_string ) { // Strict type checking as in PHP's native implementation if ( !is_string( $known_string ) ) { trigger_error( 'hash_equals(): Expected known_string to be a string, ' . gettype( $known_string ) . ' given', E_USER_WARNING ); return false; } if ( !is_string( $user_string ) ) { trigger_error( 'hash_equals(): Expected user_string to be a string, ' . gettype( $user_string ) . ' given', E_USER_WARNING ); return false; } $known_string_len = strlen( $known_string ); if ( $known_string_len !== strlen( $user_string ) ) { return false; } $result = 0; for ( $i = 0; $i < $known_string_len; $i++ ) { $result |= ord( $known_string[$i] ) ^ ord( $user_string[$i] ); } return ( $result === 0 ); } } /// @endcond /** * Load an extension * * This queues an extension to be loaded through * the ExtensionRegistry system. * * @param string $ext Name of the extension to load * @param string|null $path Absolute path of where to find the extension.json file * @since 1.25 */ function wfLoadExtension( $ext, $path = null ) { if ( !$path ) { global $wgExtensionDirectory; $path = "$wgExtensionDirectory/$ext/extension.json"; } ExtensionRegistry::getInstance()->queue( $path ); } /** * Load multiple extensions at once * * Same as wfLoadExtension, but more efficient if you * are loading multiple extensions. * * If you want to specify custom paths, you should interact with * ExtensionRegistry directly. * * @see wfLoadExtension * @param string[] $exts Array of extension names to load * @since 1.25 */ function wfLoadExtensions( array $exts ) { global $wgExtensionDirectory; $registry = ExtensionRegistry::getInstance(); foreach ( $exts as $ext ) { $registry->queue( "$wgExtensionDirectory/$ext/extension.json" ); } } /** * Load a skin * * @see wfLoadExtension * @param string $skin Name of the extension to load * @param string|null $path Absolute path of where to find the skin.json file * @since 1.25 */ function wfLoadSkin( $skin, $path = null ) { if ( !$path ) { global $wgStyleDirectory; $path = "$wgStyleDirectory/$skin/skin.json"; } ExtensionRegistry::getInstance()->queue( $path ); } /** * Load multiple skins at once * * @see wfLoadExtensions * @param string[] $skins Array of extension names to load * @since 1.25 */ function wfLoadSkins( array $skins ) { global $wgStyleDirectory; $registry = ExtensionRegistry::getInstance(); foreach ( $skins as $skin ) { $registry->queue( "$wgStyleDirectory/$skin/skin.json" ); } } /** * Like array_diff( $a, $b ) except that it works with two-dimensional arrays. * @param array $a * @param array $b * @return array */ function wfArrayDiff2( $a, $b ) { return array_udiff( $a, $b, 'wfArrayDiff2_cmp' ); } /** * @param array|string $a * @param array|string $b * @return int */ function wfArrayDiff2_cmp( $a, $b ) { if ( is_string( $a ) && is_string( $b ) ) { return strcmp( $a, $b ); } elseif ( count( $a ) !== count( $b ) ) { return count( $a ) < count( $b ) ? -1 : 1; } else { reset( $a ); reset( $b ); while ( ( list( , $valueA ) = each( $a ) ) && ( list( , $valueB ) = each( $b ) ) ) { $cmp = strcmp( $valueA, $valueB ); if ( $cmp !== 0 ) { return $cmp; } } return 0; } } /** * Appends to second array if $value differs from that in $default * * @param string|int $key * @param mixed $value * @param mixed $default * @param array $changed Array to alter * @throws MWException */ function wfAppendToArrayIfNotDefault( $key, $value, $default, &$changed ) { if ( is_null( $changed ) ) { throw new MWException( 'GlobalFunctions::wfAppendToArrayIfNotDefault got null' ); } if ( $default[$key] !== $value ) { $changed[$key] = $value; } } /** * Merge arrays in the style of getUserPermissionsErrors, with duplicate removal * e.g. * wfMergeErrorArrays( * array( array( 'x' ) ), * array( array( 'x', '2' ) ), * array( array( 'x' ) ), * array( array( 'y' ) ) * ); * returns: * array( * array( 'x', '2' ), * array( 'x' ), * array( 'y' ) * ) * * @param array $array1,... * @return array */ function wfMergeErrorArrays( /*...*/ ) { $args = func_get_args(); $out = array(); foreach ( $args as $errors ) { foreach ( $errors as $params ) { $originalParams = $params; if ( $params[0] instanceof MessageSpecifier ) { $msg = $params[0]; $params = array_merge( array( $msg->getKey() ), $msg->getParams() ); } # @todo FIXME: Sometimes get nested arrays for $params, # which leads to E_NOTICEs $spec = implode( "\t", $params ); $out[$spec] = $originalParams; } } return array_values( $out ); } /** * Insert array into another array after the specified *KEY* * * @param array $array The array. * @param array $insert The array to insert. * @param mixed $after The key to insert after * @return array */ function wfArrayInsertAfter( array $array, array $insert, $after ) { // Find the offset of the element to insert after. $keys = array_keys( $array ); $offsetByKey = array_flip( $keys ); $offset = $offsetByKey[$after]; // Insert at the specified offset $before = array_slice( $array, 0, $offset + 1, true ); $after = array_slice( $array, $offset + 1, count( $array ) - $offset, true ); $output = $before + $insert + $after; return $output; } /** * Recursively converts the parameter (an object) to an array with the same data * * @param object|array $objOrArray * @param bool $recursive * @return array */ function wfObjectToArray( $objOrArray, $recursive = true ) { $array = array(); if ( is_object( $objOrArray ) ) { $objOrArray = get_object_vars( $objOrArray ); } foreach ( $objOrArray as $key => $value ) { if ( $recursive && ( is_object( $value ) || is_array( $value ) ) ) { $value = wfObjectToArray( $value ); } $array[$key] = $value; } return $array; } /** * Get a random decimal value between 0 and 1, in a way * not likely to give duplicate values for any realistic * number of articles. * * @return string */ function wfRandom() { # The maximum random value is "only" 2^31-1, so get two random # values to reduce the chance of dupes $max = mt_getrandmax() + 1; $rand = number_format( ( mt_rand() * $max + mt_rand() ) / $max / $max, 12, '.', '' ); return $rand; } /** * Get a random string containing a number of pseudo-random hex * characters. * @note This is not secure, if you are trying to generate some sort * of token please use MWCryptRand instead. * * @param int $length The length of the string to generate * @return string * @since 1.20 */ function wfRandomString( $length = 32 ) { $str = ''; for ( $n = 0; $n < $length; $n += 7 ) { $str .= sprintf( '%07x', mt_rand() & 0xfffffff ); } return substr( $str, 0, $length ); } /** * We want some things to be included as literal characters in our title URLs * for prettiness, which urlencode encodes by default. According to RFC 1738, * all of the following should be safe: * * ;:@&=$-_.+!*'(), * * RFC 1738 says ~ is unsafe, however RFC 3986 considers it an unreserved * character which should not be encoded. More importantly, google chrome * always converts %7E back to ~, and converting it in this function can * cause a redirect loop (T105265). * * But + is not safe because it's used to indicate a space; &= are only safe in * paths and not in queries (and we don't distinguish here); ' seems kind of * scary; and urlencode() doesn't touch -_. to begin with. Plus, although / * is reserved, we don't care. So the list we unescape is: * * ;:@$!*(),/~ * * However, IIS7 redirects fail when the url contains a colon (Bug 22709), * so no fancy : for IIS7. * * %2F in the page titles seems to fatally break for some reason. * * @param string $s * @return string */ function wfUrlencode( $s ) { static $needle; if ( is_null( $s ) ) { $needle = null; return ''; } if ( is_null( $needle ) ) { $needle = array( '%3B', '%40', '%24', '%21', '%2A', '%28', '%29', '%2C', '%2F', '%7E' ); if ( !isset( $_SERVER['SERVER_SOFTWARE'] ) || ( strpos( $_SERVER['SERVER_SOFTWARE'], 'Microsoft-IIS/7' ) === false ) ) { $needle[] = '%3A'; } } $s = urlencode( $s ); $s = str_ireplace( $needle, array( ';', '@', '$', '!', '*', '(', ')', ',', '/', '~', ':' ), $s ); return $s; } /** * This function takes one or two arrays as input, and returns a CGI-style string, e.g. * "days=7&limit=100". Options in the first array override options in the second. * Options set to null or false will not be output. * * @param array $array1 ( String|Array ) * @param array|null $array2 ( String|Array ) * @param string $prefix * @return string */ function wfArrayToCgi( $array1, $array2 = null, $prefix = '' ) { if ( !is_null( $array2 ) ) { $array1 = $array1 + $array2; } $cgi = ''; foreach ( $array1 as $key => $value ) { if ( !is_null( $value ) && $value !== false ) { if ( $cgi != '' ) { $cgi .= '&'; } if ( $prefix !== '' ) { $key = $prefix . "[$key]"; } if ( is_array( $value ) ) { $firstTime = true; foreach ( $value as $k => $v ) { $cgi .= $firstTime ? '' : '&'; if ( is_array( $v ) ) { $cgi .= wfArrayToCgi( $v, null, $key . "[$k]" ); } else { $cgi .= urlencode( $key . "[$k]" ) . '=' . urlencode( $v ); } $firstTime = false; } } else { if ( is_object( $value ) ) { $value = $value->__toString(); } $cgi .= urlencode( $key ) . '=' . urlencode( $value ); } } } return $cgi; } /** * This is the logical opposite of wfArrayToCgi(): it accepts a query string as * its argument and returns the same string in array form. This allows compatibility * with legacy functions that accept raw query strings instead of nice * arrays. Of course, keys and values are urldecode()d. * * @param string $query Query string * @return string[] Array version of input */ function wfCgiToArray( $query ) { if ( isset( $query[0] ) && $query[0] == '?' ) { $query = substr( $query, 1 ); } $bits = explode( '&', $query ); $ret = array(); foreach ( $bits as $bit ) { if ( $bit === '' ) { continue; } if ( strpos( $bit, '=' ) === false ) { // Pieces like &qwerty become 'qwerty' => '' (at least this is what php does) $key = $bit; $value = ''; } else { list( $key, $value ) = explode( '=', $bit ); } $key = urldecode( $key ); $value = urldecode( $value ); if ( strpos( $key, '[' ) !== false ) { $keys = array_reverse( explode( '[', $key ) ); $key = array_pop( $keys ); $temp = $value; foreach ( $keys as $k ) { $k = substr( $k, 0, -1 ); $temp = array( $k => $temp ); } if ( isset( $ret[$key] ) ) { $ret[$key] = array_merge( $ret[$key], $temp ); } else { $ret[$key] = $temp; } } else { $ret[$key] = $value; } } return $ret; } /** * Append a query string to an existing URL, which may or may not already * have query string parameters already. If so, they will be combined. * * @param string $url * @param string|string[] $query String or associative array * @return string */ function wfAppendQuery( $url, $query ) { if ( is_array( $query ) ) { $query = wfArrayToCgi( $query ); } if ( $query != '' ) { if ( false === strpos( $url, '?' ) ) { $url .= '?'; } else { $url .= '&'; } $url .= $query; } return $url; } /** * Expand a potentially local URL to a fully-qualified URL. Assumes $wgServer * is correct. * * The meaning of the PROTO_* constants is as follows: * PROTO_HTTP: Output a URL starting with http:// * PROTO_HTTPS: Output a URL starting with https:// * PROTO_RELATIVE: Output a URL starting with // (protocol-relative URL) * PROTO_CURRENT: Output a URL starting with either http:// or https:// , depending * on which protocol was used for the current incoming request * PROTO_CANONICAL: For URLs without a domain, like /w/index.php , use $wgCanonicalServer. * For protocol-relative URLs, use the protocol of $wgCanonicalServer * PROTO_INTERNAL: Like PROTO_CANONICAL, but uses $wgInternalServer instead of $wgCanonicalServer * * @todo this won't work with current-path-relative URLs * like "subdir/foo.html", etc. * * @param string $url Either fully-qualified or a local path + query * @param string $defaultProto One of the PROTO_* constants. Determines the * protocol to use if $url or $wgServer is protocol-relative * @return string Fully-qualified URL, current-path-relative URL or false if * no valid URL can be constructed */ function wfExpandUrl( $url, $defaultProto = PROTO_CURRENT ) { global $wgServer, $wgCanonicalServer, $wgInternalServer, $wgRequest, $wgHttpsPort; if ( $defaultProto === PROTO_CANONICAL ) { $serverUrl = $wgCanonicalServer; } elseif ( $defaultProto === PROTO_INTERNAL && $wgInternalServer !== false ) { // Make $wgInternalServer fall back to $wgServer if not set $serverUrl = $wgInternalServer; } else { $serverUrl = $wgServer; if ( $defaultProto === PROTO_CURRENT ) { $defaultProto = $wgRequest->getProtocol() . '://'; } } // Analyze $serverUrl to obtain its protocol $bits = wfParseUrl( $serverUrl ); $serverHasProto = $bits && $bits['scheme'] != ''; if ( $defaultProto === PROTO_CANONICAL || $defaultProto === PROTO_INTERNAL ) { if ( $serverHasProto ) { $defaultProto = $bits['scheme'] . '://'; } else { // $wgCanonicalServer or $wgInternalServer doesn't have a protocol. // This really isn't supposed to happen. Fall back to HTTP in this // ridiculous case. $defaultProto = PROTO_HTTP; } } $defaultProtoWithoutSlashes = substr( $defaultProto, 0, -2 ); if ( substr( $url, 0, 2 ) == '//' ) { $url = $defaultProtoWithoutSlashes . $url; } elseif ( substr( $url, 0, 1 ) == '/' ) { // If $serverUrl is protocol-relative, prepend $defaultProtoWithoutSlashes, // otherwise leave it alone. $url = ( $serverHasProto ? '' : $defaultProtoWithoutSlashes ) . $serverUrl . $url; } $bits = wfParseUrl( $url ); // ensure proper port for HTTPS arrives in URL // https://phabricator.wikimedia.org/T67184 if ( $defaultProto === PROTO_HTTPS && $wgHttpsPort != 443 ) { $bits['port'] = $wgHttpsPort; } if ( $bits && isset( $bits['path'] ) ) { $bits['path'] = wfRemoveDotSegments( $bits['path'] ); return wfAssembleUrl( $bits ); } elseif ( $bits ) { # No path to expand return $url; } elseif ( substr( $url, 0, 1 ) != '/' ) { # URL is a relative path return wfRemoveDotSegments( $url ); } # Expanded URL is not valid. return false; } /** * This function will reassemble a URL parsed with wfParseURL. This is useful * if you need to edit part of a URL and put it back together. * * This is the basic structure used (brackets contain keys for $urlParts): * [scheme][delimiter][user]:[pass]@[host]:[port][path]?[query]#[fragment] * * @todo Need to integrate this into wfExpandUrl (bug 32168) * * @since 1.19 * @param array $urlParts URL parts, as output from wfParseUrl * @return string URL assembled from its component parts */ function wfAssembleUrl( $urlParts ) { $result = ''; if ( isset( $urlParts['delimiter'] ) ) { if ( isset( $urlParts['scheme'] ) ) { $result .= $urlParts['scheme']; } $result .= $urlParts['delimiter']; } if ( isset( $urlParts['host'] ) ) { if ( isset( $urlParts['user'] ) ) { $result .= $urlParts['user']; if ( isset( $urlParts['pass'] ) ) { $result .= ':' . $urlParts['pass']; } $result .= '@'; } $result .= $urlParts['host']; if ( isset( $urlParts['port'] ) ) { $result .= ':' . $urlParts['port']; } } if ( isset( $urlParts['path'] ) ) { $result .= $urlParts['path']; } if ( isset( $urlParts['query'] ) ) { $result .= '?' . $urlParts['query']; } if ( isset( $urlParts['fragment'] ) ) { $result .= '#' . $urlParts['fragment']; } return $result; } /** * Remove all dot-segments in the provided URL path. For example, * '/a/./b/../c/' becomes '/a/c/'. For details on the algorithm, please see * RFC3986 section 5.2.4. * * @todo Need to integrate this into wfExpandUrl (bug 32168) * * @param string $urlPath URL path, potentially containing dot-segments * @return string URL path with all dot-segments removed */ function wfRemoveDotSegments( $urlPath ) { $output = ''; $inputOffset = 0; $inputLength = strlen( $urlPath ); while ( $inputOffset < $inputLength ) { $prefixLengthOne = substr( $urlPath, $inputOffset, 1 ); $prefixLengthTwo = substr( $urlPath, $inputOffset, 2 ); $prefixLengthThree = substr( $urlPath, $inputOffset, 3 ); $prefixLengthFour = substr( $urlPath, $inputOffset, 4 ); $trimOutput = false; if ( $prefixLengthTwo == './' ) { # Step A, remove leading "./" $inputOffset += 2; } elseif ( $prefixLengthThree == '../' ) { # Step A, remove leading "../" $inputOffset += 3; } elseif ( ( $prefixLengthTwo == '/.' ) && ( $inputOffset + 2 == $inputLength ) ) { # Step B, replace leading "/.$" with "/" $inputOffset += 1; $urlPath[$inputOffset] = '/'; } elseif ( $prefixLengthThree == '/./' ) { # Step B, replace leading "/./" with "/" $inputOffset += 2; } elseif ( $prefixLengthThree == '/..' && ( $inputOffset + 3 == $inputLength ) ) { # Step C, replace leading "/..$" with "/" and # remove last path component in output $inputOffset += 2; $urlPath[$inputOffset] = '/'; $trimOutput = true; } elseif ( $prefixLengthFour == '/../' ) { # Step C, replace leading "/../" with "/" and # remove last path component in output $inputOffset += 3; $trimOutput = true; } elseif ( ( $prefixLengthOne == '.' ) && ( $inputOffset + 1 == $inputLength ) ) { # Step D, remove "^.$" $inputOffset += 1; } elseif ( ( $prefixLengthTwo == '..' ) && ( $inputOffset + 2 == $inputLength ) ) { # Step D, remove "^..$" $inputOffset += 2; } else { # Step E, move leading path segment to output if ( $prefixLengthOne == '/' ) { $slashPos = strpos( $urlPath, '/', $inputOffset + 1 ); } else { $slashPos = strpos( $urlPath, '/', $inputOffset ); } if ( $slashPos === false ) { $output .= substr( $urlPath, $inputOffset ); $inputOffset = $inputLength; } else { $output .= substr( $urlPath, $inputOffset, $slashPos - $inputOffset ); $inputOffset += $slashPos - $inputOffset; } } if ( $trimOutput ) { $slashPos = strrpos( $output, '/' ); if ( $slashPos === false ) { $output = ''; } else { $output = substr( $output, 0, $slashPos ); } } } return $output; } /** * Returns a regular expression of url protocols * * @param bool $includeProtocolRelative If false, remove '//' from the returned protocol list. * DO NOT USE this directly, use wfUrlProtocolsWithoutProtRel() instead * @return string */ function wfUrlProtocols( $includeProtocolRelative = true ) { global $wgUrlProtocols; // Cache return values separately based on $includeProtocolRelative static $withProtRel = null, $withoutProtRel = null; $cachedValue = $includeProtocolRelative ? $withProtRel : $withoutProtRel; if ( !is_null( $cachedValue ) ) { return $cachedValue; } // Support old-style $wgUrlProtocols strings, for backwards compatibility // with LocalSettings files from 1.5 if ( is_array( $wgUrlProtocols ) ) { $protocols = array(); foreach ( $wgUrlProtocols as $protocol ) { // Filter out '//' if !$includeProtocolRelative if ( $includeProtocolRelative || $protocol !== '//' ) { $protocols[] = preg_quote( $protocol, '/' ); } } $retval = implode( '|', $protocols ); } else { // Ignore $includeProtocolRelative in this case // This case exists for pre-1.6 compatibility, and we can safely assume // that '//' won't appear in a pre-1.6 config because protocol-relative // URLs weren't supported until 1.18 $retval = $wgUrlProtocols; } // Cache return value if ( $includeProtocolRelative ) { $withProtRel = $retval; } else { $withoutProtRel = $retval; } return $retval; } /** * Like wfUrlProtocols(), but excludes '//' from the protocol list. Use this if * you need a regex that matches all URL protocols but does not match protocol- * relative URLs * @return string */ function wfUrlProtocolsWithoutProtRel() { return wfUrlProtocols( false ); } /** * parse_url() work-alike, but non-broken. Differences: * * 1) Does not raise warnings on bad URLs (just returns false). * 2) Handles protocols that don't use :// (e.g., mailto: and news:, as well as * protocol-relative URLs) correctly. * 3) Adds a "delimiter" element to the array, either '://', ':' or '//' (see (2)). * * @param string $url A URL to parse * @return string[] Bits of the URL in an associative array, per PHP docs */ function wfParseUrl( $url ) { global $wgUrlProtocols; // Allow all protocols defined in DefaultSettings/LocalSettings.php // Protocol-relative URLs are handled really badly by parse_url(). It's so // bad that the easiest way to handle them is to just prepend 'http:' and // strip the protocol out later. $wasRelative = substr( $url, 0, 2 ) == '//'; if ( $wasRelative ) { $url = "http:$url"; } MediaWiki\suppressWarnings(); $bits = parse_url( $url ); MediaWiki\restoreWarnings(); // parse_url() returns an array without scheme for some invalid URLs, e.g. // parse_url("%0Ahttp://example.com") == array( 'host' => '%0Ahttp', 'path' => 'example.com' ) if ( !$bits || !isset( $bits['scheme'] ) ) { return false; } // parse_url() incorrectly handles schemes case-sensitively. Convert it to lowercase. $bits['scheme'] = strtolower( $bits['scheme'] ); // most of the protocols are followed by ://, but mailto: and sometimes news: not, check for it if ( in_array( $bits['scheme'] . '://', $wgUrlProtocols ) ) { $bits['delimiter'] = '://'; } elseif ( in_array( $bits['scheme'] . ':', $wgUrlProtocols ) ) { $bits['delimiter'] = ':'; // parse_url detects for news: and mailto: the host part of an url as path // We have to correct this wrong detection if ( isset( $bits['path'] ) ) { $bits['host'] = $bits['path']; $bits['path'] = ''; } } else { return false; } /* Provide an empty host for eg. file:/// urls (see bug 28627) */ if ( !isset( $bits['host'] ) ) { $bits['host'] = ''; // bug 45069 if ( isset( $bits['path'] ) ) { /* parse_url loses the third / for file:///c:/ urls (but not on variants) */ if ( substr( $bits['path'], 0, 1 ) !== '/' ) { $bits['path'] = '/' . $bits['path']; } } else { $bits['path'] = ''; } } // If the URL was protocol-relative, fix scheme and delimiter if ( $wasRelative ) { $bits['scheme'] = ''; $bits['delimiter'] = '//'; } return $bits; } /** * Take a URL, make sure it's expanded to fully qualified, and replace any * encoded non-ASCII Unicode characters with their UTF-8 original forms * for more compact display and legibility for local audiences. * * @todo handle punycode domains too * * @param string $url * @return string */ function wfExpandIRI( $url ) { return preg_replace_callback( '/((?:%[89A-F][0-9A-F])+)/i', 'wfExpandIRI_callback', wfExpandUrl( $url ) ); } /** * Private callback for wfExpandIRI * @param array $matches * @return string */ function wfExpandIRI_callback( $matches ) { return urldecode( $matches[1] ); } /** * Make URL indexes, appropriate for the el_index field of externallinks. * * @param string $url * @return array */ function wfMakeUrlIndexes( $url ) { $bits = wfParseUrl( $url ); // Reverse the labels in the hostname, convert to lower case // For emails reverse domainpart only if ( $bits['scheme'] == 'mailto' ) { $mailparts = explode( '@', $bits['host'], 2 ); if ( count( $mailparts ) === 2 ) { $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) ); } else { // No domain specified, don't mangle it $domainpart = ''; } $reversedHost = $domainpart . '@' . $mailparts[0]; } else { $reversedHost = strtolower( implode( '.', array_reverse( explode( '.', $bits['host'] ) ) ) ); } // Add an extra dot to the end // Why? Is it in wrong place in mailto links? if ( substr( $reversedHost, -1, 1 ) !== '.' ) { $reversedHost .= '.'; } // Reconstruct the pseudo-URL $prot = $bits['scheme']; $index = $prot . $bits['delimiter'] . $reversedHost; // Leave out user and password. Add the port, path, query and fragment if ( isset( $bits['port'] ) ) { $index .= ':' . $bits['port']; } if ( isset( $bits['path'] ) ) { $index .= $bits['path']; } else { $index .= '/'; } if ( isset( $bits['query'] ) ) { $index .= '?' . $bits['query']; } if ( isset( $bits['fragment'] ) ) { $index .= '#' . $bits['fragment']; } if ( $prot == '' ) { return array( "http:$index", "https:$index" ); } else { return array( $index ); } } /** * Check whether a given URL has a domain that occurs in a given set of domains * @param string $url URL * @param array $domains Array of domains (strings) * @return bool True if the host part of $url ends in one of the strings in $domains */ function wfMatchesDomainList( $url, $domains ) { $bits = wfParseUrl( $url ); if ( is_array( $bits ) && isset( $bits['host'] ) ) { $host = '.' . $bits['host']; foreach ( (array)$domains as $domain ) { $domain = '.' . $domain; if ( substr( $host, -strlen( $domain ) ) === $domain ) { return true; } } } return false; } /** * Sends a line to the debug log if enabled or, optionally, to a comment in output. * In normal operation this is a NOP. * * Controlling globals: * $wgDebugLogFile - points to the log file * $wgDebugRawPage - if false, 'action=raw' hits will not result in debug output. * $wgDebugComments - if on, some debug items may appear in comments in the HTML output. * * @since 1.25 support for additional context data * * @param string $text * @param string|bool $dest Unused * @param array $context Additional logging context data */ function wfDebug( $text, $dest = 'all', array $context = array() ) { global $wgDebugRawPage, $wgDebugLogPrefix; global $wgDebugTimestamps, $wgRequestTime; if ( !$wgDebugRawPage && wfIsDebugRawPage() ) { return; } $text = trim( $text ); // Inline logic from deprecated wfDebugTimer() if ( $wgDebugTimestamps ) { $context['seconds_elapsed'] = sprintf( '%6.4f', microtime( true ) - $wgRequestTime ); $context['memory_used'] = sprintf( '%5.1fM', ( memory_get_usage( true ) / ( 1024 * 1024 ) ) ); } if ( $wgDebugLogPrefix !== '' ) { $context['prefix'] = $wgDebugLogPrefix; } $logger = LoggerFactory::getInstance( 'wfDebug' ); $logger->debug( $text, $context ); } /** * Returns true if debug logging should be suppressed if $wgDebugRawPage = false * @return bool */ function wfIsDebugRawPage() { static $cache; if ( $cache !== null ) { return $cache; } # Check for raw action using $_GET not $wgRequest, since the latter might not be initialised yet if ( ( isset( $_GET['action'] ) && $_GET['action'] == 'raw' ) || ( isset( $_SERVER['SCRIPT_NAME'] ) && substr( $_SERVER['SCRIPT_NAME'], -8 ) == 'load.php' ) ) { $cache = true; } else { $cache = false; } return $cache; } /** * Get microsecond timestamps for debug logs * * @deprecated since 1.25 * @return string */ function wfDebugTimer() { global $wgDebugTimestamps, $wgRequestTime; wfDeprecated( __METHOD__, '1.25' ); if ( !$wgDebugTimestamps ) { return ''; } $prefix = sprintf( "%6.4f", microtime( true ) - $wgRequestTime ); $mem = sprintf( "%5.1fM", ( memory_get_usage( true ) / ( 1024 * 1024 ) ) ); return "$prefix $mem "; } /** * Send a line giving PHP memory usage. * * @param bool $exact Print exact byte values instead of kibibytes (default: false) */ function wfDebugMem( $exact = false ) { $mem = memory_get_usage(); if ( !$exact ) { $mem = floor( $mem / 1024 ) . ' KiB'; } else { $mem .= ' B'; } wfDebug( "Memory usage: $mem\n" ); } /** * Send a line to a supplementary debug log file, if configured, or main debug * log if not. * * To configure a supplementary log file, set $wgDebugLogGroups[$logGroup] to * a string filename or an associative array mapping 'destination' to the * desired filename. The associative array may also contain a 'sample' key * with an integer value, specifying a sampling factor. Sampled log events * will be emitted with a 1 in N random chance. * * @since 1.23 support for sampling log messages via $wgDebugLogGroups. * @since 1.25 support for additional context data * @since 1.25 sample behavior dependent on configured $wgMWLoggerDefaultSpi * * @param string $logGroup * @param string $text * @param string|bool $dest Destination of the message: * - 'all': both to the log and HTML (debug toolbar or HTML comments) * - 'log': only to the log and not in HTML * - 'private': only to the specific log if set in $wgDebugLogGroups and * discarded otherwise * For backward compatibility, it can also take a boolean: * - true: same as 'all' * - false: same as 'private' * @param array $context Additional logging context data */ function wfDebugLog( $logGroup, $text, $dest = 'all', array $context = array() ) { // Turn $dest into a string if it's a boolean (for b/c) if ( $dest === true ) { $dest = 'all'; } elseif ( $dest === false ) { $dest = 'private'; } $text = trim( $text ); $logger = LoggerFactory::getInstance( $logGroup ); $context['private'] = ( $dest === 'private' ); $logger->info( $text, $context ); } /** * Log for database errors * * @since 1.25 support for additional context data * * @param string $text Database error message. * @param array $context Additional logging context data */ function wfLogDBError( $text, array $context = array() ) { $logger = LoggerFactory::getInstance( 'wfLogDBError' ); $logger->error( trim( $text ), $context ); } /** * Throws a warning that $function is deprecated * * @param string $function * @param string|bool $version Version of MediaWiki that the function * was deprecated in (Added in 1.19). * @param string|bool $component Added in 1.19. * @param int $callerOffset How far up the call stack is the original * caller. 2 = function that called the function that called * wfDeprecated (Added in 1.20) * * @return null */ function wfDeprecated( $function, $version = false, $component = false, $callerOffset = 2 ) { MWDebug::deprecated( $function, $version, $component, $callerOffset + 1 ); } /** * Send a warning either to the debug log or in a PHP error depending on * $wgDevelopmentWarnings. To log warnings in production, use wfLogWarning() instead. * * @param string $msg Message to send * @param int $callerOffset Number of items to go back in the backtrace to * find the correct caller (1 = function calling wfWarn, ...) * @param int $level PHP error level; defaults to E_USER_NOTICE; * only used when $wgDevelopmentWarnings is true */ function wfWarn( $msg, $callerOffset = 1, $level = E_USER_NOTICE ) { MWDebug::warning( $msg, $callerOffset + 1, $level, 'auto' ); } /** * Send a warning as a PHP error and the debug log. This is intended for logging * warnings in production. For logging development warnings, use WfWarn instead. * * @param string $msg Message to send * @param int $callerOffset Number of items to go back in the backtrace to * find the correct caller (1 = function calling wfLogWarning, ...) * @param int $level PHP error level; defaults to E_USER_WARNING */ function wfLogWarning( $msg, $callerOffset = 1, $level = E_USER_WARNING ) { MWDebug::warning( $msg, $callerOffset + 1, $level, 'production' ); } /** * Log to a file without getting "file size exceeded" signals. * * Can also log to TCP or UDP with the syntax udp://host:port/prefix. This will * send lines to the specified port, prefixed by the specified prefix and a space. * @since 1.25 support for additional context data * * @param string $text * @param string $file Filename * @param array $context Additional logging context data * @throws MWException * @deprecated since 1.25 Use \MediaWiki\Logger\LegacyLogger::emit or UDPTransport */ function wfErrorLog( $text, $file, array $context = array() ) { wfDeprecated( __METHOD__, '1.25' ); $logger = LoggerFactory::getInstance( 'wfErrorLog' ); $context['destination'] = $file; $logger->info( trim( $text ), $context ); } /** * @todo document */ function wfLogProfilingData() { global $wgDebugLogGroups, $wgDebugRawPage; $context = RequestContext::getMain(); $request = $context->getRequest(); $profiler = Profiler::instance(); $profiler->setContext( $context ); $profiler->logData(); $config = $context->getConfig(); if ( $config->get( 'StatsdServer' ) ) { try { $statsdServer = explode( ':', $config->get( 'StatsdServer' ) ); $statsdHost = $statsdServer[0]; $statsdPort = isset( $statsdServer[1] ) ? $statsdServer[1] : 8125; $statsdSender = new SocketSender( $statsdHost, $statsdPort ); $statsdClient = new SamplingStatsdClient( $statsdSender, true, false ); $statsdClient->send( $context->getStats()->getBuffer() ); } catch ( Exception $ex ) { MWExceptionHandler::logException( $ex ); } } # Profiling must actually be enabled... if ( $profiler instanceof ProfilerStub ) { return; } if ( isset( $wgDebugLogGroups['profileoutput'] ) && $wgDebugLogGroups['profileoutput'] === false ) { // Explicitly disabled return; } if ( !$wgDebugRawPage && wfIsDebugRawPage() ) { return; } $ctx = array( 'elapsed' => $request->getElapsedTime() ); if ( !empty( $_SERVER['HTTP_X_FORWARDED_FOR'] ) ) { $ctx['forwarded_for'] = $_SERVER['HTTP_X_FORWARDED_FOR']; } if ( !empty( $_SERVER['HTTP_CLIENT_IP'] ) ) { $ctx['client_ip'] = $_SERVER['HTTP_CLIENT_IP']; } if ( !empty( $_SERVER['HTTP_FROM'] ) ) { $ctx['from'] = $_SERVER['HTTP_FROM']; } if ( isset( $ctx['forwarded_for'] ) || isset( $ctx['client_ip'] ) || isset( $ctx['from'] ) ) { $ctx['proxy'] = $_SERVER['REMOTE_ADDR']; } // Don't load $wgUser at this late stage just for statistics purposes // @todo FIXME: We can detect some anons even if it is not loaded. // See User::getId() $user = $context->getUser(); $ctx['anon'] = $user->isItemLoaded( 'id' ) && $user->isAnon(); // Command line script uses a FauxRequest object which does not have // any knowledge about an URL and throw an exception instead. try { $ctx['url'] = urldecode( $request->getRequestURL() ); } catch ( Exception $ignored ) { // no-op } $ctx['output'] = $profiler->getOutput(); $log = LoggerFactory::getInstance( 'profileoutput' ); $log->info( "Elapsed: {elapsed}; URL: <{url}>\n{output}", $ctx ); } /** * Increment a statistics counter * * @param string $key * @param int $count * @return void */ function wfIncrStats( $key, $count = 1 ) { $stats = RequestContext::getMain()->getStats(); $stats->updateCount( $key, $count ); } /** * Check whether the wiki is in read-only mode. * * @return bool */ function wfReadOnly() { return wfReadOnlyReason() !== false; } /** * Check if the site is in read-only mode and return the message if so * * This checks wfConfiguredReadOnlyReason() and the main load balancer * for slave lag. This may result in DB_SLAVE connection being made. * * @return string|bool String when in read-only mode; false otherwise */ function wfReadOnlyReason() { $readOnly = wfConfiguredReadOnlyReason(); if ( $readOnly !== false ) { return $readOnly; } static $lbReadOnly = null; if ( $lbReadOnly === null ) { // Callers use this method to be aware that data presented to a user // may be very stale and thus allowing submissions can be problematic. $lbReadOnly = wfGetLB()->getReadOnlyReason(); } return $lbReadOnly; } /** * Get the value of $wgReadOnly or the contents of $wgReadOnlyFile. * * @return string|bool String when in read-only mode; false otherwise * @since 1.27 */ function wfConfiguredReadOnlyReason() { global $wgReadOnly, $wgReadOnlyFile; if ( $wgReadOnly === null ) { // Set $wgReadOnly for faster access next time if ( is_file( $wgReadOnlyFile ) && filesize( $wgReadOnlyFile ) > 0 ) { $wgReadOnly = file_get_contents( $wgReadOnlyFile ); } else { $wgReadOnly = false; } } return $wgReadOnly; } /** * Return a Language object from $langcode * * @param Language|string|bool $langcode Either: * - a Language object * - code of the language to get the message for, if it is * a valid code create a language for that language, if * it is a string but not a valid code then make a basic * language object * - a boolean: if it's false then use the global object for * the current user's language (as a fallback for the old parameter * functionality), or if it is true then use global object * for the wiki's content language. * @return Language */ function wfGetLangObj( $langcode = false ) { # Identify which language to get or create a language object for. # Using is_object here due to Stub objects. if ( is_object( $langcode ) ) { # Great, we already have the object (hopefully)! return $langcode; } global $wgContLang, $wgLanguageCode; if ( $langcode === true || $langcode === $wgLanguageCode ) { # $langcode is the language code of the wikis content language object. # or it is a boolean and value is true return $wgContLang; } global $wgLang; if ( $langcode === false || $langcode === $wgLang->getCode() ) { # $langcode is the language code of user language object. # or it was a boolean and value is false return $wgLang; } $validCodes = array_keys( Language::fetchLanguageNames() ); if ( in_array( $langcode, $validCodes ) ) { # $langcode corresponds to a valid language. return Language::factory( $langcode ); } # $langcode is a string, but not a valid language code; use content language. wfDebug( "Invalid language code passed to wfGetLangObj, falling back to content language.\n" ); return $wgContLang; } /** * This is the function for getting translated interface messages. * * @see Message class for documentation how to use them. * @see https://www.mediawiki.org/wiki/Manual:Messages_API * * This function replaces all old wfMsg* functions. * * @param string|string[]|MessageSpecifier $key Message key, or array of keys, or a MessageSpecifier * @param mixed $params,... Normal message parameters * @return Message * * @since 1.17 * * @see Message::__construct */ function wfMessage( $key /*...*/ ) { $params = func_get_args(); array_shift( $params ); if ( isset( $params[0] ) && is_array( $params[0] ) ) { $params = $params[0]; } return new Message( $key, $params ); } /** * This function accepts multiple message keys and returns a message instance * for the first message which is non-empty. If all messages are empty then an * instance of the first message key is returned. * * @param string|string[] $keys,... Message keys * @return Message * * @since 1.18 * * @see Message::newFallbackSequence */ function wfMessageFallback( /*...*/ ) { $args = func_get_args(); return call_user_func_array( 'Message::newFallbackSequence', $args ); } /** * Get a message from anywhere, for the current user language. * * Use wfMsgForContent() instead if the message should NOT * change depending on the user preferences. * * @deprecated since 1.18 * * @param string $key Lookup key for the message, usually * defined in languages/Language.php * * Parameters to the message, which can be used to insert variable text into * it, can be passed to this function in the following formats: * - One per argument, starting at the second parameter * - As an array in the second parameter * These are not shown in the function definition. * * @return string */ function wfMsg( $key ) { wfDeprecated( __METHOD__, '1.21' ); $args = func_get_args(); array_shift( $args ); return wfMsgReal( $key, $args ); } /** * Same as above except doesn't transform the message * * @deprecated since 1.18 * * @param string $key * @return string */ function wfMsgNoTrans( $key ) { wfDeprecated( __METHOD__, '1.21' ); $args = func_get_args(); array_shift( $args ); return wfMsgReal( $key, $args, true, false, false ); } /** * Get a message from anywhere, for the current global language * set with $wgLanguageCode. * * Use this if the message should NOT change dependent on the * language set in the user's preferences. This is the case for * most text written into logs, as well as link targets (such as * the name of the copyright policy page). Link titles, on the * other hand, should be shown in the UI language. * * Note that MediaWiki allows users to change the user interface * language in their preferences, but a single installation * typically only contains content in one language. * * Be wary of this distinction: If you use wfMsg() where you should * use wfMsgForContent(), a user of the software may have to * customize potentially hundreds of messages in * order to, e.g., fix a link in every possible language. * * @deprecated since 1.18 * * @param string $key Lookup key for the message, usually * defined in languages/Language.php * @return string */ function wfMsgForContent( $key ) { wfDeprecated( __METHOD__, '1.21' ); global $wgForceUIMsgAsContentMsg; $args = func_get_args(); array_shift( $args ); $forcontent = true; if ( is_array( $wgForceUIMsgAsContentMsg ) && in_array( $key, $wgForceUIMsgAsContentMsg ) ) { $forcontent = false; } return wfMsgReal( $key, $args, true, $forcontent ); } /** * Same as above except doesn't transform the message * * @deprecated since 1.18 * * @param string $key * @return string */ function wfMsgForContentNoTrans( $key ) { wfDeprecated( __METHOD__, '1.21' ); global $wgForceUIMsgAsContentMsg; $args = func_get_args(); array_shift( $args ); $forcontent = true; if ( is_array( $wgForceUIMsgAsContentMsg ) && in_array( $key, $wgForceUIMsgAsContentMsg ) ) { $forcontent = false; } return wfMsgReal( $key, $args, true, $forcontent, false ); } /** * Really get a message * * @deprecated since 1.18 * * @param string $key Key to get. * @param array $args * @param bool $useDB * @param string|bool $forContent Language code, or false for user lang, true for content lang. * @param bool $transform Whether or not to transform the message. * @return string The requested message. */ function wfMsgReal( $key, $args, $useDB = true, $forContent = false, $transform = true ) { wfDeprecated( __METHOD__, '1.21' ); $message = wfMsgGetKey( $key, $useDB, $forContent, $transform ); $message = wfMsgReplaceArgs( $message, $args ); return $message; } /** * Fetch a message string value, but don't replace any keys yet. * * @deprecated since 1.18 * * @param string $key * @param bool $useDB * @param string|bool $langCode Code of the language to get the message for, or * behaves as a content language switch if it is a boolean. * @param bool $transform Whether to parse magic words, etc. * @return string */ function wfMsgGetKey( $key, $useDB = true, $langCode = false, $transform = true ) { wfDeprecated( __METHOD__, '1.21' ); Hooks::run( 'NormalizeMessageKey', array( &$key, &$useDB, &$langCode, &$transform ) ); $cache = MessageCache::singleton(); $message = $cache->get( $key, $useDB, $langCode ); if ( $message === false ) { $message = '<' . htmlspecialchars( $key ) . '>'; } elseif ( $transform ) { $message = $cache->transform( $message ); } return $message; } /** * Replace message parameter keys on the given formatted output. * * @param string $message * @param array $args * @return string * @private */ function wfMsgReplaceArgs( $message, $args ) { # Fix windows line-endings # Some messages are split with explode("\n", $msg) $message = str_replace( "\r", '', $message ); // Replace arguments if ( count( $args ) ) { if ( is_array( $args[0] ) ) { $args = array_values( $args[0] ); } $replacementKeys = array(); foreach ( $args as $n => $param ) { $replacementKeys['$' . ( $n + 1 )] = $param; } $message = strtr( $message, $replacementKeys ); } return $message; } /** * Return an HTML-escaped version of a message. * Parameter replacements, if any, are done *after* the HTML-escaping, * so parameters may contain HTML (eg links or form controls). Be sure * to pre-escape them if you really do want plaintext, or just wrap * the whole thing in htmlspecialchars(). * * @deprecated since 1.18 * * @param string $key * @param string $args,... Parameters * @return string */ function wfMsgHtml( $key ) { wfDeprecated( __METHOD__, '1.21' ); $args = func_get_args(); array_shift( $args ); return wfMsgReplaceArgs( htmlspecialchars( wfMsgGetKey( $key ) ), $args ); } /** * Return an HTML version of message * Parameter replacements, if any, are done *after* parsing the wiki-text message, * so parameters may contain HTML (eg links or form controls). Be sure * to pre-escape them if you really do want plaintext, or just wrap * the whole thing in htmlspecialchars(). * * @deprecated since 1.18 * * @param string $key * @param string $args,... Parameters * @return string */ function wfMsgWikiHtml( $key ) { wfDeprecated( __METHOD__, '1.21' ); $args = func_get_args(); array_shift( $args ); return wfMsgReplaceArgs( MessageCache::singleton()->parse( wfMsgGetKey( $key ), null, /* can't be set to false */ true, /* interface */ true )->getText(), $args ); } /** * Returns message in the requested format * * @deprecated since 1.18 * * @param string $key Key of the message * @param array $options Processing rules. * Can take the following options: * parse: parses wikitext to HTML * parseinline: parses wikitext to HTML and removes the surrounding * p's added by parser or tidy * escape: filters message through htmlspecialchars * escapenoentities: same, but allows entity references like through * replaceafter: parameters are substituted after parsing or escaping * parsemag: transform the message using magic phrases * content: fetch message for content language instead of interface * Also can accept a single associative argument, of the form 'language' => 'xx': * language: Language object or language code to fetch message for * (overridden by content). * Behavior for conflicting options (e.g., parse+parseinline) is undefined. * * @return string */ function wfMsgExt( $key, $options ) { wfDeprecated( __METHOD__, '1.21' ); $args = func_get_args(); array_shift( $args ); array_shift( $args ); $options = (array)$options; $validOptions = array( 'parse', 'parseinline', 'escape', 'escapenoentities', 'replaceafter', 'parsemag', 'content' ); foreach ( $options as $arrayKey => $option ) { if ( !preg_match( '/^[0-9]+|language$/', $arrayKey ) ) { // An unknown index, neither numeric nor "language" wfWarn( "wfMsgExt called with incorrect parameter key $arrayKey", 1, E_USER_WARNING ); } elseif ( preg_match( '/^[0-9]+$/', $arrayKey ) && !in_array( $option, $validOptions ) ) { // A numeric index with unknown value wfWarn( "wfMsgExt called with incorrect parameter $option", 1, E_USER_WARNING ); } } if ( in_array( 'content', $options, true ) ) { $forContent = true; $langCode = true; $langCodeObj = null; } elseif ( array_key_exists( 'language', $options ) ) { $forContent = false; $langCode = wfGetLangObj( $options['language'] ); $langCodeObj = $langCode; } else { $forContent = false; $langCode = false; $langCodeObj = null; } $string = wfMsgGetKey( $key, /*DB*/true, $langCode, /*Transform*/false ); if ( !in_array( 'replaceafter', $options, true ) ) { $string = wfMsgReplaceArgs( $string, $args ); } $messageCache = MessageCache::singleton(); $parseInline = in_array( 'parseinline', $options, true ); if ( in_array( 'parse', $options, true ) || $parseInline ) { $string = $messageCache->parse( $string, null, true, !$forContent, $langCodeObj ); if ( $string instanceof ParserOutput ) { $string = $string->getText(); } if ( $parseInline ) { $string = Parser::stripOuterParagraph( $string ); } } elseif ( in_array( 'parsemag', $options, true ) ) { $string = $messageCache->transform( $string, !$forContent, $langCodeObj ); } if ( in_array( 'escape', $options, true ) ) { $string = htmlspecialchars( $string ); } elseif ( in_array( 'escapenoentities', $options, true ) ) { $string = Sanitizer::escapeHtmlAllowEntities( $string ); } if ( in_array( 'replaceafter', $options, true ) ) { $string = wfMsgReplaceArgs( $string, $args ); } return $string; } /** * Since wfMsg() and co suck, they don't return false if the message key they * looked up didn't exist but instead the key wrapped in <>'s, this function checks for the * nonexistence of messages by checking the MessageCache::get() result directly. * * @deprecated since 1.18. Use Message::isDisabled(). * * @param string $key The message key looked up * @return bool True if the message *doesn't* exist. */ function wfEmptyMsg( $key ) { wfDeprecated( __METHOD__, '1.21' ); return MessageCache::singleton()->get( $key, /*useDB*/true, /*content*/false ) === false; } /** * Fetch server name for use in error reporting etc. * Use real server name if available, so we know which machine * in a server farm generated the current page. * * @return string */ function wfHostname() { static $host; if ( is_null( $host ) ) { # Hostname overriding global $wgOverrideHostname; if ( $wgOverrideHostname !== false ) { # Set static and skip any detection $host = $wgOverrideHostname; return $host; } if ( function_exists( 'posix_uname' ) ) { // This function not present on Windows $uname = posix_uname(); } else { $uname = false; } if ( is_array( $uname ) && isset( $uname['nodename'] ) ) { $host = $uname['nodename']; } elseif ( getenv( 'COMPUTERNAME' ) ) { # Windows computer name $host = getenv( 'COMPUTERNAME' ); } else { # This may be a virtual server. $host = $_SERVER['SERVER_NAME']; } } return $host; } /** * Returns a script tag that stores the amount of time it took MediaWiki to * handle the request in milliseconds as 'wgBackendResponseTime'. * * If $wgShowHostnames is true, the script will also set 'wgHostname' to the * hostname of the server handling the request. * * @return string */ function wfReportTime() { global $wgRequestTime, $wgShowHostnames; $responseTime = round( ( microtime( true ) - $wgRequestTime ) * 1000 ); $reportVars = array( 'wgBackendResponseTime' => $responseTime ); if ( $wgShowHostnames ) { $reportVars['wgHostname'] = wfHostname(); } return Skin::makeVariablesScript( $reportVars ); } /** * Safety wrapper for debug_backtrace(). * * Will return an empty array if debug_backtrace is disabled, otherwise * the output from debug_backtrace() (trimmed). * * @param int $limit This parameter can be used to limit the number of stack frames returned * * @return array Array of backtrace information */ function wfDebugBacktrace( $limit = 0 ) { static $disabled = null; if ( is_null( $disabled ) ) { $disabled = !function_exists( 'debug_backtrace' ); if ( $disabled ) { wfDebug( "debug_backtrace() is disabled\n" ); } } if ( $disabled ) { return array(); } if ( $limit && version_compare( PHP_VERSION, '5.4.0', '>=' ) ) { return array_slice( debug_backtrace( DEBUG_BACKTRACE_PROVIDE_OBJECT, $limit + 1 ), 1 ); } else { return array_slice( debug_backtrace(), 1 ); } } /** * Get a debug backtrace as a string * * @param bool|null $raw If true, the return value is plain text. If false, HTML. * Defaults to $wgCommandLineMode if unset. * @return string * @since 1.25 Supports $raw parameter. */ function wfBacktrace( $raw = null ) { global $wgCommandLineMode; if ( $raw === null ) { $raw = $wgCommandLineMode; } if ( $raw ) { $frameFormat = "%s line %s calls %s()\n"; $traceFormat = "%s"; } else { $frameFormat = "
' . nl2br( htmlspecialchars( $desc ) ) . "
\n"; } /** * Clear away any user-level output buffers, discarding contents. * * Suitable for 'starting afresh', for instance when streaming * relatively large amounts of data without buffering, or wanting to * output image files without ob_gzhandler's compression. * * The optional $resetGzipEncoding parameter controls suppression of * the Content-Encoding header sent by ob_gzhandler; by default it * is left. See comments for wfClearOutputBuffers() for why it would * be used. * * Note that some PHP configuration options may add output buffer * layers which cannot be removed; these are left in place. * * @param bool $resetGzipEncoding */ function wfResetOutputBuffers( $resetGzipEncoding = true ) { if ( $resetGzipEncoding ) { // Suppress Content-Encoding and Content-Length // headers from 1.10+s wfOutputHandler global $wgDisableOutputCompression; $wgDisableOutputCompression = true; } while ( $status = ob_get_status() ) { if ( isset( $status['flags'] ) ) { $flags = PHP_OUTPUT_HANDLER_CLEANABLE | PHP_OUTPUT_HANDLER_REMOVABLE; $deleteable = ( $status['flags'] & $flags ) === $flags; } elseif ( isset( $status['del'] ) ) { $deleteable = $status['del']; } else { // Guess that any PHP-internal setting can't be removed. $deleteable = $status['type'] !== 0; /* PHP_OUTPUT_HANDLER_INTERNAL */ } if ( !$deleteable ) { // Give up, and hope the result doesn't break // output behavior. break; } if ( $status['name'] === 'MediaWikiTestCase::wfResetOutputBuffersBarrier' ) { // Unit testing barrier to prevent this function from breaking PHPUnit. break; } if ( !ob_end_clean() ) { // Could not remove output buffer handler; abort now // to avoid getting in some kind of infinite loop. break; } if ( $resetGzipEncoding ) { if ( $status['name'] == 'ob_gzhandler' ) { // Reset the 'Content-Encoding' field set by this handler // so we can start fresh. header_remove( 'Content-Encoding' ); break; } } } } /** * More legible than passing a 'false' parameter to wfResetOutputBuffers(): * * Clear away output buffers, but keep the Content-Encoding header * produced by ob_gzhandler, if any. * * This should be used for HTTP 304 responses, where you need to * preserve the Content-Encoding header of the real result, but * also need to suppress the output of ob_gzhandler to keep to spec * and avoid breaking Firefox in rare cases where the headers and * body are broken over two packets. */ function wfClearOutputBuffers() { wfResetOutputBuffers( false ); } /** * Converts an Accept-* header into an array mapping string values to quality * factors * * @param string $accept * @param string $def Default * @return float[] Associative array of string => float pairs */ function wfAcceptToPrefs( $accept, $def = '*/*' ) { # No arg means accept anything (per HTTP spec) if ( !$accept ) { return array( $def => 1.0 ); } $prefs = array(); $parts = explode( ',', $accept ); foreach ( $parts as $part ) { # @todo FIXME: Doesn't deal with params like 'text/html; level=1' $values = explode( ';', trim( $part ) ); $match = array(); if ( count( $values ) == 1 ) { $prefs[$values[0]] = 1.0; } elseif ( preg_match( '/q\s*=\s*(\d*\.\d+)/', $values[1], $match ) ) { $prefs[$values[0]] = floatval( $match[1] ); } } return $prefs; } /** * Checks if a given MIME type matches any of the keys in the given * array. Basic wildcards are accepted in the array keys. * * Returns the matching MIME type (or wildcard) if a match, otherwise * NULL if no match. * * @param string $type * @param array $avail * @return string * @private */ function mimeTypeMatch( $type, $avail ) { if ( array_key_exists( $type, $avail ) ) { return $type; } else { $parts = explode( '/', $type ); if ( array_key_exists( $parts[0] . '/*', $avail ) ) { return $parts[0] . '/*'; } elseif ( array_key_exists( '*/*', $avail ) ) { return '*/*'; } else { return null; } } } /** * Returns the 'best' match between a client's requested internet media types * and the server's list of available types. Each list should be an associative * array of type to preference (preference is a float between 0.0 and 1.0). * Wildcards in the types are acceptable. * * @param array $cprefs Client's acceptable type list * @param array $sprefs Server's offered types * @return string * * @todo FIXME: Doesn't handle params like 'text/plain; charset=UTF-8' * XXX: generalize to negotiate other stuff */ function wfNegotiateType( $cprefs, $sprefs ) { $combine = array(); foreach ( array_keys( $sprefs ) as $type ) { $parts = explode( '/', $type ); if ( $parts[1] != '*' ) { $ckey = mimeTypeMatch( $type, $cprefs ); if ( $ckey ) { $combine[$type] = $sprefs[$type] * $cprefs[$ckey]; } } } foreach ( array_keys( $cprefs ) as $type ) { $parts = explode( '/', $type ); if ( $parts[1] != '*' && !array_key_exists( $type, $sprefs ) ) { $skey = mimeTypeMatch( $type, $sprefs ); if ( $skey ) { $combine[$type] = $sprefs[$skey] * $cprefs[$type]; } } } $bestq = 0; $besttype = null; foreach ( array_keys( $combine ) as $type ) { if ( $combine[$type] > $bestq ) { $besttype = $type; $bestq = $combine[$type]; } } return $besttype; } /** * Reference-counted warning suppression * * @deprecated since 1.26, use MediaWiki\suppressWarnings() directly * @param bool $end */ function wfSuppressWarnings( $end = false ) { MediaWiki\suppressWarnings( $end ); } /** * @deprecated since 1.26, use MediaWiki\restoreWarnings() directly * Restore error level to previous value */ function wfRestoreWarnings() { MediaWiki\suppressWarnings( true ); } # Autodetect, convert and provide timestamps of various types /** * Unix time - the number of seconds since 1970-01-01 00:00:00 UTC */ define( 'TS_UNIX', 0 ); /** * MediaWiki concatenated string timestamp (YYYYMMDDHHMMSS) */ define( 'TS_MW', 1 ); /** * MySQL DATETIME (YYYY-MM-DD HH:MM:SS) */ define( 'TS_DB', 2 ); /** * RFC 2822 format, for E-mail and HTTP headers */ define( 'TS_RFC2822', 3 ); /** * ISO 8601 format with no timezone: 1986-02-09T20:00:00Z * * This is used by Special:Export */ define( 'TS_ISO_8601', 4 ); /** * An Exif timestamp (YYYY:MM:DD HH:MM:SS) * * @see http://exif.org/Exif2-2.PDF The Exif 2.2 spec, see page 28 for the * DateTime tag and page 36 for the DateTimeOriginal and * DateTimeDigitized tags. */ define( 'TS_EXIF', 5 ); /** * Oracle format time. */ define( 'TS_ORACLE', 6 ); /** * Postgres format time. */ define( 'TS_POSTGRES', 7 ); /** * ISO 8601 basic format with no timezone: 19860209T200000Z. This is used by ResourceLoader */ define( 'TS_ISO_8601_BASIC', 9 ); /** * Get a timestamp string in one of various formats * * @param mixed $outputtype A timestamp in one of the supported formats, the * function will autodetect which format is supplied and act accordingly. * @param mixed $ts Optional timestamp to convert, default 0 for the current time * @return string|bool String / false The same date in the format specified in $outputtype or false */ function wfTimestamp( $outputtype = TS_UNIX, $ts = 0 ) { try { $timestamp = new MWTimestamp( $ts ); return $timestamp->getTimestamp( $outputtype ); } catch ( TimestampException $e ) { wfDebug( "wfTimestamp() fed bogus time value: TYPE=$outputtype; VALUE=$ts\n" ); return false; } } /** * Return a formatted timestamp, or null if input is null. * For dealing with nullable timestamp columns in the database. * * @param int $outputtype * @param string $ts * @return string */ function wfTimestampOrNull( $outputtype = TS_UNIX, $ts = null ) { if ( is_null( $ts ) ) { return null; } else { return wfTimestamp( $outputtype, $ts ); } } /** * Convenience function; returns MediaWiki timestamp for the present time. * * @return string */ function wfTimestampNow() { # return NOW return wfTimestamp( TS_MW, time() ); } /** * Check if the operating system is Windows * * @return bool True if it's Windows, false otherwise. */ function wfIsWindows() { static $isWindows = null; if ( $isWindows === null ) { $isWindows = strtoupper( substr( PHP_OS, 0, 3 ) ) === 'WIN'; } return $isWindows; } /** * Check if we are running under HHVM * * @return bool */ function wfIsHHVM() { return defined( 'HHVM_VERSION' ); } /** * Tries to get the system directory for temporary files. First * $wgTmpDirectory is checked, and then the TMPDIR, TMP, and TEMP * environment variables are then checked in sequence, and if none are * set try sys_get_temp_dir(). * * NOTE: When possible, use instead the tmpfile() function to create * temporary files to avoid race conditions on file creation, etc. * * @return string */ function wfTempDir() { global $wgTmpDirectory; if ( $wgTmpDirectory !== false ) { return $wgTmpDirectory; } $tmpDir = array_map( "getenv", array( 'TMPDIR', 'TMP', 'TEMP' ) ); foreach ( $tmpDir as $tmp ) { if ( $tmp && file_exists( $tmp ) && is_dir( $tmp ) && is_writable( $tmp ) ) { return $tmp; } } return sys_get_temp_dir(); } /** * Make directory, and make all parent directories if they don't exist * * @param string $dir Full path to directory to create * @param int $mode Chmod value to use, default is $wgDirectoryMode * @param string $caller Optional caller param for debugging. * @throws MWException * @return bool */ function wfMkdirParents( $dir, $mode = null, $caller = null ) { global $wgDirectoryMode; if ( FileBackend::isStoragePath( $dir ) ) { // sanity throw new MWException( __FUNCTION__ . " given storage path '$dir'." ); } if ( !is_null( $caller ) ) { wfDebug( "$caller: called wfMkdirParents($dir)\n" ); } if ( strval( $dir ) === '' || is_dir( $dir ) ) { return true; } $dir = str_replace( array( '\\', '/' ), DIRECTORY_SEPARATOR, $dir ); if ( is_null( $mode ) ) { $mode = $wgDirectoryMode; } // Turn off the normal warning, we're doing our own below MediaWiki\suppressWarnings(); $ok = mkdir( $dir, $mode, true ); // PHP5 <3 MediaWiki\restoreWarnings(); if ( !$ok ) { // directory may have been created on another request since we last checked if ( is_dir( $dir ) ) { return true; } // PHP doesn't report the path in its warning message, so add our own to aid in diagnosis. wfLogWarning( sprintf( "failed to mkdir \"%s\" mode 0%o", $dir, $mode ) ); } return $ok; } /** * Remove a directory and all its content. * Does not hide error. * @param string $dir */ function wfRecursiveRemoveDir( $dir ) { wfDebug( __FUNCTION__ . "( $dir )\n" ); // taken from http://de3.php.net/manual/en/function.rmdir.php#98622 if ( is_dir( $dir ) ) { $objects = scandir( $dir ); foreach ( $objects as $object ) { if ( $object != "." && $object != ".." ) { if ( filetype( $dir . '/' . $object ) == "dir" ) { wfRecursiveRemoveDir( $dir . '/' . $object ); } else { unlink( $dir . '/' . $object ); } } } reset( $objects ); rmdir( $dir ); } } /** * @param int $nr The number to format * @param int $acc The number of digits after the decimal point, default 2 * @param bool $round Whether or not to round the value, default true * @return string */ function wfPercent( $nr, $acc = 2, $round = true ) { $ret = sprintf( "%.${acc}f", $nr ); return $round ? round( $ret, $acc ) . '%' : "$ret%"; } /** * Safety wrapper around ini_get() for boolean settings. * The values returned from ini_get() are pre-normalized for settings * set via php.ini or php_flag/php_admin_flag... but *not* * for those set via php_value/php_admin_value. * * It's fairly common for people to use php_value instead of php_flag, * which can leave you with an 'off' setting giving a false positive * for code that just takes the ini_get() return value as a boolean. * * To make things extra interesting, setting via php_value accepts * "true" and "yes" as true, but php.ini and php_flag consider them false. :) * Unrecognized values go false... again opposite PHP's own coercion * from string to bool. * * Luckily, 'properly' set settings will always come back as '0' or '1', * so we only have to worry about them and the 'improper' settings. * * I frickin' hate PHP... :P * * @param string $setting * @return bool */ function wfIniGetBool( $setting ) { $val = strtolower( ini_get( $setting ) ); // 'on' and 'true' can't have whitespace around them, but '1' can. return $val == 'on' || $val == 'true' || $val == 'yes' || preg_match( "/^\s*[+-]?0*[1-9]/", $val ); // approx C atoi() function } /** * Windows-compatible version of escapeshellarg() * Windows doesn't recognise single-quotes in the shell, but the escapeshellarg() * function puts single quotes in regardless of OS. * * Also fixes the locale problems on Linux in PHP 5.2.6+ (bug backported to * earlier distro releases of PHP) * * @param string ... strings to escape and glue together, or a single array of strings parameter * @return string */ function wfEscapeShellArg( /*...*/ ) { wfInitShellLocale(); $args = func_get_args(); if ( count( $args ) === 1 && is_array( reset( $args ) ) ) { // If only one argument has been passed, and that argument is an array, // treat it as a list of arguments $args = reset( $args ); } $first = true; $retVal = ''; foreach ( $args as $arg ) { if ( !$first ) { $retVal .= ' '; } else { $first = false; } if ( wfIsWindows() ) { // Escaping for an MSVC-style command line parser and CMD.EXE // @codingStandardsIgnoreStart For long URLs // Refs: // * http://web.archive.org/web/20020708081031/http://mailman.lyra.org/pipermail/scite-interest/2002-March/000436.html // * http://technet.microsoft.com/en-us/library/cc723564.aspx // * Bug #13518 // * CR r63214 // Double the backslashes before any double quotes. Escape the double quotes. // @codingStandardsIgnoreEnd $tokens = preg_split( '/(\\\\*")/', $arg, -1, PREG_SPLIT_DELIM_CAPTURE ); $arg = ''; $iteration = 0; foreach ( $tokens as $token ) { if ( $iteration % 2 == 1 ) { // Delimiter, a double quote preceded by zero or more slashes $arg .= str_replace( '\\', '\\\\', substr( $token, 0, -1 ) ) . '\\"'; } elseif ( $iteration % 4 == 2 ) { // ^ in $token will be outside quotes, need to be escaped $arg .= str_replace( '^', '^^', $token ); } else { // $iteration % 4 == 0 // ^ in $token will appear inside double quotes, so leave as is $arg .= $token; } $iteration++; } // Double the backslashes before the end of the string, because // we will soon add a quote $m = array(); if ( preg_match( '/^(.*?)(\\\\+)$/', $arg, $m ) ) { $arg = $m[1] . str_replace( '\\', '\\\\', $m[2] ); } // Add surrounding quotes $retVal .= '"' . $arg . '"'; } else { $retVal .= escapeshellarg( $arg ); } } return $retVal; } /** * Check if wfShellExec() is effectively disabled via php.ini config * * @return bool|string False or one of (safemode,disabled) * @since 1.22 */ function wfShellExecDisabled() { static $disabled = null; if ( is_null( $disabled ) ) { if ( wfIniGetBool( 'safe_mode' ) ) { wfDebug( "wfShellExec can't run in safe_mode, PHP's exec functions are too broken.\n" ); $disabled = 'safemode'; } elseif ( !function_exists( 'proc_open' ) ) { wfDebug( "proc_open() is disabled\n" ); $disabled = 'disabled'; } else { $disabled = false; } } return $disabled; } /** * Execute a shell command, with time and memory limits mirrored from the PHP * configuration if supported. * * @param string|string[] $cmd If string, a properly shell-escaped command line, * or an array of unescaped arguments, in which case each value will be escaped * Example: [ 'convert', '-font', 'font name' ] would produce "'convert' '-font' 'font name'" * @param null|mixed &$retval Optional, will receive the program's exit code. * (non-zero is usually failure). If there is an error from * read, select, or proc_open(), this will be set to -1. * @param array $environ Optional environment variables which should be * added to the executed command environment. * @param array $limits Optional array with limits(filesize, memory, time, walltime) * this overwrites the global wgMaxShell* limits. * @param array $options Array of options: * - duplicateStderr: Set this to true to duplicate stderr to stdout, * including errors from limit.sh * - profileMethod: By default this function will profile based on the calling * method. Set this to a string for an alternative method to profile from * * @return string Collected stdout as a string */ function wfShellExec( $cmd, &$retval = null, $environ = array(), $limits = array(), $options = array() ) { global $IP, $wgMaxShellMemory, $wgMaxShellFileSize, $wgMaxShellTime, $wgMaxShellWallClockTime, $wgShellCgroup; $disabled = wfShellExecDisabled(); if ( $disabled ) { $retval = 1; return $disabled == 'safemode' ? 'Unable to run external programs in safe mode.' : 'Unable to run external programs, proc_open() is disabled.'; } $includeStderr = isset( $options['duplicateStderr'] ) && $options['duplicateStderr']; $profileMethod = isset( $options['profileMethod'] ) ? $options['profileMethod'] : wfGetCaller(); wfInitShellLocale(); $envcmd = ''; foreach ( $environ as $k => $v ) { if ( wfIsWindows() ) { /* Surrounding a set in quotes (method used by wfEscapeShellArg) makes the quotes themselves * appear in the environment variable, so we must use carat escaping as documented in * http://technet.microsoft.com/en-us/library/cc723564.aspx * Note however that the quote isn't listed there, but is needed, and the parentheses * are listed there but doesn't appear to need it. */ $envcmd .= "set $k=" . preg_replace( '/([&|()<>^"])/', '^\\1', $v ) . '&& '; } else { /* Assume this is a POSIX shell, thus required to accept variable assignments before the command * http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_09_01 */ $envcmd .= "$k=" . escapeshellarg( $v ) . ' '; } } if ( is_array( $cmd ) ) { $cmd = wfEscapeShellArg( $cmd ); } $cmd = $envcmd . $cmd; $useLogPipe = false; if ( is_executable( '/bin/bash' ) ) { $time = intval( isset( $limits['time'] ) ? $limits['time'] : $wgMaxShellTime ); if ( isset( $limits['walltime'] ) ) { $wallTime = intval( $limits['walltime'] ); } elseif ( isset( $limits['time'] ) ) { $wallTime = $time; } else { $wallTime = intval( $wgMaxShellWallClockTime ); } $mem = intval( isset( $limits['memory'] ) ? $limits['memory'] : $wgMaxShellMemory ); $filesize = intval( isset( $limits['filesize'] ) ? $limits['filesize'] : $wgMaxShellFileSize ); if ( $time > 0 || $mem > 0 || $filesize > 0 || $wallTime > 0 ) { $cmd = '/bin/bash ' . escapeshellarg( "$IP/includes/limit.sh" ) . ' ' . escapeshellarg( $cmd ) . ' ' . escapeshellarg( "MW_INCLUDE_STDERR=" . ( $includeStderr ? '1' : '' ) . ';' . "MW_CPU_LIMIT=$time; " . 'MW_CGROUP=' . escapeshellarg( $wgShellCgroup ) . '; ' . "MW_MEM_LIMIT=$mem; " . "MW_FILE_SIZE_LIMIT=$filesize; " . "MW_WALL_CLOCK_LIMIT=$wallTime; " . "MW_USE_LOG_PIPE=yes" ); $useLogPipe = true; } elseif ( $includeStderr ) { $cmd .= ' 2>&1'; } } elseif ( $includeStderr ) { $cmd .= ' 2>&1'; } wfDebug( "wfShellExec: $cmd\n" ); $desc = array( 0 => array( 'file', 'php://stdin', 'r' ), 1 => array( 'pipe', 'w' ), 2 => array( 'file', 'php://stderr', 'w' ) ); if ( $useLogPipe ) { $desc[3] = array( 'pipe', 'w' ); } $pipes = null; $scoped = Profiler::instance()->scopedProfileIn( __FUNCTION__ . '-' . $profileMethod ); $proc = proc_open( $cmd, $desc, $pipes ); if ( !$proc ) { wfDebugLog( 'exec', "proc_open() failed: $cmd" ); $retval = -1; return ''; } $outBuffer = $logBuffer = ''; $emptyArray = array(); $status = false; $logMsg = false; /* According to the documentation, it is possible for stream_select() * to fail due to EINTR. I haven't managed to induce this in testing * despite sending various signals. If it did happen, the error * message would take the form: * * stream_select(): unable to select [4]: Interrupted system call (max_fd=5) * * where [4] is the value of the macro EINTR and "Interrupted system * call" is string which according to the Linux manual is "possibly" * localised according to LC_MESSAGES. */ $eintr = defined( 'SOCKET_EINTR' ) ? SOCKET_EINTR : 4; $eintrMessage = "stream_select(): unable to select [$eintr]"; // Build a table mapping resource IDs to pipe FDs to work around a // PHP 5.3 issue in which stream_select() does not preserve array keys //