/**
* Perform an HTTP request
- * @param $method string HTTP method. Usually GET/POST
- * @param $url string Full URL to act on
- * @param $options options to pass to HttpRequest object
- * Possible keys for the array:
- * timeout Timeout length in seconds
- * postData An array of key-value pairs or a url-encoded form data
- * proxy The proxy to use. Will use $wgHTTPProxy (if set) otherwise.
- * noProxy Override $wgHTTPProxy (if set) and don't use any proxy at all.
- * sslVerifyHost (curl only) Verify the SSL certificate
- * caInfo (curl only) Provide CA information
- * maxRedirects Maximum number of redirects to follow (defaults to 5)
- * followRedirects Whether to follow redirects (defaults to true)
- * @returns mixed (bool)false on failure or a string on success
+ *
+ * @param $method String: HTTP method. Usually GET/POST
+ * @param $url String: full URL to act on
+ * @param $options Array: options to pass to HttpRequest object.
+ * Possible keys for the array:
+ * - timeout Timeout length in seconds
+ * - postData An array of key-value pairs or a url-encoded form data
+ * - proxy The proxy to use.
+ * Will use $wgHTTPProxy (if set) otherwise.
+ * - noProxy Override $wgHTTPProxy (if set) and don't use any proxy at all.
+ * - sslVerifyHost (curl only) Verify hostname against certificate
+ * - sslVerifyCert (curl only) Verify SSL certificate
+ * - caInfo (curl only) Provide CA information
+ * - maxRedirects Maximum number of redirects to follow (defaults to 5)
+ * - followRedirects Whether to follow redirects (defaults to false).
+ * Note: this should only be used when the target URL is trusted,
+ * to avoid attacks on intranet services accessible by HTTP.
+ * @return Mixed: (bool)false on failure or a string on success
*/
public static function request( $method, $url, $options = array() ) {
- wfDebug( "HTTP: $method: $url" );
+ $url = wfExpandUrl( $url );
+ wfDebug( "HTTP: $method: $url\n" );
$options['method'] = strtoupper( $method );
if ( !isset( $options['timeout'] ) ) {
$options['timeout'] = 'default';
/**
* Check if the URL can be served by localhost
- * @param $url string Full url to check
- * @return bool
+ *
+ * @param $url String: full url to check
+ * @return Boolean
*/
public static function isLocalURL( $url ) {
global $wgCommandLineMode, $wgConf;
/**
* A standard user-agent we can use for external requests.
- * @returns string
+ * @return String
*/
public static function userAgent() {
global $wgVersion;
/**
* Checks that the given URI is a valid one
+ *
* @param $uri Mixed: URI to check for validity
- * @returns bool
+ * @returns Boolean
*/
public static function isValidURI( $uri ) {
return preg_match(
protected $proxy = null;
protected $noProxy = false;
protected $sslVerifyHost = true;
+ protected $sslVerifyCert = true;
protected $caInfo = null;
protected $method = "GET";
protected $reqHeaders = array();
protected $parsedUrl;
protected $callback;
protected $maxRedirects = 5;
- protected $followRedirects = true;
+ protected $followRedirects = false;
protected $cookieJar;
protected $headerList = array();
protected $respVersion = "0.9";
- protected $respStatus = "0.1";
+ protected $respStatus = "200 Ok";
protected $respHeaders = array();
public $status;
/**
- * @param $url string url to use
- * @param $options array (optional) extra params to pass (see Http::request())
+ * @param $url String: url to use
+ * @param $options Array: (optional) extra params to pass (see Http::request())
*/
function __construct( $url, $options = array() ) {
global $wgHTTPTimeout;
$this->parsedUrl = parse_url( $url );
if ( !Http::isValidURI( $this->url ) ) {
- $this->status = Status::newFromFatal('http-invalid-url');
+ $this->status = Status::newFatal('http-invalid-url');
} else {
$this->status = Status::newGood( 100 ); // continue
}
}
$members = array( "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
- "method", "followRedirects", "maxRedirects" );
+ "method", "followRedirects", "maxRedirects", "sslVerifyCert" );
foreach ( $members as $o ) {
if ( isset($options[$o]) ) {
$this->$o = $options[$o];
/**
* Get the body, or content, of the response to the request
- * @return string
+ *
+ * @return String
*/
public function getContent() {
return $this->content;
}
+ /**
+ * Set the parameters of the request
+
+ * @param $args Array
+ * @todo overload the args param
+ */
+ public function setData($args) {
+ $this->postData = $args;
+ }
+
/**
* Take care of setting up the proxy
* (override in subclass)
- * @return string
+ *
+ * @return String
*/
public function proxySetup() {
global $wgHTTPProxy;
$this->proxy = 'http://localhost:80/';
} elseif ( $wgHTTPProxy ) {
$this->proxy = $wgHTTPProxy ;
+ } elseif ( getenv( "http_proxy" ) ) {
+ $this->proxy = getenv( "http_proxy" );
}
}
if( $this->cookieJar ) {
$this->reqHeaders['Cookie'] =
- $this->cookieJar->serializeToHttpRequest($this->parsedURL['path'],
- $this->parsedURL['host']);
+ $this->cookieJar->serializeToHttpRequest($this->parsedUrl['path'],
+ $this->parsedUrl['host']);
}
foreach($this->reqHeaders as $name => $value) {
$list[] = "$name: $value";
/**
* Set the callback
- * @param $callback callback
+ *
+ * @param $callback Callback
*/
public function setCallback( $callback ) {
$this->callback = $callback;
/**
* A generic callback to read the body of the response from a remote
* server.
+ *
* @param $fh handle
- * @param $content string
+ * @param $content String
*/
public function read( $fh, $content ) {
$this->content .= $content;
/**
* Take care of whatever is necessary to perform the URI request.
+ *
* @return Status
*/
public function execute() {
global $wgTitle;
+ $this->content = "";
+
if( strtoupper($this->method) == "HEAD" ) {
$this->headersOnly = true;
}
}
}
+ /**
+ * Parses the headers, including the HTTP status code and any
+ * Set-Cookie headers. This function expectes the headers to be
+ * found in an array in the member variable headerList.
+ *
+ * @return nothing
+ */
protected function parseHeader() {
$lastname = "";
foreach( $this->headerList as $header ) {
$this->parseCookies();
}
+ /**
+ * Sets the member variable status to a fatal status if the HTTP
+ * status code was not 200.
+ *
+ * @return nothing
+ */
+ protected function setStatus() {
+ if( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ if((int)$this->respStatus !== 200) {
+ list( $code, $message ) = explode(" ", $this->respStatus, 2);
+ $this->status->fatal("http-bad-status", $code, $message );
+ }
+ }
+
+
+ /**
+ * Returns true if the last status code was a redirect.
+ *
+ * @return Boolean
+ */
+ public function isRedirect() {
+ if( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ $status = (int)$this->respStatus;
+ if ( $status >= 300 && $status <= 303 ) {
+ return true;
+ }
+ return false;
+ }
+
/**
* Returns an associative array of response headers after the
* request has been executed. Because some headers
* (e.g. Set-Cookie) can appear more than once the, each value of
* the associative array is an array of the values given.
- * @return array
+ *
+ * @return Array
*/
public function getResponseHeaders() {
if( !$this->respHeaders ) {
return $this->respHeaders;
}
+ /**
+ * Returns the value of the given response header.
+ *
+ * @param $header String
+ * @return String
+ */
+ public function getResponseHeader($header) {
+ if( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+ if ( isset( $this->respHeaders[strtolower ( $header ) ] ) ) {
+ $v = $this->respHeaders[strtolower ( $header ) ];
+ return $v[count( $v ) - 1];
+ }
+ return null;
+ }
+
/**
* Tells the HttpRequest object to use this pre-loaded CookieJar.
+ *
* @param $jar CookieJar
*/
public function setCookieJar( $jar ) {
/**
* Returns the cookie jar in use.
+ *
* @returns CookieJar
*/
public function getCookieJar() {
* Parse the cookies in the response headers and store them in the cookie jar.
*/
protected function parseCookies() {
+ if( !$this->cookieJar ) {
+ $this->cookieJar = new CookieJar;
+ }
if( isset( $this->respHeaders['set-cookie'] ) ) {
- if( !$this->cookieJar ) {
- $this->cookieJar = new CookieJar;
- }
$url = parse_url( $this->getFinalUrl() );
foreach( $this->respHeaders['set-cookie'] as $cookie ) {
$this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
/**
* Returns the final URL after all redirections.
- * @returns string
+ *
+ * @return String
*/
public function getFinalUrl() {
- $finalUrl = $this->url;
- if ( isset( $this->respHeaders['location'] ) ) {
- $redir = $this->respHeaders['location'];
- $finalUrl = $redir[count($redir) - 1];
+ $location = $this->getResponseHeader("Location");
+ if ( $location ) {
+ return $location;
}
- return $finalUrl;
+ return $this->url;
+ }
+
+ /**
+ * Returns true if the backend can follow redirects. Overridden by the
+ * child classes.
+ */
+ public function canFollowRedirects() {
+ return true;
}
}
* Sets a cookie. Used before a request to set up any individual
* cookies. Used internally after a request to parse the
* Set-Cookie headers.
- * @param $name string the name of the cookie
- * @param $value string the value of the cookie
- * @param $attr array possible key/values:
+ *
+ * @param $value String: the value of the cookie
+ * @param $attr Array: possible key/values:
* expires A date string
* path The path this cookie is used on
* domain Domain this cookie is used on
$this->path = "/";
}
if( isset( $attr['domain'] ) ) {
- $this->domain = $attr['domain'];
+ if( self::validateCookieDomain( $attr['domain'] ) ) {
+ $this->domain = $attr['domain'];
+ }
} else {
throw new MWException("You must specify a domain.");
}
}
+ /**
+ * Return the true if the cookie is valid is valid. Otherwise,
+ * false. The uses a method similar to IE cookie security
+ * described here:
+ * http://kuza55.blogspot.com/2008/02/understanding-cookie-security.html
+ * A better method might be to use a blacklist like
+ * http://publicsuffix.org/
+ *
+ * @param $domain String: the domain to validate
+ * @param $originDomain String: (optional) the domain the cookie originates from
+ * @return Boolean
+ */
+ public static function validateCookieDomain( $domain, $originDomain = null) {
+ // Don't allow a trailing dot
+ if( substr( $domain, -1 ) == "." ) return false;
+
+ $dc = explode(".", $domain);
+
+ // Only allow full, valid IP addresses
+ if( preg_match( '/^[0-9.]+$/', $domain ) ) {
+ if( count( $dc ) != 4 ) return false;
+
+ if( ip2long( $domain ) === false ) return false;
+
+ if( $originDomain == null || $originDomain == $domain ) return true;
+
+ }
+
+ // Don't allow cookies for "co.uk" or "gov.uk", etc, but allow "supermarket.uk"
+ if( strrpos( $domain, "." ) - strlen( $domain ) == -3 ) {
+ if( (count($dc) == 2 && strlen( $dc[0] ) <= 2 )
+ || (count($dc) == 3 && strlen( $dc[0] ) == "" && strlen( $dc[1] ) <= 2 ) ) {
+ return false;
+ }
+ if( (count($dc) == 2 || (count($dc) == 3 && $dc[0] == "") )
+ && preg_match( '/(com|net|org|gov|edu)\...$/', $domain) ) {
+ return false;
+ }
+ }
+
+ if( $originDomain != null ) {
+ if( substr( $domain, 0, 1 ) != "." && $domain != $originDomain ) {
+ return false;
+ }
+ if( substr( $domain, 0, 1 ) == "."
+ && substr_compare( $originDomain, $domain, -strlen( $domain ),
+ strlen( $domain ), TRUE ) != 0 ) {
+ return false;
+ }
+ }
+
+ return true;
+ }
+
/**
* Serialize the cookie jar into a format useful for HTTP Request headers.
- * @param $path string the path that will be used. Required.
- * @param $domain string the domain that will be used. Required.
- * @return string
+ *
+ * @param $path String: the path that will be used. Required.
+ * @param $domain String: the domain that will be used. Required.
+ * @return String
*/
public function serializeToHttpRequest( $path, $domain ) {
$ret = "";
}
protected function canServeDomain( $domain ) {
- if( $this->domain && substr_compare( $domain, $this->domain, -strlen( $this->domain ),
- strlen( $this->domain ), TRUE ) == 0 ) {
+ if( $domain == $this->domain
+ || ( strlen( $domain) > strlen( $this->domain )
+ && substr( $this->domain, 0, 1) == "."
+ && substr_compare( $domain, $this->domain, -strlen( $this->domain ),
+ strlen( $this->domain ), TRUE ) == 0 ) ) {
return true;
}
return false;
}
class CookieJar {
- private $cookie;
+ private $cookie = array();
/**
* Set a cookie in the cookie jar. Make sure only one cookie per-name exists.
/**
* Parse the content of an Set-Cookie HTTP Response header.
- * @param $cookie string
+ *
+ * @param $cookie String
+ * @param $domain String: cookie's domain
*/
- public function parseCookieResponseHeader ( $cookie, $domain = null ) {
+ public function parseCookieResponseHeader ( $cookie, $domain ) {
$len = strlen( "Set-Cookie:" );
if ( substr_compare( "Set-Cookie:", $cookie, 0, $len, TRUE ) === 0 ) {
$cookie = substr( $cookie, $len );
}
$bit = array_map( 'trim', explode( ";", $cookie ) );
- list($name, $value) = explode( "=", array_shift( $bit ), 2 );
- $attr = array();
- foreach( $bit as $piece ) {
- $parts = explode( "=", $piece );
- if( count( $parts ) > 1 ) {
- $attr[strtolower( $parts[0] )] = $parts[1];
- } else {
- $attr[strtolower( $parts[0] )] = true;
+ if ( count($bit) >= 1 ) {
+ list($name, $value) = explode( "=", array_shift( $bit ), 2 );
+ $attr = array();
+ foreach( $bit as $piece ) {
+ $parts = explode( "=", $piece );
+ if( count( $parts ) > 1 ) {
+ $attr[strtolower( $parts[0] )] = $parts[1];
+ } else {
+ $attr[strtolower( $parts[0] )] = true;
+ }
}
- }
- if( !isset( $attr['domain'] ) ) {
- $attr['domain'] = $domain;
- } else {
- /* FIXME: Check that domain is valid */
+ if( !isset( $attr['domain'] ) ) {
+ $attr['domain'] = $domain;
+ } elseif ( !Cookie::validateCookieDomain( $attr['domain'], $domain ) ) {
+ return null;
+ }
+ $this->setCookie( $name, $value, $attr );
}
- $this->setCookie( $name, $value, $attr );
}
}
$this->curlOptions[CURLOPT_HTTP_VERSION] = CURL_HTTP_VERSION_1_0;
$this->curlOptions[CURLOPT_WRITEFUNCTION] = $this->callback;
$this->curlOptions[CURLOPT_HEADERFUNCTION] = array($this, "readHeader");
- $this->curlOptions[CURLOPT_FOLLOWLOCATION] = $this->followRedirects;
$this->curlOptions[CURLOPT_MAXREDIRS] = $this->maxRedirects;
+ $this->curlOptions[CURLOPT_ENCODING] = ""; # Enable compression
/* not sure these two are actually necessary */
if(isset($this->reqHeaders['Referer'])) {
}
$this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
- if ( $this->sslVerifyHost ) {
+ if ( isset( $this->sslVerifyHost ) ) {
$this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost;
}
+
+ if ( isset( $this->sslVerifyCert ) ) {
+ $this->curlOptions[CURLOPT_SSL_VERIFYPEER] = $this->sslVerifyCert;
+ }
if ( $this->caInfo ) {
$this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
$this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
$curlHandle = curl_init( $this->url );
- curl_setopt_array( $curlHandle, $this->curlOptions );
+ if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
+ throw new MWException("Error setting curl options.");
+ }
+ if ( $this->followRedirects && $this->canFollowRedirects() ) {
+ if ( ! @curl_setopt( $curlHandle, CURLOPT_FOLLOWLOCATION, true ) ) {
+ wfDebug( __METHOD__.": Couldn't set CURLOPT_FOLLOWLOCATION. " .
+ "Probably safe_mode or open_basedir is set.\n");
+ // Continue the processing. If it were in curl_setopt_array,
+ // processing would have halted on its entry
+ }
+ }
if ( false === curl_exec( $curlHandle ) ) {
$code = curl_error( $curlHandle );
curl_close( $curlHandle );
+ $this->parseHeader();
+ $this->setStatus();
return $this->status;
}
+
+ public function canFollowRedirects() {
+ if ( strval( ini_get( 'open_basedir' ) ) !== '' || wfIniGetBool( 'safe_mode' ) ) {
+ wfDebug( "Cannot follow redirects in safe mode\n" );
+ return false;
+ }
+ if ( !defined( 'CURLOPT_REDIR_PROTOCOLS' ) ) {
+ wfDebug( "Cannot follow redirects with libcurl < 7.19.4 due to CVE-2009-0037\n" );
+ return false;
+ }
+ return true;
+ }
}
class PhpHttpRequest extends HttpRequest {
}
public function execute() {
- if ( $this->parsedUrl['scheme'] != 'http' ) {
- $this->status->fatal( 'http-invalid-scheme', $this->parsedURL['scheme'] );
- }
-
parent::execute();
- if ( !$this->status->isOK() ) {
- return $this->status;
+
+ // At least on Centos 4.8 with PHP 5.1.6, using max_redirects to follow redirects
+ // causes a segfault
+ $manuallyRedirect = version_compare( phpversion(), '5.1.7', '<' );
+
+ if ( $this->parsedUrl['scheme'] != 'http' ) {
+ $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
}
$this->reqHeaders['Accept'] = "*/*";
$options['request_fulluri'] = true;
}
- if ( !$this->followRedirects ) {
+ if ( !$this->followRedirects || $manuallyRedirect ) {
$options['max_redirects'] = 0;
} else {
$options['max_redirects'] = $this->maxRedirects;
}
$options['method'] = $this->method;
- $options['timeout'] = $this->timeout;
$options['header'] = implode("\r\n", $this->getHeaderList());
// Note that at some future point we may want to support
// HTTP/1.1, but we'd have to write support for chunking
// in version of PHP < 5.3.1
$options['protocol_version'] = "1.0";
+ // This is how we tell PHP we want to deal with 404s (for example) ourselves.
+ // Only works on 5.2.10+
+ $options['ignore_errors'] = true;
+
if ( $this->postData ) {
$options['content'] = $this->postData;
}
$oldTimeout = false;
if ( version_compare( '5.2.1', phpversion(), '>' ) ) {
$oldTimeout = ini_set('default_socket_timeout', $this->timeout);
+ } else {
+ $options['timeout'] = $this->timeout;
}
$context = stream_context_create( array( 'http' => $options ) );
- wfSuppressWarnings();
- $fh = fopen( $this->url, "r", false, $context );
- wfRestoreWarnings();
+
+ $this->headerList = array();
+ $reqCount = 0;
+ $url = $this->url;
+ do {
+ $reqCount++;
+ wfSuppressWarnings();
+ $fh = fopen( $url, "r", false, $context );
+ wfRestoreWarnings();
+ if ( !$fh ) {
+ break;
+ }
+ $result = stream_get_meta_data( $fh );
+ $this->headerList = $result['wrapper_data'];
+ $this->parseHeader();
+ if ( !$manuallyRedirect || !$this->followRedirects ) {
+ break;
+ }
+
+ # Handle manual redirection
+ if ( !$this->isRedirect() || $reqCount > $this->maxRedirects ) {
+ break;
+ }
+ # Check security of URL
+ $url = $this->getResponseHeader("Location");
+ if ( substr( $url, 0, 7 ) !== 'http://' ) {
+ wfDebug( __METHOD__.": insecure redirection\n" );
+ break;
+ }
+ } while ( true );
+
if ( $oldTimeout !== false ) {
ini_set('default_socket_timeout', $oldTimeout);
}
+ $this->setStatus();
+
if ( $fh === false ) {
$this->status->fatal( 'http-request-error' );
return $this->status;
}
- $result = stream_get_meta_data( $fh );
if ( $result['timed_out'] ) {
$this->status->fatal( 'http-timed-out', $this->url );
return $this->status;
}
- $this->headerList = $result['wrapper_data'];
- while ( !feof( $fh ) ) {
- $buf = fread( $fh, 8192 );
- if ( $buf === false ) {
- $this->status->fatal( 'http-read-error' );
- break;
- }
- if ( strlen( $buf ) ) {
- call_user_func( $this->callback, $fh, $buf );
+ if($this->status->isOK()) {
+ while ( !feof( $fh ) ) {
+ $buf = fread( $fh, 8192 );
+ if ( $buf === false ) {
+ $this->status->fatal( 'http-read-error' );
+ break;
+ }
+ if ( strlen( $buf ) ) {
+ call_user_func( $this->callback, $fh, $buf );
+ }
}
}
fclose( $fh );