* @ingroup HTTP
*/
class Http {
+ static $httpEngine = false;
+
/**
* Perform an HTTP request
- * @param $method string HTTP method. Usually GET/POST
- * @param $url string Full URL to act on
- * @param $opts options to pass to HttpRequest object
- * @returns mixed (bool)false on failure or a string on success
- */
- public static function request( $method, $url, $opts = array() ) {
- $opts['method'] = strtoupper( $method );
- if ( !array_key_exists( 'timeout', $opts ) ) {
- $opts['timeout'] = 'default';
- }
- $req = HttpRequest::factory( $url, $opts );
- $status = $req->doRequest();
+ *
+ * @param $method String: HTTP method. Usually GET/POST
+ * @param $url String: full URL to act on. If protocol-relative, will be expanded to an http:// URL
+ * @param $options Array: options to pass to MWHttpRequest 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.
+ * Otherwise it will use $wgHTTPProxy (if set)
+ * Otherwise it will use the environment variable "http_proxy" (if set)
+ * - noProxy Don't use any proxy at all. Takes precedence over proxy value(s).
+ * - sslVerifyHost (curl only) Verify hostname against certificate
+ * - sslVerifyCert (curl only) Verify SSL certificate
+ * - caInfo (curl only) Provide CA information
+ * - maxRedirects Maximum number of redirects to follow (defaults to 5)
+ * - followRedirects Whether to follow redirects (defaults to false).
+ * Note: this should only be used when the target URL is trusted,
+ * to avoid attacks on intranet services accessible by HTTP.
+ * - userAgent A user agent, if you want to override the default
+ * MediaWiki/$wgVersion
+ * @return Mixed: (bool)false on failure or a string on success
+ */
+ public static function request( $method, $url, $options = array() ) {
+ wfDebug( "HTTP: $method: $url\n" );
+ $options['method'] = strtoupper( $method );
+
+ if ( !isset( $options['timeout'] ) ) {
+ $options['timeout'] = 'default';
+ }
+
+ $req = MWHttpRequest::factory( $url, $options );
+ $status = $req->execute();
+
if ( $status->isOK() ) {
return $req->getContent();
} else {
/**
* Simple wrapper for Http::request( 'GET' )
* @see Http::request()
+ *
+ * @param $url
+ * @param $timeout string
+ * @param $options array
+ * @return string
*/
- public static function get( $url, $timeout = 'default', $opts = array() ) {
- $opts['timeout'] = $timeout;
- return Http::request( 'GET', $url, $opts );
+ public static function get( $url, $timeout = 'default', $options = array() ) {
+ $options['timeout'] = $timeout;
+ return Http::request( 'GET', $url, $options );
}
/**
* Simple wrapper for Http::request( 'POST' )
* @see Http::request()
+ *
+ * @param $url
+ * @param $options array
+ * @return string
*/
- public static function post( $url, $opts = array() ) {
- return Http::request( 'POST', $url, $opts );
+ public static function post( $url, $options = array() ) {
+ return Http::request( 'POST', $url, $options );
}
/**
* 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;
+
if ( $wgCommandLineMode ) {
return false;
}
$domainParts = explode( '.', $host );
// Check if this domain or any superdomain is listed in $wgConf as a local virtual host
$domainParts = array_reverse( $domainParts );
+
+ $domain = '';
for ( $i = 0; $i < count( $domainParts ); $i++ ) {
$domainPart = $domainParts[$i];
if ( $i == 0 ) {
} else {
$domain = $domainPart . '.' . $domain;
}
+
if ( $wgConf->isLocalVHost( $domain ) ) {
return true;
}
}
}
+
return false;
}
/**
* 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
+ * Checks that the given URI is a valid one. Hardcoding the
+ * protocols, because we only want protocols that both cURL
+ * and php support.
+ *
+ * file:// should not be allowed here for security purpose (r67684)
+ *
+ * @fixme this is wildly inaccurate and fails to actually check most stuff
+ *
* @param $uri Mixed: URI to check for validity
- * @returns bool
+ * @return Boolean
*/
public static function isValidURI( $uri ) {
return preg_match(
- '/(ftp|http|https):\/\/(\w+:{0,1}\w*@)?(\S+)(:[0-9]+)?(\/|\/([\w#!:.?+=&%@!\-\/]))?/',
- $uri,
- $matches
+ '/^https?:\/\/[^\/\s]\S*$/D',
+ $uri
);
}
+}
+
+/**
+ * This wrapper class will call out to curl (if available) or fallback
+ * to regular PHP if necessary for handling internal HTTP requests.
+ *
+ * Renamed from HttpRequest to MWHttpRequest to avoid conflict with
+ * PHP's HTTP extension.
+ */
+class MWHttpRequest {
+ const SUPPORTS_FILE_POSTS = false;
+
+ protected $content;
+ protected $timeout = 'default';
+ protected $headersOnly = null;
+ protected $postData = null;
+ protected $proxy = null;
+ protected $noProxy = false;
+ protected $sslVerifyHost = true;
+ protected $sslVerifyCert = true;
+ protected $caInfo = null;
+ protected $method = "GET";
+ protected $reqHeaders = array();
+ protected $url;
+ protected $parsedUrl;
+ protected $callback;
+ protected $maxRedirects = 5;
+ protected $followRedirects = false;
+
/**
- * Fetch a URL, write the result to a file.
- * @params $url string url to fetch
- * @params $targetFilePath string full path (including filename) to write the file to
- * @params $async bool whether the download should be asynchronous (defaults to false)
- * @params $redirectCount int used internally to keep track of the number of redirects
- *
- * @returns Status -- for async requests this will contain the request key
+ * @var CookieJar
*/
- public static function doDownload( $url, $targetFilePath, $async = false, $redirectCount = 0 ) {
- global $wgPhpCli, $wgMaxUploadSize, $wgMaxRedirects;
+ protected $cookieJar;
- // do a quick check to HEAD to insure the file size is not > $wgMaxUploadSize
- $headRequest = HttpRequest::factory( $url, array( 'headersOnly' => true ) );
- $headResponse = $headRequest->doRequest();
- if ( !$headResponse->isOK() ) {
- return $headResponse;
- }
- $head = $headResponse->value;
+ protected $headerList = array();
+ protected $respVersion = "0.9";
+ protected $respStatus = "200 Ok";
+ protected $respHeaders = array();
+
+ public $status;
- // check for redirects:
- if ( $redirectCount < 0 ) {
- $redirectCount = 0;
+ /**
+ * @param $url String: url to use. If protocol-relative, will be expanded to an http:// URL
+ * @param $options Array: (optional) extra params to pass (see Http::request())
+ */
+ function __construct( $url, $options = array() ) {
+ global $wgHTTPTimeout;
+
+ $this->url = wfExpandUrl( $url, PROTO_HTTP );
+ $this->parsedUrl = wfParseUrl( $this->url );
+
+ if ( !$this->parsedUrl || !Http::isValidURI( $this->url ) ) {
+ $this->status = Status::newFatal( 'http-invalid-url' );
+ } else {
+ $this->status = Status::newGood( 100 ); // continue
}
- if ( isset( $head['Location'] ) && strrpos( $head[0], '302' ) !== false ) {
- if ( $redirectCount < $wgMaxRedirects ) {
- if ( self::isValidURI( $head['Location'] ) ) {
- return self::doDownload( $head['Location'], $targetFilePath,
- $async, $redirectCount++ );
- } else {
- return Status::newFatal( 'upload-proto-error' );
- }
- } else {
- return Status::newFatal( 'upload-too-many-redirects' );
- }
+
+ if ( isset( $options['timeout'] ) && $options['timeout'] != 'default' ) {
+ $this->timeout = $options['timeout'];
+ } else {
+ $this->timeout = $wgHTTPTimeout;
}
- // we did not get a 200 ok response:
- if ( strrpos( $head[0], '200 OK' ) === false ) {
- return Status::newFatal( 'upload-http-error', htmlspecialchars( $head[0] ) );
+ if( isset( $options['userAgent'] ) ) {
+ $this->setUserAgent( $options['userAgent'] );
}
- $contentLength = $head['Content-Length'];
- if ( $contentLength ) {
- if ( $contentLength > $wgMaxUploadSize ) {
- return Status::newFatal( 'requested file length ' . $contentLength .
- ' is greater than $wgMaxUploadSize: ' . $wgMaxUploadSize );
+ $members = array( "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
+ "method", "followRedirects", "maxRedirects", "sslVerifyCert", "callback" );
+
+ foreach ( $members as $o ) {
+ if ( isset( $options[$o] ) ) {
+ $this->$o = $options[$o];
}
}
- // check if we can find phpCliPath (for doing a background shell request to
- // php to do the download:
- if ( $async && $wgPhpCli && wfShellExecEnabled() ) {
- wfDebug( __METHOD__ . "\nASYNC_DOWNLOAD\n" );
- // setup session and shell call:
- return self::startBackgroundRequest( $url, $targetFilePath, $contentLength );
- } else {
- wfDebug( __METHOD__ . "\nSYNC_DOWNLOAD\n" );
- // SYNC_DOWNLOAD download as much as we can in the time we have to execute
- $opts['method'] = 'GET';
- $opts['targetFilePath'] = $mTargetFilePath;
- $req = HttpRequest::factory( $url, $opts );
- return $req->doRequest();
+ if ( $this->noProxy ) {
+ $this->proxy = ''; // noProxy takes precedence
}
}
/**
- * Start backgrounded (i.e. non blocking) request. The
- * backgrounded request will provide updates to the user's session
- * data.
- * @param $url string the URL to download
- * @param $targetFilePath string the destination for the downloaded file
- * @param $contentLength int (optional) the length of the download from the HTTP header
- *
- * @returns Status
+ * Simple function to test if we can make any sort of requests at all, using
+ * cURL or fopen()
+ * @return bool
*/
- private static function startBackgroundRequest( $url, $targetFilePath, $contentLength = null ) {
- global $IP, $wgPhpCli, $wgServer;
- $status = Status::newGood();
-
- // generate a session id with all the details for the download (pid, targetFilePath )
- $requestKey = self::createRequestKey();
- $sessionID = session_id();
-
- // store the url and target path:
- $_SESSION['wsBgRequest'][$requestKey]['url'] = $url;
- $_SESSION['wsBgRequest'][$requestKey]['targetFilePath'] = $targetFilePath;
- // since we request from the cmd line we lose the original host name pass in the session:
- $_SESSION['wsBgRequest'][$requestKey]['orgServer'] = $wgServer;
+ public static function canMakeRequests() {
+ return function_exists( 'curl_init' ) || wfIniGetBool( 'allow_url_fopen' );
+ }
- if ( $contentLength ) {
- $_SESSION['wsBgRequest'][$requestKey]['contentLength'] = $contentLength;
+ /**
+ * Generate a new request object
+ * @param $url String: url to use
+ * @param $options Array: (optional) extra params to pass (see Http::request())
+ * @return CurlHttpRequest|PhpHttpRequest
+ * @see MWHttpRequest::__construct
+ */
+ public static function factory( $url, $options = null ) {
+ if ( !Http::$httpEngine ) {
+ Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
+ } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) {
+ throw new MWException( __METHOD__ . ': curl (http://php.net/curl) is not installed, but' .
+ ' Http::$httpEngine is set to "curl"' );
+ }
+
+ switch( Http::$httpEngine ) {
+ case 'curl':
+ return new CurlHttpRequest( $url, $options );
+ case 'php':
+ if ( !wfIniGetBool( 'allow_url_fopen' ) ) {
+ throw new MWException( __METHOD__ . ': allow_url_fopen needs to be enabled for pure PHP' .
+ ' http requests to work. If possible, curl should be used instead. See http://php.net/curl.' );
+ }
+ return new PhpHttpRequest( $url, $options );
+ default:
+ throw new MWException( __METHOD__ . ': The setting of Http::$httpEngine is not valid.' );
}
+ }
- // set initial loaded bytes:
- $_SESSION['wsBgRequest'][$requestKey]['loaded'] = 0;
-
- // run the background download request:
- $cmd = $wgPhpCli . ' ' . $IP . "/maintenance/httpSessionDownload.php " .
- "--sid {$sessionID} --usk {$requestKey} --wiki " . wfWikiId();
- $pid = wfShellBackgroundExec( $cmd );
- // the pid is not of much use since we won't be visiting this same apache any-time soon.
- if ( !$pid )
- return Status::newFatal( 'http-could-not-background' );
+ /**
+ * Get the body, or content, of the response to the request
+ *
+ * @return String
+ */
+ public function getContent() {
+ return $this->content;
+ }
- // update the status value with the $requestKey (for the user to
- // check on the status of the download)
- $status->value = $requestKey;
+ /**
+ * Set the parameters of the request
- // return good status
- return $status;
+ * @param $args Array
+ * @todo overload the args param
+ */
+ public function setData( $args ) {
+ $this->postData = $args;
}
/**
- * Returns a unique, random string that can be used as an request key and
- * preloads it into the session data.
+ * Take care of setting up the proxy (do nothing if "noProxy" is set)
*
- * @returns string
+ * @return void
*/
- static function createRequestKey() {
- if ( !array_key_exists( 'wsBgRequest', $_SESSION ) ) {
- $_SESSION['wsBgRequest'] = array();
- }
+ public function proxySetup() {
+ global $wgHTTPProxy;
- $key = uniqid( 'bgrequest', true );
+ if ( $this->proxy || !$this->noProxy ) {
+ return;
+ }
- // This is probably over-defensive.
- while ( array_key_exists( $key, $_SESSION['wsBgRequest'] ) ) {
- $key = uniqid( 'bgrequest', true );
+ if ( Http::isLocalURL( $this->url ) || $this->noProxy ) {
+ $this->proxy = '';
+ } elseif ( $wgHTTPProxy ) {
+ $this->proxy = $wgHTTPProxy ;
+ } elseif ( getenv( "http_proxy" ) ) {
+ $this->proxy = getenv( "http_proxy" );
}
- $_SESSION['wsBgRequest'][$key] = array();
+ }
- return $key;
+ /**
+ * Set the refererer header
+ */
+ public function setReferer( $url ) {
+ $this->setHeader( 'Referer', $url );
}
/**
- * Recover the necessary session and request information
- * @param $sessionID string
- * @param $requestKey string the HTTP request key
- *
- * @returns array request information
+ * Set the user agent
+ * @param $UA string
+ */
+ public function setUserAgent( $UA ) {
+ $this->setHeader( 'User-Agent', $UA );
+ }
+
+ /**
+ * Set an arbitrary header
+ * @param $name
+ * @param $value
*/
- private static function recoverSession( $sessionID, $requestKey ) {
- global $wgUser, $wgServer, $wgSessionsInMemcached;
+ public function setHeader( $name, $value ) {
+ // I feel like I should normalize the case here...
+ $this->reqHeaders[$name] = $value;
+ }
- // set session to the provided key:
- session_id( $sessionID );
- // fire up mediaWiki session system:
- wfSetupSession();
+ /**
+ * Get an array of the headers
+ * @return array
+ */
+ public function getHeaderList() {
+ $list = array();
- // start the session
- if ( session_start() === false ) {
- wfDebug( __METHOD__ . ' could not start session' );
+ if ( $this->cookieJar ) {
+ $this->reqHeaders['Cookie'] =
+ $this->cookieJar->serializeToHttpRequest(
+ $this->parsedUrl['path'],
+ $this->parsedUrl['host']
+ );
}
- // get all the vars we need from session_id
- if ( !isset( $_SESSION[ 'wsBgRequest' ][ $requestKey ] ) ) {
- wfDebug( __METHOD__ . ' Error:could not find upload session' );
- exit();
+
+ foreach ( $this->reqHeaders as $name => $value ) {
+ $list[] = "$name: $value";
}
- // setup the global user from the session key we just inherited
- $wgUser = User::newFromSession();
- // grab the session data to setup the request:
- $sd =& $_SESSION['wsBgRequest'][$requestKey];
+ return $list;
+ }
- // update the wgServer var ( since cmd line thinks we are localhost
- // when we are really orgServer)
- if ( isset( $sd['orgServer'] ) && $sd['orgServer'] ) {
- $wgServer = $sd['orgServer'];
- }
- // close down the session so we can other http queries can get session
- // updates: (if not $wgSessionsInMemcached)
- if ( !$wgSessionsInMemcached ) {
- session_write_close();
+ /**
+ * Set a read callback to accept data read from the HTTP request.
+ * By default, data is appended to an internal buffer which can be
+ * retrieved through $req->getContent().
+ *
+ * To handle data as it comes in -- especially for large files that
+ * would not fit in memory -- you can instead set your own callback,
+ * in the form function($resource, $buffer) where the first parameter
+ * is the low-level resource being read (implementation specific),
+ * and the second parameter is the data buffer.
+ *
+ * You MUST return the number of bytes handled in the buffer; if fewer
+ * bytes are reported handled than were passed to you, the HTTP fetch
+ * will be aborted.
+ *
+ * @param $callback Callback
+ */
+ public function setCallback( $callback ) {
+ if ( !is_callable( $callback ) ) {
+ throw new MWException( 'Invalid MwHttpRequest callback' );
}
+ $this->callback = $callback;
+ }
- return $sd;
+ /**
+ * A generic callback to read the body of the response from a remote
+ * server.
+ *
+ * @param $fh handle
+ * @param $content String
+ * @return int
+ */
+ public function read( $fh, $content ) {
+ $this->content .= $content;
+ return strlen( $content );
}
/**
- * Update the session with the finished information.
- * @param $sessionID string
- * @param $requestKey string the HTTP request key
+ * Take care of whatever is necessary to perform the URI request.
+ *
+ * @return Status
*/
- private static function updateSession( $sessionID, $requestKey, $status ) {
+ public function execute() {
+ global $wgTitle;
+
+ $this->content = "";
- if ( session_start() === false ) {
- wfDebug( __METHOD__ . ' ERROR:: Could not start session' );
+ if ( strtoupper( $this->method ) == "HEAD" ) {
+ $this->headersOnly = true;
}
- $sd =& $_SESSION['wsBgRequest'][$requestKey];
- if ( !$status->isOK() ) {
- $sd['apiUploadResult'] = FormatJson::encode(
- array( 'error' => $status->getWikiText() )
- );
- } else {
- $sd['apiUploadResult'] = FormatJson::encode( $status->value );
+ if ( is_object( $wgTitle ) && !isset( $this->reqHeaders['Referer'] ) ) {
+ $this->setReferer( wfExpandUrl( $wgTitle->getFullURL(), PROTO_CURRENT ) );
}
- session_write_close();
+ $this->proxySetup(); // set up any proxy as needed
+
+ if ( !$this->callback ) {
+ $this->setCallback( array( $this, 'read' ) );
+ }
+
+ if ( !isset( $this->reqHeaders['User-Agent'] ) ) {
+ $this->setUserAgent( Http::userAgent() );
+ }
}
/**
- * Take care of the downloaded file
- * @param $sd array
- * @param $status Status
- *
- * @returns Status
+ * 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.
*/
- private static function doFauxRequest( $sd, $status ) {
- global $wgEnableWriteAPI;
-
- if ( $status->isOK() ) {
- $fauxReqData = $sd['mParams'];
-
- // Fix boolean parameters
- foreach ( $fauxReqData as $k => $v ) {
- if ( $v === false )
- unset( $fauxReqData[$k] );
+ protected function parseHeader() {
+ $lastname = "";
+
+ foreach ( $this->headerList as $header ) {
+ if ( preg_match( "#^HTTP/([0-9.]+) (.*)#", $header, $match ) ) {
+ $this->respVersion = $match[1];
+ $this->respStatus = $match[2];
+ } elseif ( preg_match( "#^[ \t]#", $header ) ) {
+ $last = count( $this->respHeaders[$lastname] ) - 1;
+ $this->respHeaders[$lastname][$last] .= "\r\n$header";
+ } elseif ( preg_match( "#^([^:]*):[\t ]*(.*)#", $header, $match ) ) {
+ $this->respHeaders[strtolower( $match[1] )][] = $match[2];
+ $lastname = strtolower( $match[1] );
}
+ }
- $fauxReqData['action'] = 'upload';
- $fauxReqData['format'] = 'json';
- $fauxReqData['internalhttpsession'] = $requestKey;
-
- // evil but no other clean way about it:
- $fauxReq = new FauxRequest( $fauxReqData, true );
- $processor = new ApiMain( $fauxReq, $wgEnableWriteAPI );
+ $this->parseCookies();
+ }
- // init the mUpload var for the $processor
- $processor->execute();
- $processor->getResult()->cleanUpUTF8();
- $printer = $processor->createPrinterByName( 'json' );
- $printer->initPrinter( false );
- ob_start();
- $printer->execute();
+ /**
+ * Sets HTTPRequest status member to a fatal value with the error
+ * message if the returned integer value of the status code was
+ * not successful (< 300) or a redirect (>=300 and < 400). (see
+ * RFC2616, section 10,
+ * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html for a
+ * list of status codes.)
+ */
+ protected function setStatus() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
- // the status updates runner will grab the result form the session:
- $status->value = ob_get_clean();
+ if ( (int)$this->respStatus > 399 ) {
+ list( $code, $message ) = explode( " ", $this->respStatus, 2 );
+ $this->status->fatal( "http-bad-status", $code, $message );
}
- return $status;
}
/**
- * Run a session based download.
+ * Get the integer value of the HTTP status code (e.g. 200 for "200 Ok")
+ * (see RFC2616, section 10, http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
+ * for a list of status codes.)
*
- * @param $sessionID string: the session id with the download details
- * @param $requestKey string: the key of the given upload session
- * (a given client could have started a few http uploads at once)
- */
- public static function doSessionIdDownload( $sessionID, $requestKey ) {
- global $wgAsyncHTTPTimeout;
-
- wfDebug( __METHOD__ . "\n\n doSessionIdDownload :\n\n" );
- $sd = self::recoverSession( $sessionID );
- $req = HttpRequest::factory( $sd['url'],
- array(
- 'targetFilePath' => $sd['targetFilePath'],
- 'requestKey' => $requestKey,
- 'timeout' => $wgAsyncHTTPTimeout,
- ) );
-
- // run the actual request .. (this can take some time)
- wfDebug( __METHOD__ . 'do Session Download :: ' . $sd['url'] . ' tf: ' .
- $sd['targetFilePath'] . "\n\n" );
- $status = $req->doRequest();
-
- self::updateSession( $sessionID, $requestKey,
- self::handleFauxResponse( $sd, $status ) );
+ * @return Integer
+ */
+ public function getStatus() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ return (int)$this->respStatus;
}
-}
-/**
- * This wrapper class will call out to curl (if available) or fallback
- * to regular PHP if necessary for handling internal HTTP requests.
- */
-class HttpRequest {
- private $targetFilePath;
- private $requestKey;
- protected $content;
- protected $timeout = 'default';
- protected $headersOnly = null;
- protected $postdata = null;
- protected $proxy = null;
- protected $no_proxy = false;
- protected $sslVerifyHost = true;
- protected $caInfo = null;
- protected $method = "GET";
- protected $url;
- public $status;
/**
- * @param $url string url to use
- * @param $options array (optional) extra params to pass
- * Possible keys for the array:
- * method
- * timeout
- * targetFilePath
- * requestKey
- * headersOnly
- * postdata
- * proxy
- * no_proxy
- * sslVerifyHost
- * caInfo
- */
- function __construct( $url = null, $opt ) {
- global $wgHTTPTimeout;
-
- $this->url = $url;
-
- if ( !ini_get( 'allow_url_fopen' ) ) {
- $this->status = Status::newFatal( 'allow_url_fopen needs to be enabled for http copy to work' );
- } elseif ( !Http::isValidURI( $this->url ) ) {
- $this->status = Status::newFatal( 'bad-url' );
- } else {
- $this->status = Status::newGood( 100 ); // continue
+ * Returns true if the last status code was a redirect.
+ *
+ * @return Boolean
+ */
+ public function isRedirect() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
}
- if ( array_key_exists( 'timeout', $opt ) && $opt['timeout'] != 'default' ) {
- $this->timeout = $opt['timeout'];
- } else {
- $this->timeout = $wgHTTPTimeout;
- }
+ $status = (int)$this->respStatus;
- $members = array( "targetFilePath", "requestKey", "headersOnly", "postdata",
- "proxy", "no_proxy", "sslVerifyHost", "caInfo", "method" );
- foreach ( $members as $o ) {
- if ( array_key_exists( $o, $opt ) ) {
- $this->$o = $opt[$o];
- }
+ if ( $status >= 300 && $status <= 303 ) {
+ return true;
}
- if ( is_array( $this->postdata ) ) {
- $this->postdata = wfArrayToCGI( $this->postdata );
- }
+ return false;
}
/**
- * For backwards compatibility, we provide a __toString method so
- * that any code that expects a string result from Http::Get()
- * will see the content of the request.
+ * 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
*/
- function __toString() {
- return $this->content;
+ public function getResponseHeaders() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ return $this->respHeaders;
}
/**
- * Generate a new request object
- * @see HttpRequest::__construct
+ * Returns the value of the given response header.
+ *
+ * @param $header String
+ * @return String
*/
- public static function factory( $url, $opt ) {
- global $wgForceHTTPEngine;
+ public function getResponseHeader( $header ) {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
- if ( function_exists( 'curl_init' ) && $wgForceHTTPEngine == "curl" ) {
- return new CurlHttpRequest( $url, $opt );
- } else {
- return new PhpHttpRequest( $url, $opt );
+ if ( isset( $this->respHeaders[strtolower ( $header ) ] ) ) {
+ $v = $this->respHeaders[strtolower ( $header ) ];
+ return $v[count( $v ) - 1];
}
+
+ return null;
}
- public function getContent() {
- return $this->content;
+ /**
+ * Tells the MWHttpRequest object to use this pre-loaded CookieJar.
+ *
+ * @param $jar CookieJar
+ */
+ public function setCookieJar( $jar ) {
+ $this->cookieJar = $jar;
}
- public function handleOutput() {
- // if we wrote to a target file close up or return error
- if ( $this->targetFilePath ) {
- $this->writer->close();
- if ( !$this->writer->status->isOK() ) {
- $this->status = $this->writer->status;
- return $this->status;
- }
+ /**
+ * Returns the cookie jar in use.
+ *
+ * @return CookieJar
+ */
+ public function getCookieJar() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
}
- }
- public function doRequest() {
- global $wgTitle;
+ return $this->cookieJar;
+ }
- if ( !$this->status->isOK() ) {
- return $this->status;
+ /**
+ * Sets a cookie. Used before a request to set up any individual
+ * cookies. Used internally after a request to parse the
+ * Set-Cookie headers.
+ * @see Cookie::set
+ * @param $name
+ * @param $value null
+ * @param $attr null
+ */
+ public function setCookie( $name, $value = null, $attr = null ) {
+ if ( !$this->cookieJar ) {
+ $this->cookieJar = new CookieJar;
}
- $this->initRequest();
+ $this->cookieJar->setCookie( $name, $value, $attr );
+ }
- if ( !$this->no_proxy ) {
- $this->proxySetup();
+ /**
+ * Parse the cookies in the response headers and store them in the cookie jar.
+ */
+ protected function parseCookies() {
+ if ( !$this->cookieJar ) {
+ $this->cookieJar = new CookieJar;
}
- # Set the referer to $wgTitle, even in command-line mode
- # This is useful for interwiki transclusion, where the foreign
- # server wants to know what the referring page is.
- # $_SERVER['REQUEST_URI'] gives a less reliable indication of the
- # referring page.
- if ( is_object( $wgTitle ) ) {
- $this->set_referer( $wgTitle->getFullURL() );
+ if ( isset( $this->respHeaders['set-cookie'] ) ) {
+ $url = parse_url( $this->getFinalUrl() );
+ foreach ( $this->respHeaders['set-cookie'] as $cookie ) {
+ $this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
+ }
}
+ }
- $this->setupOutputHandler();
-
- if ( $this->status->isOK() ) {
- $this->spinTheWheel();
- }
+ /**
+ * Returns the final URL after all redirections.
+ *
+ * Relative values of the "Location" header are incorrect as stated in RFC, however they do happen and modern browsers support them.
+ * This function loops backwards through all locations in order to build the proper absolute URI - Marooned at wikia-inc.com
+ *
+ * Note that the multiple Location: headers are an artifact of CURL -- they
+ * shouldn't actually get returned this way. Rewrite this when bug 29232 is
+ * taken care of (high-level redirect handling rewrite).
+ *
+ * @return string
+ */
+ public function getFinalUrl() {
+ $headers = $this->getResponseHeaders();
+
+ //return full url (fix for incorrect but handled relative location)
+ if ( isset( $headers[ 'location' ] ) ) {
+ $locations = $headers[ 'location' ];
+ $domain = '';
+ $foundRelativeURI = false;
+ $countLocations = count($locations);
+
+ for ( $i = $countLocations - 1; $i >= 0; $i-- ) {
+ $url = parse_url( $locations[ $i ] );
+
+ if ( isset($url[ 'host' ]) ) {
+ $domain = $url[ 'scheme' ] . '://' . $url[ 'host' ];
+ break; //found correct URI (with host)
+ } else {
+ $foundRelativeURI = true;
+ }
+ }
- if ( !$this->status->isOK() ) {
- return $this->status;
+ if ( $foundRelativeURI ) {
+ if ( $domain ) {
+ return $domain . $locations[ $countLocations - 1 ];
+ } else {
+ $url = parse_url( $this->url );
+ if ( isset($url[ 'host' ]) ) {
+ return $url[ 'scheme' ] . '://' . $url[ 'host' ] . $locations[ $countLocations - 1 ];
+ }
+ }
+ } else {
+ return $locations[ $countLocations - 1 ];
+ }
}
- $this->handleOutput();
-
- $this->finish();
- return $this->status;
+ return $this->url;
}
- public function setupOutputHandler() {
- if ( $this->targetFilePath ) {
- $this->writer = new SimpleFileWriter( $this->targetFilePath,
- $this->requestKey );
- if ( !$this->writer->status->isOK() ) {
- wfDebug( __METHOD__ . "ERROR in setting up SimpleFileWriter\n" );
- $this->status = $this->writer->status;
- return $this->status;
- }
- $this->setCallback( array( $this, 'readAndSave' ) );
- } else {
- $this->setCallback( array( $this, 'readOnly' ) );
- }
+ /**
+ * Returns true if the backend can follow redirects. Overridden by the
+ * child classes.
+ * @return bool
+ */
+ public function canFollowRedirects() {
+ return true;
}
}
/**
- * HttpRequest implemented using internal curl compiled into PHP
+ * MWHttpRequest implemented using internal curl compiled into PHP
*/
-class CurlHttpRequest extends HttpRequest {
- private $c;
+class CurlHttpRequest extends MWHttpRequest {
+ const SUPPORTS_FILE_POSTS = true;
- public function initRequest() {
- $this->c = curl_init( $this->url );
+ static $curlMessageMap = array(
+ 6 => 'http-host-unreachable',
+ 28 => 'http-timed-out'
+ );
+
+ protected $curlOptions = array();
+ protected $headerText = "";
+
+ /**
+ * @param $fh
+ * @param $content
+ * @return int
+ */
+ protected function readHeader( $fh, $content ) {
+ $this->headerText .= $content;
+ return strlen( $content );
}
- public function proxySetup() {
- global $wgHTTPProxy;
+ public function execute() {
+ parent::execute();
- if ( is_string( $this->proxy ) ) {
- curl_setopt( $this->c, CURLOPT_PROXY, $this->proxy );
- } else if ( Http::isLocalURL( $this->url ) ) { /* Not sure this makes any sense. */
- curl_setopt( $this->c, CURLOPT_PROXY, 'localhost:80' );
- } else if ( $wgHTTPProxy ) {
- curl_setopt( $this->c, CURLOPT_PROXY, $wgHTTPProxy );
+ if ( !$this->status->isOK() ) {
+ return $this->status;
}
- }
- public function setCallback( $cb ) {
- curl_setopt( $this->c, CURLOPT_WRITEFUNCTION, $cb );
- }
+ $this->curlOptions[CURLOPT_PROXY] = $this->proxy;
+ $this->curlOptions[CURLOPT_TIMEOUT] = $this->timeout;
+ $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_MAXREDIRS] = $this->maxRedirects;
+ $this->curlOptions[CURLOPT_ENCODING] = ""; # Enable compression
- public function spinTheWheel() {
- curl_setopt( $this->c, CURLOPT_TIMEOUT, $this->timeout );
- curl_setopt( $this->c, CURLOPT_USERAGENT, Http::userAgent() );
- curl_setopt( $this->c, CURLOPT_HTTP_VERSION, CURL_HTTP_VERSION_1_0 );
+ /* not sure these two are actually necessary */
+ if ( isset( $this->reqHeaders['Referer'] ) ) {
+ $this->curlOptions[CURLOPT_REFERER] = $this->reqHeaders['Referer'];
+ }
+ $this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
- if ( $this->sslVerifyHost ) {
- curl_setopt( $this->c, CURLOPT_SSL_VERIFYHOST, $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 ) {
- curl_setopt( $this->c, CURLOPT_CAINFO, $this->caInfo );
+ $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
}
if ( $this->headersOnly ) {
- curl_setopt( $this->c, CURLOPT_NOBODY, true );
- curl_setopt( $this->c, CURLOPT_HEADER, true );
+ $this->curlOptions[CURLOPT_NOBODY] = true;
+ $this->curlOptions[CURLOPT_HEADER] = true;
} elseif ( $this->method == 'POST' ) {
- curl_setopt( $this->c, CURLOPT_POST, true );
- curl_setopt( $this->c, CURLOPT_POSTFIELDS, $this->postdata );
+ $this->curlOptions[CURLOPT_POST] = true;
+ $this->curlOptions[CURLOPT_POSTFIELDS] = $this->postData;
// Suppress 'Expect: 100-continue' header, as some servers
// will reject it with a 417 and Curl won't auto retry
// with HTTP 1.0 fallback
- curl_setopt( $this->c, CURLOPT_HTTPHEADER, array( 'Expect:' ) );
+ $this->reqHeaders['Expect'] = '';
} else {
- curl_setopt( $this->c, CURLOPT_CUSTOMREQUEST, $this->method );
+ $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
}
- try {
- if ( false === curl_exec( $this->c ) ) {
- $error_txt = 'Error sending request: #' . curl_errno( $this->c ) . ' ' . curl_error( $this->c );
- wfDebug( __METHOD__ . $error_txt . "\n" );
- $this->status->fatal( $error_txt ); /* i18n? */
- }
- } catch ( Exception $e ) {
- $errno = curl_errno( $this->c );
- if ( $errno != CURLE_OK ) {
- $errstr = curl_error( $this->c );
- wfDebug( __METHOD__ . ": CURL error code $errno: $errstr\n" );
- $this->status->fatal( "CURL error code $errno: $errstr\n" ); /* i18n? */
- }
- }
- }
+ $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
- public function readOnly( $curlH, $content ) {
- $this->content .= $content;
- return strlen( $content );
- }
-
- public function readAndSave( $curlH, $content ) {
- return $this->writer->write( $content );
- }
+ $curlHandle = curl_init( $this->url );
- public function getCode() {
- # Don't return truncated output
- $code = curl_getinfo( $this->c, CURLINFO_HTTP_CODE );
- if ( $code < 400 ) {
- $this->status->setResult( true, $code );
- } else {
- $this->status->setResult( false, $code );
+ if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
+ throw new MWException( "Error setting curl options." );
}
- }
-
- public function finish() {
- curl_close( $this->c );
- }
-
-}
-class PhpHttpRequest extends HttpRequest {
- private $reqHeaders;
- private $callback;
- private $fh;
-
- public function initRequest() {
- $this->reqHeaders[] = "User-Agent: " . Http::userAgent();
- $this->reqHeaders[] = "Accept: */*";
- if ( $this->method == 'POST' ) {
- // Required for HTTP 1.0 POSTs
- $this->reqHeaders[] = "Content-Length: " . strlen( $this->postdata );
- $this->reqHeaders[] = "Content-type: application/x-www-form-urlencoded";
+ if ( $this->followRedirects && $this->canFollowRedirects() ) {
+ wfSuppressWarnings();
+ 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
+ }
+ wfRestoreWarnings();
}
- }
- public function proxySetup() {
- global $wgHTTPProxy;
+ if ( false === curl_exec( $curlHandle ) ) {
+ $code = curl_error( $curlHandle );
- if ( $this->proxy ) {
- $this->proxy = 'tcp://' . $this->proxy;
- } elseif ( Http::isLocalURL( $this->url ) ) {
- $this->proxy = 'tcp://localhost:80';
- } elseif ( $wgHTTPProxy ) {
- $this->proxy = 'tcp://' . $wgHTTPProxy ;
+ if ( isset( self::$curlMessageMap[$code] ) ) {
+ $this->status->fatal( self::$curlMessageMap[$code] );
+ } else {
+ $this->status->fatal( 'http-curl-error', curl_error( $curlHandle ) );
+ }
+ } else {
+ $this->headerList = explode( "\r\n", $this->headerText );
}
- }
- public function setReferrer( $url ) {
- $this->reqHeaders[] = "Referer: $url";
- }
+ curl_close( $curlHandle );
+
+ $this->parseHeader();
+ $this->setStatus();
- public function setCallback( $cb ) {
- $this->callback = $cb;
+ return $this->status;
}
- public function readOnly( $contents ) {
- if ( $this->headersOnly ) {
+ /**
+ * @return bool
+ */
+ public function canFollowRedirects() {
+ if ( strval( ini_get( 'open_basedir' ) ) !== '' || wfIniGetBool( 'safe_mode' ) ) {
+ wfDebug( "Cannot follow redirects in safe mode\n" );
return false;
}
- $this->content .= $contents;
- return strlen( $contents );
- }
-
- public function readAndSave( $contents ) {
- if ( $this->headersOnly ) {
+ if ( !defined( 'CURLOPT_REDIR_PROTOCOLS' ) ) {
+ wfDebug( "Cannot follow redirects with libcurl < 7.19.4 due to CVE-2009-0037\n" );
return false;
}
- return $this->writer->write( $content );
+
+ return true;
}
+}
+
+class PhpHttpRequest extends MWHttpRequest {
- public function finish() {
- fclose( $this->fh );
+ /**
+ * @param $url string
+ * @return string
+ */
+ protected function urlToTcp( $url ) {
+ $parsedUrl = parse_url( $url );
+
+ return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
}
- public function spinTheWheel() {
- $opts = array();
- if ( $this->proxy && !$this->no_proxy ) {
- $opts['proxy'] = $this->proxy;
- $opts['request_fulluri'] = true;
+ public function execute() {
+ parent::execute();
+
+ if ( is_array( $this->postData ) ) {
+ $this->postData = wfArrayToCGI( $this->postData );
}
- $opts['method'] = $this->method;
- $opts['timeout'] = $this->timeout;
- $opts['header'] = implode( "\r\n", $this->reqHeaders );
- if ( version_compare( "5.3.0", phpversion(), ">" ) ) {
- $opts['protocol_version'] = "1.0";
- } else {
- $opts['protocol_version'] = "1.1";
+ if ( $this->parsedUrl['scheme'] != 'http' &&
+ $this->parsedUrl['scheme'] != 'https' ) {
+ $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
}
- if ( $this->postdata ) {
- $opts['content'] = $this->postdata;
+ $this->reqHeaders['Accept'] = "*/*";
+ if ( $this->method == 'POST' ) {
+ // Required for HTTP 1.0 POSTs
+ $this->reqHeaders['Content-Length'] = strlen( $this->postData );
+ $this->reqHeaders['Content-type'] = "application/x-www-form-urlencoded";
}
- $context = stream_context_create( array( 'http' => $opts ) );
- $this->fh = fopen( $this->url, "r", false, $context );
- $result = stream_get_meta_data( $this->fh );
+ $options = array();
+ if ( $this->proxy ) {
+ $options['proxy'] = $this->urlToTCP( $this->proxy );
+ $options['request_fulluri'] = true;
+ }
- if ( $result['timed_out'] ) {
- $this->status->error( __CLASS__ . '::timed-out-in-headers' );
+ if ( !$this->followRedirects ) {
+ $options['max_redirects'] = 0;
+ } else {
+ $options['max_redirects'] = $this->maxRedirects;
}
- $this->headers = $result['wrapper_data'];
+ $options['method'] = $this->method;
+ $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";
- $end = false;
- $size = 8192;
- while ( !$end ) {
- $contents = fread( $this->fh, $size );
- $size = call_user_func( $this->callback, $contents );
- $end = ( $size == 0 ) || feof( $this->fh );
+ // 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;
}
- }
-}
-/**
- * SimpleFileWriter with session id updates
- */
-class SimpleFileWriter {
- private $targetFilePath = null;
- private $status = null;
- private $sessionId = null;
- private $sessionUpdateInterval = 0; // how often to update the session while downloading
- private $currentFileSize = 0;
- private $requestKey = null;
- private $prevTime = 0;
- private $fp = null;
+ $options['timeout'] = $this->timeout;
- /**
- * @param $targetFilePath string the path to write the file out to
- * @param $requestKey string the request to update
- */
- function __construct__( $targetFilePath, $requestKey ) {
- $this->targetFilePath = $targetFilePath;
- $this->requestKey = $requestKey;
- $this->status = Status::newGood();
- // open the file:
- $this->fp = fopen( $this->targetFilePath, 'w' );
- if ( $this->fp === false ) {
- $this->status = Status::newFatal( 'HTTP::could-not-open-file-for-writing' );
- }
- // true start time
- $this->prevTime = time();
- }
+ $context = stream_context_create( array( 'http' => $options ) );
- public function write( $dataPacket ) {
- global $wgMaxUploadSize, $wgLang;
+ $this->headerList = array();
+ $reqCount = 0;
+ $url = $this->url;
- if ( !$this->status->isOK() ) {
- return false;
- }
+ $result = array();
- // write out the content
- if ( fwrite( $this->fp, $dataPacket ) === false ) {
- wfDebug( __METHOD__ . " ::could-not-write-to-file\n" );
- $this->status = Status::newFatal( 'HTTP::could-not-write-to-file' );
- return false;
- }
+ do {
+ $reqCount++;
+ wfSuppressWarnings();
+ $fh = fopen( $url, "r", false, $context );
+ wfRestoreWarnings();
- // check file size:
- clearstatcache();
- $this->currentFileSize = filesize( $this->targetFilePath );
+ if ( !$fh ) {
+ break;
+ }
- if ( $this->currentFileSize > $wgMaxUploadSize ) {
- wfDebug( __METHOD__ . " ::http-download-too-large\n" );
- $this->status = Status::newFatal( 'HTTP::file-has-grown-beyond-upload-limit-killing: ' . /* i18n? */
- 'downloaded more than ' .
- $wgLang->formatSize( $wgMaxUploadSize ) . ' ' );
- return false;
- }
- // if more than session_update_interval second have passed updateProgress
- if ( $this->requestKey &&
- ( ( time() - $this->prevTime ) > $this->sessionUpdateInterval ) ) {
- $this->prevTime = time();
- $session_status = $this->updateProgress();
- if ( !$session_status->isOK() ) {
- $this->status = $session_status;
- wfDebug( __METHOD__ . ' update session failed or was canceled' );
- return false;
+ $result = stream_get_meta_data( $fh );
+ $this->headerList = $result['wrapper_data'];
+ $this->parseHeader();
+
+ if ( !$this->followRedirects ) {
+ break;
}
- }
- return strlen( $dataPacket );
- }
- public function updateProgress() {
- global $wgSessionsInMemcached;
+ # Handle manual redirection
+ if ( !$this->isRedirect() || $reqCount > $this->maxRedirects ) {
+ break;
+ }
+ # Check security of URL
+ $url = $this->getResponseHeader( "Location" );
- // start the session (if necessary)
- if ( !$wgSessionsInMemcached ) {
- wfSuppressWarnings();
- if ( session_start() === false ) {
- wfDebug( __METHOD__ . ' could not start session' );
- exit( 0 );
+ if ( !Http::isValidURI( $url ) ) {
+ wfDebug( __METHOD__ . ": insecure redirection\n" );
+ break;
}
- wfRestoreWarnings();
- }
- $sd =& $_SESSION['wsBgRequest'][ $this->requestKey ];
- // check if the user canceled the request:
- if ( $sd['userCancel'] ) {
- // @@todo kill the download
- return Status::newFatal( 'user-canceled-request' );
+ } while ( true );
+
+ $this->setStatus();
+
+ if ( $fh === false ) {
+ $this->status->fatal( 'http-request-error' );
+ return $this->status;
}
- // update the progress bytes download so far:
- $sd['loaded'] = $this->currentFileSize;
- // close down the session so we can other http queries can get session updates:
- if ( !$wgSessionsInMemcached )
- session_write_close();
+ if ( $result['timed_out'] ) {
+ $this->status->fatal( 'http-timed-out', $this->url );
+ return $this->status;
+ }
- return Status::newGood();
- }
+ // If everything went OK, or we recieved some error code
+ // get the response body content.
+ if ( $this->status->isOK()
+ || (int)$this->respStatus >= 300) {
+ while ( !feof( $fh ) ) {
+ $buf = fread( $fh, 8192 );
- public function close() {
- $this->updateProgress();
+ if ( $buf === false ) {
+ $this->status->fatal( 'http-read-error' );
+ break;
+ }
- // close up the file handle:
- if ( false === fclose( $this->fp ) ) {
- $this->status = Status::newFatal( 'HTTP::could-not-close-file' );
+ if ( strlen( $buf ) ) {
+ call_user_func( $this->callback, $fh, $buf );
+ }
+ }
}
- }
+ fclose( $fh );
+ return $this->status;
+ }
}
dépôts / lhc / web / wiklou.git / blobdiff