<?php
/**
- * HTTP handling class
* @defgroup HTTP HTTP
- * @file
- * @ingroup HTTP
*/
+/**
+ * Various HTTP related functions
+ * @ingroup HTTP
+ */
class Http {
- const SYNC_DOWNLOAD = 1; // syncronys upload (in a single request)
- const ASYNC_DOWNLOAD = 2; // asynchronous upload we should spawn out another process and monitor progress if possible)
+ static $httpEngine = false;
+
+ /**
+ * Perform an HTTP request
+ *
+ * @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.
+ * Will use $wgHTTPProxy (if set) otherwise.
+ * - noProxy Override $wgHTTPProxy (if set) and don't use any proxy at all.
+ * - sslVerifyHost (curl only) Verify hostname against certificate
+ * - sslVerifyCert (curl only) Verify SSL certificate
+ * - caInfo (curl only) Provide CA information
+ * - maxRedirects Maximum number of redirects to follow (defaults to 5)
+ * - followRedirects Whether to follow redirects (defaults to false).
+ * Note: this should only be used when the target URL is trusted,
+ * to avoid attacks on intranet services accessible by HTTP.
+ * - userAgent A user agent, if you want to override the default
+ * MediaWiki/$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';
+ }
- var $body = '';
+ $req = MWHttpRequest::factory( $url, $options );
+ if( isset( $options['userAgent'] ) ) {
+ $req->setUserAgent( $options['userAgent'] );
+ }
+ $status = $req->execute();
- public static function request( $method, $url, $opts = array() ){
- $opts['method'] = ( strtoupper( $method ) == 'GET' || strtoupper( $method ) == 'POST' ) ? strtoupper( $method ) : null;
- $req = new HttpRequest( $url, $opts );
- $status = $req->doRequest();
- if( $status->isOK() ){
- return $status->value;
+ if ( $status->isOK() ) {
+ return $req->getContent();
} else {
- wfDebug( 'http error: ' . $status->getWikiText() );
return false;
}
}
/**
* 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 = false ) {
- $opts = array();
- if( $timeout )
- $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 );
}
- public static function doDownload( $url, $target_file_path, $dl_mode = self::SYNC_DOWNLOAD, $redirectCount = 0 ){
- global $wgPhpCli, $wgMaxUploadSize, $wgMaxRedirects;
- // do a quick check to HEAD to insure the file size is not > $wgMaxUploadSize
- $headRequest = new HttpRequest( $url, array( 'headers_only' => true ) );
- $headResponse = $headRequest->doRequest();
- if( !$headResponse->isOK() ){
- return $headResponse;
+ /**
+ * Check if the URL can be served by localhost
+ *
+ * @param $url String: full url to check
+ * @return Boolean
+ */
+ public static function isLocalURL( $url ) {
+ global $wgCommandLineMode, $wgConf;
+
+ if ( $wgCommandLineMode ) {
+ return false;
}
- $head = $headResponse->value;
- // check for redirects:
- if( isset( $head['Location'] ) && strrpos( $head[0], '302' ) !== false ){
- if( $redirectCount < $wgMaxRedirects ){
- if( UploadFromUrl::isValidURI( $head['Location'] ) ){
- return self::doDownload( $head['Location'], $target_file_path, $dl_mode, $redirectCount++ );
+ // Extract host part
+ $matches = array();
+ if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
+ $host = $matches[1];
+ // Split up dotwise
+ $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 ) {
+ $domain = $domainPart;
} else {
- return Status::newFatal( 'upload-proto-error' );
+ $domain = $domainPart . '.' . $domain;
+ }
+
+ if ( $wgConf->isLocalVHost( $domain ) ) {
+ return true;
}
- } else {
- return Status::newFatal( 'upload-too-many-redirects' );
}
}
- // we did not get a 200 ok response:
- if( strrpos( $head[0], '200 OK' ) === false ){
- return Status::newFatal( 'upload-http-error', htmlspecialchars( $head[0] ) );
+
+ return false;
+ }
+
+ /**
+ * A standard user-agent we can use for external requests.
+ * @return String
+ */
+ public static function userAgent() {
+ global $wgVersion;
+ return "MediaWiki/$wgVersion";
+ }
+
+ /**
+ * 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
+ * @return Boolean
+ */
+ public static function isValidURI( $uri ) {
+ return preg_match(
+ '/^https?:\/\/[^\/\s]\S*$/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;
+
+ /**
+ * @var CookieJar
+ */
+ protected $cookieJar;
+
+ protected $headerList = array();
+ protected $respVersion = "0.9";
+ protected $respStatus = "200 Ok";
+ protected $respHeaders = array();
+
+ public $status;
+
+ /**
+ * @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( $options['timeout'] ) && $options['timeout'] != 'default' ) {
+ $this->timeout = $options['timeout'];
+ } else {
+ $this->timeout = $wgHTTPTimeout;
}
- $content_length = ( isset( $head['Content-Length'] ) ) ? $head['Content-Length'] : null;
- if( $content_length ){
- if( $content_length > $wgMaxUploadSize ){
- return Status::newFatal( 'requested file length ' . $content_length . ' 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( $wgPhpCli && wfShellExecEnabled() && $dl_mode == self::ASYNC_DOWNLOAD ){
- wfDebug( __METHOD__ . "\ASYNC_DOWNLOAD\n" );
- // setup session and shell call:
- return self::initBackgroundDownload( $url, $target_file_path, $content_length );
- } 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['target_file_path'] = $target_file_path;
- $req = new HttpRequest( $url, $opts );
- return $req->doRequest();
+ /**
+ * Simple function to test if we can make any sort of requests at all, using
+ * cURL or fopen()
+ * @return bool
+ */
+ public static function canMakeRequests() {
+ return function_exists( 'curl_init' ) || wfIniGetBool( 'allow_url_fopen' );
+ }
+
+ /**
+ * 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.' );
}
}
/**
- * a non blocking request (generally an exit point in the application)
- * should write to a file location and give updates
+ * Get the body, or content, of the response to the request
+ *
+ * @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
+ */
+ public function proxySetup() {
+ global $wgHTTPProxy;
+
+ if ( $this->proxy ) {
+ return;
+ }
+
+ if ( Http::isLocalURL( $this->url ) ) {
+ $this->proxy = '';
+ } elseif ( $wgHTTPProxy ) {
+ $this->proxy = $wgHTTPProxy ;
+ } elseif ( getenv( "http_proxy" ) ) {
+ $this->proxy = getenv( "http_proxy" );
+ }
+ }
+
+ /**
+ * Set the refererer header
+ */
+ public function setReferer( $url ) {
+ $this->setHeader( 'Referer', $url );
+ }
+
+ /**
+ * Set the user agent
+ * @param $UA string
*/
- private static function initBackgroundDownload( $url, $target_file_path, $content_length = null ){
- global $wgMaxUploadSize, $IP, $wgPhpCli, $wgServer;
- $status = Status::newGood();
+ public function setUserAgent( $UA ) {
+ $this->setHeader( 'User-Agent', $UA );
+ }
- // generate a session id with all the details for the download (pid, target_file_path )
- $upload_session_key = self::getUploadSessionKey();
- $session_id = session_id();
+ /**
+ * Set an arbitrary header
+ * @param $name
+ * @param $value
+ */
+ public function setHeader( $name, $value ) {
+ // I feel like I should normalize the case here...
+ $this->reqHeaders[$name] = $value;
+ }
- // store the url and target path:
- $_SESSION['wsDownload'][$upload_session_key]['url'] = $url;
- $_SESSION['wsDownload'][$upload_session_key]['target_file_path'] = $target_file_path;
- // since we request from the cmd line we lose the original host name pass in the session:
- $_SESSION['wsDownload'][$upload_session_key]['orgServer'] = $wgServer;
+ /**
+ * Get an array of the headers
+ * @return array
+ */
+ public function getHeaderList() {
+ $list = array();
- if( $content_length )
- $_SESSION['wsDownload'][$upload_session_key]['content_length'] = $content_length;
+ if ( $this->cookieJar ) {
+ $this->reqHeaders['Cookie'] =
+ $this->cookieJar->serializeToHttpRequest(
+ $this->parsedUrl['path'],
+ $this->parsedUrl['host']
+ );
+ }
- // set initial loaded bytes:
- $_SESSION['wsDownload'][$upload_session_key]['loaded'] = 0;
+ foreach ( $this->reqHeaders as $name => $value ) {
+ $list[] = "$name: $value";
+ }
- // run the background download request:
- $cmd = $wgPhpCli . ' ' . $IP . "/maintenance/http_session_download.php --sid {$session_id} --usk {$upload_session_key}";
- $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( 'could not run background shell exec' );
+ return $list;
+ }
- // update the status value with the $upload_session_key (for the user to check on the status of the upload)
- $status->value = $upload_session_key;
+ /**
+ * 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 good status
- return $status;
+ /**
+ * A generic callback to read the body of the response from a remote
+ * server.
+ *
+ * @param $fh handle
+ * @param $content String
+ */
+ public function read( $fh, $content ) {
+ $this->content .= $content;
+ return strlen( $content );
}
- static function getUploadSessionKey(){
- $key = mt_rand( 0, 0x7fffffff );
- $_SESSION['wsUploadData'][$key] = array();
- return $key;
+ /**
+ * 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;
+ }
+
+ if ( is_object( $wgTitle ) && !isset( $this->reqHeaders['Referer'] ) ) {
+ $this->setReferer( wfExpandUrl( $wgTitle->getFullURL(), PROTO_CURRENT ) );
+ }
+
+ if ( !$this->noProxy ) {
+ $this->proxySetup();
+ }
+
+ if ( !$this->callback ) {
+ $this->setCallback( array( $this, 'read' ) );
+ }
+
+ if ( !isset( $this->reqHeaders['User-Agent'] ) ) {
+ $this->setUserAgent( Http::userAgent() );
+ }
}
/**
- * used to run a session based download. Is initiated via the shell.
+ * 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.
*
- * @param $session_id String: the session id to grab download details from
- * @param $upload_session_key String: the key of the given upload session
- * (a given client could have started a few http uploads at once)
- */
- public static function doSessionIdDownload( $session_id, $upload_session_key ){
- global $wgUser, $wgEnableWriteAPI, $wgAsyncHTTPTimeout, $wgServer;
- wfDebug( __METHOD__ . "\n\ndoSessionIdDownload:\n\n" );
-
- // set session to the provided key:
- session_id( $session_id );
- // start the session
- if( session_start() === false ){
- wfDebug( __METHOD__ . ' could not start session' );
- }
- // get all the vars we need from session_id
- if( !isset( $_SESSION[ 'wsDownload' ][$upload_session_key] ) ){
- wfDebug( __METHOD__ . ' Error:could not find upload session');
- exit();
- }
- // setup the global user from the session key we just inherited
- $wgUser = User::newFromSession();
-
- // grab the session data to setup the request:
- $sd =& $_SESSION['wsDownload'][$upload_session_key];
-
- // 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:
- session_write_close();
-
- $req = new HttpRequest( $sd['url'], array(
- 'target_file_path' => $sd['target_file_path'],
- 'upload_session_key'=> $upload_session_key,
- 'timeout' => $wgAsyncHTTPTimeout,
- 'do_close_session_update' => true
- ) );
- // run the actual request .. (this can take some time)
- wfDebug( __METHOD__ . 'do Request: ' . $sd['url'] . ' tf: ' . $sd['target_file_path'] );
- $status = $req->doRequest();
- //wfDebug("done with req status is: ". $status->isOK(). ' '.$status->getWikiText(). "\n");
-
- // start up the session again:
- if( session_start() === false ){
- wfDebug( __METHOD__ . ' ERROR:: Could not start session');
- }
- // grab the updated session data pointer
- $sd =& $_SESSION['wsDownload'][$upload_session_key];
- // if error update status:
- if( !$status->isOK() ){
- $sd['apiUploadResult'] = ApiFormatJson::getJsonEncode(
- array( 'error' => $status->getWikiText() )
- );
- }
- // if status okay process upload using fauxReq to api:
- if( $status->isOK() ){
- // setup the FauxRequest
- $fauxReqData = $sd['mParams'];
-
- // Fix boolean parameters
- foreach( $fauxReqData as $k => $v ) {
- if( $v === false )
- unset( $fauxReqData[$k] );
+ * @return nothing
+ */
+ 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'] = $upload_session_key;
- // evil but no other clean way about it:
- $faxReq = new FauxRequest( $fauxReqData, true );
- $processor = new ApiMain( $faxReq, $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();
- $apiUploadResult = ob_get_clean();
+ /**
+ * Sets HTTPRequest status member to a fatal value with the error
+ * message if the returned integer value of the status code was
+ * not successful (< 300) or a redirect (>=300 and < 400). (see
+ * RFC2616, section 10,
+ * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html for a
+ * list of status codes.)
+ *
+ * @return nothing
+ */
+ protected function setStatus() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
- wfDebug( __METHOD__ . "\n\n got api result:: $apiUploadResult \n" );
- // the status updates runner will grab the result form the session:
- $sd['apiUploadResult'] = $apiUploadResult;
+ if ( (int)$this->respStatus > 399 ) {
+ list( $code, $message ) = explode( " ", $this->respStatus, 2 );
+ $this->status->fatal( "http-bad-status", $code, $message );
}
- // close the session:
- session_write_close();
}
/**
- * Check if the URL can be served by localhost
- * @param $url string Full url to check
- * @return bool
+ * 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.)
+ *
+ * @return Integer
*/
- public static function isLocalURL( $url ) {
- global $wgCommandLineMode, $wgConf;
- if ( $wgCommandLineMode ) {
- return false;
+ public function getStatus() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
}
- // Extract host part
- $matches = array();
- if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
- $host = $matches[1];
- // Split up dotwise
- $domainParts = explode( '.', $host );
- // Check if this domain or any superdomain is listed in $wgConf as a local virtual host
- $domainParts = array_reverse( $domainParts );
- for ( $i = 0; $i < count( $domainParts ); $i++ ) {
- $domainPart = $domainParts[$i];
- if ( $i == 0 ) {
- $domain = $domainPart;
+ return (int)$this->respStatus;
+ }
+
+
+ /**
+ * 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 && $status <= 303 ) {
+ return true;
+ }
+
+ return false;
+ }
+
+ /**
+ * Returns an associative array of response headers after the
+ * request has been executed. Because some headers
+ * (e.g. Set-Cookie) can appear more than once the, each value of
+ * the associative array is an array of the values given.
+ *
+ * @return Array
+ */
+ public function getResponseHeaders() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ 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 MWHttpRequest object to use this pre-loaded CookieJar.
+ *
+ * @param $jar CookieJar
+ */
+ public function setCookieJar( $jar ) {
+ $this->cookieJar = $jar;
+ }
+
+ /**
+ * Returns the cookie jar in use.
+ *
+ * @return CookieJar
+ */
+ public function getCookieJar() {
+ if ( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ return $this->cookieJar;
+ }
+
+ /**
+ * 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->cookieJar->setCookie( $name, $value, $attr );
+ }
+
+ /**
+ * Parse the cookies in the response headers and store them in the cookie jar.
+ */
+ protected function parseCookies() {
+ if ( !$this->cookieJar ) {
+ $this->cookieJar = new CookieJar;
+ }
+
+ if ( isset( $this->respHeaders['set-cookie'] ) ) {
+ $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.
+ *
+ * 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 {
- $domain = $domainPart . '.' . $domain;
+ $foundRelativeURI = true;
}
- if ( $wgConf->isLocalVHost( $domain ) ) {
- return true;
+ }
+
+ 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 ];
}
}
- return false;
+
+ return $this->url;
}
/**
- * Return a standard user-agent we can use for external requests.
+ * Returns true if the backend can follow redirects. Overridden by the
+ * child classes.
+ * @return bool
*/
- public static function userAgent() {
- global $wgVersion;
- return "MediaWiki/$wgVersion";
+ public function canFollowRedirects() {
+ return true;
}
}
-class HttpRequest {
- var $target_file_path;
- var $upload_session_key;
+/**
+ * MWHttpRequest implemented using internal curl compiled into PHP
+ */
+class CurlHttpRequest extends MWHttpRequest {
+ const SUPPORTS_FILE_POSTS = true;
- function __construct( $url, $opt ){
- global $wgSyncHTTPTimeout;
- // double check that it's a valid url:
- $this->url = $url;
+ static $curlMessageMap = array(
+ 6 => 'http-host-unreachable',
+ 28 => 'http-timed-out'
+ );
- // set the timeout to default sync timeout (unless the timeout option is provided)
- $this->timeout = ( isset( $opt['timeout'] ) ) ? $opt['timeout'] : $wgSyncHTTPTimeout;
- $this->method = ( isset( $opt['method'] ) ) ? $opt['method'] : 'GET';
- $this->target_file_path = ( isset( $opt['target_file_path'] ) ) ? $opt['target_file_path'] : false;
- $this->upload_session_key = ( isset( $opt['upload_session_key'] ) ) ? $opt['upload_session_key'] : false;
- $this->headers_only = ( isset( $opt['headers_only'] ) ) ? $opt['headers_only'] : false;
- $this->do_close_session_update = isset( $opt['do_close_session_update'] );
- }
+ protected $curlOptions = array();
+ protected $headerText = "";
/**
- * Get the contents of a file by HTTP
- * @param $url string Full URL to act on
- * @param $Opt associative array Optional array of options:
- * 'method' => 'GET', 'POST' etc.
- * 'target_file_path' => if curl should output to a target file
- * 'adapter' => 'curl', 'soket'
+ * @param $fh
+ * @param $content
+ * @return int
*/
- public function doRequest() {
- # Make sure we have a valid url
- if( !UploadFromUrl::isValidURI( $this->url ) )
- return Status::newFatal('bad-url');
+ protected function readHeader( $fh, $content ) {
+ $this->headerText .= $content;
+ return strlen( $content );
+ }
- # Use curl if available
- if ( function_exists( 'curl_init' ) ) {
- return $this->doCurlReq();
- } else {
- return $this->doPhpReq();
+ public function execute() {
+ parent::execute();
+
+ if ( !$this->status->isOK() ) {
+ return $this->status;
}
- }
- private function doCurlReq(){
- global $wgHTTPProxy, $wgTitle;
+ $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
- $status = Status::newGood();
- $c = curl_init( $this->url );
+ /* 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'];
- // proxy setup:
- if ( Http::isLocalURL( $this->url ) ) {
- curl_setopt( $c, CURLOPT_PROXY, 'localhost:80' );
- } else if ( $wgHTTPProxy ) {
- curl_setopt( $c, CURLOPT_PROXY, $wgHTTPProxy );
+ if ( isset( $this->sslVerifyHost ) ) {
+ $this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost;
+ }
+
+ if ( isset( $this->sslVerifyCert ) ) {
+ $this->curlOptions[CURLOPT_SSL_VERIFYPEER] = $this->sslVerifyCert;
}
- curl_setopt( $c, CURLOPT_TIMEOUT, $this->timeout );
- curl_setopt( $c, CURLOPT_USERAGENT, Http::userAgent() );
+ if ( $this->caInfo ) {
+ $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
+ }
- if ( $this->headers_only ) {
- curl_setopt( $c, CURLOPT_NOBODY, true );
- curl_setopt( $c, CURLOPT_HEADER, true );
+ if ( $this->headersOnly ) {
+ $this->curlOptions[CURLOPT_NOBODY] = true;
+ $this->curlOptions[CURLOPT_HEADER] = true;
} elseif ( $this->method == 'POST' ) {
- curl_setopt( $c, CURLOPT_POST, true );
- curl_setopt( $c, CURLOPT_POSTFIELDS, '' );
+ $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
+ $this->reqHeaders['Expect'] = '';
} else {
- curl_setopt( $c, CURLOPT_CUSTOMREQUEST, $this->method );
- }
-
- # 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 ) ) {
- curl_setopt( $c, CURLOPT_REFERER, $wgTitle->getFullURL() );
- }
-
- // set the write back function (if we are writing to a file)
- if( $this->target_file_path ){
- $cwrite = new simpleFileWriter( $this->target_file_path,
- $this->upload_session_key,
- $this->do_close_session_update
- );
- if( !$cwrite->status->isOK() ){
- wfDebug( __METHOD__ . "ERROR in setting up simpleFileWriter\n" );
- $status = $cwrite->status;
- return $status;
- }
- curl_setopt( $c, CURLOPT_WRITEFUNCTION, array( $cwrite, 'callbackWriteBody' ) );
+ $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
}
- // start output grabber:
- if( !$this->target_file_path )
- ob_start();
+ $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
- //run the actual curl_exec:
- try {
- if ( false === curl_exec( $c ) ) {
- $error_txt ='Error sending request: #' . curl_errno( $c ) .' '. curl_error( $c );
- wfDebug( __METHOD__ . $error_txt . "\n" );
- $status = Status::newFatal( $error_txt );
- }
- } catch ( Exception $e ) {
- // do something with curl exec error?
- }
- // if direct request output the results to the stats value:
- if( !$this->target_file_path && $status->isOK() ){
- $status->value = ob_get_contents();
- ob_end_clean();
- }
- // if we wrote to a target file close up or return error
- if( $this->target_file_path ){
- $cwrite->close();
- if( !$cwrite->status->isOK() ){
- return $cwrite->status;
+ $curlHandle = curl_init( $this->url );
+
+ if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
+ throw new MWException( "Error setting curl options." );
+ }
+
+ 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();
}
- if ( $this->headers_only ) {
- $headers = explode( "\n", $status->value );
- $headerArray = array();
- foreach ( $headers as $header ) {
- if ( !strlen( trim( $header ) ) )
- continue;
- $headerParts = explode( ':', $header, 2 );
- if ( count( $headerParts ) == 1 ) {
- $headerArray[] = trim( $header );
- } else {
- list( $key, $val ) = $headerParts;
- $headerArray[trim( $key )] = trim( $val );
- }
+ if ( false === curl_exec( $curlHandle ) ) {
+ $code = curl_error( $curlHandle );
+
+ if ( isset( self::$curlMessageMap[$code] ) ) {
+ $this->status->fatal( self::$curlMessageMap[$code] );
+ } else {
+ $this->status->fatal( 'http-curl-error', curl_error( $curlHandle ) );
}
- $status->value = $headerArray;
} else {
- # Don't return the text of error messages, return false on error
- $retcode = curl_getinfo( $c, CURLINFO_HTTP_CODE );
- if ( $retcode != 200 ) {
- wfDebug( __METHOD__ . ": HTTP return code $retcode\n" );
- $status = Status::newFatal( "HTTP return code $retcode\n" );
- }
- # Don't return truncated output
- $errno = curl_errno( $c );
- if ( $errno != CURLE_OK ) {
- $errstr = curl_error( $c );
- wfDebug( __METHOD__ . ": CURL error code $errno: $errstr\n" );
- $status = Status::newFatal( " CURL error code $errno: $errstr\n" );
- }
+ $this->headerList = explode( "\r\n", $this->headerText );
}
- curl_close( $c );
+ curl_close( $curlHandle );
+
+ $this->parseHeader();
+ $this->setStatus();
- // return the result obj
- return $status;
+ return $this->status;
}
- public function doPhpReq(){
- global $wgTitle, $wgHTTPProxy;
+ /**
+ * @return bool
+ */
+ public function canFollowRedirects() {
+ if ( strval( ini_get( 'open_basedir' ) ) !== '' || wfIniGetBool( 'safe_mode' ) ) {
+ wfDebug( "Cannot follow redirects in safe mode\n" );
+ return false;
+ }
- # Check for php.ini allow_url_fopen
- if( !ini_get( 'allow_url_fopen' ) ){
- return Status::newFatal( 'allow_url_fopen needs to be enabled for http copy to work' );
+ if ( !defined( 'CURLOPT_REDIR_PROTOCOLS' ) ) {
+ wfDebug( "Cannot follow redirects with libcurl < 7.19.4 due to CVE-2009-0037\n" );
+ return false;
}
- // start with good status:
- $status = Status::newGood();
+ return true;
+ }
+}
+
+class PhpHttpRequest extends MWHttpRequest {
+
+ /**
+ * @param $url string
+ * @return string
+ */
+ protected function urlToTcp( $url ) {
+ $parsedUrl = parse_url( $url );
+
+ return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
+ }
+
+ public function execute() {
+ parent::execute();
- if ( $this->headers_only ) {
- $status->value = get_headers( $this->url, 1 );
- return $status;
+ if ( is_array( $this->postData ) ) {
+ $this->postData = wfArrayToCGI( $this->postData );
}
- // setup the headers
- $headers = array( "User-Agent: " . Http::userAgent() );
- if ( is_object( $wgTitle ) ) {
- $headers[] = "Referer: ". $wgTitle->getFullURL();
+ if ( $this->parsedUrl['scheme'] != 'http' &&
+ $this->parsedUrl['scheme'] != 'https' ) {
+ $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
}
- if( strcasecmp( $this->method, 'post' ) == 0 ) {
+ $this->reqHeaders['Accept'] = "*/*";
+ if ( $this->method == 'POST' ) {
// Required for HTTP 1.0 POSTs
- $headers[] = "Content-Length: 0";
- }
- $fcontext = stream_context_create ( array(
- 'http' => array(
- 'method' => $this->method,
- 'header' => implode( "\r\n", $headers ),
- 'timeout' => $this->timeout )
- )
- );
+ $this->reqHeaders['Content-Length'] = strlen( $this->postData );
+ $this->reqHeaders['Content-type'] = "application/x-www-form-urlencoded";
+ }
+
+ $options = array();
+ if ( $this->proxy && !$this->noProxy ) {
+ $options['proxy'] = $this->urlToTCP( $this->proxy );
+ $options['request_fulluri'] = true;
+ }
+
+ if ( !$this->followRedirects ) {
+ $options['max_redirects'] = 0;
+ } else {
+ $options['max_redirects'] = $this->maxRedirects;
+ }
- $fh = fopen( $this->url, "r", false, $fcontext);
+ $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";
- // set the write back function (if we are writing to a file)
- if( $this->target_file_path ){
- $cwrite = new simpleFileWriter( $this->target_file_path, $this->upload_session_key, $this->do_close_session_update );
- if( !$cwrite->status->isOK() ){
- wfDebug( __METHOD__ . "ERROR in setting up simpleFileWriter\n" );
- $status = $cwrite->status;
- return $status;
+ // 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;
+ }
+
+ $options['timeout'] = $this->timeout;
+
+ $context = stream_context_create( array( 'http' => $options ) );
+
+ $this->headerList = array();
+ $reqCount = 0;
+ $url = $this->url;
+
+ $result = array();
+
+ do {
+ $reqCount++;
+ wfSuppressWarnings();
+ $fh = fopen( $url, "r", false, $context );
+ wfRestoreWarnings();
+
+ if ( !$fh ) {
+ break;
}
- // read $fh into the simpleFileWriter (grab in 64K chunks since its likely a media file)
- while ( !feof( $fh ) ) {
- $contents = fread( $fh, 65536 );
- $cwrite->callbackWriteBody( $fh, $contents );
+ $result = stream_get_meta_data( $fh );
+ $this->headerList = $result['wrapper_data'];
+ $this->parseHeader();
+
+ if ( !$this->followRedirects ) {
+ break;
}
- $cwrite->close();
- // check for simpleFileWriter error:
- if( !$cwrite->status->isOK() ){
- return $cwrite->status;
+ # Handle manual redirection
+ if ( !$this->isRedirect() || $reqCount > $this->maxRedirects ) {
+ break;
}
- } else {
- // read $fh into status->value
- $status->value = @stream_get_contents( $fh );
+ # Check security of URL
+ $url = $this->getResponseHeader( "Location" );
+
+ if ( !Http::isValidURI( $url ) ) {
+ wfDebug( __METHOD__ . ": insecure redirection\n" );
+ break;
+ }
+ } while ( true );
+
+ $this->setStatus();
+
+ if ( $fh === false ) {
+ $this->status->fatal( 'http-request-error' );
+ return $this->status;
}
- //close the url file wrapper
- fclose( $fh );
- // check for "false"
- if( $status->value === false ){
- $status->error( 'file_get_contents-failed' );
+ if ( $result['timed_out'] ) {
+ $this->status->fatal( 'http-timed-out', $this->url );
+ return $this->status;
}
- return $status;
- }
-}
+ // 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 );
-/**
- * a simpleFileWriter with session id updates
- */
-class simpleFileWriter {
- var $target_file_path;
- var $status = null;
- var $session_id = null;
- var $session_update_interval = 0; // how often to update the session while downloading
-
- function simpleFileWriter( $target_file_path, $upload_session_key, $do_close_session_update = false ){
- $this->target_file_path = $target_file_path;
- $this->upload_session_key = $upload_session_key;
- $this->status = Status::newGood();
- $this->do_close_session_update = $do_close_session_update;
- // open the file:
- $this->fp = fopen( $this->target_file_path, 'w' );
- if( $this->fp === false ){
- $this->status = Status::newFatal( 'HTTP::could-not-open-file-for-writing' );
- }
- // true start time
- $this->prevTime = time();
- }
-
- public function callbackWriteBody( $ch, $data_packet ){
- global $wgMaxUploadSize, $wgLang;
-
- // write out the content
- if( fwrite( $this->fp, $data_packet ) === false ){
- wfDebug( __METHOD__ ." ::could-not-write-to-file\n" );
- $this->status = Status::newFatal( 'HTTP::could-not-write-to-file' );
- return 0;
- }
-
- // check file size:
- clearstatcache();
- $this->current_fsize = filesize( $this->target_file_path );
-
- if( $this->current_fsize > $wgMaxUploadSize ){
- wfDebug( __METHOD__ . " ::http download too large\n" );
- $this->status = Status::newFatal( 'HTTP::file-has-grown-beyond-upload-limit-killing: downloaded more than ' .
- $wgLang->formatSize( $wgMaxUploadSize ) . ' ' );
- return 0;
- }
-
- // if more than session_update_interval second have passed update_session_progress
- if( $this->do_close_session_update && $this->upload_session_key &&
- ( ( time() - $this->prevTime ) > $this->session_update_interval ) ) {
- $this->prevTime = time();
- $session_status = $this->update_session_progress();
- if( !$session_status->isOK() ){
- $this->status = $session_status;
- wfDebug( __METHOD__ . ' update session failed or was canceled');
- return 0;
+ if ( $buf === false ) {
+ $this->status->fatal( 'http-read-error' );
+ break;
}
- }
- return strlen( $data_packet );
- }
- public function update_session_progress(){
- $status = Status::newGood();
- // start the session
- if( session_start() === false){
- wfDebug( __METHOD__ . ' could not start session' );
- exit( 0 );
- }
- $sd =& $_SESSION['wsDownload'][$this->upload_session_key];
- // check if the user canceled the request:
- if( isset( $sd['user_cancel'] ) && $sd['user_cancel'] == true ){
- // kill the download
- return Status::newFatal( 'user-canceled-request' );
+ if ( strlen( $buf ) ) {
+ call_user_func( $this->callback, $fh, $buf );
+ }
+ }
}
- // update the progress bytes download so far:
- $sd['loaded'] = $this->current_fsize;
- wfDebug( __METHOD__ . ': set session loaded amount to: ' . $sd['loaded'] . "\n");
- // close down the session so we can other http queries can get session updates:
- session_write_close();
- return $status;
- }
+ fclose( $fh );
- public function close(){
- // do a final session update:
- $this->update_session_progress();
- // close up the file handle:
- if( false === fclose( $this->fp ) ){
- $this->status = Status::newFatal( 'HTTP::could-not-close-file' );
- }
+ return $this->status;
}
-
}
dépôts / lhc / web / wiklou.git / blobdiff