7 * Various HTTP related functions
11 static $httpEngine = false;
14 * Perform an HTTP request
15 * @param $method string HTTP method. Usually GET/POST
16 * @param $url string Full URL to act on
17 * @param $options options to pass to HttpRequest object
18 * Possible keys for the array:
19 * timeout Timeout length in seconds
20 * postData An array of key-value pairs or a url-encoded form data
21 * proxy The proxy to use. Will use $wgHTTPProxy (if set) otherwise.
22 * noProxy Override $wgHTTPProxy (if set) and don't use any proxy at all.
23 * sslVerifyHost (curl only) Verify the SSL certificate
24 * caInfo (curl only) Provide CA information
25 * maxRedirects Maximum number of redirects to follow (defaults to 5)
26 * followRedirects Whether to follow redirects (defaults to true)
27 * @returns mixed (bool)false on failure or a string on success
29 public static function request( $method, $url, $options = array() ) {
30 wfDebug( "HTTP: $method: $url" );
31 $options['method'] = strtoupper( $method );
32 if ( !isset( $options['timeout'] ) ) {
33 $options['timeout'] = 'default';
35 $req = HttpRequest::factory( $url, $options );
36 $status = $req->execute();
37 if ( $status->isOK() ) {
38 return $req->getContent();
45 * Simple wrapper for Http::request( 'GET' )
46 * @see Http::request()
48 public static function get( $url, $timeout = 'default', $options = array() ) {
49 $options['timeout'] = $timeout;
50 return Http::request( 'GET', $url, $options );
54 * Simple wrapper for Http::request( 'POST' )
55 * @see Http::request()
57 public static function post( $url, $options = array() ) {
58 return Http::request( 'POST', $url, $options );
62 * Check if the URL can be served by localhost
63 * @param $url string Full url to check
66 public static function isLocalURL( $url ) {
67 global $wgCommandLineMode, $wgConf;
68 if ( $wgCommandLineMode ) {
74 if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
77 $domainParts = explode( '.', $host );
78 // Check if this domain or any superdomain is listed in $wgConf as a local virtual host
79 $domainParts = array_reverse( $domainParts );
80 for ( $i = 0; $i < count( $domainParts ); $i++ ) {
81 $domainPart = $domainParts[$i];
83 $domain = $domainPart;
85 $domain = $domainPart . '.' . $domain;
87 if ( $wgConf->isLocalVHost( $domain ) ) {
96 * A standard user-agent we can use for external requests.
99 public static function userAgent() {
101 return "MediaWiki/$wgVersion";
105 * Checks that the given URI is a valid one
106 * @param $uri Mixed: URI to check for validity
109 public static function isValidURI( $uri ) {
111 '/(ftp|http|https):\/\/(\w+:{0,1}\w*@)?(\S+)(:[0-9]+)?(\/|\/([\w#!:.?+=&%@!\-\/]))?/',
119 * This wrapper class will call out to curl (if available) or fallback
120 * to regular PHP if necessary for handling internal HTTP requests.
124 protected $timeout = 'default';
125 protected $headersOnly = null;
126 protected $postData = null;
127 protected $proxy = null;
128 protected $noProxy = false;
129 protected $sslVerifyHost = true;
130 protected $caInfo = null;
131 protected $method = "GET";
132 protected $reqHeaders = array();
134 protected $parsedUrl;
136 protected $maxRedirects = 5;
137 protected $followRedirects = true;
139 protected $cookieJar;
141 protected $headerList = array();
142 protected $respVersion = "0.9";
143 protected $respStatus = "200 Ok";
144 protected $respHeaders = array();
149 * @param $url string url to use
150 * @param $options array (optional) extra params to pass (see Http::request())
152 function __construct( $url, $options = array() ) {
153 global $wgHTTPTimeout;
156 $this->parsedUrl = parse_url( $url );
158 if ( !Http::isValidURI( $this->url ) ) {
159 $this->status = Status::newFatal('http-invalid-url');
161 $this->status = Status::newGood( 100 ); // continue
164 if ( isset($options['timeout']) && $options['timeout'] != 'default' ) {
165 $this->timeout = $options['timeout'];
167 $this->timeout = $wgHTTPTimeout;
170 $members = array( "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
171 "method", "followRedirects", "maxRedirects" );
172 foreach ( $members as $o ) {
173 if ( isset($options[$o]) ) {
174 $this->$o = $options[$o];
180 * Generate a new request object
181 * @see HttpRequest::__construct
183 public static function factory( $url, $options = null ) {
184 if ( !Http::$httpEngine ) {
185 Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
186 } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) {
187 throw new MWException( __METHOD__.': curl (http://php.net/curl) is not installed, but'.
188 ' Http::$httpEngine is set to "curl"' );
191 switch( Http::$httpEngine ) {
193 return new CurlHttpRequest( $url, $options );
195 if ( !wfIniGetBool( 'allow_url_fopen' ) ) {
196 throw new MWException( __METHOD__.': allow_url_fopen needs to be enabled for pure PHP'.
197 ' http requests to work. If possible, curl should be used instead. See http://php.net/curl.' );
199 return new PhpHttpRequest( $url, $options );
201 throw new MWException( __METHOD__.': The setting of Http::$httpEngine is not valid.' );
206 * Get the body, or content, of the response to the request
209 public function getContent() {
210 return $this->content;
214 * Take care of setting up the proxy
215 * (override in subclass)
218 public function proxySetup() {
221 if ( $this->proxy ) {
224 if ( Http::isLocalURL( $this->url ) ) {
225 $this->proxy = 'http://localhost:80/';
226 } elseif ( $wgHTTPProxy ) {
227 $this->proxy = $wgHTTPProxy ;
228 } elseif ( getenv( "http_proxy" ) ) {
229 $this->proxy = getenv( "http_proxy" );
234 * Set the refererer header
236 public function setReferer( $url ) {
237 $this->setHeader('Referer', $url);
243 public function setUserAgent( $UA ) {
244 $this->setHeader('User-Agent', $UA);
248 * Set an arbitrary header
250 public function setHeader($name, $value) {
251 // I feel like I should normalize the case here...
252 $this->reqHeaders[$name] = $value;
256 * Get an array of the headers
258 public function getHeaderList() {
261 if( $this->cookieJar ) {
262 $this->reqHeaders['Cookie'] =
263 $this->cookieJar->serializeToHttpRequest($this->parsedUrl['path'],
264 $this->parsedUrl['host']);
266 foreach($this->reqHeaders as $name => $value) {
267 $list[] = "$name: $value";
274 * @param $callback callback
276 public function setCallback( $callback ) {
277 $this->callback = $callback;
281 * A generic callback to read the body of the response from a remote
284 * @param $content string
286 public function read( $fh, $content ) {
287 $this->content .= $content;
288 return strlen( $content );
292 * Take care of whatever is necessary to perform the URI request.
295 public function execute() {
298 if( strtoupper($this->method) == "HEAD" ) {
299 $this->headersOnly = true;
302 if ( is_array( $this->postData ) ) {
303 $this->postData = wfArrayToCGI( $this->postData );
306 if ( is_object( $wgTitle ) && !isset($this->reqHeaders['Referer']) ) {
307 $this->setReferer( $wgTitle->getFullURL() );
310 if ( !$this->noProxy ) {
314 if ( !$this->callback ) {
315 $this->setCallback( array( $this, 'read' ) );
318 if ( !isset($this->reqHeaders['User-Agent']) ) {
319 $this->setUserAgent(Http::userAgent());
324 * Parses the headers, including the HTTP status code and any
325 * Set-Cookie headers. This function expectes the headers to be
326 * found in an array in the member variable headerList.
329 protected function parseHeader() {
331 foreach( $this->headerList as $header ) {
332 if( preg_match( "#^HTTP/([0-9.]+) (.*)#", $header, $match ) ) {
333 $this->respVersion = $match[1];
334 $this->respStatus = $match[2];
335 } elseif( preg_match( "#^[ \t]#", $header ) ) {
336 $last = count($this->respHeaders[$lastname]) - 1;
337 $this->respHeaders[$lastname][$last] .= "\r\n$header";
338 } elseif( preg_match( "#^([^:]*):[\t ]*(.*)#", $header, $match ) ) {
339 $this->respHeaders[strtolower( $match[1] )][] = $match[2];
340 $lastname = strtolower( $match[1] );
344 $this->parseCookies();
348 * Sets the member variable status to a fatal status if the HTTP
349 * status code was not 200.
352 protected function setStatus() {
353 if( !$this->respHeaders ) {
354 $this->parseHeader();
357 if((int)$this->respStatus !== 200) {
358 list( $code, $message ) = explode(" ", $this->respStatus, 2);
359 $this->status->fatal("http-bad-status", $code, $message );
365 * Returns true if the last status code was a redirect.
368 public function isRedirect() {
369 if( !$this->respHeaders ) {
370 $this->parseHeader();
373 $status = (int)$this->respStatus;
374 if ( $status >= 300 && $status < 400 ) {
381 * Returns an associative array of response headers after the
382 * request has been executed. Because some headers
383 * (e.g. Set-Cookie) can appear more than once the, each value of
384 * the associative array is an array of the values given.
387 public function getResponseHeaders() {
388 if( !$this->respHeaders ) {
389 $this->parseHeader();
391 return $this->respHeaders;
395 * Returns the value of the given response header.
396 * @param $header string
399 public function getResponseHeader($header) {
400 if( !$this->respHeaders ) {
401 $this->parseHeader();
403 if ( isset( $this->respHeaders[strtolower ( $header ) ] ) ) {
404 $v = $this->respHeaders[strtolower ( $header ) ];
405 return $v[count( $v ) - 1];
411 * Tells the HttpRequest object to use this pre-loaded CookieJar.
412 * @param $jar CookieJar
414 public function setCookieJar( $jar ) {
415 $this->cookieJar = $jar;
419 * Returns the cookie jar in use.
422 public function getCookieJar() {
423 if( !$this->respHeaders ) {
424 $this->parseHeader();
426 return $this->cookieJar;
430 * Sets a cookie. Used before a request to set up any individual
431 * cookies. Used internally after a request to parse the
432 * Set-Cookie headers.
435 public function setCookie( $name, $value = null, $attr = null) {
436 if( !$this->cookieJar ) {
437 $this->cookieJar = new CookieJar;
439 $this->cookieJar->setCookie($name, $value, $attr);
443 * Parse the cookies in the response headers and store them in the cookie jar.
445 protected function parseCookies() {
446 if( !$this->cookieJar ) {
447 $this->cookieJar = new CookieJar;
449 if( isset( $this->respHeaders['set-cookie'] ) ) {
450 $url = parse_url( $this->getFinalUrl() );
451 foreach( $this->respHeaders['set-cookie'] as $cookie ) {
452 $this->cookieJar->parseCookieResponseHeader( $cookie, $url['host'] );
458 * Returns the final URL after all redirections.
461 public function getFinalUrl() {
462 $location = $this->getResponseHeader("Location");
478 protected $isSessionKey = true;
479 // TO IMPLEMENT protected $secure
480 // TO IMPLEMENT? protected $maxAge (add onto expires)
481 // TO IMPLEMENT? protected $version
482 // TO IMPLEMENT? protected $comment
484 function __construct( $name, $value, $attr ) {
486 $this->set( $value, $attr );
490 * Sets a cookie. Used before a request to set up any individual
491 * cookies. Used internally after a request to parse the
492 * Set-Cookie headers.
493 * @param $name string the name of the cookie
494 * @param $value string the value of the cookie
495 * @param $attr array possible key/values:
496 * expires A date string
497 * path The path this cookie is used on
498 * domain Domain this cookie is used on
500 public function set( $value, $attr ) {
501 $this->value = $value;
502 if( isset( $attr['expires'] ) ) {
503 $this->isSessionKey = false;
504 $this->expires = strtotime( $attr['expires'] );
506 if( isset( $attr['path'] ) ) {
507 $this->path = $attr['path'];
511 if( isset( $attr['domain'] ) ) {
512 if( self::validateCookieDomain( $attr['domain'] ) ) {
513 $this->domain = $attr['domain'];
516 throw new MWException("You must specify a domain.");
521 * Return the true if the cookie is valid is valid. Otherwise,
522 * false. The uses a method similar to IE cookie security
524 * http://kuza55.blogspot.com/2008/02/understanding-cookie-security.html
525 * A better method might be to use a blacklist like
526 * http://publicsuffix.org/
528 * @param $domain string the domain to validate
529 * @param $originDomain string (optional) the domain the cookie originates from
532 public static function validateCookieDomain( $domain, $originDomain = null) {
533 // Don't allow a trailing dot
534 if( substr( $domain, -1 ) == "." ) return false;
536 $dc = explode(".", $domain);
538 // Don't allow cookies for "localhost", "ls" or other dot-less hosts
539 if( count($dc) < 2 ) return false;
541 // Only allow full, valid IP addresses
542 if( preg_match( '/^[0-9.]+$/', $domain ) ) {
543 if( count( $dc ) != 4 ) return false;
545 if( ip2long( $domain ) === false ) return false;
547 if( $originDomain == null || $originDomain == $domain ) return true;
551 // Don't allow cookies for "co.uk" or "gov.uk", etc, but allow "supermarket.uk"
552 if( strrpos( $domain, "." ) - strlen( $domain ) == -3 ) {
553 if( (count($dc) == 2 && strlen( $dc[0] ) <= 2 )
554 || (count($dc) == 3 && strlen( $dc[0] ) == "" && strlen( $dc[1] ) <= 2 ) ) {
557 if( (count($dc) == 2 || (count($dc) == 3 && $dc[0] == "") )
558 && preg_match( '/(com|net|org|gov|edu)\...$/', $domain) ) {
563 if( $originDomain != null ) {
564 if( substr( $domain, 0, 1 ) != "." && $domain != $originDomain ) {
567 if( substr( $domain, 0, 1 ) == "."
568 && substr_compare( $originDomain, $domain, -strlen( $domain ),
569 strlen( $domain ), TRUE ) != 0 ) {
578 * Serialize the cookie jar into a format useful for HTTP Request headers.
579 * @param $path string the path that will be used. Required.
580 * @param $domain string the domain that will be used. Required.
583 public function serializeToHttpRequest( $path, $domain ) {
586 if( $this->canServeDomain( $domain )
587 && $this->canServePath( $path )
588 && $this->isUnExpired() ) {
589 $ret = $this->name ."=". $this->value;
595 protected function canServeDomain( $domain ) {
596 if( $domain == $this->domain
597 || ( strlen( $domain) > strlen( $this->domain )
598 && substr( $this->domain, 0, 1) == "."
599 && substr_compare( $domain, $this->domain, -strlen( $this->domain ),
600 strlen( $this->domain ), TRUE ) == 0 ) ) {
606 protected function canServePath( $path ) {
607 if( $this->path && substr_compare( $this->path, $path, 0, strlen( $this->path ) ) == 0 ) {
613 protected function isUnExpired() {
614 if( $this->isSessionKey || $this->expires > time() ) {
623 private $cookie = array();
626 * Set a cookie in the cookie jar. Make sure only one cookie per-name exists.
629 public function setCookie ($name, $value, $attr) {
630 /* cookies: case insensitive, so this should work.
631 * We'll still send the cookies back in the same case we got them, though.
633 $index = strtoupper($name);
634 if( isset( $this->cookie[$index] ) ) {
635 $this->cookie[$index]->set( $value, $attr );
637 $this->cookie[$index] = new Cookie( $name, $value, $attr );
642 * @see Cookie::serializeToHttpRequest
644 public function serializeToHttpRequest( $path, $domain ) {
647 foreach( $this->cookie as $c ) {
648 $serialized = $c->serializeToHttpRequest( $path, $domain );
649 if ( $serialized ) $cookies[] = $serialized;
652 return implode("; ", $cookies);
656 * Parse the content of an Set-Cookie HTTP Response header.
657 * @param $cookie string
659 public function parseCookieResponseHeader ( $cookie, $domain ) {
660 $len = strlen( "Set-Cookie:" );
661 if ( substr_compare( "Set-Cookie:", $cookie, 0, $len, TRUE ) === 0 ) {
662 $cookie = substr( $cookie, $len );
665 $bit = array_map( 'trim', explode( ";", $cookie ) );
666 if ( count($bit) >= 1 ) {
667 list($name, $value) = explode( "=", array_shift( $bit ), 2 );
669 foreach( $bit as $piece ) {
670 $parts = explode( "=", $piece );
671 if( count( $parts ) > 1 ) {
672 $attr[strtolower( $parts[0] )] = $parts[1];
674 $attr[strtolower( $parts[0] )] = true;
678 if( !isset( $attr['domain'] ) ) {
679 $attr['domain'] = $domain;
680 } elseif ( !Cookie::validateCookieDomain( $attr['domain'], $domain ) ) {
684 $this->setCookie( $name, $value, $attr );
691 * HttpRequest implemented using internal curl compiled into PHP
693 class CurlHttpRequest extends HttpRequest {
694 static $curlMessageMap = array(
695 6 => 'http-host-unreachable',
696 28 => 'http-timed-out'
699 protected $curlOptions = array();
700 protected $headerText = "";
702 protected function readHeader( $fh, $content ) {
703 $this->headerText .= $content;
704 return strlen( $content );
707 public function execute() {
709 if ( !$this->status->isOK() ) {
710 return $this->status;
712 $this->curlOptions[CURLOPT_PROXY] = $this->proxy;
713 $this->curlOptions[CURLOPT_TIMEOUT] = $this->timeout;
714 $this->curlOptions[CURLOPT_HTTP_VERSION] = CURL_HTTP_VERSION_1_0;
715 $this->curlOptions[CURLOPT_WRITEFUNCTION] = $this->callback;
716 $this->curlOptions[CURLOPT_HEADERFUNCTION] = array($this, "readHeader");
717 $this->curlOptions[CURLOPT_MAXREDIRS] = $this->maxRedirects;
719 /* not sure these two are actually necessary */
720 if(isset($this->reqHeaders['Referer'])) {
721 $this->curlOptions[CURLOPT_REFERER] = $this->reqHeaders['Referer'];
723 $this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
725 if ( $this->sslVerifyHost ) {
726 $this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost;
729 if ( $this->caInfo ) {
730 $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
733 if ( $this->headersOnly ) {
734 $this->curlOptions[CURLOPT_NOBODY] = true;
735 $this->curlOptions[CURLOPT_HEADER] = true;
736 } elseif ( $this->method == 'POST' ) {
737 $this->curlOptions[CURLOPT_POST] = true;
738 $this->curlOptions[CURLOPT_POSTFIELDS] = $this->postData;
739 // Suppress 'Expect: 100-continue' header, as some servers
740 // will reject it with a 417 and Curl won't auto retry
741 // with HTTP 1.0 fallback
742 $this->reqHeaders['Expect'] = '';
744 $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
747 $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
749 $curlHandle = curl_init( $this->url );
750 if ( !curl_setopt_array( $curlHandle, $this->curlOptions ) ) {
751 throw new MWException("Error setting curl options.");
753 if ( ! @curl_setopt( $curlHandle, CURLOPT_FOLLOWLOCATION, $this->followRedirects ) ) {
754 wfDebug("Couldn't set CURLOPT_FOLLOWLOCATION. Probably safe_mode or open_basedir is set.");
755 /* Continue the processing. If it were in curl_setopt_array, processing would have halted on its entry */
758 if ( false === curl_exec( $curlHandle ) ) {
759 $code = curl_error( $curlHandle );
761 if ( isset( self::$curlMessageMap[$code] ) ) {
762 $this->status->fatal( self::$curlMessageMap[$code] );
764 $this->status->fatal( 'http-curl-error', curl_error( $curlHandle ) );
767 $this->headerList = explode("\r\n", $this->headerText);
770 curl_close( $curlHandle );
772 $this->parseHeader();
774 return $this->status;
778 class PhpHttpRequest extends HttpRequest {
779 protected $manuallyRedirect = false;
781 protected function urlToTcp( $url ) {
782 $parsedUrl = parse_url( $url );
784 return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
787 public function execute() {
790 // At least on Centos 4.8 with PHP 5.1.6, using max_redirects to follow redirects
792 if ( version_compare( '5.1.7', phpversion(), '>' ) ) {
793 $this->manuallyRedirect = true;
796 if ( $this->parsedUrl['scheme'] != 'http' ) {
797 $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
800 $this->reqHeaders['Accept'] = "*/*";
801 if ( $this->method == 'POST' ) {
802 // Required for HTTP 1.0 POSTs
803 $this->reqHeaders['Content-Length'] = strlen( $this->postData );
804 $this->reqHeaders['Content-type'] = "application/x-www-form-urlencoded";
808 if ( $this->proxy && !$this->noProxy ) {
809 $options['proxy'] = $this->urlToTCP( $this->proxy );
810 $options['request_fulluri'] = true;
813 if ( !$this->followRedirects || $this->manuallyRedirect ) {
814 $options['max_redirects'] = 0;
816 $options['max_redirects'] = $this->maxRedirects;
819 $options['method'] = $this->method;
820 $options['header'] = implode("\r\n", $this->getHeaderList());
821 // Note that at some future point we may want to support
822 // HTTP/1.1, but we'd have to write support for chunking
823 // in version of PHP < 5.3.1
824 $options['protocol_version'] = "1.0";
826 // This is how we tell PHP we want to deal with 404s (for example) ourselves.
827 // Only works on 5.2.10+
828 $options['ignore_errors'] = true;
830 if ( $this->postData ) {
831 $options['content'] = $this->postData;
835 if ( version_compare( '5.2.1', phpversion(), '>' ) ) {
836 $oldTimeout = ini_set('default_socket_timeout', $this->timeout);
838 $options['timeout'] = $this->timeout;
841 $context = stream_context_create( array( 'http' => $options ) );
843 $this->headerList = array();
849 wfSuppressWarnings();
850 $fh = fopen( $url, "r", false, $context );
853 $result = stream_get_meta_data( $fh );
854 $this->headerList = $result['wrapper_data'];
855 $this->parseHeader();
856 $url = $this->getResponseHeader("Location");
857 $again = $this->manuallyRedirect && $this->followRedirects && $url
858 && $this->isRedirect() && $this->maxRedirects > $reqCount;
862 if ( $oldTimeout !== false ) {
863 ini_set('default_socket_timeout', $oldTimeout);
867 if ( $fh === false ) {
868 $this->status->fatal( 'http-request-error' );
869 return $this->status;
872 if ( $result['timed_out'] ) {
873 $this->status->fatal( 'http-timed-out', $this->url );
874 return $this->status;
877 if($this->status->isOK()) {
878 while ( !feof( $fh ) ) {
879 $buf = fread( $fh, 8192 );
880 if ( $buf === false ) {
881 $this->status->fatal( 'http-read-error' );
884 if ( strlen( $buf ) ) {
885 call_user_func( $this->callback, $fh, $buf );
891 return $this->status;