* @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
+ * @param $options options to pass to HttpRequest object
+ * Possible keys for the array:
+ * timeout Timeout length in seconds
+ * postData An array of key-value pairs or a url-encoded form data
+ * proxy The proxy to use. Will use $wgHTTPProxy (if set) otherwise.
+ * noProxy Override $wgHTTPProxy (if set) and don't use any proxy at all.
+ * sslVerifyHost (curl only) Verify the SSL certificate
+ * caInfo (curl only) Provide CA information
+ * maxRedirects Maximum number of redirects to follow (defaults to 5)
+ * followRedirects Whether to follow redirects (defaults to true)
* @returns mixed (bool)false on failure or a string on success
*/
- public static function request( $method, $url, $opts = array() ) {
- $opts['method'] = strtoupper( $method );
- if ( !array_key_exists( 'timeout', $opts ) ) {
- $opts['timeout'] = 'default';
+ public static function request( $method, $url, $options = array() ) {
+ $url = wfExpandUrl( $url );
+ wfDebug( "HTTP: $method: $url" );
+ $options['method'] = strtoupper( $method );
+ if ( !isset( $options['timeout'] ) ) {
+ $options['timeout'] = 'default';
}
- $req = HttpRequest::factory( $url, $opts );
+ $req = HttpRequest::factory( $url, $options );
$status = $req->execute();
if ( $status->isOK() ) {
- return $req;
+ return $req->getContent();
} else {
return false;
}
* Simple wrapper for Http::request( 'GET' )
* @see Http::request()
*/
- 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()
*/
- 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 );
}
/**
protected $content;
protected $timeout = 'default';
protected $headersOnly = null;
- protected $postdata = null;
+ protected $postData = null;
protected $proxy = null;
- protected $no_proxy = false;
+ protected $noProxy = false;
protected $sslVerifyHost = true;
protected $caInfo = null;
protected $method = "GET";
+ protected $reqHeaders = array();
protected $url;
- protected $parsed_url;
+ protected $parsedUrl;
+ protected $callback;
+ protected $maxRedirects = 5;
+ protected $followRedirects = true;
+
+ protected $cookieJar;
+
+ protected $headerList = array();
+ protected $respVersion = "0.9";
+ protected $respStatus = "200 Ok";
+ protected $respHeaders = array();
+
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 = array()) {
- global $wgHTTPTimeout, $wgTitle;
+ * @param $options array (optional) extra params to pass (see Http::request())
+ */
+ function __construct( $url, $options = array() ) {
+ global $wgHTTPTimeout;
$this->url = $url;
- $this->parsed_url = parse_url($url);
+ $this->parsedUrl = parse_url( $url );
- if ( !ini_get( 'allow_url_fopen' ) ) {
- throw new MWException( 'allow_url_fopen needs to be enabled for http requests to work' );
- } elseif ( !Http::isValidURI( $this->url ) ) {
- throw new MWException( 'bad-uri' );
+ if ( !Http::isValidURI( $this->url ) ) {
+ $this->status = Status::newFatal('http-invalid-url');
} else {
$this->status = Status::newGood( 100 ); // continue
}
- if ( array_key_exists( 'timeout', $opt ) && $opt['timeout'] != 'default' ) {
- $this->timeout = $opt['timeout'];
+ if ( isset($options['timeout']) && $options['timeout'] != 'default' ) {
+ $this->timeout = $options['timeout'];
} else {
$this->timeout = $wgHTTPTimeout;
}
- $members = array( "targetFilePath", "requestKey", "headersOnly", "postdata",
- "proxy", "no_proxy", "sslVerifyHost", "caInfo", "method" );
+ $members = array( "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo",
+ "method", "followRedirects", "maxRedirects" );
foreach ( $members as $o ) {
- if ( array_key_exists( $o, $opt ) ) {
- $this->$o = $opt[$o];
+ if ( isset($options[$o]) ) {
+ $this->$o = $options[$o];
}
}
+ }
- if ( is_array( $this->postdata ) ) {
- $this->postdata = wfArrayToCGI( $this->postdata );
+ /**
+ * Generate a new request object
+ * @see HttpRequest::__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"' );
}
- $this->initRequest();
+ switch( Http::$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.' );
+ }
+ }
- if ( !$this->no_proxy ) {
- $this->proxySetup();
+ /**
+ * 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 $params array
+ * @todo overload the args param
+ */
+ public function setData($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 = 'http://localhost:80/';
+ } 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
+ */
+ public function setUserAgent( $UA ) {
+ $this->setHeader('User-Agent', $UA);
+ }
+
+ /**
+ * Set an arbitrary header
+ */
+ public function setHeader($name, $value) {
+ // I feel like I should normalize the case here...
+ $this->reqHeaders[$name] = $value;
+ }
- # 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->setReferrer( $wgTitle->getFullURL() );
+ /**
+ * Get an array of the headers
+ */
+ public function getHeaderList() {
+ $list = array();
+
+ if( $this->cookieJar ) {
+ $this->reqHeaders['Cookie'] =
+ $this->cookieJar->serializeToHttpRequest($this->parsedUrl['path'],
+ $this->parsedUrl['host']);
}
+ foreach($this->reqHeaders as $name => $value) {
+ $list[] = "$name: $value";
+ }
+ return $list;
}
/**
- * 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.
+ * Set the callback
+ * @param $callback callback
*/
- function __toString() {
- return $this->content;
+ public function setCallback( $callback ) {
+ $this->callback = $callback;
}
/**
- * Generate a new request object
- * @see HttpRequest::__construct
+ * A generic callback to read the body of the response from a remote
+ * server.
+ * @param $fh handle
+ * @param $content string
*/
- public static function factory( $url, $opt ) {
- global $wgHTTPEngine;
- $engine = $wgHTTPEngine;
+ public function read( $fh, $content ) {
+ $this->content .= $content;
+ return strlen( $content );
+ }
+
+ /**
+ * Take care of whatever is necessary to perform the URI request.
+ * @return Status
+ */
+ public function execute() {
+ global $wgTitle;
+
+ $this->content = "";
- if ( !$wgHTTPEngine ) {
- $wgHTTPEngine = function_exists( 'curl_init' ) ? 'curl' : 'php';
- } elseif ( $wgHTTPEngine == 'curl' && !function_exists( 'curl_init' ) ) {
- throw new MWException( 'FIXME' );
+ if( strtoupper($this->method) == "HEAD" ) {
+ $this->headersOnly = true;
}
- switch( $wgHTTPEngine ) {
- case 'curl':
- return new CurlHttpRequest( $url, $opt );
- case 'php':
- return new PhpHttpRequest( $url, $opt );
- default:
- throw new MWException( 'FIXME' );
+ if ( is_array( $this->postData ) ) {
+ $this->postData = wfArrayToCGI( $this->postData );
}
- }
- public function getContent() {
- return $this->content;
+ if ( is_object( $wgTitle ) && !isset($this->reqHeaders['Referer']) ) {
+ $this->setReferer( $wgTitle->getFullURL() );
+ }
+
+ if ( !$this->noProxy ) {
+ $this->proxySetup();
+ }
+
+ if ( !$this->callback ) {
+ $this->setCallback( array( $this, 'read' ) );
+ }
+
+ if ( !isset($this->reqHeaders['User-Agent']) ) {
+ $this->setUserAgent(Http::userAgent());
+ }
}
- public function initRequest() {}
- public function proxySetup() {}
- public function setReferrer( $url ) {}
- public function setCallback( $cb ) {}
- public function read($fh, $content) {}
- public function getCode() {}
- public function execute() {}
-}
+ /**
+ * 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.
+ * @returns 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] );
+ }
+ }
-/**
- * HttpRequest implemented using internal curl compiled into PHP
- */
-class CurlHttpRequest extends HttpRequest {
- protected $curlHandle;
- protected $curlCBSet;
+ $this->parseCookies();
+ }
- public function initRequest() {
- $this->curlHandle = curl_init( $this->url );
+ /**
+ * Sets the member variable status to a fatal status if the HTTP
+ * status code was not 200.
+ * @returns nothing
+ */
+ protected function setStatus() {
+ if( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ if((int)$this->respStatus !== 200) {
+ list( $code, $message ) = explode(" ", $this->respStatus, 2);
+ $this->status->fatal("http-bad-status", $code, $message );
+ }
}
- public function proxySetup() {
- global $wgHTTPProxy;
- if ( is_string( $this->proxy ) ) {
- curl_setopt( $this->curlHandle, CURLOPT_PROXY, $this->proxy );
- } else if ( Http::isLocalURL( $this->url ) ) { /* Not sure this makes any sense. */
- curl_setopt( $this->curlHandle, CURLOPT_PROXY, 'localhost:80' );
- } else if ( $wgHTTPProxy ) {
- curl_setopt( $this->curlHandle, CURLOPT_PROXY, $wgHTTPProxy );
+ /**
+ * Returns true if the last status code was a redirect.
+ * @return bool
+ */
+ public function isRedirect() {
+ if( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+
+ $status = (int)$this->respStatus;
+ if ( $status >= 300 && $status < 400 ) {
+ return true;
}
+ return false;
}
- public function setCallback( $cb ) {
- if ( !$this->curlCBSet ) {
- $this->curlCBSet = true;
- curl_setopt( $this->curlHandle, CURLOPT_WRITEFUNCTION, $cb );
+ /**
+ * Returns an associative array of response headers after the
+ * request has been executed. Because some headers
+ * (e.g. Set-Cookie) can appear more than once the, each value of
+ * the associative array is an array of the values given.
+ * @return array
+ */
+ public function getResponseHeaders() {
+ if( !$this->respHeaders ) {
+ $this->parseHeader();
}
+ return $this->respHeaders;
}
- public function execute() {
- if( !$this->status->isOK() ) {
- return $this->status;
+ /**
+ * 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;
+ }
- $this->setCallback( array($this, 'read') );
+ /**
+ * Tells the HttpRequest object to use this pre-loaded CookieJar.
+ * @param $jar CookieJar
+ */
+ public function setCookieJar( $jar ) {
+ $this->cookieJar = $jar;
+ }
- curl_setopt( $this->curlHandle, CURLOPT_TIMEOUT, $this->timeout );
- curl_setopt( $this->curlHandle, CURLOPT_USERAGENT, Http::userAgent() );
- curl_setopt( $this->curlHandle, CURLOPT_HTTP_VERSION, CURL_HTTP_VERSION_1_0 );
+ /**
+ * Returns the cookie jar in use.
+ * @returns CookieJar
+ */
+ public function getCookieJar() {
+ if( !$this->respHeaders ) {
+ $this->parseHeader();
+ }
+ return $this->cookieJar;
+ }
- if ( $this->sslVerifyHost ) {
- curl_setopt( $this->curlHandle, CURLOPT_SSL_VERIFYHOST, $this->sslVerifyHost );
+ /**
+ * 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
+ */
+ public function setCookie( $name, $value = null, $attr = null) {
+ if( !$this->cookieJar ) {
+ $this->cookieJar = new CookieJar;
}
+ $this->cookieJar->setCookie($name, $value, $attr);
+ }
- if ( $this->caInfo ) {
- curl_setopt( $this->curlHandle, CURLOPT_CAINFO, $this->caInfo );
+ /**
+ * Parse the cookies in the response headers and store them in the cookie jar.
+ */
+ protected function parseCookies() {
+ if( !$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'] );
+ }
+ }
+ }
- if ( $this->headersOnly ) {
- curl_setopt( $this->curlHandle, CURLOPT_NOBODY, true );
- curl_setopt( $this->curlHandle, CURLOPT_HEADER, true );
- } elseif ( $this->method == 'POST' ) {
- curl_setopt( $this->curlHandle, CURLOPT_POST, true );
- curl_setopt( $this->curlHandle, 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->curlHandle, CURLOPT_HTTPHEADER, array( 'Expect:' ) );
+ /**
+ * Returns the final URL after all redirections.
+ * @returns string
+ */
+ public function getFinalUrl() {
+ $location = $this->getResponseHeader("Location");
+ if ( $location ) {
+ return $location;
+ }
+
+ return $this->url;
+ }
+}
+
+
+class Cookie {
+ protected $name;
+ protected $value;
+ protected $expires;
+ protected $path;
+ protected $domain;
+ protected $isSessionKey = true;
+ // TO IMPLEMENT protected $secure
+ // TO IMPLEMENT? protected $maxAge (add onto expires)
+ // TO IMPLEMENT? protected $version
+ // TO IMPLEMENT? protected $comment
+
+ function __construct( $name, $value, $attr ) {
+ $this->name = $name;
+ $this->set( $value, $attr );
+ }
+
+ /**
+ * Sets a cookie. Used before a request to set up any individual
+ * cookies. Used internally after a request to parse the
+ * Set-Cookie headers.
+ * @param $name string the name of the cookie
+ * @param $value string the value of the cookie
+ * @param $attr array possible key/values:
+ * expires A date string
+ * path The path this cookie is used on
+ * domain Domain this cookie is used on
+ */
+ public function set( $value, $attr ) {
+ $this->value = $value;
+ if( isset( $attr['expires'] ) ) {
+ $this->isSessionKey = false;
+ $this->expires = strtotime( $attr['expires'] );
+ }
+ if( isset( $attr['path'] ) ) {
+ $this->path = $attr['path'];
} else {
- curl_setopt( $this->curlHandle, CURLOPT_CUSTOMREQUEST, $this->method );
+ $this->path = "/";
}
+ if( isset( $attr['domain'] ) ) {
+ if( self::validateCookieDomain( $attr['domain'] ) ) {
+ $this->domain = $attr['domain'];
+ }
+ } else {
+ throw new MWException("You must specify a domain.");
+ }
+ }
+
+ /**
+ * Return the true if the cookie is valid is valid. Otherwise,
+ * false. The uses a method similar to IE cookie security
+ * described here:
+ * http://kuza55.blogspot.com/2008/02/understanding-cookie-security.html
+ * A better method might be to use a blacklist like
+ * http://publicsuffix.org/
+ *
+ * @param $domain string the domain to validate
+ * @param $originDomain string (optional) the domain the cookie originates from
+ * @return bool
+ */
+ public static function validateCookieDomain( $domain, $originDomain = null) {
+ // Don't allow a trailing dot
+ if( substr( $domain, -1 ) == "." ) return false;
- try {
- if ( false === curl_exec( $this->curlHandle ) ) {
- $error_txt = 'Error sending request: #' . curl_errno( $this->curlHandle ) . ' ' .
- curl_error( $this->curlHandle );
- wfDebug( __METHOD__ . $error_txt . "\n" );
- $this->status->fatal( $error_txt ); /* i18n? */
+ $dc = explode(".", $domain);
+
+ // Don't allow cookies for "localhost", "ls" or other dot-less hosts
+ if( count($dc) < 2 ) return false;
+
+ // Only allow full, valid IP addresses
+ if( preg_match( '/^[0-9.]+$/', $domain ) ) {
+ if( count( $dc ) != 4 ) return false;
+
+ if( ip2long( $domain ) === false ) return false;
+
+ if( $originDomain == null || $originDomain == $domain ) return true;
+
+ }
+
+ // Don't allow cookies for "co.uk" or "gov.uk", etc, but allow "supermarket.uk"
+ if( strrpos( $domain, "." ) - strlen( $domain ) == -3 ) {
+ if( (count($dc) == 2 && strlen( $dc[0] ) <= 2 )
+ || (count($dc) == 3 && strlen( $dc[0] ) == "" && strlen( $dc[1] ) <= 2 ) ) {
+ return false;
}
- } catch ( Exception $e ) {
- $errno = curl_errno( $this->curlHandle );
- if ( $errno != CURLE_OK ) {
- $errstr = curl_error( $this->curlHandle );
- wfDebug( __METHOD__ . ": CURL error code $errno: $errstr\n" );
- $this->status->fatal( "CURL error code $errno: $errstr\n" ); /* i18n? */
+ if( (count($dc) == 2 || (count($dc) == 3 && $dc[0] == "") )
+ && preg_match( '/(com|net|org|gov|edu)\...$/', $domain) ) {
+ return false;
}
}
- curl_close( $this->curlHandle );
+ if( $originDomain != null ) {
+ if( substr( $domain, 0, 1 ) != "." && $domain != $originDomain ) {
+ return false;
+ }
+ if( substr( $domain, 0, 1 ) == "."
+ && substr_compare( $originDomain, $domain, -strlen( $domain ),
+ strlen( $domain ), TRUE ) != 0 ) {
+ return false;
+ }
+ }
- return $this->status;
+ return true;
}
- public function read( $curlH, $content ) {
- $this->content .= $content;
- return strlen( $content );
+ /**
+ * Serialize the cookie jar into a format useful for HTTP Request headers.
+ * @param $path string the path that will be used. Required.
+ * @param $domain string the domain that will be used. Required.
+ * @return string
+ */
+ public function serializeToHttpRequest( $path, $domain ) {
+ $ret = "";
+
+ if( $this->canServeDomain( $domain )
+ && $this->canServePath( $path )
+ && $this->isUnExpired() ) {
+ $ret = $this->name ."=". $this->value;
+ }
+
+ return $ret;
}
- public function getCode() {
- # Don't return truncated output
- $code = curl_getinfo( $this->curlHandle, CURLINFO_HTTP_CODE );
- if ( $code < 400 ) {
- $this->status->setResult( true, $code );
- } else {
- $this->status->setResult( false, $code );
+ protected function canServeDomain( $domain ) {
+ if( $domain == $this->domain
+ || ( strlen( $domain) > strlen( $this->domain )
+ && substr( $this->domain, 0, 1) == "."
+ && substr_compare( $domain, $this->domain, -strlen( $this->domain ),
+ strlen( $this->domain ), TRUE ) == 0 ) ) {
+ return true;
}
+ return false;
}
+
+ protected function canServePath( $path ) {
+ if( $this->path && substr_compare( $this->path, $path, 0, strlen( $this->path ) ) == 0 ) {
+ return true;
+ }
+ return false;
+ }
+
+ protected function isUnExpired() {
+ if( $this->isSessionKey || $this->expires > time() ) {
+ return true;
+ }
+ return false;
+ }
+
}
-class PhpHttpRequest extends HttpRequest {
- private $reqHeaders;
- private $callback;
- private $fh;
+class CookieJar {
+ private $cookie = array();
+
+ /**
+ * Set a cookie in the cookie jar. Make sure only one cookie per-name exists.
+ * @see Cookie::set()
+ */
+ public function setCookie ($name, $value, $attr) {
+ /* cookies: case insensitive, so this should work.
+ * We'll still send the cookies back in the same case we got them, though.
+ */
+ $index = strtoupper($name);
+ if( isset( $this->cookie[$index] ) ) {
+ $this->cookie[$index]->set( $value, $attr );
+ } else {
+ $this->cookie[$index] = new Cookie( $name, $value, $attr );
+ }
+ }
- public function initRequest() {
- $this->setCallback( array( $this, 'read' ) );
+ /**
+ * @see Cookie::serializeToHttpRequest
+ */
+ public function serializeToHttpRequest( $path, $domain ) {
+ $cookies = array();
- $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";
+ foreach( $this->cookie as $c ) {
+ $serialized = $c->serializeToHttpRequest( $path, $domain );
+ if ( $serialized ) $cookies[] = $serialized;
}
- if( $this->parsed_url['scheme'] != 'http' ) {
- $this->status->fatal( "Only http:// is supported currently." );
- }
+ return implode("; ", $cookies);
}
- protected function urlToTcp($url) {
- $parsed_url = parse_url($url);
+ /**
+ * Parse the content of an Set-Cookie HTTP Response header.
+ * @param $cookie string
+ */
+ public function parseCookieResponseHeader ( $cookie, $domain ) {
+ $len = strlen( "Set-Cookie:" );
+ if ( substr_compare( "Set-Cookie:", $cookie, 0, $len, TRUE ) === 0 ) {
+ $cookie = substr( $cookie, $len );
+ }
- return 'tcp://'.$parsed_url['host'].':'.$parsed_url['port'];
- }
+ $bit = array_map( 'trim', explode( ";", $cookie ) );
+ if ( count($bit) >= 1 ) {
+ list($name, $value) = explode( "=", array_shift( $bit ), 2 );
+ $attr = array();
+ foreach( $bit as $piece ) {
+ $parts = explode( "=", $piece );
+ if( count( $parts ) > 1 ) {
+ $attr[strtolower( $parts[0] )] = $parts[1];
+ } else {
+ $attr[strtolower( $parts[0] )] = true;
+ }
+ }
- public function proxySetup() {
- global $wgHTTPProxy;
+ if( !isset( $attr['domain'] ) ) {
+ $attr['domain'] = $domain;
+ } elseif ( !Cookie::validateCookieDomain( $attr['domain'], $domain ) ) {
+ return null;
+ }
- if ( Http::isLocalURL( $this->url ) ) {
- $this->proxy = 'http://localhost:80/';
- } elseif ( $wgHTTPProxy ) {
- $this->proxy = $wgHTTPProxy ;
+ $this->setCookie( $name, $value, $attr );
}
}
+}
- public function setReferrer( $url ) {
- $this->reqHeaders[] = "Referer: $url";
- }
- public function setCallback( $cb ) {
- $this->callback = $cb;
+/**
+ * HttpRequest implemented using internal curl compiled into PHP
+ */
+class CurlHttpRequest extends HttpRequest {
+ static $curlMessageMap = array(
+ 6 => 'http-host-unreachable',
+ 28 => 'http-timed-out'
+ );
+
+ protected $curlOptions = array();
+ protected $headerText = "";
+
+ protected function readHeader( $fh, $content ) {
+ $this->headerText .= $content;
+ return strlen( $content );
}
- public function read( $fh, $contents ) {
+ public function execute() {
+ parent::execute();
+ if ( !$this->status->isOK() ) {
+ return $this->status;
+ }
+ $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_FOLLOWLOCATION] = $this->followRedirects;
+ $this->curlOptions[CURLOPT_MAXREDIRS] = $this->maxRedirects;
+
+ /* not sure these two are actually necessary */
+ if(isset($this->reqHeaders['Referer'])) {
+ $this->curlOptions[CURLOPT_REFERER] = $this->reqHeaders['Referer'];
+ }
+ $this->curlOptions[CURLOPT_USERAGENT] = $this->reqHeaders['User-Agent'];
+
+ if ( $this->sslVerifyHost ) {
+ $this->curlOptions[CURLOPT_SSL_VERIFYHOST] = $this->sslVerifyHost;
+ }
+
+ if ( $this->caInfo ) {
+ $this->curlOptions[CURLOPT_CAINFO] = $this->caInfo;
+ }
+
if ( $this->headersOnly ) {
- return false;
+ $this->curlOptions[CURLOPT_NOBODY] = true;
+ $this->curlOptions[CURLOPT_HEADER] = true;
+ } elseif ( $this->method == 'POST' ) {
+ $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 {
+ $this->curlOptions[CURLOPT_CUSTOMREQUEST] = $this->method;
+ }
+
+ $this->curlOptions[CURLOPT_HTTPHEADER] = $this->getHeaderList();
+
+ $curlHandle = curl_init( $this->url );
+ curl_setopt_array( $curlHandle, $this->curlOptions );
+
+ 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 ) );
+ }
+ } else {
+ $this->headerList = explode("\r\n", $this->headerText);
}
- $this->content .= $contents;
- return strlen( $contents );
+ curl_close( $curlHandle );
+
+ $this->parseHeader();
+ $this->setStatus();
+ return $this->status;
+ }
+}
+
+class PhpHttpRequest extends HttpRequest {
+ protected $manuallyRedirect = false;
+
+ protected function urlToTcp( $url ) {
+ $parsedUrl = parse_url( $url );
+
+ return 'tcp://' . $parsedUrl['host'] . ':' . $parsedUrl['port'];
}
public function execute() {
- if( !$this->status->isOK() ) {
- return $this->status;
+ parent::execute();
+
+ // At least on Centos 4.8 with PHP 5.1.6, using max_redirects to follow redirects
+ // causes a segfault
+ if ( version_compare( '5.1.7', phpversion(), '>' ) ) {
+ $this->manuallyRedirect = true;
+ }
+
+ if ( $this->parsedUrl['scheme'] != 'http' ) {
+ $this->status->fatal( 'http-invalid-scheme', $this->parsedUrl['scheme'] );
+ }
+
+ $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";
}
- $opts = array();
- if ( $this->proxy && !$this->no_proxy ) {
- $opts['proxy'] = $this->urlToTCP($this->proxy);
- $opts['request_fulluri'] = true;
+ $options = array();
+ if ( $this->proxy && !$this->noProxy ) {
+ $options['proxy'] = $this->urlToTCP( $this->proxy );
+ $options['request_fulluri'] = true;
}
- $opts['method'] = $this->method;
- $opts['timeout'] = $this->timeout;
- $opts['header'] = implode( "\r\n", $this->reqHeaders );
- // FOR NOW: Force everyone to HTTP 1.0
- /* if ( version_compare( "5.3.0", phpversion(), ">" ) ) { */
- $opts['protocol_version'] = "1.0";
- /* } else { */
- /* $opts['protocol_version'] = "1.1"; */
- /* } */
+ if ( !$this->followRedirects || $this->manuallyRedirect ) {
+ $options['max_redirects'] = 0;
+ } else {
+ $options['max_redirects'] = $this->maxRedirects;
+ }
+
+ $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";
+
+ // 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 ) {
- $opts['content'] = $this->postdata;
+ if ( $this->postData ) {
+ $options['content'] = $this->postData;
}
- $context = stream_context_create( array( 'http' => $opts ) );
- try {
- $this->fh = fopen( $this->url, "r", false, $context );
- } catch (Exception $e) {
- $this->status->fatal($e->getMessage());
+ $oldTimeout = false;
+ if ( version_compare( '5.2.1', phpversion(), '>' ) ) {
+ $oldTimeout = ini_set('default_socket_timeout', $this->timeout);
+ } else {
+ $options['timeout'] = $this->timeout;
+ }
+
+ $context = stream_context_create( array( 'http' => $options ) );
+
+ $this->headerList = array();
+ $reqCount = 0;
+ $url = $this->url;
+ do {
+ $again = false;
+ $reqCount++;
+ wfSuppressWarnings();
+ $fh = fopen( $url, "r", false, $context );
+ wfRestoreWarnings();
+ if ( $fh ) {
+ $result = stream_get_meta_data( $fh );
+ $this->headerList = $result['wrapper_data'];
+ $this->parseHeader();
+ $url = $this->getResponseHeader("Location");
+ $again = $this->manuallyRedirect && $this->followRedirects && $url
+ && $this->isRedirect() && $this->maxRedirects > $reqCount;
+ }
+ } while ( $again );
+
+ if ( $oldTimeout !== false ) {
+ ini_set('default_socket_timeout', $oldTimeout);
+ }
+ $this->setStatus();
+
+ if ( $fh === false ) {
+ $this->status->fatal( 'http-request-error' );
return $this->status;
}
- $result = stream_get_meta_data( $this->fh );
if ( $result['timed_out'] ) {
- $this->status->error( __CLASS__ . '::timed-out-in-headers' );
+ $this->status->fatal( 'http-timed-out', $this->url );
+ return $this->status;
}
- $this->headers = $result['wrapper_data'];
-
- $end = false;
- while ( !$end ) {
- $contents = fread( $this->fh, 8192 );
- $size = call_user_func_array( $this->callback, array( $this->fh, $contents ) );
- $end = ( $size == 0 ) || feof( $this->fh );
+ if($this->status->isOK()) {
+ while ( !feof( $fh ) ) {
+ $buf = fread( $fh, 8192 );
+ if ( $buf === false ) {
+ $this->status->fatal( 'http-read-error' );
+ break;
+ }
+ if ( strlen( $buf ) ) {
+ call_user_func( $this->callback, $fh, $buf );
+ }
+ }
}
- fclose( $this->fh );
+ fclose( $fh );
return $this->status;
}