X-Git-Url: https://scripts.mit.edu/gitweb/autoinstallsdev/mediawiki.git/blobdiff_plain/19e297c21b10b1b8a3acad5e73fc71dcb35db44a..6932310fd58ebef145fa01eb76edf7150284d8ea:/includes/http/MWHttpRequest.php diff --git a/includes/http/MWHttpRequest.php b/includes/http/MWHttpRequest.php new file mode 100644 index 00000000..0f0118ce --- /dev/null +++ b/includes/http/MWHttpRequest.php @@ -0,0 +1,670 @@ +url = wfExpandUrl( $url, PROTO_HTTP ); + $this->parsedUrl = wfParseUrl( $this->url ); + + if ( isset( $options['logger'] ) ) { + $this->logger = $options['logger']; + } else { + $this->logger = new NullLogger(); + } + + if ( !$this->parsedUrl || !Http::isValidURI( $this->url ) ) { + $this->status = StatusValue::newFatal( 'http-invalid-url', $url ); + } else { + $this->status = StatusValue::newGood( 100 ); // continue + } + + if ( isset( $options['timeout'] ) && $options['timeout'] != 'default' ) { + $this->timeout = $options['timeout']; + } else { + $this->timeout = $wgHTTPTimeout; + } + if ( isset( $options['connectTimeout'] ) && $options['connectTimeout'] != 'default' ) { + $this->connectTimeout = $options['connectTimeout']; + } else { + $this->connectTimeout = $wgHTTPConnectTimeout; + } + if ( isset( $options['userAgent'] ) ) { + $this->setUserAgent( $options['userAgent'] ); + } + if ( isset( $options['username'] ) && isset( $options['password'] ) ) { + $this->setHeader( + 'Authorization', + 'Basic ' . base64_encode( $options['username'] . ':' . $options['password'] ) + ); + } + if ( isset( $options['originalRequest'] ) ) { + $this->setOriginalRequest( $options['originalRequest'] ); + } + + $members = [ "postData", "proxy", "noProxy", "sslVerifyHost", "caInfo", + "method", "followRedirects", "maxRedirects", "sslVerifyCert", "callback" ]; + + foreach ( $members as $o ) { + if ( isset( $options[$o] ) ) { + // ensure that MWHttpRequest::method is always + // uppercased. T38137 + if ( $o == 'method' ) { + $options[$o] = strtoupper( $options[$o] ); + } + $this->$o = $options[$o]; + } + } + + if ( $this->noProxy ) { + $this->proxy = ''; // noProxy takes precedence + } + + // Profile based on what's calling us + $this->profiler = $profiler; + $this->profileName = $caller; + } + + /** + * @param LoggerInterface $logger + */ + public function setLogger( LoggerInterface $logger ) { + $this->logger = $logger; + } + + /** + * 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 string $url Url to use + * @param array $options (optional) extra params to pass (see Http::request()) + * @param string $caller The method making this request, for profiling + * @throws DomainException + * @return CurlHttpRequest|PhpHttpRequest + * @see MWHttpRequest::__construct + */ + public static function factory( $url, $options = null, $caller = __METHOD__ ) { + if ( !Http::$httpEngine ) { + Http::$httpEngine = function_exists( 'curl_init' ) ? 'curl' : 'php'; + } elseif ( Http::$httpEngine == 'curl' && !function_exists( 'curl_init' ) ) { + throw new DomainException( __METHOD__ . ': curl (http://php.net/curl) is not installed, but' . + ' Http::$httpEngine is set to "curl"' ); + } + + if ( !is_array( $options ) ) { + $options = []; + } + + if ( !isset( $options['logger'] ) ) { + $options['logger'] = LoggerFactory::getInstance( 'http' ); + } + + switch ( Http::$httpEngine ) { + case 'curl': + return new CurlHttpRequest( $url, $options, $caller, Profiler::instance() ); + case 'php': + if ( !wfIniGetBool( 'allow_url_fopen' ) ) { + throw new DomainException( __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, $caller, Profiler::instance() ); + default: + throw new DomainException( __METHOD__ . ': The setting of Http::$httpEngine is not valid.' ); + } + } + + /** + * 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 array $args + * @todo overload the args param + */ + public function setData( $args ) { + $this->postData = $args; + } + + /** + * Take care of setting up the proxy (do nothing if "noProxy" is set) + * + * @return void + */ + protected function proxySetup() { + // If there is an explicit proxy set and proxies are not disabled, then use it + if ( $this->proxy && !$this->noProxy ) { + return; + } + + // Otherwise, fallback to $wgHTTPProxy if this is not a machine + // local URL and proxies are not disabled + if ( self::isLocalURL( $this->url ) || $this->noProxy ) { + $this->proxy = ''; + } else { + $this->proxy = Http::getProxy(); + } + } + + /** + * Check if the URL can be served by localhost + * + * @param string $url Full url to check + * @return bool + */ + private static function isLocalURL( $url ) { + global $wgCommandLineMode, $wgLocalVirtualHosts; + + if ( $wgCommandLineMode ) { + return false; + } + + // Extract host part + $matches = []; + if ( preg_match( '!^https?://([\w.-]+)[/:].*$!', $url, $matches ) ) { + $host = $matches[1]; + // Split up dotwise + $domainParts = explode( '.', $host ); + // Check if this domain or any superdomain is listed as a local virtual host + $domainParts = array_reverse( $domainParts ); + + $domain = ''; + $countParts = count( $domainParts ); + for ( $i = 0; $i < $countParts; $i++ ) { + $domainPart = $domainParts[$i]; + if ( $i == 0 ) { + $domain = $domainPart; + } else { + $domain = $domainPart . '.' . $domain; + } + + if ( in_array( $domain, $wgLocalVirtualHosts ) ) { + return true; + } + } + } + + return false; + } + + /** + * Set the user agent + * @param string $UA + */ + public function setUserAgent( $UA ) { + $this->setHeader( 'User-Agent', $UA ); + } + + /** + * Set an arbitrary header + * @param string $name + * @param string $value + */ + public function setHeader( $name, $value ) { + // I feel like I should normalize the case here... + $this->reqHeaders[$name] = $value; + } + + /** + * Get an array of the headers + * @return array + */ + protected function getHeaderList() { + $list = []; + + 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; + } + + /** + * 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 callable|null $callback + * @throws InvalidArgumentException + */ + public function setCallback( $callback ) { + if ( is_null( $callback ) ) { + $callback = [ $this, 'read' ]; + } elseif ( !is_callable( $callback ) ) { + throw new InvalidArgumentException( __METHOD__ . ': invalid callback' ); + } + $this->callback = $callback; + } + + /** + * A generic callback to read the body of the response from a remote + * server. + * + * @param resource $fh + * @param string $content + * @return int + * @internal + */ + public function read( $fh, $content ) { + $this->content .= $content; + return strlen( $content ); + } + + /** + * Take care of whatever is necessary to perform the URI request. + * + * @return StatusValue + * @note currently returns Status for B/C + */ + public function execute() { + throw new LogicException( 'children must override this' ); + } + + protected function prepare() { + $this->content = ""; + + if ( strtoupper( $this->method ) == "HEAD" ) { + $this->headersOnly = true; + } + + $this->proxySetup(); // set up any proxy as needed + + if ( !$this->callback ) { + $this->setCallback( null ); + } + + if ( !isset( $this->reqHeaders['User-Agent'] ) ) { + $this->setUserAgent( Http::userAgent() ); + } + } + + /** + * Parses the headers, including the HTTP status code and any + * Set-Cookie headers. This function expects the headers to be + * found in an array in the member variable headerList. + */ + 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] ); + } + } + + $this->parseCookies(); + } + + /** + * Sets HTTPRequest status member to a fatal value with the error + * message if the returned integer value of the status code was + * not successful (< 300) or a redirect (>=300 and < 400). (see + * RFC2616, section 10, + * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html for a + * list of status codes.) + */ + protected function setStatus() { + if ( !$this->respHeaders ) { + $this->parseHeader(); + } + + if ( (int)$this->respStatus > 399 ) { + list( $code, $message ) = explode( " ", $this->respStatus, 2 ); + $this->status->fatal( "http-bad-status", $code, $message ); + } + } + + /** + * Get the integer value of the HTTP status code (e.g. 200 for "200 Ok") + * (see RFC2616, section 10, http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html + * for a list of status codes.) + * + * @return int + */ + public function getStatus() { + if ( !$this->respHeaders ) { + $this->parseHeader(); + } + + return (int)$this->respStatus; + } + + /** + * 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 && $status <= 303 ) { + return true; + } + + return false; + } + + /** + * Returns an associative array of response headers after the + * request has been executed. Because some headers + * (e.g. Set-Cookie) can appear more than once the, each value of + * the associative array is an array of the values given. + * + * @return array + */ + public function getResponseHeaders() { + if ( !$this->respHeaders ) { + $this->parseHeader(); + } + + return $this->respHeaders; + } + + /** + * Returns the value of the given response header. + * + * @param string $header + * @return string|null + */ + 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. + * + * To read response cookies from the jar, getCookieJar must be called first. + * + * @param CookieJar $jar + */ + 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 string $name + * @param string $value + * @param array $attr + */ + public function setCookie( $name, $value, $attr = [] ) { + if ( !$this->cookieJar ) { + $this->cookieJar = new CookieJar; + } + + if ( $this->parsedUrl && !isset( $attr['domain'] ) ) { + $attr['domain'] = $this->parsedUrl['host']; + } + + $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 T31232 is taken care of (high-level redirect + * handling rewrite). + * + * @return string + */ + public function getFinalUrl() { + $headers = $this->getResponseHeaders(); + + // return full url (fix for incorrect but handled relative location) + if ( isset( $headers['location'] ) ) { + $locations = $headers['location']; + $domain = ''; + $foundRelativeURI = false; + $countLocations = count( $locations ); + + for ( $i = $countLocations - 1; $i >= 0; $i-- ) { + $url = parse_url( $locations[$i] ); + + if ( isset( $url['host'] ) ) { + $domain = $url['scheme'] . '://' . $url['host']; + break; // found correct URI (with host) + } else { + $foundRelativeURI = true; + } + } + + if ( !$foundRelativeURI ) { + return $locations[$countLocations - 1]; + } + if ( $domain ) { + return $domain . $locations[$countLocations - 1]; + } + $url = parse_url( $this->url ); + if ( isset( $url['host'] ) ) { + return $url['scheme'] . '://' . $url['host'] . + $locations[$countLocations - 1]; + } + } + + return $this->url; + } + + /** + * Returns true if the backend can follow redirects. Overridden by the + * child classes. + * @return bool + */ + public function canFollowRedirects() { + return true; + } + + /** + * Set information about the original request. This can be useful for + * endpoints/API modules which act as a proxy for some service, and + * throttling etc. needs to happen in that service. + * Calling this will result in the X-Forwarded-For and X-Original-User-Agent + * headers being set. + * @param WebRequest|array $originalRequest When in array form, it's + * expected to have the keys 'ip' and 'userAgent'. + * @note IP/user agent is personally identifiable information, and should + * only be set when the privacy policy of the request target is + * compatible with that of the MediaWiki installation. + */ + public function setOriginalRequest( $originalRequest ) { + if ( $originalRequest instanceof WebRequest ) { + $originalRequest = [ + 'ip' => $originalRequest->getIP(), + 'userAgent' => $originalRequest->getHeader( 'User-Agent' ), + ]; + } elseif ( + !is_array( $originalRequest ) + || array_diff( [ 'ip', 'userAgent' ], array_keys( $originalRequest ) ) + ) { + throw new InvalidArgumentException( __METHOD__ . ': $originalRequest must be a ' + . "WebRequest or an array with 'ip' and 'userAgent' keys" ); + } + + $this->reqHeaders['X-Forwarded-For'] = $originalRequest['ip']; + $this->reqHeaders['X-Original-User-Agent'] = $originalRequest['userAgent']; + } +}