X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FNetwork%2FHTTPRequest.php;h=c37db4c9a06a7aa5abe6a14e98076853c63f4001;hb=16f5965fd444a4f5d28906623a5ceba5e41cc138;hp=1a0048b2a135d9e22e3e9668680873eb7823d396;hpb=57587efe58e1ef9aed8c218f264696b10b331fea;p=friendica.git

diff --git a/src/Network/HTTPRequest.php b/src/Network/HTTPRequest.php
index 1a0048b2a1..c37db4c9a0 100644
--- a/src/Network/HTTPRequest.php
+++ b/src/Network/HTTPRequest.php
@@ -21,18 +21,25 @@
 
 namespace Friendica\Network;
 
+use DOMDocument;
+use DomXPath;
 use Friendica\App;
 use Friendica\Core\Config\IConfig;
-use Friendica\Core\Logger;
 use Friendica\Core\System;
 use Friendica\Util\Network;
 use Friendica\Util\Profiler;
+use GuzzleHttp\Client;
+use GuzzleHttp\Exception\RequestException;
+use GuzzleHttp\Exception\TransferException;
+use Psr\Http\Message\RequestInterface;
+use Psr\Http\Message\ResponseInterface;
+use Psr\Http\Message\UriInterface;
 use Psr\Log\LoggerInterface;
 
 /**
  * Performs HTTP requests to a given URL
  */
-class HTTPRequest
+class HTTPRequest implements IHTTPRequest
 {
 	/** @var LoggerInterface */
 	private $logger;
@@ -52,25 +59,9 @@ class HTTPRequest
 	}
 
 	/**
-	 * fetches an URL.
-	 *
-	 * @param string $url        URL to fetch
-	 * @param bool   $binary     default false
-	 *                           TRUE if asked to return binary results (file download)
-	 * @param array  $opts       (optional parameters) assoziative array with:
-	 *                           'accept_content' => supply Accept: header with 'accept_content' as the value
-	 *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
-	 *                           'http_auth' => username:password
-	 *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
-	 *                           'nobody' => only return the header
-	 *                           'cookiejar' => path to cookie jar file
-	 *                           'header' => header array
-	 * @param int    $redirects  The recursion counter for internal use - default 0
-	 *
-	 * @return CurlResult
-	 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
+	 * {@inheritDoc}
 	 */
-	public function curl(string $url, bool $binary = false, array $opts = [], int &$redirects = 0)
+	public function get(string $url, bool $binary = false, array $opts = [])
 	{
 		$stamp1 = microtime(true);
 
@@ -97,134 +88,147 @@ class HTTPRequest
 			return CurlResult::createErrorCurl($url);
 		}
 
-		$ch = @curl_init($url);
+		$curlOptions = [];
 
-		if (($redirects > 8) || (!$ch)) {
-			return CurlResult::createErrorCurl($url);
-		}
-
-		@curl_setopt($ch, CURLOPT_HEADER, true);
+		$curlOptions[CURLOPT_HEADER] = true;
 
 		if (!empty($opts['cookiejar'])) {
-			curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
-			curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
+			$curlOptions[CURLOPT_COOKIEJAR] = $opts["cookiejar"];
+			$curlOptions[CURLOPT_COOKIEFILE] = $opts["cookiejar"];
 		}
 
 		// These settings aren't needed. We're following the location already.
-		//	@curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
-		//	@curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
+		//	$curlOptions[CURLOPT_FOLLOWLOCATION] =true;
+		//	$curlOptions[CURLOPT_MAXREDIRS] = 5;
 
 		if (!empty($opts['accept_content'])) {
-			curl_setopt(
-				$ch,
-				CURLOPT_HTTPHEADER,
-				['Accept: ' . $opts['accept_content']]
-			);
+			if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
+				$curlOptions[CURLOPT_HTTPHEADER] = [];
+			}
+			array_push($curlOptions[CURLOPT_HTTPHEADER], 'Accept: ' . $opts['accept_content']);
 		}
 
 		if (!empty($opts['header'])) {
-			curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['header']);
+			if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
+				$curlOptions[CURLOPT_HTTPHEADER] = [];
+			}
+			$curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['header'], $curlOptions[CURLOPT_HTTPHEADER]);
 		}
 
-		@curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
-		@curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
+		$curlOptions[CURLOPT_RETURNTRANSFER] = true;
+		$curlOptions[CURLOPT_USERAGENT] = $this->getUserAgent();
 
 		$range = intval($this->config->get('system', 'curl_range_bytes', 0));
 
 		if ($range > 0) {
-			@curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
+			$curlOptions[CURLOPT_RANGE] = '0-' . $range;
 		}
 
 		// Without this setting it seems as if some webservers send compressed content
 		// This seems to confuse curl so that it shows this uncompressed.
 		/// @todo  We could possibly set this value to "gzip" or something similar
-		curl_setopt($ch, CURLOPT_ENCODING, '');
+		$curlOptions[CURLOPT_ENCODING] = '';
 
 		if (!empty($opts['headers'])) {
-			@curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
+			if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
+				$curlOptions[CURLOPT_HTTPHEADER] = [];
+			}
+			$curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['headers'], $curlOptions[CURLOPT_HTTPHEADER]);
 		}
 
 		if (!empty($opts['nobody'])) {
-			@curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
+			$curlOptions[CURLOPT_NOBODY] = $opts['nobody'];
 		}
 
+		$curlOptions[CURLOPT_CONNECTTIMEOUT] = 10;
+
 		if (!empty($opts['timeout'])) {
-			@curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
+			$curlOptions[CURLOPT_TIMEOUT] = $opts['timeout'];
 		} else {
 			$curl_time = $this->config->get('system', 'curl_timeout', 60);
-			@curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
+			$curlOptions[CURLOPT_TIMEOUT] = intval($curl_time);
 		}
 
 		// by default we will allow self-signed certs
 		// but you can override this
 
 		$check_cert = $this->config->get('system', 'verifyssl');
-		@curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
+		$curlOptions[CURLOPT_SSL_VERIFYPEER] = ($check_cert) ? true : false;
 
 		if ($check_cert) {
-			@curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
+			$curlOptions[CURLOPT_SSL_VERIFYHOST] = 2;
 		}
 
 		$proxy = $this->config->get('system', 'proxy');
 
 		if (!empty($proxy)) {
-			@curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
-			@curl_setopt($ch, CURLOPT_PROXY, $proxy);
+			$curlOptions[CURLOPT_HTTPPROXYTUNNEL] = 1;
+			$curlOptions[CURLOPT_PROXY] = $proxy;
 			$proxyuser = $this->config->get('system', 'proxyuser');
 
 			if (!empty($proxyuser)) {
-				@curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
+				$curlOptions[CURLOPT_PROXYUSERPWD] = $proxyuser;
 			}
 		}
 
 		if ($this->config->get('system', 'ipv4_resolve', false)) {
-			curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
+			$curlOptions[CURLOPT_IPRESOLVE] = CURL_IPRESOLVE_V4;
 		}
 
 		if ($binary) {
-			@curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
+			$curlOptions[CURLOPT_BINARYTRANSFER] = 1;
 		}
 
-		// don't let curl abort the entire application
-		// if it throws any errors.
+		$logger = $this->logger;
 
-		$s         = @curl_exec($ch);
-		$curl_info = @curl_getinfo($ch);
+		$onRedirect = function(
+			RequestInterface $request,
+			ResponseInterface $response,
+			UriInterface $uri
+		) use ($logger) {
+			$logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
+		};
 
-		// Special treatment for HTTP Code 416
-		// See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
-		if (($curl_info['http_code'] == 416) && ($range > 0)) {
-			@curl_setopt($ch, CURLOPT_RANGE, '');
-			$s         = @curl_exec($ch);
-			$curl_info = @curl_getinfo($ch);
-		}
-
-		$curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
-
-		if ($curlResponse->isRedirectUrl()) {
-			$redirects++;
-			$this->logger->notice('Curl redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
-			@curl_close($ch);
-			return self::curl($curlResponse->getRedirectUrl(), $binary, $opts, $redirects);
+		$onHeaders = function (ResponseInterface $response) use ($opts) {
+			if (!empty($opts['content_length']) &&
+				$response->getHeaderLine('Content-Length') > $opts['content_length']) {
+				throw new TransferException('The file is too big!');
+			}
+		};
+
+		$client = new Client([
+			'allow_redirect' => [
+				'max' => 8,
+				'on_redirect' => $onRedirect,
+				'track_redirect' => true,
+				'strict' => true,
+				'referer' => true,
+			],
+			'on_headers' => $onHeaders,
+			'sink' => tempnam(get_temppath(), 'guzzle'),
+			'curl' => $curlOptions
+		]);
+
+		try {
+			$response = $client->get($url);
+			return new GuzzleResponse($response, $url);
+		} catch (TransferException $exception) {
+			if ($exception instanceof RequestException &&
+				$exception->hasResponse()) {
+				return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), '');
+			} else {
+				return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), '');
+			}
+		} finally {
+			$this->profiler->saveTimestamp($stamp1, 'network');
 		}
-
-		@curl_close($ch);
-
-		$this->profiler->saveTimestamp($stamp1, 'network', System::callstack());
-
-		return $curlResponse;
 	}
 
 	/**
-	 * Send POST request to $url
+	 * {@inheritDoc}
 	 *
-	 * @param string $url       URL to post
-	 * @param mixed  $params    array of POST variables
-	 * @param array  $headers   HTTP headers
-	 * @param int    $redirects Recursion counter for internal use - default = 0
-	 * @param int    $timeout   The timeout in seconds, default system config value or 60 seconds
+	 * @param int $redirects The recursion counter for internal use - default 0
 	 *
-	 * @return CurlResult The content
 	 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 	 */
 	public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
@@ -248,12 +252,14 @@ class HTTPRequest
 		curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 		curl_setopt($ch, CURLOPT_POST, 1);
 		curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
-		curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
+		curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 
 		if ($this->config->get('system', 'ipv4_resolve', false)) {
 			curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 		}
 
+		@curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
+
 		if (intval($timeout)) {
 			curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 		} else {
@@ -292,16 +298,16 @@ class HTTPRequest
 
 		$curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 
-		if ($curlResponse->isRedirectUrl()) {
+		if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 			$redirects++;
 			$this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 			curl_close($ch);
-			return self::post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
+			return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 		}
 
 		curl_close($ch);
 
-		$this->profiler->saveTimestamp($stamp1, 'network', System::callstack());
+		$this->profiler->saveTimestamp($stamp1, 'network');
 
 		// Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 		if ($curlResponse->getReturnCode() == 417) {
@@ -314,75 +320,168 @@ class HTTPRequest
 					array_push($headers, 'Expect:');
 				}
 			}
-			Logger::info('Server responds with 417, applying workaround', ['url' => $url]);
-			return self::post($url, $params, $headers, $redirects, $timeout);
+			$this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
+			return $this->post($url, $params, $headers, $redirects, $timeout);
 		}
 
-		Logger::log('post_url: end ' . $url, Logger::DATA);
+		$this->logger->debug('Post_url: End.', ['url' => $url]);
 
 		return $curlResponse;
 	}
 
 	/**
-	 * Curl wrapper
-	 *
-	 * If binary flag is true, return binary results.
-	 * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
-	 * to preserve cookies from one request to the next.
+	 * {@inheritDoc}
+	 */
+	public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
+	{
+		if (Network::isUrlBlocked($url)) {
+			$this->logger->info('Domain is blocked.', ['url' => $url]);
+			return $url;
+		}
+
+		if (Network::isRedirectBlocked($url)) {
+			$this->logger->info('Domain should not be redirected.', ['url' => $url]);
+			return $url;
+		}
+
+		$url = Network::stripTrackingQueryParams($url);
+
+		if ($depth > 10) {
+			return $url;
+		}
+
+		$url = trim($url, "'");
+
+		$stamp1 = microtime(true);
+
+		$ch = curl_init();
+		curl_setopt($ch, CURLOPT_URL, $url);
+		curl_setopt($ch, CURLOPT_HEADER, 1);
+		curl_setopt($ch, CURLOPT_NOBODY, 1);
+		curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
+		curl_setopt($ch, CURLOPT_TIMEOUT, 10);
+		curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
+		curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
+
+		curl_exec($ch);
+		$curl_info = @curl_getinfo($ch);
+		$http_code = $curl_info['http_code'];
+		curl_close($ch);
+
+		$this->profiler->saveTimestamp($stamp1, "network");
+
+		if ($http_code == 0) {
+			return $url;
+		}
+
+		if (in_array($http_code, ['301', '302'])) {
+			if (!empty($curl_info['redirect_url'])) {
+				return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
+			} elseif (!empty($curl_info['location'])) {
+				return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
+			}
+		}
+
+		// Check for redirects in the meta elements of the body if there are no redirects in the header.
+		if (!$fetchbody) {
+			return $this->finalUrl($url, ++$depth, true);
+		}
+
+		// if the file is too large then exit
+		if ($curl_info["download_content_length"] > 1000000) {
+			return $url;
+		}
+
+		// if it isn't a HTML file then exit
+		if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
+			return $url;
+		}
+
+		$stamp1 = microtime(true);
+
+		$ch = curl_init();
+		curl_setopt($ch, CURLOPT_URL, $url);
+		curl_setopt($ch, CURLOPT_HEADER, 0);
+		curl_setopt($ch, CURLOPT_NOBODY, 0);
+		curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
+		curl_setopt($ch, CURLOPT_TIMEOUT, 10);
+		curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
+		curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
+
+		$body = curl_exec($ch);
+		curl_close($ch);
+
+		$this->profiler->saveTimestamp($stamp1, "network");
+
+		if (trim($body) == "") {
+			return $url;
+		}
+
+		// Check for redirect in meta elements
+		$doc = new DOMDocument();
+		@$doc->loadHTML($body);
+
+		$xpath = new DomXPath($doc);
+
+		$list = $xpath->query("//meta[@content]");
+		foreach ($list as $node) {
+			$attr = [];
+			if ($node->attributes->length) {
+				foreach ($node->attributes as $attribute) {
+					$attr[$attribute->name] = $attribute->value;
+				}
+			}
+
+			if (@$attr["http-equiv"] == 'refresh') {
+				$path = $attr["content"];
+				$pathinfo = explode(";", $path);
+				foreach ($pathinfo as $value) {
+					if (substr(strtolower($value), 0, 4) == "url=") {
+						return $this->finalUrl(substr($value, 4), ++$depth);
+					}
+				}
+			}
+		}
+
+		return $url;
+	}
+
+	/**
+	 * {@inheritDoc}
 	 *
-	 * @param string $url             URL to fetch
-	 * @param bool   $binary          default false
-	 *                                TRUE if asked to return binary results (file download)
-	 * @param int    $timeout         Timeout in seconds, default system config value or 60 seconds
-	 * @param string $accept_content  supply Accept: header with 'accept_content' as the value
-	 * @param string $cookiejar       Path to cookie jar file
-	 * @param int    $redirects       The recursion counter for internal use - default 0
+	 * @param int $redirects The recursion counter for internal use - default 0
 	 *
-	 * @return string The fetched content
 	 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 	 */
-	public function fetchUrl(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
+	public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 	{
-		$ret = $this->fetchUrlFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
+		$ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 
 		return $ret->getBody();
 	}
 
 	/**
-	 * Curl wrapper with array of return values.
-	 *
-	 * Inner workings and parameters are the same as @ref fetchUrl but returns an array with
-	 * all the information collected during the fetch.
+	 * {@inheritDoc}
 	 *
-	 * @param string $url             URL to fetch
-	 * @param bool   $binary          default false
-	 *                                TRUE if asked to return binary results (file download)
-	 * @param int    $timeout         Timeout in seconds, default system config value or 60 seconds
-	 * @param string $accept_content  supply Accept: header with 'accept_content' as the value
-	 * @param string $cookiejar       Path to cookie jar file
-	 * @param int    $redirects       The recursion counter for internal use - default 0
+	 * @param int $redirects The recursion counter for internal use - default 0
 	 *
-	 * @return CurlResult With all relevant information, 'body' contains the actual fetched content.
 	 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 	 */
-	public function fetchUrlFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
+	public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 	{
-		return $this->curl(
+		return $this->get(
 			$url,
 			$binary,
 			[
 				'timeout'        => $timeout,
 				'accept_content' => $accept_content,
 				'cookiejar'      => $cookiejar
-			],
-			$redirects
+			]
 		);
 	}
 
 	/**
-	 * Returns the current UserAgent as a String
-	 *
-	 * @return string the UserAgent as a String
+	 * {@inheritDoc}
 	 */
 	public function getUserAgent()
 	{