Do only follow HTTP and HTTPS redirects

We do not want to follow redirects to other protocols since they might allow an adversary to bypass network restrictions. (i.e. a redirect to ftp:// might be used to access files of a FTP server which might be in a secure zone and not be reachable from the net but from the ownCloud server)

Get final redirect manually using get_headers()

Migrate to HTTPHelper class and add unit tests

1 files changed, 177 insertions, 0 deletions

diff --git a/lib/private/httphelper.php b/lib/private/httphelper.php
new file mode 100644
index 00000000000..8b7aebb3d4d
--- /dev/null
+++ b/lib/private/httphelper.php
@@ -0,0 +1,177 @@
+<?php
+/**
+ * Copyright (c) 2014 Lukas Reschke <lukas@owncloud.com>
+ * This file is licensed under the Affero General Public License version 3 or
+ * later.
+ * See the COPYING-README file.
+ */
+
+namespace OC;
+
+class HTTPHelper {
+	const USER_AGENT = 'ownCloud Server Crawler';
+
+	/** @var \OC\AllConfig */
+	private $config;
+
+	/**
+	 * @param \OC\AllConfig $config
+	 */
+	public function __construct(AllConfig $config) {
+		$this->config = $config;
+	}
+
+	/**
+	 * Returns the default context array
+	 * @return array
+	 */
+	public function getDefaultContextArray() {
+		return array(
+			'http' => array(
+				'header' => 'User-Agent: ' . self::USER_AGENT . "\r\n",
+				'timeout' => 10,
+				'follow_location' => false, // Do not follow the location since we can't limit the protocol
+			),
+			'ssl' => array(
+				'disable_compression' => true
+			)
+		);
+	}
+
+	/**
+	 * Get URL content
+	 * @param string $url Url to get content
+	 * @throws \Exception If the URL does not start with http:// or https://
+	 * @return string of the response or false on error
+	 * This function get the content of a page via curl, if curl is enabled.
+	 * If not, file_get_contents is used.
+	 */
+	public function getUrlContent($url) {
+		if (!$this->isHTTPURL($url)) {
+			throw new \Exception('$url must start with https:// or http://', 1);
+		}
+
+		$proxy = $this->config->getSystemValue('proxy', null);
+		$proxyUserPwd = $this->config->getSystemValue('proxyuserpwd', null);
+		if (function_exists('curl_init')) {
+			$curl = curl_init();
+			$max_redirects = 10;
+
+			curl_setopt($curl, CURLOPT_HEADER, 0);
+			curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1);
+			curl_setopt($curl, CURLOPT_CONNECTTIMEOUT, 10);
+			curl_setopt($curl, CURLOPT_URL, $url);
+			curl_setopt($curl, CURLOPT_PROTOCOLS,  CURLPROTO_HTTP | CURLPROTO_HTTPS);
+			curl_setopt($curl, CURLOPT_REDIR_PROTOCOLS,  CURLPROTO_HTTP | CURLPROTO_HTTPS);
+
+			curl_setopt($curl, CURLOPT_USERAGENT, self::USER_AGENT);
+			if ($proxy !== null) {
+				curl_setopt($curl, CURLOPT_PROXY, $proxy);
+			}
+			if ($proxyUserPwd !== null) {
+				curl_setopt($curl, CURLOPT_PROXYUSERPWD, $proxyUserPwd);
+			}
+
+			if (ini_get('open_basedir') === '' && (ini_get('safe_mode') === false) || strtolower(ini_get('safe_mode')) === 'off') {
+				curl_setopt($curl, CURLOPT_FOLLOWLOCATION, true);
+				curl_setopt($curl, CURLOPT_MAXREDIRS, $max_redirects);
+				$data = curl_exec($curl);
+			} else {
+				curl_setopt($curl, CURLOPT_FOLLOWLOCATION, false);
+				$mr = $max_redirects;
+				if ($mr > 0) {
+					$newURL = curl_getinfo($curl, CURLINFO_EFFECTIVE_URL);
+					$rcurl = curl_copy_handle($curl);
+					curl_setopt($rcurl, CURLOPT_HEADER, true);
+					curl_setopt($rcurl, CURLOPT_NOBODY, true);
+					curl_setopt($rcurl, CURLOPT_FORBID_REUSE, false);
+					curl_setopt($rcurl, CURLOPT_RETURNTRANSFER, true);
+					curl_setopt($rcurl, CURLOPT_USERAGENT, self::USER_AGENT);
+					do {
+						curl_setopt($rcurl, CURLOPT_URL, $newURL);
+						$header = curl_exec($rcurl);
+						if (curl_errno($rcurl)) {
+							$code = 0;
+						} else {
+							$code = curl_getinfo($rcurl, CURLINFO_HTTP_CODE);
+							if ($code == 301 || $code == 302) {
+								preg_match('/Location:(.*?)\n/', $header, $matches);
+								$newURL = trim(array_pop($matches));
+							} else {
+								$code = 0;
+							}
+						}
+					} while ($code && --$mr);
+					curl_close($rcurl);
+					if ($mr > 0) {
+						curl_setopt($curl, CURLOPT_URL, $newURL);
+					}
+				}
+
+				if ($mr == 0 && $max_redirects > 0) {
+					$data = false;
+				} else {
+					$data = curl_exec($curl);
+				}
+			}
+			curl_close($curl);
+		} else {
+			$url = $this->getFinalLocationOfURL($url);
+			$contextArray = $this->getDefaultContextArray();
+
+			if ($proxy !== null) {
+				$contextArray['http']['proxy'] = $proxy;
+			}
+
+			$ctx = stream_context_create(
+				$contextArray
+			);
+			$data = @file_get_contents($url, 0, $ctx);
+
+		}
+		return $data;
+	}
+
+	/**
+	 * Returns the response headers of a HTTP URL without following redirects
+	 * @param string $location Needs to be a HTTPS or HTTP URL
+	 * @return array
+	 */
+	public function getHeaders($location) {
+		stream_context_set_default($this->getDefaultContextArray());
+		return get_headers($location, 1);
+	}
+
+	/**
+	 * Checks whether the supplied URL begins with HTTPS:// or HTTP:// (case insensitive)
+	 * @param string $url
+	 * @return bool
+	 */
+	public function isHTTPURL($url) {
+		return stripos($url, 'https://') === 0 || stripos($url, 'http://') === 0;
+	}
+
+	/**
+	 * Returns the last HTTP or HTTPS site the request has been redirected too using the Location HTTP header
+	 * This is a very ugly workaround about the missing functionality to restrict fopen() to protocols
+	 * @param string $location Needs to be a HTTPS or HTTP URL
+	 * @throws \Exception In case the initial URL is not a HTTP or HTTPS one
+	 * @return string
+	 */
+	public function getFinalLocationOfURL($location) {
+		if(!$this->isHTTPURL($location)) {
+			throw new \Exception('URL must begin with HTTPS or HTTP.');
+		}
+		$headerArray = $this->getHeaders($location, 1);
+
+		if($headerArray !== false && isset($headerArray['Location'])) {
+			while($this->isHTTPURL($headerArray['Location'])) {
+				$location = $headerArray['Location'];
+				$headerArray = $this->getHeaders($location);
+			}
+		}
+
+		return $location;
+	}
+
+}