Merge remote-tracking branch 'FreshRSS/master'

1 files changed, 148 insertions, 66 deletions

diff --git a/lib/SimplePie/SimplePie.php b/lib/SimplePie/SimplePie.php
index d7aaeb0c5..0f2fdbb87 100644
--- a/lib/SimplePie/SimplePie.php
+++ b/lib/SimplePie/SimplePie.php
@@ -75,6 +75,12 @@ define('SIMPLEPIE_USERAGENT', SIMPLEPIE_NAME . '/' . SIMPLEPIE_VERSION . ' (Feed
 define('SIMPLEPIE_LINKBACK', '<a href="' . SIMPLEPIE_URL . '" title="' . SIMPLEPIE_NAME . ' ' . SIMPLEPIE_VERSION . '">' . SIMPLEPIE_NAME . '</a>');
 
 /**
+ * Use syslog to report HTTP requests done by SimplePie.
+ * @see SimplePie::set_syslog()
+ */
+define('SIMPLEPIE_SYSLOG', true);	//FreshRSS
+
+/**
  * No Autodiscovery
  * @see SimplePie::set_autodiscovery_level()
  */
@@ -446,6 +452,13 @@ class SimplePie
 	public $feed_url;
 
 	/**
+	 * @var string Original feed URL, or new feed URL iff HTTP 301 Moved Permanently
+	 * @see SimplePie::subscribe_url()
+	 * @access private
+	 */
+	public $permanent_url = null;
+
+	/**
 	 * @var object Instance of SimplePie_File to use as a feed
 	 * @see SimplePie::set_file()
 	 * @access private
@@ -467,6 +480,13 @@ class SimplePie
 	public $timeout = 10;
 
 	/**
+	 * @var array Custom curl options
+	 * @see SimplePie::set_curl_options()
+	 * @access private
+	 */
+	public $curl_options = array();
+
+	/**
 	 * @var bool Forces fsockopen() to be used for remote files instead
 	 * of cURL, even if a new enough version is installed
 	 * @see SimplePie::force_fsockopen()
@@ -616,6 +636,12 @@ class SimplePie
 	public $strip_htmltags = array('base', 'blink', 'body', 'doctype', 'embed', 'font', 'form', 'frame', 'frameset', 'html', 'iframe', 'input', 'marquee', 'meta', 'noscript', 'object', 'param', 'script', 'style');
 
 	/**
+	 * Use syslog to report HTTP requests done by SimplePie.
+	 * @see SimplePie::set_syslog()
+	 */
+	public $syslog_enabled = SIMPLEPIE_SYSLOG;
+
+	/**
 	 * The SimplePie class contains feed level data and options
 	 *
 	 * To use SimplePie, create the SimplePie object with no parameters. You can
@@ -735,6 +761,7 @@ class SimplePie
 		else
 		{
 			$this->feed_url = $this->registry->call('Misc', 'fix_protocol', array($url, 1));
+			$this->permanent_url = $this->feed_url;
 		}
 	}
 
@@ -749,6 +776,7 @@ class SimplePie
 		if ($file instanceof SimplePie_File)
 		{
 			$this->feed_url = $file->url;
+			$this->permanent_url = $this->feed_url;
 			$this->file =& $file;
 			return true;
 		}
@@ -786,6 +814,19 @@ class SimplePie
 	{
 		$this->timeout = (int) $timeout;
 	}
+    
+	/**
+	 * Set custom curl options
+	 *
+	 * This allows you to change default curl options
+	 *
+	 * @since 1.0 Beta 3
+	 * @param array $curl_options Curl options to add to default settings
+	 */
+	public function set_curl_options(array $curl_options = array())
+	{
+		$this->curl_options = $curl_options;
+	}
 
 	/**
 	 * Force SimplePie to use fsockopen() instead of cURL
@@ -1082,6 +1123,7 @@ class SimplePie
 			$this->strip_attributes(false);
 			$this->add_attributes(false);
 			$this->set_image_handler(false);
+			$this->set_https_domains(array());
 		}
 	}
 
@@ -1127,7 +1169,7 @@ class SimplePie
 		$this->sanitize->strip_attributes($attribs);
 	}
 
-	public function add_attributes($attribs = '')
+	public function add_attributes($attribs = '')	//FreshRSS
 	{
 		if ($attribs === '')
 		{
@@ -1137,6 +1179,14 @@ class SimplePie
 	}
 
 	/**
+	 * Use syslog to report HTTP requests done by SimplePie.
+	 */
+	public function set_syslog($value = SIMPLEPIE_SYSLOG)	//FreshRSS
+	{
+		$this->syslog_enabled = $value == true;
+	}
+
+	/**
 	 * Set the output encoding
 	 *
 	 * Allows you to override SimplePie's output to match that of your webpage.
@@ -1185,6 +1235,19 @@ class SimplePie
 	}
 
 	/**
+	 * Set the list of domains for which force HTTPS.
+	 * @see SimplePie_Sanitize::set_https_domains()
+	 * FreshRSS
+	 */
+	public function set_https_domains($domains = array())
+	{
+		if (is_array($domains))
+		{
+			$this->sanitize->set_https_domains($domains);
+		}
+	}
+
+	/**
 	 * Set the handler to enable the display of cached images.
 	 *
 	 * @param str $page Web-accessible path to the handler_image.php file.
@@ -1222,7 +1285,8 @@ class SimplePie
 		$this->enable_exceptions = $enable;
 	}
 
-	function cleanMd5($rss) {	//FreshRSS
+	function cleanMd5($rss)
+	{
 		return md5(preg_replace(array('#<(lastBuildDate|pubDate|updated|feedDate|dc:date|slash:comments)>[^<]+</\\1>#', '#<!--.+?-->#s'), '', $rss));
 	}
 
@@ -1240,6 +1304,7 @@ class SimplePie
 		// Check absolute bare minimum requirements.
 		if (!extension_loaded('xml') || !extension_loaded('pcre'))
 		{
+			$this->error = 'XML or PCRE extensions not loaded!';
 			return false;
 		}
 		// Then check the xml extension is sane (i.e., libxml 2.7.x issue on PHP < 5.2.9 and libxml 2.7.0 to 2.7.2 on any version) if we don't have xmlreader.
@@ -1267,7 +1332,7 @@ class SimplePie
 		// Pass whatever was set with config options over to the sanitizer.
 		// Pass the classes in for legacy support; new classes should use the registry instead
 		$this->sanitize->pass_cache_data($this->cache, $this->cache_location, $this->cache_name_function, $this->registry->get_class('Cache'));
-		$this->sanitize->pass_file_data($this->registry->get_class('File'), $this->timeout, $this->useragent, $this->force_fsockopen);
+		$this->sanitize->pass_file_data($this->registry->get_class('File'), $this->timeout, $this->useragent, $this->force_fsockopen, $this->curl_options);
 
 		if (!empty($this->multifeed_url))
 		{
@@ -1312,7 +1377,7 @@ class SimplePie
 			// Fetch the data via SimplePie_File into $this->raw_data
 			if (($fetched = $this->fetch_data($cache)) === true)
 			{
-				return $this->data['mtime'];	//FreshRSS
+				return $this->data['mtime'];
 			}
 			elseif ($fetched === false) {
 				return false;
@@ -1320,7 +1385,8 @@ class SimplePie
 
 			list($headers, $sniffed) = $fetched;
 
-			if (isset($this->data['md5'])) {	//FreshRSS
+			if (isset($this->data['md5']))
+			{
 				$md5 = $this->data['md5'];
 			}
 		}
@@ -1331,7 +1397,7 @@ class SimplePie
 		// First check to see if input has been overridden.
 		if ($this->input_encoding !== false)
 		{
-			$encodings[] = strtoupper($this->input_encoding);	//FreshRSS
+			$encodings[] = strtoupper($this->input_encoding);
 		}
 
 		$application_types = array('application/xml', 'application/xml-dtd', 'application/xml-external-parsed-entity');
@@ -1355,7 +1421,7 @@ class SimplePie
 			{
 				if (isset($headers['content-type']) && preg_match('/;\x20?charset=([^;]*)/i', $headers['content-type'], $charset))
 				{
-					$encodings[] = strtoupper($charset[1]);	//FreshRSS
+					$encodings[] = strtoupper($charset[1]);
 				}
 				else
 				{
@@ -1404,8 +1470,8 @@ class SimplePie
 						$this->data['headers'] = $headers;
 					}
 					$this->data['build'] = SIMPLEPIE_BUILD;
-					$this->data['mtime'] = time();	//FreshRSS
-					$this->data['md5'] = empty($md5) ? $this->cleanMd5($this->raw_data) : $md5;	//FreshRSS
+					$this->data['mtime'] = time();
+					$this->data['md5'] = empty($md5) ? $this->cleanMd5($this->raw_data) : $md5;
 
 					// Cache the file if caching is enabled
 					if ($cache && !$cache->save($this))
@@ -1420,7 +1486,7 @@ class SimplePie
 		if (isset($parser))
 		{
 			// We have an error, just set SimplePie_Misc::error to it and quit
-			$this->error = sprintf('This XML document is invalid, likely due to invalid characters. XML error: %s at line %d, column %d', $parser->get_error_string(), $parser->get_current_line(), $parser->get_current_column());
+			$this->error = sprintf('This XML document is invalid, likely due to invalid characters. XML error: %s at line %d, column %d, encoding %s, URL: %s', $parser->get_error_string(), $parser->get_current_line(), $parser->get_current_column(), $encoding, $this->feed_url);
 		}
 		else
 		{
@@ -1446,7 +1512,12 @@ class SimplePie
 		{
 			// Load the Cache
 			$this->data = $cache->load();
-			if (!empty($this->data))
+			if ($cache->mtime() + $this->cache_duration > time())
+			{
+				$this->raw_data = false;
+				return true;	// If the cache is still valid, just return true
+			}
+			elseif (!empty($this->data))
 			{
 				// If the cache is for an outdated build of SimplePie
 				if (!isset($this->data['build']) || $this->data['build'] !== SIMPLEPIE_BUILD)
@@ -1478,63 +1549,58 @@ class SimplePie
 					}
 				}
 				// Check if the cache has been updated
-				elseif ($cache->mtime() + $this->cache_duration < time())
+				else
 				{
-					// If we have last-modified and/or etag set
-					//if (isset($this->data['headers']['last-modified']) || isset($this->data['headers']['etag']))	//FreshRSS removed
+					$headers = array(
+						'Accept' => 'application/atom+xml, application/rss+xml, application/rdf+xml;q=0.9, application/xml;q=0.8, text/xml;q=0.8, text/html;q=0.7, unknown/unknown;q=0.1, application/unknown;q=0.1, */*;q=0.1',
+					);
+					if (isset($this->data['headers']['last-modified']))
 					{
-						$headers = array(
-							'Accept' => 'application/atom+xml, application/rss+xml, application/rdf+xml;q=0.9, application/xml;q=0.8, text/xml;q=0.8, text/html;q=0.7, unknown/unknown;q=0.1, application/unknown;q=0.1, */*;q=0.1',
-						);
-						if (isset($this->data['headers']['last-modified']))
-						{
-							$headers['if-modified-since'] = $this->data['headers']['last-modified'];
-						}
-						if (isset($this->data['headers']['etag']))
-						{
-							$headers['if-none-match'] = $this->data['headers']['etag'];
-						}
+						$headers['if-modified-since'] = $this->data['headers']['last-modified'];
+					}
+					if (isset($this->data['headers']['etag']))
+					{
+						$headers['if-none-match'] = $this->data['headers']['etag'];
+					}
 
-						$file = $this->registry->create('File', array($this->feed_url, $this->timeout, 5, $headers, $this->useragent, $this->force_fsockopen));	//FreshRSS
+					$file = $this->registry->create('File', array($this->feed_url, $this->timeout, 5, $headers, $this->useragent, $this->force_fsockopen, $this->curl_options));
 
-						if ($file->success)
+					if ($file->success)
+					{
+						if ($file->status_code === 304)
 						{
-							if ($file->status_code === 304)
-							{
-								$cache->touch();
-								return true;
-							}
+							$cache->touch();
+							return true;
 						}
-						else
+					}
+					else
+					{
+						$cache->touch();
+						$this->error = $file->error;
+						return !empty($this->data);
+					}
+
+					$md5 = $this->cleanMd5($file->body);
+					if ($this->data['md5'] === $md5) {
+						if ($this->syslog_enabled)
 						{
-							$this->error = $file->error;	//FreshRSS
-							return !empty($this->data);	//FreshRSS
-							//unset($file);	//FreshRSS removed
+							syslog(LOG_DEBUG, 'SimplePie MD5 cache match for ' . SimplePie_Misc::url_remove_credentials($this->feed_url));
 						}
-					}
-					{	//FreshRSS
-						$md5 = $this->cleanMd5($file->body);
-						if ($this->data['md5'] === $md5) {
-							syslog(LOG_DEBUG, 'SimplePie MD5 cache match for ' . $this->feed_url);
-							$cache->touch();
-							return true;	//Content unchanged even though server did not send a 304
-						} else {
-							syslog(LOG_DEBUG, 'SimplePie MD5 cache no match for ' . $this->feed_url);
-							$this->data['md5'] = $md5;
+						$cache->touch();
+						return true;	//Content unchanged even though server did not send a 304
+					} else {
+						if ($this->syslog_enabled)
+						{
+							syslog(LOG_DEBUG, 'SimplePie MD5 cache no match for ' . SimplePie_Misc::url_remove_credentials($this->feed_url));
 						}
+						$this->data['md5'] = $md5;
 					}
 				}
-				// If the cache is still valid, just return true
-				else
-				{
-					$this->raw_data = false;
-					return true;
-				}
 			}
-			// If the cache is empty, delete it
+			// If the cache is empty
 			else
 			{
-				$cache->unlink();
+				$cache->touch();	//To keep the date/time of the last tentative update
 				$this->data = array();
 			}
 		}
@@ -1550,7 +1616,7 @@ class SimplePie
 				$headers = array(
 					'Accept' => 'application/atom+xml, application/rss+xml, application/rdf+xml;q=0.9, application/xml;q=0.8, text/xml;q=0.8, text/html;q=0.7, unknown/unknown;q=0.1, application/unknown;q=0.1, */*;q=0.1',
 				);
-				$file = $this->registry->create('File', array($this->feed_url, $this->timeout, 5, $headers, $this->useragent, $this->force_fsockopen));
+				$file = $this->registry->create('File', array($this->feed_url, $this->timeout, 5, $headers, $this->useragent, $this->force_fsockopen, $this->curl_options));
 			}
 		}
 		// If the file connection has an error, set SimplePie::error to that and quit
@@ -1567,13 +1633,15 @@ class SimplePie
 
 			if (!$locate->is_feed($file))
 			{
+				$copyStatusCode = $file->status_code;
+				$copyContentType = $file->headers['content-type'];
 				// We need to unset this so that if SimplePie::set_file() has been called that object is untouched
 				unset($file);
 				try
 				{
 					if (!($file = $locate->find($this->autodiscovery, $this->all_discovered_feeds)))
 					{
-						$this->error = "A feed could not be found at $this->feed_url. A feed with an invalid mime type may fall victim to this error, or " . SIMPLEPIE_NAME . " was unable to auto-discover it.. Use force_feed() if you are certain this URL is a real feed.";
+						$this->error = "A feed could not be found at `$this->feed_url`; the status code is `$copyStatusCode` and content-type is `$copyContentType`";
 						$this->registry->call('Misc', 'error', array($this->error, E_USER_NOTICE, __FILE__, __LINE__));
 						return false;
 					}
@@ -1588,8 +1656,8 @@ class SimplePie
 				if ($cache)
 				{
 					$this->data = array('url' => $this->feed_url, 'feed_url' => $file->url, 'build' => SIMPLEPIE_BUILD);
-					$this->data['mtime'] = time();	//FreshRSS
-					$this->data['md5'] = empty($md5) ? $this->cleanMd5($file->body) : $md5;	//FreshRSS
+					$this->data['mtime'] = time();
+					$this->data['md5'] = empty($md5) ? $this->cleanMd5($file->body) : $md5;
 					if (!$cache->save($this))
 					{
 						trigger_error("$this->cache_location is not writeable. Make sure you've set the correct relative or absolute path, and that the location is server-writable.", E_USER_WARNING);
@@ -1601,8 +1669,9 @@ class SimplePie
 			$locate = null;
 		}
 
+		$file->body = trim($file->body);
 		$this->raw_data = $file->body;
-
+		$this->permanent_url = $file->permanent_url;
 		$headers = $file->headers;
 		$sniffer = $this->registry->create('Content_Type_Sniffer', array(&$file));
 		$sniffed = $sniffer->get_type();
@@ -1788,26 +1857,39 @@ class SimplePie
 
 	/**
 	 * Get the URL for the feed
+	 * 
+	 * When the 'permanent' mode is enabled, returns the original feed URL,
+	 * except in the case of an `HTTP 301 Moved Permanently` status response,
+	 * in which case the location of the first redirection is returned.
 	 *
-	 * May or may not be different from the URL passed to {@see set_feed_url()},
+	 * When the 'permanent' mode is disabled (default),
+	 * may or may not be different from the URL passed to {@see set_feed_url()},
 	 * depending on whether auto-discovery was used.
 	 *
 	 * @since Preview Release (previously called `get_feed_url()` since SimplePie 0.8.)
-	 * @todo If we have a perm redirect we should return the new URL
-	 * @todo When we make the above change, let's support <itunes:new-feed-url> as well
+	 * @todo Support <itunes:new-feed-url>
 	 * @todo Also, |atom:link|@rel=self
+	 * @param bool $permanent Permanent mode to return only the original URL or the first redirection
+	 *  iff it is a 301 redirection
 	 * @return string|null
 	 */
-	public function subscribe_url()
+	public function subscribe_url($permanent = false)
 	{
-		if ($this->feed_url !== null)
+		if ($permanent)
 		{
-			return $this->sanitize($this->feed_url, SIMPLEPIE_CONSTRUCT_IRI);
+			if ($this->permanent_url !== null)
+			{
+				return $this->sanitize($this->permanent_url, SIMPLEPIE_CONSTRUCT_IRI);
+			}
 		}
 		else
 		{
-			return null;
+			if ($this->feed_url !== null)
+			{
+				return $this->sanitize($this->feed_url, SIMPLEPIE_CONSTRUCT_IRI);
+			}
 		}
+		return null;
 	}
 
 	/**