Merge pull request #14229 from annando/probing

Reduce probing / handle exception for invalid url / always check for https

src/Content/Text/BBCode.php

@@ -2124,7 +2124,12 @@ class BBCode
 		}
 
 		$parts['host'] = idn_to_ascii(urldecode($parts['host']));
+		try {
 			return (string)Uri::fromParts($parts);
+		} catch (\Throwable $th) {
+			Logger::notice('Exception on unparsing url', ['url' => $url, 'parts' => $parts, 'code' => $th->getCode(), 'message' => $th->getMessage()]);
+			return $url;
+		}
 	}
 
 	private static function unifyLinks(string $text): string

src/Model/Contact.php

@@ -2653,6 +2653,14 @@ class Contact
 
 		$data = Probe::uri($contact['url'], $network, $contact['uid']);
 
+		if (in_array($data['network'], Protocol::FEDERATED) && (parse_url($data['url'], PHP_URL_SCHEME) == 'http')) {
+			$ssl_url = str_replace('http://', 'https://', $contact['url']);
+			$ssl_data = Probe::uri($ssl_url, $network, $contact['uid']);
+			if (($ssl_data['network'] == $data['network']) && (parse_url($ssl_data['url'], PHP_URL_SCHEME) != 'http')) {
+				$data = $ssl_data;
+			}
+		}
+
 		if ($data['network'] == Protocol::DIASPORA) {
 			try {
 				DI::dsprContact()->updateFromProbeArray($data);
@@ -2811,7 +2819,7 @@ class Contact
 		// We must not try to update relay contacts via probe. They are no real contacts.
 		// See Relay::updateContact() for more details.
 		// We check after the probing to be able to correct falsely detected contact types.
-		if (($contact['contact-type'] == self::TYPE_RELAY) && Strings::compareLink($contact['url'], $contact['baseurl']) &&
+		if (($contact['contact-type'] == self::TYPE_RELAY) && Strings::compareLink($contact['url'], $contact['baseurl'] ?? '') &&
 			(!Strings::compareLink($ret['url'], $contact['url']) || in_array($ret['network'], [Protocol::FEED, Protocol::PHANTOM]))
 		) {
 			if (GServer::reachable($contact)) {

src/Network/Probe.php

@@ -244,7 +244,7 @@ class Probe
 			return [];
 		}
 
-		if (!is_object($xrd) && !empty($url)) {
+		if ($ssl_connection_error && !is_object($xrd) && !empty($url)) {
 			$curlResult = DI::httpClient()->get($url, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::REQUEST => HttpClientRequest::CONTACTINFO]);
 			$connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 			if ($curlResult->isTimeout()) {
@@ -568,7 +568,7 @@ class Probe
 			}
 
 			$webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
-			if (empty($webfinger)) {
+			if (empty($webfinger) && !is_null($webfinger)) {
 				$lrdd = self::hostMeta($host);
 			}
 
@@ -582,7 +582,7 @@ class Probe
 					}
 
 					$webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
-					if (empty($webfinger)) {
+					if (empty($webfinger) && !is_null($webfinger)) {
 						$lrdd = self::hostMeta($host);
 					}
 				}
@@ -604,11 +604,12 @@ class Probe
 				return [];
 			}
 
-			if (empty($webfinger)) {
+			if (is_null($webfinger)) {
 				$webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
-				if (self::$isTimeout) {
+				if (self::$isTimeout || is_null($webfinger)) {
 					return [];
 				}
+				$baseurl = 'http://' . $host;
 			} else {
 				$baseurl = 'https://' . $host;
 			}
@@ -619,8 +620,6 @@ class Probe
 					return [];
 				}
 				$baseurl = self::$baseurl;
-			} else {
-				$baseurl = 'http://' . $host;
 			}
 		} else {
 			Logger::info('URI was not detectable', ['uri' => $uri]);
@@ -661,11 +660,11 @@ class Probe
 	 *
 	 * @return array webfinger results
 	 */
-	private static function getWebfinger(string $template, string $type, string $uri, string $addr): array
+	private static function getWebfinger(string $template, string $type, string $uri, string $addr): ?array
 	{
 		if (Network::isUrlBlocked($template)) {
 			Logger::info('Domain is blocked', ['url' => $template]);
-			return [];
+			return null;
 		}
 
 		// First try the address because this is the primary purpose of webfinger
@@ -673,8 +672,8 @@ class Probe
 			$detected = $addr;
 			$path = str_replace('{uri}', urlencode('acct:' . $addr), $template);
 			$webfinger = self::webfinger($path, $type);
-			if (self::$isTimeout) {
+			if (is_null($webfinger)) {
-				return [];
+				return null;
 			}
 		}
 
@@ -683,8 +682,8 @@ class Probe
 			$detected = $uri;
 			$path = str_replace('{uri}', urlencode($uri), $template);
 			$webfinger = self::webfinger($path, $type);
-			if (self::$isTimeout) {
+			if (is_null($webfinger)) {
-				return [];
+				return null;
 			}
 		}
 
@@ -1001,7 +1000,7 @@ class Probe
 	 * @return array webfinger data
 	 * @throws HTTPException\InternalServerErrorException
 	 */
-	private static function webfinger(string $url, string $type): array
+	private static function webfinger(string $url, string $type): ?array
 	{
 		try {
 			$curlResult = DI::httpClient()->get(
@@ -1011,12 +1010,12 @@ class Probe
 			);
 		} catch (\Throwable $e) {
 			Logger::notice($e->getMessage(), ['url' => $url, 'type' => $type, 'class' => get_class($e)]);
-			return [];
+			return null;
 		}
 
 		if ($curlResult->isTimeout()) {
 			self::$isTimeout = true;
-			return [];
+			return null;
 		}
 		$data = $curlResult->getBodyString();