Merge pull request #6455 from annando/issue-6282

Issue 6282: Update the contact data regularly (including the network)

src/Core/Protocol.php

@@ -14,25 +14,27 @@ use Friendica\Util\Network;
 class Protocol
 {
 	// Native support
-	const ACTIVITYPUB = 'apub';    // ActivityPub
+	const ACTIVITYPUB = 'apub';    // ActivityPub (Pleroma, Mastodon, Osada, ...)
 	const DFRN        = 'dfrn';    // Friendica, Mistpark, other DFRN implementations
-	const DIASPORA    = 'dspr';    // Diaspora
+	const DIASPORA    = 'dspr';    // Diaspora, Hubzilla, Socialhome, Ganggo
 	const FEED        = 'feed';    // RSS/Atom feeds with no known "post/notify" protocol
 	const MAIL        = 'mail';    // IMAP/POP
-	const OSTATUS     = 'stat';    // GNU-social, Pleroma, Mastodon, other OStatus implementations
+	const OSTATUS     = 'stat';    // GNU Social and other OStatus implementations
 
 	const NATIVE_SUPPORT = [self::DFRN, self::DIASPORA, self::OSTATUS, self::FEED, self::MAIL, self::ACTIVITYPUB];
 
 	// Supported through a connector
-	const APPNET    = 'apdn';    // app.net - Dead protocol
 	const DIASPORA2 = 'dspc';    // Diaspora connector
-	const FACEBOOK  = 'face';    // Facebook API
-	const GPLUS     = 'goog';    // Google+
 	const LINKEDIN  = 'lnkd';    // LinkedIn
 	const PUMPIO    = 'pump';    // pump.io
 	const STATUSNET = 'stac';    // Statusnet connector
 	const TWITTER   = 'twit';    // Twitter
 
+	// Dead protocols
+	const APPNET    = 'apdn';    // app.net - Dead protocol
+	const FACEBOOK  = 'face';    // Facebook API - Not working anymore, API is closed
+	const GPLUS     = 'goog';    // Google+ - Dead in 2019
+
 	// Currently unsupported
 	const ICALENDAR = 'ical';    // iCalendar
 	const MYSPACE   = 'mysp';    // MySpace

src/Model/Contact.php

@@ -634,6 +634,8 @@ class Contact extends BaseObject
 			Logger::log('Empty contact: ' . json_encode($contact) . ' - ' . System::callstack(20), Logger::DEBUG);
 		}
 
+		Logger::log('Contact '.$contact['id'].' is marked for archival', Logger::DEBUG);
+
 		// Contact already archived or "self" contact? => nothing to do
 		if ($contact['archive'] || $contact['self']) {
 			return;
@@ -682,6 +684,8 @@ class Contact extends BaseObject
 			return;
 		}
 
+		Logger::log('Contact '.$contact['id'].' is marked as vital again', Logger::DEBUG);
+
 		if (!isset($contact['url']) && !empty($contact['id'])) {
 			$fields = ['id', 'url', 'batch'];
 			$contact = DBA::selectFirst('contact', [], ['id' => $contact['id']]);
@@ -1571,8 +1575,8 @@ class Contact extends BaseObject
 
 		$ret = Probe::uri($contact["url"], $network);
 
-		// If Probe::uri fails the network code will be different
+		// If Probe::uri fails the network code will be different (mostly "feed" or "unkn")
-		if (($ret["network"] != $contact["network"]) && !in_array($ret["network"], [Protocol::ACTIVITYPUB, $network])) {
+		if (($ret["network"] != $contact["network"]) && !in_array($ret["network"], [Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, $network])) {
 			return false;
 		}
 

src/Model/Item.php

@@ -2370,7 +2370,7 @@ class Item extends BaseObject
 		$update = (!$arr['private'] && ((defaults($arr, 'author-link', '') === defaults($arr, 'owner-link', '')) || ($arr["parent-uri"] === $arr["uri"])));
 
 		// Is it a forum? Then we don't care about the rules from above
-		if (!$update && ($arr["network"] == Protocol::DFRN) && ($arr["parent-uri"] === $arr["uri"])) {
+		if (!$update && in_array($arr["network"], [Protocol::ACTIVITYPUB, Protocol::DFRN]) && ($arr["parent-uri"] === $arr["uri"])) {
 			if (DBA::exists('contact', ['id' => $arr['contact-id'], 'forum' => true])) {
 				$update = true;
 			}

src/Protocol/PortableContact.php

@@ -334,7 +334,7 @@ class PortableContact
 			$server_url = Strings::normaliseLink(self::detectServer($profile));
 		}
 
-		if (!in_array($gcontacts[0]["network"], [Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::FEED, Protocol::OSTATUS, ""])) {
+		if (!in_array($gcontacts[0]["network"], [Protocol::DFRN, Protocol::DIASPORA, Protocol::FEED, Protocol::OSTATUS, ""])) {
 			Logger::log("Profile ".$profile.": Network type ".$gcontacts[0]["network"]." can't be checked", Logger::DEBUG);
 			return false;
 		}

src/Worker/OnePoll.php

@@ -28,7 +28,7 @@ class OnePoll
 	{
 		$a = BaseObject::getApp();
 
-		Logger::log('start');
+		Logger::log('Start for contact ' . $contact_id);
 
 		$manual_id  = 0;
 		$generation = 0;
@@ -63,29 +63,23 @@ class OnePoll
 
 		// These three networks can be able to speak AP, so we are trying to fetch AP profile data here
 		if (in_array($contact['network'], [Protocol::ACTIVITYPUB, Protocol::DIASPORA, Protocol::DFRN])) {
-			APContact::getByURL($contact['url']);
+			$apcontact = APContact::getByURL($contact['url'], true);
-		}
 
-		// We currently don't do anything more with AP here
+			$updated = DateTimeFormat::utcNow();
-		if ($contact['network'] === Protocol::ACTIVITYPUB) {
+			if (($contact['network'] === Protocol::ACTIVITYPUB) && empty($apcontact)) {
+				self::updateContact($contact, ['last-update' => $updated, 'failure_update' => $updated]);
+				Contact::markForArchival($contact);
+				Logger::log('Contact archived');
 				return;
-		}
+			} elseif (!empty($apcontact)) {
-
+				$fields = ['last-update' => $updated, 'success_update' => $updated];
-		// load current friends if possible.
+				self::updateContact($contact, $fields);
-		if (($contact['poco'] != "") && ($contact['success_update'] > $contact['failure_update'])) {
+				Contact::unmarkForArchival($contact);
-			$r = q("SELECT count(*) AS total FROM glink
-				WHERE `cid` = %d AND updated > UTC_TIMESTAMP() - INTERVAL 1 DAY",
-				intval($contact['id'])
-			);
-			if (DBA::isResult($r)) {
-				if (!$r[0]['total']) {
-					PortableContact::loadWorker($contact['id'], $importer_uid, 0, $contact['poco']);
-				}
 			}
 		}
 
 		// Diaspora users, archived users and followers are only checked if they still exist.
-		if ($contact['archive'] || ($contact["network"] == Protocol::DIASPORA) || ($contact["rel"] == Contact::FOLLOWER)) {
+		if (($contact['network'] != Protocol::ACTIVITYPUB) && ($contact['archive'] || ($contact["network"] == Protocol::DIASPORA) || ($contact["rel"] == Contact::FOLLOWER))) {
 			$last_updated = PortableContact::lastUpdated($contact["url"], true);
 			$updated = DateTimeFormat::utcNow();
 
@@ -103,56 +97,64 @@ class OnePoll
 			} else {
 				self::updateContact($contact, ['last-update' => $updated, 'failure_update' => $updated]);
 				Contact::markForArchival($contact);
-				Logger::log('Contact '.$contact['id'].' is marked for archival', Logger::DEBUG);
+				Logger::log('Contact archived');
-			}
-
 				return;
 			}
-
-		$xml = false;
-
-		$t = $contact['last-update'];
-
-		if ($contact['subhub']) {
-			$poll_interval = Config::get('system', 'pushpoll_frequency', 3);
-			$contact['priority'] = intval($poll_interval);
-			$hub_update = false;
-
-			if (DateTimeFormat::utcNow() > DateTimeFormat::utc($t . " + 1 day")) {
-				$hub_update = true;
 		}
-		} else {
-			$hub_update = false;
-		}
-
-		$last_update = (($contact['last-update'] <= DBA::NULL_DATETIME)
-			? DateTimeFormat::utc('now - 7 days', DateTimeFormat::ATOM)
-			: DateTimeFormat::utc($contact['last-update'], DateTimeFormat::ATOM)
-		);
 
 		// Update the contact entry
-		if (($contact['network'] === Protocol::OSTATUS) || ($contact['network'] === Protocol::DIASPORA) || ($contact['network'] === Protocol::DFRN)) {
+		if (in_array($contact['network'], [Protocol::ACTIVITYPUB, Protocol::OSTATUS, Protocol::DIASPORA, Protocol::DFRN])) {
-			if (!PortableContact::reachable($contact['url'])) {
+			$updated = DateTimeFormat::utcNow();
+			// Currently we can't check every AP implementation, so we don't do it at all
+			if (($contact['network' != Protocol::ACTIVITYPUB]) && !PortableContact::reachable($contact['url'])) {
 				Logger::log("Skipping probably dead contact ".$contact['url']);
 
 				// set the last-update so we don't keep polling
-				DBA::update('contact', ['last-update' => DateTimeFormat::utcNow()], ['id' => $contact['id']]);
+				self::updateContact($contact, ['last-update' => $updated]);
 				return;
 			}
 
 			if (!Contact::updateFromProbe($contact["id"])) {
-				Contact::markForArchival($contact);
-				Logger::log('Contact is marked dead');
-
 				// set the last-update so we don't keep polling
-				DBA::update('contact', ['last-update' => DateTimeFormat::utcNow()], ['id' => $contact['id']]);
+				self::updateContact($contact, ['last-update' => $updated]);
+				Contact::markForArchival($contact);
+				Logger::log('Contact archived');
 				return;
 			} else {
+				$fields = ['last-update' => $updated, 'success_update' => $updated];
+				self::updateContact($contact, $fields);
 				Contact::unmarkForArchival($contact);
 			}
 		}
 
+		// load current friends if possible.
+		if (!empty($contact['poco']) && ($contact['success_update'] > $contact['failure_update'])) {
+			$r = q("SELECT count(*) AS total FROM glink
+				WHERE `cid` = %d AND updated > UTC_TIMESTAMP() - INTERVAL 1 DAY",
+				intval($contact['id'])
+			);
+			if (DBA::isResult($r)) {
+				if (!$r[0]['total']) {
+					PortableContact::loadWorker($contact['id'], $importer_uid, 0, $contact['poco']);
+				}
+			}
+		}
+
+		// We don't poll our followers
+		if ($contact["rel"] == Contact::FOLLOWER) {
+			Logger::log("Don't poll follower");
+			return;
+		}
+
+		// Don't poll if polling is deactivated (But we poll feeds and mails anyway)
 		if (!in_array($contact['network'], [Protocol::FEED, Protocol::MAIL]) && Config::get('system', 'disable_polling')) {
+			Logger::log('Polling is disabled');
+			return;
+		}
+
+		// We don't poll AP contacts by now
+		if ($contact['network'] === Protocol::ACTIVITYPUB) {
+			Logger::log("Don't poll AP contact");
 			return;
 		}
 
@@ -178,6 +180,24 @@ class OnePoll
 
 		$importer = $r[0];
 		$url = '';
+		$xml = false;
+
+		if ($contact['subhub']) {
+			$poll_interval = Config::get('system', 'pushpoll_frequency', 3);
+			$contact['priority'] = intval($poll_interval);
+			$hub_update = false;
+
+			if (DateTimeFormat::utcNow() > DateTimeFormat::utc($contact['last-update'] . " + 1 day")) {
+				$hub_update = true;
+			}
+		} else {
+			$hub_update = false;
+		}
+
+		$last_update = (($contact['last-update'] <= DBA::NULL_DATETIME)
+			? DateTimeFormat::utc('now - 7 days', DateTimeFormat::ATOM)
+			: DateTimeFormat::utc($contact['last-update'], DateTimeFormat::ATOM)
+		);
 
 		Logger::log("poll: ({$contact['network']}-{$contact['id']}) IMPORTER: {$importer['name']}, CONTACT: {$contact['name']}");
 
@@ -209,8 +229,9 @@ class OnePoll
 
 			if (!$curlResult->isSuccess() && ($curlResult->getErrorNumber() == CURLE_OPERATION_TIMEDOUT)) {
 				// set the last-update so we don't keep polling
-				DBA::update('contact', ['last-update' => DateTimeFormat::utcNow()], ['id' => $contact['id']]);
+				self::updateContact($contact, ['last-update' => DateTimeFormat::utcNow()]);
 				Contact::markForArchival($contact);
+				Logger::log('Contact archived');
 				return;
 			}
 
@@ -220,27 +241,24 @@ class OnePoll
 			Logger::log('handshake with url ' . $url . ' returns xml: ' . $handshake_xml, Logger::DATA);
 
 			if (!strlen($handshake_xml) || ($html_code >= 400) || !$html_code) {
-				Logger::log("$url appears to be dead - marking for death ");
-
 				// dead connection - might be a transient event, or this might
 				// mean the software was uninstalled or the domain expired.
 				// Will keep trying for one month.
-
+				Logger::log("$url appears to be dead - marking for death ");
-				Contact::markForArchival($contact);
 
 				// set the last-update so we don't keep polling
 				$fields = ['last-update' => DateTimeFormat::utcNow(), 'failure_update' => DateTimeFormat::utcNow()];
 				self::updateContact($contact, $fields);
+				Contact::markForArchival($contact);
 				return;
 			}
 
 			if (!strstr($handshake_xml, '<')) {
 				Logger::log('response from ' . $url . ' did not contain XML.');
 
-				Contact::markForArchival($contact);
-
 				$fields = ['last-update' => DateTimeFormat::utcNow(), 'failure_update' => DateTimeFormat::utcNow()];
 				self::updateContact($contact, $fields);
+				Contact::markForArchival($contact);
 				return;
 			}
 
@@ -248,22 +266,21 @@ class OnePoll
 			$res = XML::parseString($handshake_xml);
 
 			if (intval($res->status) == 1) {
+				// we may not be friends anymore. Will keep trying for one month.
 				Logger::log("$url replied status 1 - marking for death ");
 
-				// we may not be friends anymore. Will keep trying for one month.
 				// set the last-update so we don't keep polling
 				$fields = ['last-update' => DateTimeFormat::utcNow(), 'failure_update' => DateTimeFormat::utcNow()];
 				self::updateContact($contact, $fields);
-
 				Contact::markForArchival($contact);
 			} elseif ($contact['term-date'] > DBA::NULL_DATETIME) {
-				Logger::log("$url back from the dead - removing mark for death");
 				Contact::unmarkForArchival($contact);
 			}
 
 			if ((intval($res->status) != 0) || !strlen($res->challenge) || !strlen($res->dfrn_id)) {
 				// set the last-update so we don't keep polling
 				DBA::update('contact', ['last-update' => DateTimeFormat::utcNow()], ['id' => $contact['id']]);
+				Logger::log('Contact status is ' . $res->status);
 				return;
 			}
 
@@ -293,6 +310,13 @@ class OnePoll
 				$final_dfrn_id = substr($final_dfrn_id, 2);
 			}
 
+			// There are issues with the legacy DFRN transport layer.
+			// Since we mostly don't use it anyway, we won't dig into it deeper, but simply ignore it.
+			if (empty($final_dfrn_id) || empty($orig_id)) {
+				Logger::log('Contact has got no ID - quitting');
+				return;
+			}
+
 			if ($final_dfrn_id != $orig_id) {
 				// did not decode properly - cannot trust this site
 				Logger::log('ID did not decode: ' . $contact['id'] . ' orig: ' . $orig_id . ' final: ' . $final_dfrn_id);
@@ -330,10 +354,10 @@ class OnePoll
 			}
 
 			// Are we allowed to import from this person?
-
 			if ($contact['rel'] == Contact::FOLLOWER || $contact['blocked']) {
 				// set the last-update so we don't keep polling
 				DBA::update('contact', ['last-update' => DateTimeFormat::utcNow()], ['id' => $contact['id']]);
+				Logger::log('Contact is blocked or only a follower');
 				return;
 			}
 
@@ -343,8 +367,9 @@ class OnePoll
 
 			if ($curlResult->isTimeout()) {
 				// set the last-update so we don't keep polling
-				DBA::update('contact', ['last-update' => DateTimeFormat::utcNow()], ['id' => $contact['id']]);
+				self::updateContact($contact, ['last-update' => DateTimeFormat::utcNow()]);
 				Contact::markForArchival($contact);
+				Logger::log('Contact archived');
 				return;
 			}
 
@@ -356,8 +381,9 @@ class OnePoll
 			$mail_disabled = ((function_exists('imap_open') && !Config::get('system', 'imap_disabled')) ? 0 : 1);
 			if ($mail_disabled) {
 				// set the last-update so we don't keep polling
-				DBA::update('contact', ['last-update' => DateTimeFormat::utcNow()], ['id' => $contact['id']]);
+				self::updateContact($contact, ['last-update' => DateTimeFormat::utcNow()]);
 				Contact::markForArchival($contact);
+				Logger::log('Contact archived');
 				return;
 			}
 
@@ -652,10 +678,10 @@ class OnePoll
 			DBA::update('gcontact', ['last_failure' => $updated], ['nurl' => $contact['nurl']]);
 			Contact::markForArchival($contact);
 		} else {
-			$updated = DateTimeFormat::utcNow();
+			self::updateContact($contact, ['last-update' => DateTimeFormat::utcNow()]);
-			DBA::update('contact', ['last-update' => $updated], ['id' => $contact['id']]);
 		}
 
+		Logger::log('End');
 		return;
 	}