Merge pull request #6988 from annando/fetchprofile

Fetch profile data from different contact tables
2024-11-09 23:02:54 +00:00 · 2019-04-09 09:06:11 -04:00 · 2019-04-09 09:06:11 -04:00 · 3aa8a255c9
commit 3aa8a255c9
parent 4b80b7caea 8748aff110
5 changed files with 194 additions and 92 deletions
--- a/config/dbstructure.config.php
+++ b/config/dbstructure.config.php
@ -34,7 +34,7 @@
 use Friendica\Database\DBA;

 if (!defined('DB_UPDATE_VERSION')) {
-	define('DB_UPDATE_VERSION', 1309);
+	define('DB_UPDATE_VERSION', 1310);
 }

 return [
@ -180,6 +180,7 @@ return [
 			"id" => ["type" => "int unsigned", "not null" => "1", "extra" => "auto_increment", "primary" => "1", "comment" => "sequential ID"],
 			"uid" => ["type" => "mediumint unsigned", "not null" => "1", "default" => "0", "relation" => ["user" => "uid"], "comment" => "Owner User id"],
 			"created" => ["type" => "datetime", "not null" => "1", "default" => DBA::NULL_DATETIME, "comment" => ""],
+			"updated" => ["type" => "datetime", "default" => DBA::NULL_DATETIME, "comment" => "Date of last contact update"],
 			"self" => ["type" => "boolean", "not null" => "1", "default" => "0", "comment" => "1 if the contact is the user him/her self"],
 			"remote_self" => ["type" => "boolean", "not null" => "1", "default" => "0", "comment" => ""],
 			"rel" => ["type" => "tinyint unsigned", "not null" => "1", "default" => "0", "comment" => "The kind of the relation between the user and the contact"],
@ -187,7 +188,7 @@ return [
 			"network" => ["type" => "char(4)", "not null" => "1", "default" => "", "comment" => "Network protocol of the contact"],
 			"name" => ["type" => "varchar(255)", "not null" => "1", "default" => "", "comment" => "Name that this contact is known by"],
 			"nick" => ["type" => "varchar(255)", "not null" => "1", "default" => "", "comment" => "Nick- and user name of the contact"],
-			"location" => ["type" => "varchar(255)", "not null" => "1", "default" => "", "comment" => ""],
+			"location" => ["type" => "varchar(255)", "default" => "", "comment" => ""],
 			"about" => ["type" => "text", "comment" => ""],
 			"keywords" => ["type" => "text", "comment" => "public keywords (interests) of the contact"],
 			"gender" => ["type" => "varchar(32)", "not null" => "1", "default" => "", "comment" => ""],
--- a/database.sql
+++ b/database.sql
@ -1,6 +1,6 @@
 -- ------------------------------------------
 -- Friendica 2019.06-dev (Dalmatian Bellflower)
-- DB_UPDATE_VERSION 1309
+-- DB_UPDATE_VERSION 1310
 -- ------------------------------------------


@ -138,6 +138,7 @@ CREATE TABLE IF NOT EXISTS `contact` (
 	`id` int unsigned NOT NULL auto_increment COMMENT 'sequential ID',
 	`uid` mediumint unsigned NOT NULL DEFAULT 0 COMMENT 'Owner User id',
 	`created` datetime NOT NULL DEFAULT '0001-01-01 00:00:00' COMMENT '',
+	`updated` datetime DEFAULT '0001-01-01 00:00:00' COMMENT 'Date of last contact update',
 	`self` boolean NOT NULL DEFAULT '0' COMMENT '1 if the contact is the user him/her self',
 	`remote_self` boolean NOT NULL DEFAULT '0' COMMENT '',
 	`rel` tinyint unsigned NOT NULL DEFAULT 0 COMMENT 'The kind of the relation between the user and the contact',
@ -145,7 +146,7 @@ CREATE TABLE IF NOT EXISTS `contact` (
 	`network` char(4) NOT NULL DEFAULT '' COMMENT 'Network protocol of the contact',
 	`name` varchar(255) NOT NULL DEFAULT '' COMMENT 'Name that this contact is known by',
 	`nick` varchar(255) NOT NULL DEFAULT '' COMMENT 'Nick- and user name of the contact',
-	`location` varchar(255) NOT NULL DEFAULT '' COMMENT '',
+	`location` varchar(255) DEFAULT '' COMMENT '',
 	`about` text COMMENT '',
 	`keywords` text COMMENT 'public keywords (interests) of the contact',
 	`gender` varchar(32) NOT NULL DEFAULT '' COMMENT '',
--- a/src/Model/Contact.php
+++ b/src/Model/Contact.php
@ -599,7 +599,10 @@ class Contact extends BaseObject
 		}

 		if ($update) {
-			$fields['name-date'] = DateTimeFormat::utcNow();
+			if ($fields['name'] != $self['name']) {
+				$fields['name-date'] = DateTimeFormat::utcNow();
+			}
+			$fields['updated'] = DateTimeFormat::utcNow();
 			DBA::update('contact', $fields, ['id' => $self['id']]);

 			// Update the public contact as well
@ -1114,6 +1117,90 @@ class Contact extends BaseObject
 			   )", intval($uid), intval($uid));
 	}

+	/**
+	 * Have a look at all contact tables for a given profile url.
+	 * This function works as a replacement for probing the contact.
+	 *
+	 * @param string  $url Contact URL
+	 * @param integer $cid Contact ID
+	 *
+	 * @return array Contact array in the "probe" structure
+	*/
+	private static function getProbeDataFromDatabase($url, $cid)
+	{
+		// The link could be provided as http although we stored it as https
+		$ssl_url = str_replace('http://', 'https://', $url);
+
+		$fields = ['url', 'addr', 'alias', 'notify', 'poll', 'name', 'nick',
+			'photo', 'keywords', 'location', 'about', 'network',
+			'priority', 'batch', 'request', 'confirm', 'poco'];
+
+		if (!empty($cid)) {
+			$data = DBA::selectFirst('contact', $fields, ['id' => $cid]);
+			if (DBA::isResult($data)) {
+				return $data;
+			}
+		}
+
+		$data = DBA::selectFirst('contact', $fields, ['nurl' => Strings::normaliseLink($url)]);
+
+		if (!DBA::isResult($data)) {
+			$condition = ['alias' => [$url, Strings::normaliseLink($url), $ssl_url]];
+			$data = DBA::selectFirst('contact', $fields, $condition);
+		}
+
+		if (DBA::isResult($data)) {
+			// For security reasons we don't fetch key data from our users
+			$data["pubkey"] = '';
+			return $data;
+		}
+
+		$fields = ['url', 'addr', 'alias', 'notify', 'name', 'nick',
+			'photo', 'keywords', 'location', 'about', 'network'];
+		$data = DBA::selectFirst('gcontact', $fields, ['nurl' => Strings::normaliseLink($url)]);
+
+		if (!DBA::isResult($data)) {
+			$condition = ['alias' => [$url, Strings::normaliseLink($url), $ssl_url]];
+			$data = DBA::selectFirst('contact', $fields, $condition);
+		}
+
+		if (DBA::isResult($data)) {
+			$data["pubkey"] = '';
+			$data["poll"] = '';
+			$data["priority"] = 0;
+			$data["batch"] = '';
+			$data["request"] = '';
+			$data["confirm"] = '';
+			$data["poco"] = '';
+			return $data;
+		}
+
+		$data = ActivityPub::probeProfile($url, false);
+		if (!empty($data)) {
+			return $data;
+		}
+
+		$fields = ['url', 'addr', 'alias', 'notify', 'poll', 'name', 'nick',
+			'photo', 'network', 'priority', 'batch', 'request', 'confirm'];
+		$data = DBA::selectFirst('fcontact', $fields, ['url' => $url]);
+
+		if (!DBA::isResult($data)) {
+			$condition = ['alias' => [$url, Strings::normaliseLink($url), $ssl_url]];
+			$data = DBA::selectFirst('contact', $fields, $condition);
+		}
+
+		if (DBA::isResult($data)) {
+			$data["pubkey"] = '';
+			$data["keywords"] = '';
+			$data["location"] = '';
+			$data["about"] = '';
+			$data["poco"] = '';
+			return $data;
+		}
+
+		return [];
+	}
+
 	/**
 	 * @brief Fetch the contact id for a given URL and user
 	 *
@ -1154,11 +1241,11 @@ class Contact extends BaseObject

 		/// @todo Verify if we can't use Contact::getDetailsByUrl instead of the following
 		// We first try the nurl (http://server.tld/nick), most common case
-		$contact = DBA::selectFirst('contact', ['id', 'avatar', 'avatar-date'], ['nurl' => Strings::normaliseLink($url), 'uid' => $uid, 'deleted' => false]);
+		$contact = DBA::selectFirst('contact', ['id', 'avatar', 'updated', 'network'], ['nurl' => Strings::normaliseLink($url), 'uid' => $uid, 'deleted' => false]);

 		// Then the addr (nick@server.tld)
 		if (!DBA::isResult($contact)) {
-			$contact = DBA::selectFirst('contact', ['id', 'avatar', 'avatar-date'], ['addr' => $url, 'uid' => $uid, 'deleted' => false]);
+			$contact = DBA::selectFirst('contact', ['id', 'avatar', 'updated', 'network'], ['addr' => $url, 'uid' => $uid, 'deleted' => false]);
 		}

 		// Then the alias (which could be anything)
@ -1166,19 +1253,25 @@ class Contact extends BaseObject
 			// The link could be provided as http although we stored it as https
 			$ssl_url = str_replace('http://', 'https://', $url);
 			$condition = ['`alias` IN (?, ?, ?) AND `uid` = ? AND NOT `deleted`', $url, Strings::normaliseLink($url), $ssl_url, $uid];
-			$contact = DBA::selectFirst('contact', ['id', 'avatar', 'avatar-date'], $condition);
+			$contact = DBA::selectFirst('contact', ['id', 'avatar', 'updated', 'network'], $condition);
 		}

 		if (DBA::isResult($contact)) {
 			$contact_id = $contact["id"];

 			// Update the contact every 7 days
-			$update_contact = ($contact['avatar-date'] < DateTimeFormat::utc('now -7 days'));
+			$update_contact = ($contact['updated'] < DateTimeFormat::utc('now -7 days'));

 			// We force the update if the avatar is empty
 			if (empty($contact['avatar'])) {
 				$update_contact = true;
 			}
+
+			// Update the contact in the background if needed but it is called by the frontend
+			if ($update_contact && $no_update && in_array($contact['network'], Protocol::NATIVE_SUPPORT)) {
+				Worker::add(PRIORITY_LOW, "UpdateContact", $contact_id);
+			}
+
 			if (!$update_contact || $no_update) {
 				return $contact_id;
 			}
@ -1187,19 +1280,13 @@ class Contact extends BaseObject
 			return 0;
 		}

-		// When we don't want to update, we look if some of our users already know this contact
-		if ($no_update) {
-			$fields = ['url', 'addr', 'alias', 'notify', 'poll', 'name', 'nick',
-				'photo', 'keywords', 'location', 'about', 'network',
-				'priority', 'batch', 'request', 'confirm', 'poco'];
-			$data = DBA::selectFirst('contact', $fields, ['nurl' => Strings::normaliseLink($url)]);
-
-			if (DBA::isResult($data)) {
-				// For security reasons we don't fetch key data from our users
-				$data["pubkey"] = '';
-			}
+		// When we don't want to update, we look if we know this contact in any way
+		if ($no_update && empty($default)) {
+			$data = self::getProbeDataFromDatabase($url, $contact_id);
+			$background_update = true;
 		} else {
 			$data = [];
+			$background_update = false;
 		}

 		if (empty($data)) {
@ -1217,44 +1304,16 @@ class Contact extends BaseObject
 				return 0;
 			}

-			// Get data from the gcontact table
-			$fields = ['name', 'nick', 'url', 'photo', 'addr', 'alias', 'network'];
-			$contact = DBA::selectFirst('gcontact', $fields, ['nurl' => Strings::normaliseLink($url)]);
-			if (!DBA::isResult($contact)) {
-				$contact = DBA::selectFirst('contact', $fields, ['nurl' => Strings::normaliseLink($url)]);
-			}
-
-			if (!DBA::isResult($contact)) {
-				$fields = ['url', 'addr', 'alias', 'notify', 'poll', 'name', 'nick',
-					'photo', 'keywords', 'location', 'about', 'network',
-					'priority', 'batch', 'request', 'confirm', 'poco'];
-				$contact = DBA::selectFirst('contact', $fields, ['addr' => $url]);
-			}
-
-			// The link could be provided as http although we stored it as https
-			$ssl_url = str_replace('http://', 'https://', $url);
-
-			if (!DBA::isResult($contact)) {
-				$condition = ['alias' => [$url, Strings::normaliseLink($url), $ssl_url]];
-				$contact = DBA::selectFirst('contact', $fields, $condition);
-			}
-
-			if (!DBA::isResult($contact)) {
-				$fields = ['url', 'addr', 'alias', 'notify', 'poll', 'name', 'nick',
-					'photo', 'network', 'priority', 'batch', 'request', 'confirm'];
-				$condition = ['url' => [$url, Strings::normaliseLink($url), $ssl_url]];
-				$contact = DBA::selectFirst('fcontact', $fields, $condition);
-			}
-
-			if (!empty($default)) {
-				$contact = $default;
-			}
-
-			if (!DBA::isResult($contact)) {
+			$contact = array_merge(self::getProbeDataFromDatabase($url, $contact_id), $default);
+			if (empty($contact)) {
 				return 0;
-			} else {
-				$data = array_merge($data, $contact);
 			}
+
+			$data = array_merge($data, $contact);
+		}
+
+		if (empty($data)) {
+			return 0;
 		}

 		if (!$contact_id && ($data["alias"] != '') && ($data["alias"] != $url) && !$in_loop) {
@ -1265,26 +1324,26 @@ class Contact extends BaseObject
 			$fields = [
 				'uid'       => $uid,
 				'created'   => DateTimeFormat::utcNow(),
-				'url'       => $data["url"],
-				'nurl'      => Strings::normaliseLink($data["url"]),
-				'addr'      => $data["addr"],
-				'alias'     => $data["alias"],
-				'notify'    => $data["notify"],
-				'poll'      => $data["poll"],
-				'name'      => $data["name"],
-				'nick'      => $data["nick"],
-				'photo'     => $data["photo"],
-				'keywords'  => $data["keywords"],
-				'location'  => $data["location"],
-				'about'     => $data["about"],
-				'network'   => $data["network"],
-				'pubkey'    => $data["pubkey"],
+				'url'       => $data['url'],
+				'nurl'      => Strings::normaliseLink($data['url']),
+				'addr'      => defaults($data, 'addr', ''),
+				'alias'     => defaults($data, 'alias', ''),
+				'notify'    => defaults($data, 'notify', ''),
+				'poll'      => defaults($data, 'poll', ''),
+				'name'      => defaults($data, 'name', ''),
+				'nick'      => defaults($data, 'nick', ''),
+				'photo'     => defaults($data, 'photo', ''),
+				'keywords'  => defaults($data, 'keywords', ''),
+				'location'  => defaults($data, 'location', ''),
+				'about'     => defaults($data, 'about', ''),
+				'network'   => $data['network'],
+				'pubkey'    => defaults($data, 'pubkey', ''),
 				'rel'       => self::SHARING,
-				'priority'  => $data["priority"],
-				'batch'     => $data["batch"],
-				'request'   => $data["request"],
-				'confirm'   => $data["confirm"],
-				'poco'      => $data["poco"],
+				'priority'  => defaults($data, 'priority', 0),
+				'batch'     => defaults($data, 'batch', ''),
+				'request'   => defaults($data, 'request', ''),
+				'confirm'   => defaults($data, 'confirm', ''),
+				'poco'      => defaults($data, 'poco', ''),
 				'name-date' => DateTimeFormat::utcNow(),
 				'uri-date'  => DateTimeFormat::utcNow(),
 				'avatar-date' => DateTimeFormat::utcNow(),
@ -1305,17 +1364,22 @@ class Contact extends BaseObject

 			$contact_id = $contacts[0]["id"];

+			// Update in the background when we fetched the data solely from the database
+			if ($background_update) {
+				Worker::add(PRIORITY_LOW, "UpdateContact", $contact_id);
+			}
+
 			// Update the newly created contact from data in the gcontact table
 			$gcontact = DBA::selectFirst('gcontact', ['location', 'about', 'keywords', 'gender'], ['nurl' => Strings::normaliseLink($data["url"])]);
 			if (DBA::isResult($gcontact)) {
 				// Only use the information when the probing hadn't fetched these values
-				if ($data['keywords'] != '') {
+				if (!empty($data['keywords'])) {
 					unset($gcontact['keywords']);
 				}
-				if ($data['location'] != '') {
+				if (!empty($data['location'])) {
 					unset($gcontact['location']);
 				}
-				if ($data['about'] != '') {
+				if (!empty($data['about'])) {
 					unset($gcontact['about']);
 				}
 				DBA::update('contact', $gcontact, ['id' => $contact_id]);
@ -1329,7 +1393,9 @@ class Contact extends BaseObject
 			}
 		}

-		self::updateAvatar($data["photo"], $uid, $contact_id);
+		if (!empty($data['photo'])) {
+			self::updateAvatar($data['photo'], $uid, $contact_id);
+		}

 		$fields = ['url', 'nurl', 'addr', 'alias', 'name', 'nick', 'keywords', 'location', 'about', 'avatar-date', 'pubkey'];
 		$contact = DBA::selectFirst('contact', $fields, ['id' => $contact_id]);
@ -1346,30 +1412,30 @@ class Contact extends BaseObject
 			'name' => $data['name'],
 			'nick' => $data['nick']];

-		if ($data['keywords'] != '') {
+		if (!empty($data['keywords'])) {
 			$updated['keywords'] = $data['keywords'];
 		}
-		if ($data['location'] != '') {
+		if (!empty($data['location'])) {
 			$updated['location'] = $data['location'];
 		}

 		// Update the technical stuff as well - if filled
-		if ($data['notify'] != '') {
+		if (!empty($data['notify'])) {
 			$updated['notify'] = $data['notify'];
 		}
-		if ($data['poll'] != '') {
+		if (!empty($data['poll'])) {
 			$updated['poll'] = $data['poll'];
 		}
-		if ($data['batch'] != '') {
+		if (!empty($data['batch'])) {
 			$updated['batch'] = $data['batch'];
 		}
-		if ($data['request'] != '') {
+		if (!empty($data['request'])) {
 			$updated['request'] = $data['request'];
 		}
-		if ($data['confirm'] != '') {
+		if (!empty($data['confirm'])) {
 			$updated['confirm'] = $data['confirm'];
 		}
-		if ($data['poco'] != '') {
+		if (!empty($data['poco'])) {
 			$updated['poco'] = $data['poco'];
 		}

@ -1385,7 +1451,7 @@ class Contact extends BaseObject
 			$updated['name-date'] = DateTimeFormat::utcNow();
 		}

-		$updated['avatar-date'] = DateTimeFormat::utcNow();
+		$updated['updated'] = DateTimeFormat::utcNow();

 		DBA::update('contact', $updated, ['id' => $contact_id], $contact);

@ -1662,7 +1728,11 @@ class Contact extends BaseObject
 		$ret = Probe::uri($contact['url'], $network, $uid, !$force);

 		// If Probe::uri fails the network code will be different (mostly "feed" or "unkn")
-		if ((in_array($ret['network'], [Protocol::FEED, Protocol::PHANTOM])) && ($ret['network'] != $contact['network'])) {
+		if (in_array($ret['network'], [Protocol::FEED, Protocol::PHANTOM]) && ($ret['network'] != $contact['network'])) {
+			return false;
+		}
+
+		if (!in_array($ret['network'], Protocol::NATIVE_SUPPORT)) {
 			return false;
 		}

@ -1684,6 +1754,7 @@ class Contact extends BaseObject
 		}

 		$ret['nurl'] = Strings::normaliseLink($ret['url']);
+		$ret['updated'] = DateTimeFormat::utcNow();

 		self::updateAvatar($ret['photo'], $uid, $id, true);

--- a/src/Protocol/ActivityPub.php
+++ b/src/Protocol/ActivityPub.php
@ -89,14 +89,15 @@ class ActivityPub
 	/**
 	 * Fetches a profile from the given url into an array that is compatible to Probe::uri
 	 *
-	 * @param string $url profile url
+	 * @param string  $url    profile url
+	 * @param boolean $update Update the profile
 	 * @return array
 	 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 	 * @throws \ImagickException
 	 */
-	public static function probeProfile($url)
+	public static function probeProfile($url, $update = true)
 	{
-		$apcontact = APContact::getByURL($url, true);
+		$apcontact = APContact::getByURL($url, $update);
 		if (empty($apcontact)) {
 			return false;
 		}
--- a/src/Worker/UpdateContact.php
+++ b/src/Worker/UpdateContact.php
@ -0,0 +1,28 @@
+<?php
+
+/**
+ * @file src/Worker/UpdateContact.php
+ */
+
+namespace Friendica\Worker;
+
+use Friendica\Core\Logger;
+use Friendica\Model\Contact;
+use Friendica\Util\DateTimeFormat;
+use Friendica\Database\DBA;
+
+class UpdateContact
+{
+	public static function execute($contact_id)
+	{
+		$success = Contact::updateFromProbe($contact_id);
+		// Update the "updated" field if the contact could be probed.
+		// We don't do this in the function above, since we don't want to
+		// update the contact whenever that function is called from anywhere.
+		if ($success) {
+			DBA::update('contact', ['updated' => DateTimeFormat::utcNow()], ['id' => $contact_id]);
+		}
+
+		Logger::info('Updated from probe', ['id' => $contact_id, 'success' => $success]);
+	}
+}