friendica/src/Model/GServer.php
2020-01-19 22:50:44 +01:00

1435 lines
43 KiB
PHP

<?php
/**
* @file src/Model/GServer.php
* This file includes the GServer class to handle with servers
*/
namespace Friendica\Model;
use DOMDocument;
use DOMXPath;
use Friendica\Core\Protocol;
use Friendica\Core\Worker;
use Friendica\Database\DBA;
use Friendica\DI;
use Friendica\Module\Register;
use Friendica\Network\CurlResult;
use Friendica\Util\Network;
use Friendica\Util\DateTimeFormat;
use Friendica\Util\Strings;
use Friendica\Util\XML;
use Friendica\Core\Logger;
use Friendica\Protocol\PortableContact;
use Friendica\Protocol\Diaspora;
use Friendica\Network\Probe;
/**
* This class handles GServer related functions
*/
class GServer
{
// Directory types
const DT_NONE = 0;
const DT_POCO = 1;
const DT_MASTODON = 2;
/**
* Checks if the given server is reachable
*
* @param string $profile URL of the given profile
* @param string $server URL of the given server (If empty, taken from profile)
* @param string $network Network value that is used, when detection failed
* @param boolean $force Force an update.
*
* @return boolean 'true' if server seems vital
*/
public static function reachable(string $profile, string $server = '', string $network = '', bool $force = false)
{
if ($server == '') {
$server = GContact::getBasepath($profile);
}
if ($server == '') {
return true;
}
return self::check($server, $network, $force);
}
/**
* Decides if a server needs to be updated, based upon several date fields
*
* @param date $created Creation date of that server entry
* @param date $updated When had the server entry be updated
* @param date $last_failure Last failure when contacting that server
* @param date $last_contact Last time the server had been contacted
*
* @return boolean Does the server record needs an update?
*/
public static function updateNeeded($created, $updated, $last_failure, $last_contact)
{
$now = strtotime(DateTimeFormat::utcNow());
if ($updated > $last_contact) {
$contact_time = strtotime($updated);
} else {
$contact_time = strtotime($last_contact);
}
$failure_time = strtotime($last_failure);
$created_time = strtotime($created);
// If there is no "created" time then use the current time
if ($created_time <= 0) {
$created_time = $now;
}
// If the last contact was less than 24 hours then don't update
if (($now - $contact_time) < (60 * 60 * 24)) {
return false;
}
// If the last failure was less than 24 hours then don't update
if (($now - $failure_time) < (60 * 60 * 24)) {
return false;
}
// If the last contact was less than a week ago and the last failure is older than a week then don't update
//if ((($now - $contact_time) < (60 * 60 * 24 * 7)) && ($contact_time > $failure_time))
// return false;
// If the last contact time was more than a week ago and the contact was created more than a week ago, then only try once a week
if ((($now - $contact_time) > (60 * 60 * 24 * 7)) && (($now - $created_time) > (60 * 60 * 24 * 7)) && (($now - $failure_time) < (60 * 60 * 24 * 7))) {
return false;
}
// If the last contact time was more than a month ago and the contact was created more than a month ago, then only try once a month
if ((($now - $contact_time) > (60 * 60 * 24 * 30)) && (($now - $created_time) > (60 * 60 * 24 * 30)) && (($now - $failure_time) < (60 * 60 * 24 * 30))) {
return false;
}
return true;
}
/**
* Checks the state of the given server.
*
* @param string $server_url URL of the given server
* @param string $network Network value that is used, when detection failed
* @param boolean $force Force an update.
*
* @return boolean 'true' if server seems vital
*/
public static function check(string $server_url, string $network = '', bool $force = false)
{
// Unify the server address
$server_url = trim($server_url, '/');
$server_url = str_replace('/index.php', '', $server_url);
if ($server_url == '') {
return false;
}
$gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
if (DBA::isResult($gserver)) {
if ($gserver['created'] <= DBA::NULL_DATETIME) {
$fields = ['created' => DateTimeFormat::utcNow()];
$condition = ['nurl' => Strings::normaliseLink($server_url)];
DBA::update('gserver', $fields, $condition);
}
$last_contact = $gserver['last_contact'];
$last_failure = $gserver['last_failure'];
// See discussion under https://forum.friendi.ca/display/0b6b25a8135aabc37a5a0f5684081633
// It can happen that a zero date is in the database, but storing it again is forbidden.
if ($last_contact < DBA::NULL_DATETIME) {
$last_contact = DBA::NULL_DATETIME;
}
if ($last_failure < DBA::NULL_DATETIME) {
$last_failure = DBA::NULL_DATETIME;
}
if (!$force && !self::updateNeeded($gserver['created'], '', $last_failure, $last_contact)) {
Logger::info('No update needed', ['server' => $server_url]);
return ($last_contact >= $last_failure);
}
Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force, 'Created' => $gserver['created'], 'Failure' => $last_failure, 'Contact' => $last_contact]);
} else {
Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
}
return self::detect($server_url, $network);
}
/**
* Detect server data (type, protocol, version number, ...)
* The detected data is then updated or inserted in the gserver table.
*
* @param string $url URL of the given server
* @param string $network Network value that is used, when detection failed
*
* @return boolean 'true' if server could be detected
*/
public static function detect(string $url, string $network = '')
{
Logger::info('Detect server type', ['server' => $url]);
$serverdata = [];
$original_url = $url;
// Remove URL content that is not supposed to exist for a server url
$urlparts = parse_url($url);
unset($urlparts['user']);
unset($urlparts['pass']);
unset($urlparts['query']);
unset($urlparts['fragment']);
$url = Network::unparseURL($urlparts);
// If the URL missmatches, then we mark the old entry as failure
if ($url != $original_url) {
DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => Strings::normaliseLink($original_url)]);
}
// When a nodeinfo is present, we don't need to dig further
$xrd_timeout = DI::config()->get('system', 'xrd_timeout');
$curlResult = Network::curl($url . '/.well-known/nodeinfo', false, ['timeout' => $xrd_timeout]);
if ($curlResult->isTimeout()) {
DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => Strings::normaliseLink($url)]);
return false;
}
$nodeinfo = self::fetchNodeinfo($url, $curlResult);
// When nodeinfo isn't present, we use the older 'statistics.json' endpoint
if (empty($nodeinfo)) {
$nodeinfo = self::fetchStatistics($url);
}
// If that didn't work out well, we use some protocol specific endpoints
// For Friendica and Zot based networks we have to dive deeper to reveal more details
if (empty($nodeinfo['network']) || in_array($nodeinfo['network'], [Protocol::DFRN, Protocol::ZOT])) {
// Fetch the landing page, possibly it reveals some data
if (empty($nodeinfo['network'])) {
$curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout]);
if ($curlResult->isSuccess()) {
$serverdata = self::analyseRootHeader($curlResult, $serverdata);
$serverdata = self::analyseRootBody($curlResult, $serverdata, $url);
}
if (!$curlResult->isSuccess() || empty($curlResult->getBody()) || self::invalidBody($curlResult->getBody())) {
DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => Strings::normaliseLink($url)]);
return false;
}
}
if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ACTIVITYPUB)) {
$serverdata = self::detectMastodonAlikes($url, $serverdata);
}
// All following checks are done for systems that always have got a "host-meta" endpoint.
// With this check we don't have to waste time and ressources for dead systems.
// Also this hopefully prevents us from receiving abuse messages.
if (empty($serverdata['network']) && !self::validHostMeta($url)) {
DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => Strings::normaliseLink($url)]);
return false;
}
if (empty($serverdata['network']) || in_array($serverdata['network'], [Protocol::DFRN, Protocol::ACTIVITYPUB])) {
$serverdata = self::detectFriendica($url, $serverdata);
}
// the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ZOT)) {
$serverdata = self::fetchSiteinfo($url, $serverdata);
}
// The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations
if (empty($serverdata['network'])) {
$serverdata = self::detectHubzilla($url, $serverdata);
}
if (empty($serverdata['network'])) {
$serverdata = self::detectNextcloud($url, $serverdata);
}
if (empty($serverdata['network'])) {
$serverdata = self::detectGNUSocial($url, $serverdata);
}
} else {
$serverdata = $nodeinfo;
}
// Detect the directory type
$serverdata['directory-type'] = self::DT_NONE;
$serverdata = self::checkPoCo($url, $serverdata);
$serverdata = self::checkMastodonDirectory($url, $serverdata);
// We can't detect the network type. Possibly it is some system that we don't know yet
if (empty($serverdata['network'])) {
$serverdata['network'] = Protocol::PHANTOM;
}
// When we hadn't been able to detect the network type, we use the hint from the parameter
if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
$serverdata['network'] = $network;
}
$serverdata['url'] = $url;
$serverdata['nurl'] = Strings::normaliseLink($url);
// We take the highest number that we do find
$registeredUsers = $serverdata['registered-users'] ?? 0;
// On an active server there has to be at least a single user
if (($serverdata['network'] != Protocol::PHANTOM) && ($registeredUsers == 0)) {
$registeredUsers = 1;
}
if ($serverdata['network'] != Protocol::PHANTOM) {
$gcontacts = DBA::count('gcontact', ['server_url' => [$url, $serverdata['nurl']]]);
$apcontacts = DBA::count('apcontact', ['baseurl' => [$url, $serverdata['nurl']]]);
$contacts = DBA::count('contact', ['uid' => 0, 'baseurl' => [$url, $serverdata['nurl']]]);
$serverdata['registered-users'] = max($gcontacts, $apcontacts, $contacts, $registeredUsers);
} else {
$serverdata['registered-users'] = $registeredUsers;
$serverdata = self::detectNetworkViaContacts($url, $serverdata);
}
$serverdata['last_contact'] = DateTimeFormat::utcNow();
$gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
if (!DBA::isResult($gserver)) {
$serverdata['created'] = DateTimeFormat::utcNow();
$ret = DBA::insert('gserver', $serverdata);
} else {
// Don't override the network with 'unknown' when there had been a valid entry before
if (($serverdata['network'] == Protocol::PHANTOM) && !empty($gserver['network'])) {
unset($serverdata['network']);
}
$ret = DBA::update('gserver', $serverdata, ['nurl' => $serverdata['nurl']]);
}
if (!empty($serverdata['network']) && in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
self::discoverRelay($url);
}
return $ret;
}
/**
* Fetch relay data from a given server url
*
* @param string $server_url address of the server
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
private static function discoverRelay(string $server_url)
{
Logger::info('Discover relay data', ['server' => $server_url]);
$curlResult = Network::curl($server_url . '/.well-known/x-social-relay');
if (!$curlResult->isSuccess()) {
return;
}
$data = json_decode($curlResult->getBody(), true);
if (!is_array($data)) {
return;
}
$gserver = DBA::selectFirst('gserver', ['id', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
if (!DBA::isResult($gserver)) {
return;
}
if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
$fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
DBA::update('gserver', $fields, ['id' => $gserver['id']]);
}
DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
if ($data['scope'] == 'tags') {
// Avoid duplicates
$tags = [];
foreach ($data['tags'] as $tag) {
$tag = mb_strtolower($tag);
if (strlen($tag) < 100) {
$tags[$tag] = $tag;
}
}
foreach ($tags as $tag) {
DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], true);
}
}
// Create or update the relay contact
$fields = [];
if (isset($data['protocols'])) {
if (isset($data['protocols']['diaspora'])) {
$fields['network'] = Protocol::DIASPORA;
if (isset($data['protocols']['diaspora']['receive'])) {
$fields['batch'] = $data['protocols']['diaspora']['receive'];
} elseif (is_string($data['protocols']['diaspora'])) {
$fields['batch'] = $data['protocols']['diaspora'];
}
}
if (isset($data['protocols']['dfrn'])) {
$fields['network'] = Protocol::DFRN;
if (isset($data['protocols']['dfrn']['receive'])) {
$fields['batch'] = $data['protocols']['dfrn']['receive'];
} elseif (is_string($data['protocols']['dfrn'])) {
$fields['batch'] = $data['protocols']['dfrn'];
}
}
}
Diaspora::setRelayContact($server_url, $fields);
}
/**
* Fetch server data from '/statistics.json' on the given server
*
* @param string $url URL of the given server
*
* @return array server data
*/
private static function fetchStatistics(string $url)
{
$curlResult = Network::curl($url . '/statistics.json');
if (!$curlResult->isSuccess()) {
return [];
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data)) {
return [];
}
$serverdata = [];
if (!empty($data['version'])) {
$serverdata['version'] = $data['version'];
// Version numbers on statistics.json are presented with additional info, e.g.:
// 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
$serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
}
if (!empty($data['name'])) {
$serverdata['site_name'] = $data['name'];
}
if (!empty($data['network'])) {
$serverdata['platform'] = strtolower($data['network']);
if ($serverdata['platform'] == 'diaspora') {
$serverdata['network'] = Protocol::DIASPORA;
} elseif ($serverdata['platform'] == 'friendica') {
$serverdata['network'] = Protocol::DFRN;
} elseif ($serverdata['platform'] == 'hubzilla') {
$serverdata['network'] = Protocol::ZOT;
} elseif ($serverdata['platform'] == 'redmatrix') {
$serverdata['network'] = Protocol::ZOT;
}
}
if (!empty($data['registrations_open'])) {
$serverdata['register_policy'] = Register::OPEN;
} else {
$serverdata['register_policy'] = Register::CLOSED;
}
return $serverdata;
}
/**
* Detect server type by using the nodeinfo data
*
* @param string $url address of the server
* @param CurlResult $curlResult
* @return array Server data
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
private static function fetchNodeinfo(string $url, CurlResult $curlResult)
{
$nodeinfo = json_decode($curlResult->getBody(), true);
if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
return [];
}
$nodeinfo1_url = '';
$nodeinfo2_url = '';
foreach ($nodeinfo['links'] as $link) {
if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
Logger::info('Invalid nodeinfo format', ['url' => $url]);
continue;
}
if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
$nodeinfo1_url = $link['href'];
} elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
$nodeinfo2_url = $link['href'];
}
}
if ($nodeinfo1_url . $nodeinfo2_url == '') {
return [];
}
$server = [];
// When the nodeinfo url isn't on the same host, then there is obviously something wrong
if (!empty($nodeinfo2_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo2_url, PHP_URL_HOST))) {
$server = self::parseNodeinfo2($nodeinfo2_url);
}
// When the nodeinfo url isn't on the same host, then there is obviously something wrong
if (empty($server) && !empty($nodeinfo1_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo1_url, PHP_URL_HOST))) {
$server = self::parseNodeinfo1($nodeinfo1_url);
}
return $server;
}
/**
* Parses Nodeinfo 1
*
* @param string $nodeinfo_url address of the nodeinfo path
* @return array Server data
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
private static function parseNodeinfo1(string $nodeinfo_url)
{
$curlResult = Network::curl($nodeinfo_url);
if (!$curlResult->isSuccess()) {
return [];
}
$nodeinfo = json_decode($curlResult->getBody(), true);
if (!is_array($nodeinfo)) {
return [];
}
$server = [];
$server['register_policy'] = Register::CLOSED;
if (!empty($nodeinfo['openRegistrations'])) {
$server['register_policy'] = Register::OPEN;
}
if (is_array($nodeinfo['software'])) {
if (!empty($nodeinfo['software']['name'])) {
$server['platform'] = strtolower($nodeinfo['software']['name']);
}
if (!empty($nodeinfo['software']['version'])) {
$server['version'] = $nodeinfo['software']['version'];
// Version numbers on Nodeinfo are presented with additional info, e.g.:
// 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
$server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
}
}
if (!empty($nodeinfo['metadata']['nodeName'])) {
$server['site_name'] = $nodeinfo['metadata']['nodeName'];
}
if (!empty($nodeinfo['usage']['users']['total'])) {
$server['registered-users'] = $nodeinfo['usage']['users']['total'];
}
if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
$protocols = [];
foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
$protocols[$protocol] = true;
}
if (!empty($protocols['friendica'])) {
$server['network'] = Protocol::DFRN;
} elseif (!empty($protocols['activitypub'])) {
$server['network'] = Protocol::ACTIVITYPUB;
} elseif (!empty($protocols['diaspora'])) {
$server['network'] = Protocol::DIASPORA;
} elseif (!empty($protocols['ostatus'])) {
$server['network'] = Protocol::OSTATUS;
} elseif (!empty($protocols['gnusocial'])) {
$server['network'] = Protocol::OSTATUS;
} elseif (!empty($protocols['zot'])) {
$server['network'] = Protocol::ZOT;
}
}
if (empty($server)) {
return [];
}
return $server;
}
/**
* Parses Nodeinfo 2
*
* @param string $nodeinfo_url address of the nodeinfo path
* @return array Server data
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
private static function parseNodeinfo2(string $nodeinfo_url)
{
$curlResult = Network::curl($nodeinfo_url);
if (!$curlResult->isSuccess()) {
return [];
}
$nodeinfo = json_decode($curlResult->getBody(), true);
if (!is_array($nodeinfo)) {
return [];
}
$server = [];
$server['register_policy'] = Register::CLOSED;
if (!empty($nodeinfo['openRegistrations'])) {
$server['register_policy'] = Register::OPEN;
}
if (is_array($nodeinfo['software'])) {
if (!empty($nodeinfo['software']['name'])) {
$server['platform'] = strtolower($nodeinfo['software']['name']);
}
if (!empty($nodeinfo['software']['version'])) {
$server['version'] = $nodeinfo['software']['version'];
// Version numbers on Nodeinfo are presented with additional info, e.g.:
// 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
$server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
}
}
if (!empty($nodeinfo['metadata']['nodeName'])) {
$server['site_name'] = $nodeinfo['metadata']['nodeName'];
}
if (!empty($nodeinfo['usage']['users']['total'])) {
$server['registered-users'] = $nodeinfo['usage']['users']['total'];
}
if (!empty($nodeinfo['protocols'])) {
$protocols = [];
foreach ($nodeinfo['protocols'] as $protocol) {
$protocols[$protocol] = true;
}
if (!empty($protocols['dfrn'])) {
$server['network'] = Protocol::DFRN;
} elseif (!empty($protocols['activitypub'])) {
$server['network'] = Protocol::ACTIVITYPUB;
} elseif (!empty($protocols['diaspora'])) {
$server['network'] = Protocol::DIASPORA;
} elseif (!empty($protocols['ostatus'])) {
$server['network'] = Protocol::OSTATUS;
} elseif (!empty($protocols['gnusocial'])) {
$server['network'] = Protocol::OSTATUS;
} elseif (!empty($protocols['zot'])) {
$server['network'] = Protocol::ZOT;
}
}
if (empty($server)) {
return [];
}
return $server;
}
/**
* Fetch server information from a 'siteinfo.json' file on the given server
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function fetchSiteinfo(string $url, array $serverdata)
{
$curlResult = Network::curl($url . '/siteinfo.json');
if (!$curlResult->isSuccess()) {
return $serverdata;
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data)) {
return $serverdata;
}
if (!empty($data['url'])) {
$serverdata['platform'] = strtolower($data['platform']);
$serverdata['version'] = $data['version'];
}
if (!empty($data['plugins'])) {
if (in_array('pubcrawl', $data['plugins'])) {
$serverdata['network'] = Protocol::ACTIVITYPUB;
} elseif (in_array('diaspora', $data['plugins'])) {
$serverdata['network'] = Protocol::DIASPORA;
} elseif (in_array('gnusoc', $data['plugins'])) {
$serverdata['network'] = Protocol::OSTATUS;
} else {
$serverdata['network'] = Protocol::ZOT;
}
}
if (!empty($data['site_name'])) {
$serverdata['site_name'] = $data['site_name'];
}
if (!empty($data['channels_total'])) {
$serverdata['registered-users'] = $data['channels_total'];
}
if (!empty($data['register_policy'])) {
switch ($data['register_policy']) {
case 'REGISTER_OPEN':
$serverdata['register_policy'] = Register::OPEN;
break;
case 'REGISTER_APPROVE':
$serverdata['register_policy'] = Register::APPROVE;
break;
case 'REGISTER_CLOSED':
default:
$serverdata['register_policy'] = Register::CLOSED;
break;
}
}
return $serverdata;
}
/**
* Checks if the server contains a valid host meta file
*
* @param string $url URL of the given server
*
* @return boolean 'true' if the server seems to be vital
*/
private static function validHostMeta(string $url)
{
$xrd_timeout = DI::config()->get('system', 'xrd_timeout');
$curlResult = Network::curl($url . '/.well-known/host-meta', false, ['timeout' => $xrd_timeout]);
if (!$curlResult->isSuccess()) {
return false;
}
$xrd = XML::parseString($curlResult->getBody(), false);
if (!is_object($xrd)) {
return false;
}
$elements = XML::elementToArray($xrd);
if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
return false;
}
$valid = false;
foreach ($elements['xrd']['link'] as $link) {
// When there is more than a single "link" element, the array looks slightly different
if (!empty($link['@attributes'])) {
$link = $link['@attributes'];
}
if (empty($link['rel']) || empty($link['template'])) {
continue;
}
if ($link['rel'] == 'lrdd') {
// When the webfinger host is the same like the system host, it should be ok.
$valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
}
}
return $valid;
}
/**
* Detect the network of the given server via their known contacts
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function detectNetworkViaContacts(string $url, array $serverdata)
{
$contacts = [];
$gcontacts = DBA::select('gcontact', ['url', 'nurl'], ['server_url' => [$url, $serverdata['nurl']]]);
while ($gcontact = DBA::fetch($gcontacts)) {
$contacts[$gcontact['nurl']] = $gcontact['url'];
}
DBA::close($gcontacts);
$apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $serverdata['nurl']]]);
while ($gcontact = DBA::fetch($gcontacts)) {
$contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
}
DBA::close($apcontacts);
$pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $serverdata['nurl']]]);
while ($gcontact = DBA::fetch($gcontacts)) {
$contacts[$pcontact['nurl']] = $pcontact['url'];
}
DBA::close($pcontacts);
if (empty($contacts)) {
return $serverdata;
}
foreach ($contacts as $contact) {
$probed = Probe::uri($contact);
if (in_array($probed['network'], Protocol::FEDERATED)) {
$serverdata['network'] = $probed['network'];
break;
}
}
$serverdata['registered-users'] = max($serverdata['registered-users'], count($contacts));
return $serverdata;
}
/**
* Checks if the given server does have a '/poco' endpoint.
* This is used for the 'PortableContact' functionality,
* which is used by both Friendica and Hubzilla.
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function checkPoCo(string $url, array $serverdata)
{
$serverdata['poco'] = '';
$curlResult = Network::curl($url. '/poco');
if (!$curlResult->isSuccess()) {
return $serverdata;
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data)) {
return $serverdata;
}
if (!empty($data['totalResults'])) {
$registeredUsers = $serverdata['registered-users'] ?? 0;
$serverdata['registered-users'] = max($data['totalResults'], $registeredUsers);
$serverdata['directory-type'] = self::DT_POCO;
$serverdata['poco'] = $url . '/poco';
}
return $serverdata;
}
/**
* Checks if the given server does have a Mastodon style directory endpoint.
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
public static function checkMastodonDirectory(string $url, array $serverdata)
{
$curlResult = Network::curl($url . '/api/v1/directory?limit=1');
if (!$curlResult->isSuccess()) {
return $serverdata;
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data)) {
return $serverdata;
}
if (count($data) == 1) {
$serverdata['directory-type'] = self::DT_MASTODON;
}
return $serverdata;
}
/**
* Detects the version number of a given server when it was a NextCloud installation
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function detectNextcloud(string $url, array $serverdata)
{
$curlResult = Network::curl($url . '/status.php');
if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
return $serverdata;
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data)) {
return $serverdata;
}
if (!empty($data['version'])) {
$serverdata['platform'] = 'nextcloud';
$serverdata['version'] = $data['version'];
$serverdata['network'] = Protocol::ACTIVITYPUB;
}
return $serverdata;
}
/**
* Detects data from a given server url if it was a mastodon alike system
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function detectMastodonAlikes(string $url, array $serverdata)
{
$curlResult = Network::curl($url . '/api/v1/instance');
if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
return $serverdata;
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data)) {
return $serverdata;
}
if (!empty($data['version'])) {
$serverdata['platform'] = 'mastodon';
$serverdata['version'] = $data['version'] ?? '';
$serverdata['network'] = Protocol::ACTIVITYPUB;
}
if (!empty($data['title'])) {
$serverdata['site_name'] = $data['title'];
}
if (!empty($data['title']) && empty($serverdata['platform']) && empty($serverdata['network'])) {
$serverdata['platform'] = 'mastodon';
$serverdata['network'] = Protocol::ACTIVITYPUB;
}
if (!empty($data['description'])) {
$serverdata['info'] = trim($data['description']);
}
if (!empty($data['stats']['user_count'])) {
$serverdata['registered-users'] = $data['stats']['user_count'];
}
if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
$serverdata['platform'] = strtolower($matches[1]);
$serverdata['version'] = $matches[2];
}
if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
$serverdata['platform'] = 'pleroma';
$serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
}
if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
$serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
$serverdata['platform'] = 'pleroma';
}
return $serverdata;
}
/**
* Detects data from typical Hubzilla endpoints
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function detectHubzilla(string $url, array $serverdata)
{
$curlResult = Network::curl($url . '/api/statusnet/config.json');
if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
return $serverdata;
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data)) {
return $serverdata;
}
if (!empty($data['site']['name'])) {
$serverdata['site_name'] = $data['site']['name'];
}
if (!empty($data['site']['platform'])) {
$serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
$serverdata['version'] = $data['site']['platform']['STD_VERSION'];
$serverdata['network'] = Protocol::ZOT;
}
if (!empty($data['site']['hubzilla'])) {
$serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
$serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
$serverdata['network'] = Protocol::ZOT;
}
if (!empty($data['site']['redmatrix'])) {
if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
$serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
} elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
$serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
}
$serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
$serverdata['network'] = Protocol::ZOT;
}
$private = false;
$inviteonly = false;
$closed = false;
if (!empty($data['site']['closed'])) {
$closed = self::toBoolean($data['site']['closed']);
}
if (!empty($data['site']['private'])) {
$private = self::toBoolean($data['site']['private']);
}
if (!empty($data['site']['inviteonly'])) {
$inviteonly = self::toBoolean($data['site']['inviteonly']);
}
if (!$closed && !$private and $inviteonly) {
$register_policy = Register::APPROVE;
} elseif (!$closed && !$private) {
$register_policy = Register::OPEN;
} else {
$register_policy = Register::CLOSED;
}
return $serverdata;
}
/**
* Converts input value to a boolean value
*
* @param string|integer $val
*
* @return boolean
*/
private static function toBoolean($val)
{
if (($val == 'true') || ($val == 1)) {
return true;
} elseif (($val == 'false') || ($val == 0)) {
return false;
}
return $val;
}
/**
* Detect if the URL belongs to a GNU Social server
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function detectGNUSocial(string $url, array $serverdata)
{
// Test for GNU Social
$curlResult = Network::curl($url . '/api/gnusocial/version.json');
if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
$serverdata['platform'] = 'gnusocial';
// Remove junk that some GNU Social servers return
$serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
$serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
$serverdata['version'] = trim($serverdata['version'], '"');
$serverdata['network'] = Protocol::OSTATUS;
return $serverdata;
}
// Test for Statusnet
$curlResult = Network::curl($url . '/api/statusnet/version.json');
if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
// Remove junk that some GNU Social servers return
$serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
$serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
$serverdata['version'] = trim($serverdata['version'], '"');
if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
$serverdata['platform'] = 'pleroma';
$serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
$serverdata['network'] = Protocol::ACTIVITYPUB;
} else {
$serverdata['platform'] = 'statusnet';
$serverdata['network'] = Protocol::OSTATUS;
}
}
return $serverdata;
}
/**
* Detect if the URL belongs to a Friendica server
*
* @param string $url URL of the given server
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function detectFriendica(string $url, array $serverdata)
{
$curlResult = Network::curl($url . '/friendica/json');
if (!$curlResult->isSuccess()) {
$curlResult = Network::curl($url . '/friendika/json');
}
if (!$curlResult->isSuccess()) {
return $serverdata;
}
$data = json_decode($curlResult->getBody(), true);
if (empty($data) || empty($data['version'])) {
return $serverdata;
}
$serverdata['network'] = Protocol::DFRN;
$serverdata['version'] = $data['version'];
if (!empty($data['no_scrape_url'])) {
$serverdata['noscrape'] = $data['no_scrape_url'];
}
if (!empty($data['site_name'])) {
$serverdata['site_name'] = $data['site_name'];
}
if (!empty($data['info'])) {
$serverdata['info'] = trim($data['info']);
}
$register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
switch ($register_policy) {
case 'REGISTER_OPEN':
$serverdata['register_policy'] = Register::OPEN;
break;
case 'REGISTER_APPROVE':
$serverdata['register_policy'] = Register::APPROVE;
break;
case 'REGISTER_CLOSED':
case 'REGISTER_INVITATION':
$serverdata['register_policy'] = Register::CLOSED;
break;
default:
Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
$serverdata['register_policy'] = Register::CLOSED;
break;
}
$serverdata['platform'] = strtolower($data['platform'] ?? '');
return $serverdata;
}
/**
* Analyses the landing page of a given server for hints about type and system of that server
*
* @param object $curlResult result of curl execution
* @param array $serverdata array with server data
* @param string $url Server URL
*
* @return array server data
*/
private static function analyseRootBody($curlResult, array $serverdata, string $url)
{
$doc = new DOMDocument();
@$doc->loadHTML($curlResult->getBody());
$xpath = new DOMXPath($doc);
$title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
if (!empty($title)) {
$serverdata['site_name'] = $title;
}
$list = $xpath->query('//meta[@name]');
foreach ($list as $node) {
$attr = [];
if ($node->attributes->length) {
foreach ($node->attributes as $attribute) {
$value = trim($attribute->value);
if (empty($value)) {
continue;
}
$attr[$attribute->name] = $value;
}
if (empty($attr['name']) || empty($attr['content'])) {
continue;
}
}
if ($attr['name'] == 'description') {
$serverdata['info'] = $attr['content'];
}
if ($attr['name'] == 'application-name') {
$serverdata['platform'] = strtolower($attr['content']);
if (in_array($attr['content'], ['Misskey', 'Write.as'])) {
$serverdata['network'] = Protocol::ACTIVITYPUB;
}
}
if (($attr['name'] == 'generator') && (empty($serverdata['platform']) || (substr(strtolower($attr['content']), 0, 9) == 'wordpress'))) {
$serverdata['platform'] = strtolower($attr['content']);
$version_part = explode(' ', $attr['content']);
if (count($version_part) == 2) {
if (in_array($version_part[0], ['WordPress'])) {
$serverdata['platform'] = strtolower($version_part[0]);
$serverdata['version'] = $version_part[1];
// We still do need a reliable test if some AP plugin is activated
if (DBA::exists('apcontact', ['baseurl' => $url])) {
$serverdata['network'] = Protocol::ACTIVITYPUB;
} else {
$serverdata['network'] = Protocol::FEED;
}
}
if (in_array($version_part[0], ['Friendika', 'Friendica'])) {
$serverdata['platform'] = strtolower($version_part[0]);
$serverdata['version'] = $version_part[1];
$serverdata['network'] = Protocol::DFRN;
}
}
}
}
$list = $xpath->query('//meta[@property]');
foreach ($list as $node) {
$attr = [];
if ($node->attributes->length) {
foreach ($node->attributes as $attribute) {
$value = trim($attribute->value);
if (empty($value)) {
continue;
}
$attr[$attribute->name] = $value;
}
if (empty($attr['property']) || empty($attr['content'])) {
continue;
}
}
if ($attr['property'] == 'og:site_name') {
$serverdata['site_name'] = $attr['content'];
}
if ($attr['property'] == 'og:description') {
$serverdata['info'] = $attr['content'];
}
if ($attr['property'] == 'og:platform') {
$serverdata['platform'] = strtolower($attr['content']);
if (in_array($attr['content'], ['PeerTube'])) {
$serverdata['network'] = Protocol::ACTIVITYPUB;
}
}
if ($attr['property'] == 'generator') {
$serverdata['platform'] = strtolower($attr['content']);
if (in_array($attr['content'], ['hubzilla'])) {
// We later check which compatible protocol modules are loaded.
$serverdata['network'] = Protocol::ZOT;
}
}
}
return $serverdata;
}
/**
* Analyses the header data of a given server for hints about type and system of that server
*
* @param object $curlResult result of curl execution
* @param array $serverdata array with server data
*
* @return array server data
*/
private static function analyseRootHeader($curlResult, array $serverdata)
{
if ($curlResult->getHeader('server') == 'Mastodon') {
$serverdata['platform'] = 'mastodon';
$serverdata['network'] = $network = Protocol::ACTIVITYPUB;
} elseif ($curlResult->inHeader('x-diaspora-version')) {
$serverdata['platform'] = 'diaspora';
$serverdata['network'] = $network = Protocol::DIASPORA;
$serverdata['version'] = $curlResult->getHeader('x-diaspora-version');
} elseif ($curlResult->inHeader('x-friendica-version')) {
$serverdata['platform'] = 'friendica';
$serverdata['network'] = $network = Protocol::DFRN;
$serverdata['version'] = $curlResult->getHeader('x-friendica-version');
}
return $serverdata;
}
/**
* Test if the body contains valid content
*
* @param string $body
* @return boolean
*/
private static function invalidBody(string $body)
{
// Currently we only test for a HTML element.
// Possibly we enhance this in the future.
return !strpos($body, '>');
}
/**
* Update the user directory of a given gserver record
*
* @param array $gserver gserver record
*/
public static function updateDirectory(array $gserver)
{
/// @todo Add Mastodon API directory
if (!empty($gserver['poco'])) {
PortableContact::discoverSingleServer($gserver['id']);
}
}
/**
* Update GServer entries
*/
public static function discover()
{
// Update the server list
self::discoverFederation();
$no_of_queries = 5;
$requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
if ($requery_days == 0) {
$requery_days = 7;
}
$last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
$gservers = DBA::p("SELECT `id`, `url`, `nurl`, `network`, `poco`
FROM `gserver`
WHERE `last_contact` >= `last_failure`
AND `poco` != ''
AND `last_poco_query` < ?
ORDER BY RAND()", $last_update
);
while ($gserver = DBA::fetch($gservers)) {
if (!GServer::check($gserver['url'], $gserver['network'])) {
// The server is not reachable? Okay, then we will try it later
$fields = ['last_poco_query' => DateTimeFormat::utcNow()];
DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
continue;
}
Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
Worker::add(PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
if (--$no_of_queries == 0) {
break;
}
}
DBA::close($gservers);
}
/**
* Discover federated servers
*/
private static function discoverFederation()
{
$last = DI::config()->get('poco', 'last_federation_discovery');
if ($last) {
$next = $last + (24 * 60 * 60);
if ($next > time()) {
return;
}
}
// Discover federated servers
$curlResult = Network::fetchUrl("http://the-federation.info/pods.json");
if (!empty($curlResult)) {
$servers = json_decode($curlResult, true);
if (!empty($servers['pods'])) {
foreach ($servers['pods'] as $server) {
Worker::add(PRIORITY_LOW, 'UpdateGServer', 'https://' . $server['host']);
}
}
}
// Disvover Mastodon servers
$accesstoken = DI::config()->get('system', 'instances_social_key');
if (!empty($accesstoken)) {
$api = 'https://instances.social/api/1.0/instances/list?count=0';
$header = ['Authorization: Bearer '.$accesstoken];
$curlResult = Network::curl($api, false, ['headers' => $header]);
if ($curlResult->isSuccess()) {
$servers = json_decode($curlResult->getBody(), true);
foreach ($servers['instances'] as $server) {
$url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
Worker::add(PRIORITY_LOW, 'UpdateGServer', $url);
}
}
}
DI::config()->set('poco', 'last_federation_discovery', time());
}
}