src/Protocol/PortableContact.php

   1 <?php
   2 /**
   3  * @file src/Protocol/PortableContact.php
   4  *
   5  * @todo Move GNU Social URL schemata (http://server.tld/user/number) to http://server.tld/username
   6  * @todo Fetch profile data from profile page for Redmatrix users
   7  * @todo Detect if it is a forum
   8  */
   9
  10 namespace Friendica\Protocol;
  11
  12 use DOMDocument;
  13 use DOMXPath;
  14 use Exception;
  15 use Friendica\Content\Text\HTML;
  16 use Friendica\Core\Config;
  17 use Friendica\Core\Logger;
  18 use Friendica\Core\Protocol;
  19 use Friendica\Core\Worker;
  20 use Friendica\Database\DBA;
  21 use Friendica\Model\GContact;
  22 use Friendica\Model\Profile;
  23 use Friendica\Network\Probe;
  24 use Friendica\Util\DateTimeFormat;
  25 use Friendica\Util\Network;
  26 use Friendica\Util\XML;
  27
  28 require_once 'include/dba.php';
  29
  30 class PortableContact
  31 {
  32         /**
  33          * @brief Fetch POCO data
  34          *
  35          * @param integer $cid  Contact ID
  36          * @param integer $uid  User ID
  37          * @param integer $zcid Global Contact ID
  38          * @param integer $url  POCO address that should be polled
  39          *
  40          * Given a contact-id (minimum), load the PortableContacts friend list for that contact,
  41          * and add the entries to the gcontact (Global Contact) table, or update existing entries
  42          * if anything (name or photo) has changed.
  43          * We use normalised urls for comparison which ignore http vs https and www.domain vs domain
  44          *
  45          * Once the global contact is stored add (if necessary) the contact linkage which associates
  46          * the given uid, cid to the global contact entry. There can be many uid/cid combinations
  47          * pointing to the same global contact id.
  48          *
  49          */
  50         public static function loadWorker($cid, $uid = 0, $zcid = 0, $url = null)
  51         {
  52                 // Call the function "load" via the worker
  53                 Worker::add(PRIORITY_LOW, "DiscoverPoCo", "load", (int)$cid, (int)$uid, (int)$zcid, $url);
  54         }
  55
  56         /**
  57          * @brief Fetch POCO data from the worker
  58          *
  59          * @param integer $cid  Contact ID
  60          * @param integer $uid  User ID
  61          * @param integer $zcid Global Contact ID
  62          * @param integer $url  POCO address that should be polled
  63          *
  64          */
  65         public static function load($cid, $uid, $zcid, $url)
  66         {
  67                 $a = get_app();
  68
  69                 if ($cid) {
  70                         if (!$url || !$uid) {
  71                                 $contact = DBA::selectFirst('contact', ['poco', 'uid'], ['id' => $cid]);
  72                                 if (DBA::isResult($contact)) {
  73                                         $url = $contact['poco'];
  74                                         $uid = $contact['uid'];
  75                                 }
  76                         }
  77                         if (!$uid) {
  78                                 return;
  79                         }
  80                 }
  81
  82                 if (!$url) {
  83                         return;
  84                 }
  85
  86                 $url = $url . (($uid) ? '/@me/@all?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation' : '?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation') ;
  87
  88                 Logger::log('load: ' . $url, Logger::DEBUG);
  89
  90                 $fetchresult = Network::fetchUrlFull($url);
  91                 $s = $fetchresult->getBody();
  92
  93                 Logger::log('load: returns ' . $s, Logger::DATA);
  94
  95                 Logger::log('load: return code: ' . $fetchresult->getReturnCode(), Logger::DEBUG);
  96
  97                 if (($fetchresult->getReturnCode() > 299) || (! $s)) {
  98                         return;
  99                 }
 100
 101                 $j = json_decode($s, true);
 102
 103                 Logger::log('load: json: ' . print_r($j, true), Logger::DATA);
 104
 105                 if (!isset($j['entry'])) {
 106                         return;
 107                 }
 108
 109                 $total = 0;
 110                 foreach ($j['entry'] as $entry) {
 111                         $total ++;
 112                         $profile_url = '';
 113                         $profile_photo = '';
 114                         $connect_url = '';
 115                         $name = '';
 116                         $network = '';
 117                         $updated = DBA::NULL_DATETIME;
 118                         $location = '';
 119                         $about = '';
 120                         $keywords = '';
 121                         $gender = '';
 122                         $contact_type = -1;
 123                         $generation = 0;
 124
 125                         if (!empty($entry['displayName'])) {
 126                                 $name = $entry['displayName'];
 127                         }
 128
 129                         if (isset($entry['urls'])) {
 130                                 foreach ($entry['urls'] as $url) {
 131                                         if ($url['type'] == 'profile') {
 132                                                 $profile_url = $url['value'];
 133                                                 continue;
 134                                         }
 135                                         if ($url['type'] == 'webfinger') {
 136                                                 $connect_url = str_replace('acct:', '', $url['value']);
 137                                                 continue;
 138                                         }
 139                                 }
 140                         }
 141                         if (isset($entry['photos'])) {
 142                                 foreach ($entry['photos'] as $photo) {
 143                                         if ($photo['type'] == 'profile') {
 144                                                 $profile_photo = $photo['value'];
 145                                                 continue;
 146                                         }
 147                                 }
 148                         }
 149
 150                         if (isset($entry['updated'])) {
 151                                 $updated = date(DateTimeFormat::MYSQL, strtotime($entry['updated']));
 152                         }
 153
 154                         if (isset($entry['network'])) {
 155                                 $network = $entry['network'];
 156                         }
 157
 158                         if (isset($entry['currentLocation'])) {
 159                                 $location = $entry['currentLocation'];
 160                         }
 161
 162                         if (isset($entry['aboutMe'])) {
 163                                 $about = HTML::toBBCode($entry['aboutMe']);
 164                         }
 165
 166                         if (isset($entry['gender'])) {
 167                                 $gender = $entry['gender'];
 168                         }
 169
 170                         if (isset($entry['generation']) && ($entry['generation'] > 0)) {
 171                                 $generation = ++$entry['generation'];
 172                         }
 173
 174                         if (isset($entry['tags'])) {
 175                                 foreach ($entry['tags'] as $tag) {
 176                                         $keywords = implode(", ", $tag);
 177                                 }
 178                         }
 179
 180                         if (isset($entry['contactType']) && ($entry['contactType'] >= 0)) {
 181                                 $contact_type = $entry['contactType'];
 182                         }
 183
 184                         $gcontact = ["url" => $profile_url,
 185                                         "name" => $name,
 186                                         "network" => $network,
 187                                         "photo" => $profile_photo,
 188                                         "about" => $about,
 189                                         "location" => $location,
 190                                         "gender" => $gender,
 191                                         "keywords" => $keywords,
 192                                         "connect" => $connect_url,
 193                                         "updated" => $updated,
 194                                         "contact-type" => $contact_type,
 195                                         "generation" => $generation];
 196
 197                         try {
 198                                 $gcontact = GContact::sanitize($gcontact);
 199                                 $gcid = GContact::update($gcontact);
 200
 201                                 GContact::link($gcid, $uid, $cid, $zcid);
 202                         } catch (Exception $e) {
 203                                 Logger::log($e->getMessage(), Logger::DEBUG);
 204                         }
 205                 }
 206                 Logger::log("load: loaded $total entries", Logger::DEBUG);
 207
 208                 $condition = ["`cid` = ? AND `uid` = ? AND `zcid` = ? AND `updated` < UTC_TIMESTAMP - INTERVAL 2 DAY", $cid, $uid, $zcid];
 209                 DBA::delete('glink', $condition);
 210         }
 211
 212         public static function reachable($profile, $server = "", $network = "", $force = false)
 213         {
 214                 if ($server == "") {
 215                         $server = self::detectServer($profile);
 216                 }
 217
 218                 if ($server == "") {
 219                         return true;
 220                 }
 221
 222                 return self::checkServer($server, $network, $force);
 223         }
 224
 225         public static function detectServer($profile)
 226         {
 227                 // Try to detect the server path based upon some known standard paths
 228                 $server_url = "";
 229
 230                 if ($server_url == "") {
 231                         $friendica = preg_replace("=(https?://)(.*)/profile/(.*)=ism", "$1$2", $profile);
 232                         if ($friendica != $profile) {
 233                                 $server_url = $friendica;
 234                                 $network = Protocol::DFRN;
 235                         }
 236                 }
 237
 238                 if ($server_url == "") {
 239                         $diaspora = preg_replace("=(https?://)(.*)/u/(.*)=ism", "$1$2", $profile);
 240                         if ($diaspora != $profile) {
 241                                 $server_url = $diaspora;
 242                                 $network = Protocol::DIASPORA;
 243                         }
 244                 }
 245
 246                 if ($server_url == "") {
 247                         $red = preg_replace("=(https?://)(.*)/channel/(.*)=ism", "$1$2", $profile);
 248                         if ($red != $profile) {
 249                                 $server_url = $red;
 250                                 $network = Protocol::DIASPORA;
 251                         }
 252                 }
 253
 254                 // Mastodon
 255                 if ($server_url == "") {
 256                         $mastodon = preg_replace("=(https?://)(.*)/users/(.*)=ism", "$1$2", $profile);
 257                         if ($mastodon != $profile) {
 258                                 $server_url = $mastodon;
 259                                 $network = Protocol::OSTATUS;
 260                         }
 261                 }
 262
 263                 // Numeric OStatus variant
 264                 if ($server_url == "") {
 265                         $ostatus = preg_replace("=(https?://)(.*)/user/(.*)=ism", "$1$2", $profile);
 266                         if ($ostatus != $profile) {
 267                                 $server_url = $ostatus;
 268                                 $network = Protocol::OSTATUS;
 269                         }
 270                 }
 271
 272                 // Wild guess
 273                 if ($server_url == "") {
 274                         $base = preg_replace("=(https?://)(.*?)/(.*)=ism", "$1$2", $profile);
 275                         if ($base != $profile) {
 276                                 $server_url = $base;
 277                                 $network = Protocol::PHANTOM;
 278                         }
 279                 }
 280
 281                 if ($server_url == "") {
 282                         return "";
 283                 }
 284
 285                 $r = q(
 286                         "SELECT `id` FROM `gserver` WHERE `nurl` = '%s' AND `last_contact` > `last_failure`",
 287                         DBA::escape(normalise_link($server_url))
 288                 );
 289
 290                 if (DBA::isResult($r)) {
 291                         return $server_url;
 292                 }
 293
 294                 // Fetch the host-meta to check if this really is a server
 295                 $curlResult = Network::curl($server_url."/.well-known/host-meta");
 296                 if (!$curlResult->isSuccess()) {
 297                         return "";
 298                 }
 299
 300                 return $server_url;
 301         }
 302
 303         public static function alternateOStatusUrl($url)
 304         {
 305                 return(preg_match("=https?://.+/user/\d+=ism", $url, $matches));
 306         }
 307
 308         public static function lastUpdated($profile, $force = false)
 309         {
 310                 $gcontacts = q(
 311                         "SELECT * FROM `gcontact` WHERE `nurl` = '%s'",
 312                         DBA::escape(normalise_link($profile))
 313                 );
 314
 315                 if (!DBA::isResult($gcontacts)) {
 316                         return false;
 317                 }
 318
 319                 $contact = ["url" => $profile];
 320
 321                 if ($gcontacts[0]["created"] <= DBA::NULL_DATETIME) {
 322                         $contact['created'] = DateTimeFormat::utcNow();
 323                 }
 324
 325                 $server_url = '';
 326                 if ($force) {
 327                         $server_url = normalise_link(self::detectServer($profile));
 328                 }
 329
 330                 if (($server_url == '') && ($gcontacts[0]["server_url"] != "")) {
 331                         $server_url = $gcontacts[0]["server_url"];
 332                 }
 333
 334                 if (!$force && (($server_url == '') || ($gcontacts[0]["server_url"] == $gcontacts[0]["nurl"]))) {
 335                         $server_url = normalise_link(self::detectServer($profile));
 336                 }
 337
 338                 if (!in_array($gcontacts[0]["network"], [Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::FEED, Protocol::OSTATUS, ""])) {
 339                         Logger::log("Profile ".$profile.": Network type ".$gcontacts[0]["network"]." can't be checked", Logger::DEBUG);
 340                         return false;
 341                 }
 342
 343                 if ($server_url != "") {
 344                         if (!self::checkServer($server_url, $gcontacts[0]["network"], $force)) {
 345                                 if ($force) {
 346                                         $fields = ['last_failure' => DateTimeFormat::utcNow()];
 347                                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 348                                 }
 349
 350                                 Logger::log("Profile ".$profile.": Server ".$server_url." wasn't reachable.", Logger::DEBUG);
 351                                 return false;
 352                         }
 353                         $contact['server_url'] = $server_url;
 354                 }
 355
 356                 if (in_array($gcontacts[0]["network"], ["", Protocol::FEED])) {
 357                         $server = q(
 358                                 "SELECT `network` FROM `gserver` WHERE `nurl` = '%s' AND `network` != ''",
 359                                 DBA::escape(normalise_link($server_url))
 360                         );
 361
 362                         if ($server) {
 363                                 $contact['network'] = $server[0]["network"];
 364                         } else {
 365                                 return false;
 366                         }
 367                 }
 368
 369                 // noscrape is really fast so we don't cache the call.
 370                 if (($server_url != "") && ($gcontacts[0]["nick"] != "")) {
 371                         //  Use noscrape if possible
 372                         $server = q("SELECT `noscrape`, `network` FROM `gserver` WHERE `nurl` = '%s' AND `noscrape` != ''", DBA::escape(normalise_link($server_url)));
 373
 374                         if ($server) {
 375                                 $curlResult = Network::curl($server[0]["noscrape"]."/".$gcontacts[0]["nick"]);
 376
 377                                 if ($curlResult->isSuccess() && ($curlResult->getBody() != "")) {
 378                                         $noscrape = json_decode($curlResult->getBody(), true);
 379
 380                                         if (is_array($noscrape)) {
 381                                                 $contact["network"] = $server[0]["network"];
 382
 383                                                 if (isset($noscrape["fn"])) {
 384                                                         $contact["name"] = $noscrape["fn"];
 385                                                 }
 386                                                 if (isset($noscrape["comm"])) {
 387                                                         $contact["community"] = $noscrape["comm"];
 388                                                 }
 389                                                 if (isset($noscrape["tags"])) {
 390                                                         $keywords = implode(" ", $noscrape["tags"]);
 391                                                         if ($keywords != "") {
 392                                                                 $contact["keywords"] = $keywords;
 393                                                         }
 394                                                 }
 395
 396                                                 $location = Profile::formatLocation($noscrape);
 397                                                 if ($location) {
 398                                                         $contact["location"] = $location;
 399                                                 }
 400                                                 if (isset($noscrape["dfrn-notify"])) {
 401                                                         $contact["notify"] = $noscrape["dfrn-notify"];
 402                                                 }
 403                                                 // Remove all fields that are not present in the gcontact table
 404                                                 unset($noscrape["fn"]);
 405                                                 unset($noscrape["key"]);
 406                                                 unset($noscrape["homepage"]);
 407                                                 unset($noscrape["comm"]);
 408                                                 unset($noscrape["tags"]);
 409                                                 unset($noscrape["locality"]);
 410                                                 unset($noscrape["region"]);
 411                                                 unset($noscrape["country-name"]);
 412                                                 unset($noscrape["contacts"]);
 413                                                 unset($noscrape["dfrn-request"]);
 414                                                 unset($noscrape["dfrn-confirm"]);
 415                                                 unset($noscrape["dfrn-notify"]);
 416                                                 unset($noscrape["dfrn-poll"]);
 417
 418                                                 // Set the date of the last contact
 419                                                 /// @todo By now the function "update_gcontact" doesn't work with this field
 420                                                 //$contact["last_contact"] = DateTimeFormat::utcNow();
 421
 422                                                 $contact = array_merge($contact, $noscrape);
 423
 424                                                 GContact::update($contact);
 425
 426                                                 if (!empty($noscrape["updated"])) {
 427                                                         $fields = ['last_contact' => DateTimeFormat::utcNow()];
 428                                                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 429
 430                                                         Logger::log("Profile ".$profile." was last updated at ".$noscrape["updated"]." (noscrape)", Logger::DEBUG);
 431
 432                                                         return $noscrape["updated"];
 433                                                 }
 434                                         }
 435                                 }
 436                         }
 437                 }
 438
 439                 // If we only can poll the feed, then we only do this once a while
 440                 if (!$force && !self::updateNeeded($gcontacts[0]["created"], $gcontacts[0]["updated"], $gcontacts[0]["last_failure"], $gcontacts[0]["last_contact"])) {
 441                         Logger::log("Profile ".$profile." was last updated at ".$gcontacts[0]["updated"]." (cached)", Logger::DEBUG);
 442
 443                         GContact::update($contact);
 444                         return $gcontacts[0]["updated"];
 445                 }
 446
 447                 $data = Probe::uri($profile);
 448
 449                 // Is the profile link the alternate OStatus link notation? (http://domain.tld/user/4711)
 450                 // Then check the other link and delete this one
 451                 if (($data["network"] == Protocol::OSTATUS) && self::alternateOStatusUrl($profile)
 452                         && (normalise_link($profile) == normalise_link($data["alias"]))
 453                         && (normalise_link($profile) != normalise_link($data["url"]))
 454                 ) {
 455                         // Delete the old entry
 456                         DBA::delete('gcontact', ['nurl' => normalise_link($profile)]);
 457
 458                         $gcontact = array_merge($gcontacts[0], $data);
 459
 460                         $gcontact["server_url"] = $data["baseurl"];
 461
 462                         try {
 463                                 $gcontact = GContact::sanitize($gcontact);
 464                                 GContact::update($gcontact);
 465
 466                                 self::lastUpdated($data["url"], $force);
 467                         } catch (Exception $e) {
 468                                 Logger::log($e->getMessage(), Logger::DEBUG);
 469                         }
 470
 471                         Logger::log("Profile ".$profile." was deleted", Logger::DEBUG);
 472                         return false;
 473                 }
 474
 475                 if (($data["poll"] == "") || (in_array($data["network"], [Protocol::FEED, Protocol::PHANTOM]))) {
 476                         $fields = ['last_failure' => DateTimeFormat::utcNow()];
 477                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 478
 479                         Logger::log("Profile ".$profile." wasn't reachable (profile)", Logger::DEBUG);
 480                         return false;
 481                 }
 482
 483                 $contact = array_merge($contact, $data);
 484
 485                 $contact["server_url"] = $data["baseurl"];
 486
 487                 GContact::update($contact);
 488
 489                 $curlResult = Network::curl($data["poll"]);
 490
 491                 if (!$curlResult->isSuccess()) {
 492                         $fields = ['last_failure' => DateTimeFormat::utcNow()];
 493                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 494
 495                         Logger::log("Profile ".$profile." wasn't reachable (no feed)", Logger::DEBUG);
 496                         return false;
 497                 }
 498
 499                 $doc = new DOMDocument();
 500                 /// @TODO Avoid error supression here
 501                 @$doc->loadXML($curlResult->getBody());
 502
 503                 $xpath = new DOMXPath($doc);
 504                 $xpath->registerNamespace('atom', "http://www.w3.org/2005/Atom");
 505
 506                 $entries = $xpath->query('/atom:feed/atom:entry');
 507
 508                 $last_updated = "";
 509
 510                 foreach ($entries as $entry) {
 511                         $published = DateTimeFormat::utc($xpath->query('atom:published/text()', $entry)->item(0)->nodeValue);
 512                         $updated   = DateTimeFormat::utc($xpath->query('atom:updated/text()'  , $entry)->item(0)->nodeValue);
 513
 514                         if ($last_updated < $published) {
 515                                 $last_updated = $published;
 516                         }
 517
 518                         if ($last_updated < $updated) {
 519                                 $last_updated = $updated;
 520                         }
 521                 }
 522
 523                 // Maybe there aren't any entries. Then check if it is a valid feed
 524                 if ($last_updated == "") {
 525                         if ($xpath->query('/atom:feed')->length > 0) {
 526                                 $last_updated = DBA::NULL_DATETIME;
 527                         }
 528                 }
 529
 530                 $fields = ['last_contact' => DateTimeFormat::utcNow()];
 531
 532                 if (!empty($last_updated)) {
 533                         $fields['updated'] = $last_updated;
 534                 }
 535
 536                 DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 537
 538                 if (($gcontacts[0]["generation"] == 0)) {
 539                         $fields = ['generation' => 9];
 540                         DBA::update('gcontact', $fields, ['nurl' => normalise_link($profile)]);
 541                 }
 542
 543                 Logger::log("Profile ".$profile." was last updated at ".$last_updated, Logger::DEBUG);
 544
 545                 return $last_updated;
 546         }
 547
 548         public static function updateNeeded($created, $updated, $last_failure, $last_contact)
 549         {
 550                 $now = strtotime(DateTimeFormat::utcNow());
 551
 552                 if ($updated > $last_contact) {
 553                         $contact_time = strtotime($updated);
 554                 } else {
 555                         $contact_time = strtotime($last_contact);
 556                 }
 557
 558                 $failure_time = strtotime($last_failure);
 559                 $created_time = strtotime($created);
 560
 561                 // If there is no "created" time then use the current time
 562                 if ($created_time <= 0) {
 563                         $created_time = $now;
 564                 }
 565
 566                 // If the last contact was less than 24 hours then don't update
 567                 if (($now - $contact_time) < (60 * 60 * 24)) {
 568                         return false;
 569                 }
 570
 571                 // If the last failure was less than 24 hours then don't update
 572                 if (($now - $failure_time) < (60 * 60 * 24)) {
 573                         return false;
 574                 }
 575
 576                 // If the last contact was less than a week ago and the last failure is older than a week then don't update
 577                 //if ((($now - $contact_time) < (60 * 60 * 24 * 7)) && ($contact_time > $failure_time))
 578                 //      return false;
 579
 580                 // If the last contact time was more than a week ago and the contact was created more than a week ago, then only try once a week
 581                 if ((($now - $contact_time) > (60 * 60 * 24 * 7)) && (($now - $created_time) > (60 * 60 * 24 * 7)) && (($now - $failure_time) < (60 * 60 * 24 * 7))) {
 582                         return false;
 583                 }
 584
 585                 // If the last contact time was more than a month ago and the contact was created more than a month ago, then only try once a month
 586                 if ((($now - $contact_time) > (60 * 60 * 24 * 30)) && (($now - $created_time) > (60 * 60 * 24 * 30)) && (($now - $failure_time) < (60 * 60 * 24 * 30))) {
 587                         return false;
 588                 }
 589
 590                 return true;
 591         }
 592
 593         /// @TODO Maybe move this out to an utilities class?
 594         private static function toBoolean($val)
 595         {
 596                 if (($val == "true") || ($val == 1)) {
 597                         return true;
 598                 } elseif (($val == "false") || ($val == 0)) {
 599                         return false;
 600                 }
 601
 602                 return $val;
 603         }
 604
 605         /**
 606          * @brief Detect server type (Hubzilla or Friendica) via the poco data
 607          *
 608          * @param array $data POCO data
 609          * @return array Server data
 610          */
 611         private static function detectPocoData(array $data)
 612         {
 613                 $server = false;
 614
 615                 if (!isset($data['entry'])) {
 616                         return false;
 617                 }
 618
 619                 if (count($data['entry']) == 0) {
 620                         return false;
 621                 }
 622
 623                 if (!isset($data['entry'][0]['urls'])) {
 624                         return false;
 625                 }
 626
 627                 if (count($data['entry'][0]['urls']) == 0) {
 628                         return false;
 629                 }
 630
 631                 foreach ($data['entry'][0]['urls'] as $url) {
 632                         if ($url['type'] == 'zot') {
 633                                 $server = [];
 634                                 $server["platform"] = 'Hubzilla';
 635                                 $server["network"] = Protocol::DIASPORA;
 636                                 return $server;
 637                         }
 638                 }
 639                 return false;
 640         }
 641
 642         /**
 643          * @brief Detect server type by using the nodeinfo data
 644          *
 645          * @param string $server_url address of the server
 646          * @return array Server data
 647          */
 648         private static function fetchNodeinfo($server_url)
 649         {
 650                 $curlResult = Network::curl($server_url."/.well-known/nodeinfo");
 651                 if (!$curlResult->isSuccess()) {
 652                         return false;
 653                 }
 654
 655                 $nodeinfo = json_decode($curlResult->getBody(), true);
 656
 657                 if (!is_array($nodeinfo) || !isset($nodeinfo['links'])) {
 658                         return false;
 659                 }
 660
 661                 $nodeinfo1_url = '';
 662                 $nodeinfo2_url = '';
 663
 664                 foreach ($nodeinfo['links'] as $link) {
 665                         if (!is_array($link) || empty($link['rel'])) {
 666                                 Logger::log('Invalid nodeinfo format for ' . $server_url, Logger::DEBUG);
 667                                 continue;
 668                         }
 669                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 670                                 $nodeinfo1_url = $link['href'];
 671                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 672                                 $nodeinfo2_url = $link['href'];
 673                         }
 674                 }
 675
 676                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 677                         return false;
 678                 }
 679
 680                 $server = [];
 681
 682                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 683                 if (!empty($nodeinfo2_url) && (parse_url($server_url, PHP_URL_HOST) == parse_url($nodeinfo2_url, PHP_URL_HOST))) {
 684                         $server = self::parseNodeinfo2($nodeinfo2_url);
 685                 }
 686
 687                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 688                 if (empty($server) && !empty($nodeinfo1_url) && (parse_url($server_url, PHP_URL_HOST) == parse_url($nodeinfo1_url, PHP_URL_HOST))) {
 689                         $server = self::parseNodeinfo1($nodeinfo1_url);
 690                 }
 691
 692                 return $server;
 693         }
 694
 695         /**
 696          * @brief Parses Nodeinfo 1
 697          *
 698          * @param string $nodeinfo_url address of the nodeinfo path
 699          * @return array Server data
 700          */
 701         private static function parseNodeinfo1($nodeinfo_url)
 702         {
 703                 $curlResult = Network::curl($nodeinfo_url);
 704
 705                 if (!$curlResult->isSuccess()) {
 706                         return false;
 707                 }
 708
 709                 $nodeinfo = json_decode($curlResult->getBody(), true);
 710
 711                 if (!is_array($nodeinfo)) {
 712                         return false;
 713                 }
 714
 715                 $server = [];
 716
 717                 $server['register_policy'] = REGISTER_CLOSED;
 718
 719                 if (is_bool($nodeinfo['openRegistrations']) && $nodeinfo['openRegistrations']) {
 720                         $server['register_policy'] = REGISTER_OPEN;
 721                 }
 722
 723                 if (is_array($nodeinfo['software'])) {
 724                         if (isset($nodeinfo['software']['name'])) {
 725                                 $server['platform'] = $nodeinfo['software']['name'];
 726                         }
 727
 728                         if (isset($nodeinfo['software']['version'])) {
 729                                 $server['version'] = $nodeinfo['software']['version'];
 730                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 731                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 732                                 $server['version'] = preg_replace("=(.+)-(.{4,})=ism", "$1", $server['version']);
 733                         }
 734                 }
 735
 736                 if (is_array($nodeinfo['metadata']) && isset($nodeinfo['metadata']['nodeName'])) {
 737                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 738                 }
 739
 740                 if (!empty($nodeinfo['usage']['users']['total'])) {
 741                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 742                 }
 743
 744                 $diaspora = false;
 745                 $friendica = false;
 746                 $gnusocial = false;
 747
 748                 if (is_array($nodeinfo['protocols']['inbound'])) {
 749                         foreach ($nodeinfo['protocols']['inbound'] as $inbound) {
 750                                 if ($inbound == 'diaspora') {
 751                                         $diaspora = true;
 752                                 }
 753                                 if ($inbound == 'friendica') {
 754                                         $friendica = true;
 755                                 }
 756                                 if ($inbound == 'gnusocial') {
 757                                         $gnusocial = true;
 758                                 }
 759                         }
 760                 }
 761
 762                 if ($gnusocial) {
 763                         $server['network'] = Protocol::OSTATUS;
 764                 }
 765                 if ($diaspora) {
 766                         $server['network'] = Protocol::DIASPORA;
 767                 }
 768                 if ($friendica) {
 769                         $server['network'] = Protocol::DFRN;
 770                 }
 771
 772                 if (!$server) {
 773                         return false;
 774                 }
 775
 776                 return $server;
 777         }
 778
 779         /**
 780          * @brief Parses Nodeinfo 2
 781          *
 782          * @param string $nodeinfo_url address of the nodeinfo path
 783          * @return array Server data
 784          */
 785         private static function parseNodeinfo2($nodeinfo_url)
 786         {
 787                 $curlResult = Network::curl($nodeinfo_url);
 788                 if (!$curlResult->isSuccess()) {
 789                         return false;
 790                 }
 791
 792                 $nodeinfo = json_decode($curlResult->getBody(), true);
 793
 794                 if (!is_array($nodeinfo)) {
 795                         return false;
 796                 }
 797
 798                 $server = [];
 799
 800                 $server['register_policy'] = REGISTER_CLOSED;
 801
 802                 if (is_bool($nodeinfo['openRegistrations']) && $nodeinfo['openRegistrations']) {
 803                         $server['register_policy'] = REGISTER_OPEN;
 804                 }
 805
 806                 if (is_array($nodeinfo['software'])) {
 807                         if (isset($nodeinfo['software']['name'])) {
 808                                 $server['platform'] = $nodeinfo['software']['name'];
 809                         }
 810
 811                         if (isset($nodeinfo['software']['version'])) {
 812                                 $server['version'] = $nodeinfo['software']['version'];
 813                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 814                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 815                                 $server['version'] = preg_replace("=(.+)-(.{4,})=ism", "$1", $server['version']);
 816                         }
 817                 }
 818
 819                 if (is_array($nodeinfo['metadata']) && isset($nodeinfo['metadata']['nodeName'])) {
 820                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 821                 }
 822
 823                 if (!empty($nodeinfo['usage']['users']['total'])) {
 824                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 825                 }
 826
 827                 $diaspora = false;
 828                 $friendica = false;
 829                 $gnusocial = false;
 830
 831                 if (!empty($nodeinfo['protocols'])) {
 832                         foreach ($nodeinfo['protocols'] as $protocol) {
 833                                 if ($protocol == 'diaspora') {
 834                                         $diaspora = true;
 835                                 } elseif ($protocol == 'friendica') {
 836                                         $friendica = true;
 837                                 } elseif ($protocol == 'gnusocial') {
 838                                         $gnusocial = true;
 839                                 }
 840                         }
 841                 }
 842
 843                 if ($gnusocial) {
 844                         $server['network'] = Protocol::OSTATUS;
 845                 } elseif ($diaspora) {
 846                         $server['network'] = Protocol::DIASPORA;
 847                 } elseif ($friendica) {
 848                         $server['network'] = Protocol::DFRN;
 849                 }
 850
 851                 if (empty($server)) {
 852                         return false;
 853                 }
 854
 855                 return $server;
 856         }
 857
 858         /**
 859          * @brief Detect server type (Hubzilla or Friendica) via the front page body
 860          *
 861          * @param string $body Front page of the server
 862          * @return array Server data
 863          */
 864         private static function detectServerType($body)
 865         {
 866                 $server = false;
 867
 868                 $doc = new DOMDocument();
 869                 /// @TODO Acoid supressing error
 870                 @$doc->loadHTML($body);
 871                 $xpath = new DOMXPath($doc);
 872
 873                 $list = $xpath->query("//meta[@name]");
 874
 875                 foreach ($list as $node) {
 876                         $attr = [];
 877                         if ($node->attributes->length) {
 878                                 foreach ($node->attributes as $attribute) {
 879                                         $attr[$attribute->name] = $attribute->value;
 880                                 }
 881                         }
 882                         if ($attr['name'] == 'generator') {
 883                                 $version_part = explode(" ", $attr['content']);
 884                                 if (count($version_part) == 2) {
 885                                         if (in_array($version_part[0], ["Friendika", "Friendica"])) {
 886                                                 $server = [];
 887                                                 $server["platform"] = $version_part[0];
 888                                                 $server["version"] = $version_part[1];
 889                                                 $server["network"] = Protocol::DFRN;
 890                                         }
 891                                 }
 892                         }
 893                 }
 894
 895                 if (!$server) {
 896                         $list = $xpath->query("//meta[@property]");
 897
 898                         foreach ($list as $node) {
 899                                 $attr = [];
 900                                 if ($node->attributes->length) {
 901                                         foreach ($node->attributes as $attribute) {
 902                                                 $attr[$attribute->name] = $attribute->value;
 903                                         }
 904                                 }
 905                                 if ($attr['property'] == 'generator' && in_array($attr['content'], ["hubzilla", "BlaBlaNet"])) {
 906                                         $server = [];
 907                                         $server["platform"] = $attr['content'];
 908                                         $server["version"] = "";
 909                                         $server["network"] = Protocol::DIASPORA;
 910                                 }
 911                         }
 912                 }
 913
 914                 if (!$server) {
 915                         return false;
 916                 }
 917
 918                 $server["site_name"] = XML::getFirstNodeValue($xpath, '//head/title/text()');
 919
 920                 return $server;
 921         }
 922
 923         public static function checkServer($server_url, $network = "", $force = false)
 924         {
 925                 // Unify the server address
 926                 $server_url = trim($server_url, "/");
 927                 $server_url = str_replace("/index.php", "", $server_url);
 928
 929                 if ($server_url == "") {
 930                         return false;
 931                 }
 932
 933                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => normalise_link($server_url)]);
 934                 if (DBA::isResult($gserver)) {
 935                         if ($gserver["created"] <= DBA::NULL_DATETIME) {
 936                                 $fields = ['created' => DateTimeFormat::utcNow()];
 937                                 $condition = ['nurl' => normalise_link($server_url)];
 938                                 DBA::update('gserver', $fields, $condition);
 939                         }
 940                         $poco = $gserver["poco"];
 941                         $noscrape = $gserver["noscrape"];
 942
 943                         if ($network == "") {
 944                                 $network = $gserver["network"];
 945                         }
 946
 947                         $last_contact = $gserver["last_contact"];
 948                         $last_failure = $gserver["last_failure"];
 949                         $version = $gserver["version"];
 950                         $platform = $gserver["platform"];
 951                         $site_name = $gserver["site_name"];
 952                         $info = $gserver["info"];
 953                         $register_policy = $gserver["register_policy"];
 954                         $registered_users = $gserver["registered-users"];
 955
 956                         // See discussion under https://forum.friendi.ca/display/0b6b25a8135aabc37a5a0f5684081633
 957                         // It can happen that a zero date is in the database, but storing it again is forbidden.
 958                         if ($last_contact < DBA::NULL_DATETIME) {
 959                                 $last_contact = DBA::NULL_DATETIME;
 960                         }
 961
 962                         if ($last_failure < DBA::NULL_DATETIME) {
 963                                 $last_failure = DBA::NULL_DATETIME;
 964                         }
 965
 966                         if (!$force && !self::updateNeeded($gserver["created"], "", $last_failure, $last_contact)) {
 967                                 Logger::log("Use cached data for server ".$server_url, Logger::DEBUG);
 968                                 return ($last_contact >= $last_failure);
 969                         }
 970                 } else {
 971                         $poco = "";
 972                         $noscrape = "";
 973                         $version = "";
 974                         $platform = "";
 975                         $site_name = "";
 976                         $info = "";
 977                         $register_policy = -1;
 978                         $registered_users = 0;
 979
 980                         $last_contact = DBA::NULL_DATETIME;
 981                         $last_failure = DBA::NULL_DATETIME;
 982                 }
 983                 Logger::log("Server ".$server_url." is outdated or unknown. Start discovery. Force: ".$force." Created: ".$gserver["created"]." Failure: ".$last_failure." Contact: ".$last_contact, Logger::DEBUG);
 984
 985                 $failure = false;
 986                 $possible_failure = false;
 987                 $orig_last_failure = $last_failure;
 988                 $orig_last_contact = $last_contact;
 989
 990                 // Mastodon uses the "@" for user profiles.
 991                 // But this can be misunderstood.
 992                 if (parse_url($server_url, PHP_URL_USER) != '') {
 993                         DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => normalise_link($server_url)]);
 994                         return false;
 995                 }
 996
 997                 // Check if the page is accessible via SSL.
 998                 $orig_server_url = $server_url;
 999                 $server_url = str_replace("http://", "https://", $server_url);
1000
1001                 // We set the timeout to 20 seconds since this operation should be done in no time if the server was vital
1002                 $curlResult = Network::curl($server_url."/.well-known/host-meta", false, $redirects, ['timeout' => 20]);
1003
1004                 // Quit if there is a timeout.
1005                 // But we want to make sure to only quit if we are mostly sure that this server url fits.
1006                 if (DBA::isResult($gserver) && ($orig_server_url == $server_url) &&
1007                         ($curlResult->isTimeout())) {
1008                         Logger::log("Connection to server ".$server_url." timed out.", Logger::DEBUG);
1009                         DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => normalise_link($server_url)]);
1010                         return false;
1011                 }
1012
1013                 // Maybe the page is unencrypted only?
1014                 $xmlobj = @simplexml_load_string($curlResult->getBody(), 'SimpleXMLElement', 0, "http://docs.oasis-open.org/ns/xri/xrd-1.0");
1015                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == "") || empty($xmlobj) || !is_object($xmlobj)) {
1016                         $server_url = str_replace("https://", "http://", $server_url);
1017
1018                         // We set the timeout to 20 seconds since this operation should be done in no time if the server was vital
1019                         $curlResult = Network::curl($server_url."/.well-known/host-meta", false, $redirects, ['timeout' => 20]);
1020
1021                         // Quit if there is a timeout
1022                         if ($curlResult->isTimeout()) {
1023                                 Logger::log("Connection to server " . $server_url . " timed out.", Logger::DEBUG);
1024                                 DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => normalise_link($server_url)]);
1025                                 return false;
1026                         }
1027
1028                         $xmlobj = @simplexml_load_string($curlResult->getBody(), 'SimpleXMLElement', 0, "http://docs.oasis-open.org/ns/xri/xrd-1.0");
1029                 }
1030
1031                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == "") || empty($xmlobj) || !is_object($xmlobj)) {
1032                         // Workaround for bad configured servers (known nginx problem)
1033                         if (!empty($curlResult->getInfo()) && !in_array($curlResult->getInfo()["http_code"], ["403", "404"])) {
1034                                 $failure = true;
1035                         }
1036
1037                         $possible_failure = true;
1038                 }
1039
1040                 // If the server has no possible failure we reset the cached data
1041                 if (!$possible_failure) {
1042                         $version = "";
1043                         $platform = "";
1044                         $site_name = "";
1045                         $info = "";
1046                         $register_policy = -1;
1047                 }
1048
1049                 if (!$failure) {
1050                         // This will be too low, but better than no value at all.
1051                         $registered_users = DBA::count('gcontact', ['server_url' => normalise_link($server_url)]);
1052                 }
1053
1054                 // Look for poco
1055                 if (!$failure) {
1056                         $curlResult = Network::curl($server_url."/poco");
1057
1058                         if ($curlResult->isSuccess()) {
1059                                 $data = json_decode($curlResult->getBody(), true);
1060
1061                                 if (isset($data['totalResults'])) {
1062                                         $registered_users = $data['totalResults'];
1063                                         $poco = $server_url . "/poco";
1064                                         $server = self::detectPocoData($data);
1065
1066                                         if (!empty($server)) {
1067                                                 $platform = $server['platform'];
1068                                                 $network = $server['network'];
1069                                                 $version = '';
1070                                                 $site_name = '';
1071                                         }
1072                                 }
1073
1074                                 /*
1075                                  * There are servers out there who don't return 404 on a failure
1076                                  * We have to be sure that don't misunderstand this
1077                                  */
1078                                 if (is_null($data)) {
1079                                         $poco = "";
1080                                         $noscrape = "";
1081                                         $network = "";
1082                                 }
1083                         }
1084                 }
1085
1086                 if (!$failure) {
1087                         // Test for Diaspora, Hubzilla, Mastodon or older Friendica servers
1088                         $curlResult = Network::curl($server_url);
1089
1090                         if (!$curlResult->isSuccess() || ($curlResult->getBody() == "")) {
1091                                 $failure = true;
1092                         } else {
1093                                 $server = self::detectServerType($curlResult->getBody());
1094
1095                                 if (!empty($server)) {
1096                                         $platform = $server['platform'];
1097                                         $network = $server['network'];
1098                                         $version = $server['version'];
1099                                         $site_name = $server['site_name'];
1100                                 }
1101
1102                                 $lines = explode("\n", $curlResult->getHeader());
1103
1104                                 if (count($lines)) {
1105                                         foreach ($lines as $line) {
1106                                                 $line = trim($line);
1107
1108                                                 if (stristr($line, 'X-Diaspora-Version:')) {
1109                                                         $platform = "Diaspora";
1110                                                         $version = trim(str_replace("X-Diaspora-Version:", "", $line));
1111                                                         $version = trim(str_replace("x-diaspora-version:", "", $version));
1112                                                         $network = Protocol::DIASPORA;
1113                                                         $versionparts = explode("-", $version);
1114                                                         $version = $versionparts[0];
1115                                                 }
1116
1117                                                 if (stristr($line, 'Server: Mastodon')) {
1118                                                         $platform = "Mastodon";
1119                                                         $network = Protocol::OSTATUS;
1120                                                 }
1121                                         }
1122                                 }
1123                         }
1124                 }
1125
1126                 if (!$failure && ($poco == "")) {
1127                         // Test for Statusnet
1128                         // Will also return data for Friendica and GNU Social - but it will be overwritten later
1129                         // The "not implemented" is a special treatment for really, really old Friendica versions
1130                         $curlResult = Network::curl($server_url."/api/statusnet/version.json");
1131
1132                         if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1133                                 ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1134                                 $platform = "StatusNet";
1135                                 // Remove junk that some GNU Social servers return
1136                                 $version = str_replace(chr(239).chr(187).chr(191), "", $curlResult->getBody());
1137                                 $version = trim($version, '"');
1138                                 $network = Protocol::OSTATUS;
1139                         }
1140
1141                         // Test for GNU Social
1142                         $curlResult = Network::curl($server_url."/api/gnusocial/version.json");
1143
1144                         if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1145                                 ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1146                                 $platform = "GNU Social";
1147                                 // Remove junk that some GNU Social servers return
1148                                 $version = str_replace(chr(239) . chr(187) . chr(191), "", $curlResult->getBody());
1149                                 $version = trim($version, '"');
1150                                 $network = Protocol::OSTATUS;
1151                         }
1152
1153                         // Test for Mastodon
1154                         $orig_version = $version;
1155                         $curlResult = Network::curl($server_url . "/api/v1/instance");
1156
1157                         if ($curlResult->isSuccess() && ($curlResult->getBody() != '')) {
1158                                 $data = json_decode($curlResult->getBody(), true);
1159
1160                                 if (isset($data['version'])) {
1161                                         $platform = "Mastodon";
1162                                         $version = defaults($data, 'version', '');
1163                                         $site_name = defaults($data, 'title', '');
1164                                         $info = defaults($data, 'description', '');
1165                                         $network = Protocol::OSTATUS;
1166                                 }
1167
1168                                 if (!empty($data['stats']['user_count'])) {
1169                                         $registered_users = $data['stats']['user_count'];
1170                                 }
1171                         }
1172
1173                         if (strstr($orig_version . $version, 'Pleroma')) {
1174                                 $platform = 'Pleroma';
1175                                 $version = trim(str_replace('Pleroma', '', $version));
1176                         }
1177                 }
1178
1179                 if (!$failure) {
1180                         // Test for Hubzilla and Red
1181                         $curlResult = Network::curl($server_url . "/siteinfo.json");
1182
1183                         if ($curlResult->isSuccess()) {
1184                                 $data = json_decode($curlResult->getBody(), true);
1185
1186                                 if (isset($data['url'])) {
1187                                         $platform = $data['platform'];
1188                                         $version = $data['version'];
1189                                         $network = Protocol::DIASPORA;
1190                                 }
1191
1192                                 if (!empty($data['site_name'])) {
1193                                         $site_name = $data['site_name'];
1194                                 }
1195
1196                                 if (!empty($data['channels_total'])) {
1197                                         $registered_users = $data['channels_total'];
1198                                 }
1199
1200                                 if (!empty($data['register_policy'])) {
1201                                         switch ($data['register_policy']) {
1202                                                 case "REGISTER_OPEN":
1203                                                         $register_policy = REGISTER_OPEN;
1204                                                         break;
1205
1206                                                 case "REGISTER_APPROVE":
1207                                                         $register_policy = REGISTER_APPROVE;
1208                                                         break;
1209
1210                                                 case "REGISTER_CLOSED":
1211                                                 default:
1212                                                         $register_policy = REGISTER_CLOSED;
1213                                                         break;
1214                                         }
1215                                 }
1216                         } else {
1217                                 // Test for Hubzilla, Redmatrix or Friendica
1218                                 $curlResult = Network::curl($server_url."/api/statusnet/config.json");
1219
1220                                 if ($curlResult->isSuccess()) {
1221                                         $data = json_decode($curlResult->getBody(), true);
1222
1223                                         if (isset($data['site']['server'])) {
1224                                                 if (isset($data['site']['platform'])) {
1225                                                         $platform = $data['site']['platform']['PLATFORM_NAME'];
1226                                                         $version = $data['site']['platform']['STD_VERSION'];
1227                                                         $network = Protocol::DIASPORA;
1228                                                 }
1229
1230                                                 if (isset($data['site']['BlaBlaNet'])) {
1231                                                         $platform = $data['site']['BlaBlaNet']['PLATFORM_NAME'];
1232                                                         $version = $data['site']['BlaBlaNet']['STD_VERSION'];
1233                                                         $network = Protocol::DIASPORA;
1234                                                 }
1235
1236                                                 if (isset($data['site']['hubzilla'])) {
1237                                                         $platform = $data['site']['hubzilla']['PLATFORM_NAME'];
1238                                                         $version = $data['site']['hubzilla']['RED_VERSION'];
1239                                                         $network = Protocol::DIASPORA;
1240                                                 }
1241
1242                                                 if (isset($data['site']['redmatrix'])) {
1243                                                         if (isset($data['site']['redmatrix']['PLATFORM_NAME'])) {
1244                                                                 $platform = $data['site']['redmatrix']['PLATFORM_NAME'];
1245                                                         } elseif (isset($data['site']['redmatrix']['RED_PLATFORM'])) {
1246                                                                 $platform = $data['site']['redmatrix']['RED_PLATFORM'];
1247                                                         }
1248
1249                                                         $version = $data['site']['redmatrix']['RED_VERSION'];
1250                                                         $network = Protocol::DIASPORA;
1251                                                 }
1252
1253                                                 if (isset($data['site']['friendica'])) {
1254                                                         $platform = $data['site']['friendica']['FRIENDICA_PLATFORM'];
1255                                                         $version = $data['site']['friendica']['FRIENDICA_VERSION'];
1256                                                         $network = Protocol::DFRN;
1257                                                 }
1258
1259                                                 $site_name = $data['site']['name'];
1260
1261                                                 $private = false;
1262                                                 $inviteonly = false;
1263                                                 $closed = false;
1264
1265                                                 if (!empty($data['site']['closed'])) {
1266                                                         $closed = self::toBoolean($data['site']['closed']);
1267                                                 }
1268
1269                                                 if (!empty($data['site']['private'])) {
1270                                                         $private = self::toBoolean($data['site']['private']);
1271                                                 }
1272
1273                                                 if (!empty($data['site']['inviteonly'])) {
1274                                                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1275                                                 }
1276
1277                                                 if (!$closed && !$private and $inviteonly) {
1278                                                         $register_policy = REGISTER_APPROVE;
1279                                                 } elseif (!$closed && !$private) {
1280                                                         $register_policy = REGISTER_OPEN;
1281                                                 } else {
1282                                                         $register_policy = REGISTER_CLOSED;
1283                                                 }
1284                                         }
1285                                 }
1286                         }
1287                 }
1288
1289                 // Query statistics.json. Optional package for Diaspora, Friendica and Redmatrix
1290                 if (!$failure) {
1291                         $curlResult = Network::curl($server_url . "/statistics.json");
1292
1293                         if ($curlResult->isSuccess()) {
1294                                 $data = json_decode($curlResult->getBody(), true);
1295
1296                                 if (isset($data['version'])) {
1297                                         $version = $data['version'];
1298                                         // Version numbers on statistics.json are presented with additional info, e.g.:
1299                                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1300                                         $version = preg_replace("=(.+)-(.{4,})=ism", "$1", $version);
1301                                 }
1302
1303                                 if (!empty($data['name'])) {
1304                                         $site_name = $data['name'];
1305                                 }
1306
1307                                 if (!empty($data['network'])) {
1308                                         $platform = $data['network'];
1309                                 }
1310
1311                                 if ($platform == "Diaspora") {
1312                                         $network = Protocol::DIASPORA;
1313                                 }
1314
1315                                 if (!empty($data['registrations_open']) && $data['registrations_open']) {
1316                                         $register_policy = REGISTER_OPEN;
1317                                 } else {
1318                                         $register_policy = REGISTER_CLOSED;
1319                                 }
1320                         }
1321                 }
1322
1323                 // Query nodeinfo. Working for (at least) Diaspora and Friendica.
1324                 if (!$failure) {
1325                         $server = self::fetchNodeinfo($server_url);
1326
1327                         if (!empty($server)) {
1328                                 $register_policy = $server['register_policy'];
1329
1330                                 if (isset($server['platform'])) {
1331                                         $platform = $server['platform'];
1332                                 }
1333
1334                                 if (isset($server['network'])) {
1335                                         $network = $server['network'];
1336                                 }
1337
1338                                 if (isset($server['version'])) {
1339                                         $version = $server['version'];
1340                                 }
1341
1342                                 if (isset($server['site_name'])) {
1343                                         $site_name = $server['site_name'];
1344                                 }
1345
1346                                 if (isset($server['registered-users'])) {
1347                                         $registered_users = $server['registered-users'];
1348                                 }
1349                         }
1350                 }
1351
1352                 // Check for noscrape
1353                 // Friendica servers could be detected as OStatus servers
1354                 if (!$failure && in_array($network, [Protocol::DFRN, Protocol::OSTATUS])) {
1355                         $curlResult = Network::curl($server_url . "/friendica/json");
1356
1357                         if (!$curlResult->isSuccess()) {
1358                                 $curlResult = Network::curl($server_url . "/friendika/json");
1359                         }
1360
1361                         if ($curlResult->isSuccess()) {
1362                                 $data = json_decode($curlResult->getBody(), true);
1363
1364                                 if (isset($data['version'])) {
1365                                         $network = Protocol::DFRN;
1366
1367                                         if (!empty($data['no_scrape_url'])) {
1368                                                 $noscrape = $data['no_scrape_url'];
1369                                         }
1370
1371                                         $version = $data['version'];
1372
1373                                         if (!empty($data['site_name'])) {
1374                                                 $site_name = $data['site_name'];
1375                                         }
1376
1377                                         $info = $data['info'];
1378                                         if (in_array($data['register_policy'], ['REGISTER_CLOSED', 'REGISTER_APPROVE', 'REGISTER_OPEN'])) {
1379                                                 $register_policy = constant($data['register_policy']);
1380                                         } else {
1381                                                 Logger::log("Register policy '".$data['register_policy']."' from $server_url is invalid.");
1382                                                 $register_policy = REGISTER_CLOSED; // set a default value
1383                                         }
1384                                         $platform = $data['platform'];
1385                                 }
1386                         }
1387                 }
1388
1389                 // Every server has got at least an admin account
1390                 if (!$failure && ($registered_users == 0)) {
1391                         $registered_users = 1;
1392                 }
1393
1394                 if ($possible_failure && !$failure) {
1395                         $failure = true;
1396                 }
1397
1398                 if ($failure) {
1399                         $last_contact = $orig_last_contact;
1400                         $last_failure = DateTimeFormat::utcNow();
1401                 } else {
1402                         $last_contact = DateTimeFormat::utcNow();
1403                         $last_failure = $orig_last_failure;
1404                 }
1405
1406                 if (($last_contact <= $last_failure) && !$failure) {
1407                         Logger::log("Server ".$server_url." seems to be alive, but last contact wasn't set - could be a bug", Logger::DEBUG);
1408                 } elseif (($last_contact >= $last_failure) && $failure) {
1409                         Logger::log("Server ".$server_url." seems to be dead, but last failure wasn't set - could be a bug", Logger::DEBUG);
1410                 }
1411
1412                 // Check again if the server exists
1413                 $found = DBA::exists('gserver', ['nurl' => normalise_link($server_url)]);
1414
1415                 $version = strip_tags($version);
1416                 $site_name = strip_tags($site_name);
1417                 $info = strip_tags($info);
1418                 $platform = strip_tags($platform);
1419
1420                 $fields = ['url' => $server_url, 'version' => $version,
1421                                 'site_name' => $site_name, 'info' => $info, 'register_policy' => $register_policy,
1422                                 'poco' => $poco, 'noscrape' => $noscrape, 'network' => $network,
1423                                 'platform' => $platform, 'registered-users' => $registered_users,
1424                                 'last_contact' => $last_contact, 'last_failure' => $last_failure];
1425
1426                 if ($found) {
1427                         DBA::update('gserver', $fields, ['nurl' => normalise_link($server_url)]);
1428                 } elseif (!$failure) {
1429                         $fields['nurl'] = normalise_link($server_url);
1430                         $fields['created'] = DateTimeFormat::utcNow();
1431                         DBA::insert('gserver', $fields);
1432                 }
1433
1434                 if (!$failure && in_array($fields['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
1435                         self::discoverRelay($server_url);
1436                 }
1437
1438                 Logger::log("End discovery for server " . $server_url, Logger::DEBUG);
1439
1440                 return !$failure;
1441         }
1442
1443         /**
1444          * @brief Fetch relay data from a given server url
1445          *
1446          * @param string $server_url address of the server
1447          */
1448         private static function discoverRelay($server_url)
1449         {
1450                 Logger::log("Discover relay data for server " . $server_url, Logger::DEBUG);
1451
1452                 $curlResult = Network::curl($server_url . "/.well-known/x-social-relay");
1453
1454                 if (!$curlResult->isSuccess()) {
1455                         return;
1456                 }
1457
1458                 $data = json_decode($curlResult->getBody(), true);
1459
1460                 if (!is_array($data)) {
1461                         return;
1462                 }
1463
1464                 $gserver = DBA::selectFirst('gserver', ['id', 'relay-subscribe', 'relay-scope'], ['nurl' => normalise_link($server_url)]);
1465
1466                 if (!DBA::isResult($gserver)) {
1467                         return;
1468                 }
1469
1470                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
1471                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
1472                         DBA::update('gserver', $fields, ['id' => $gserver['id']]);
1473                 }
1474
1475                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
1476
1477                 if ($data['scope'] == 'tags') {
1478                         // Avoid duplicates
1479                         $tags = [];
1480                         foreach ($data['tags'] as $tag) {
1481                                 $tag = mb_strtolower($tag);
1482                                 if (strlen($tag) < 100) {
1483                                         $tags[$tag] = $tag;
1484                                 }
1485                         }
1486
1487                         foreach ($tags as $tag) {
1488                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], true);
1489                         }
1490                 }
1491
1492                 // Create or update the relay contact
1493                 $fields = [];
1494                 if (isset($data['protocols'])) {
1495                         if (isset($data['protocols']['diaspora'])) {
1496                                 $fields['network'] = Protocol::DIASPORA;
1497
1498                                 if (isset($data['protocols']['diaspora']['receive'])) {
1499                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
1500                                 } elseif (is_string($data['protocols']['diaspora'])) {
1501                                         $fields['batch'] = $data['protocols']['diaspora'];
1502                                 }
1503                         }
1504
1505                         if (isset($data['protocols']['dfrn'])) {
1506                                 $fields['network'] = Protocol::DFRN;
1507
1508                                 if (isset($data['protocols']['dfrn']['receive'])) {
1509                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
1510                                 } elseif (is_string($data['protocols']['dfrn'])) {
1511                                         $fields['batch'] = $data['protocols']['dfrn'];
1512                                 }
1513                         }
1514                 }
1515                 Diaspora::setRelayContact($server_url, $fields);
1516         }
1517
1518         /**
1519          * @brief Returns a list of all known servers
1520          * @return array List of server urls
1521          */
1522         public static function serverlist()
1523         {
1524                 $r = q(
1525                         "SELECT `url`, `site_name` AS `displayName`, `network`, `platform`, `version` FROM `gserver`
1526                         WHERE `network` IN ('%s', '%s', '%s') AND `last_contact` > `last_failure`
1527                         ORDER BY `last_contact`
1528                         LIMIT 1000",
1529                         DBA::escape(Protocol::DFRN),
1530                         DBA::escape(Protocol::DIASPORA),
1531                         DBA::escape(Protocol::OSTATUS)
1532                 );
1533
1534                 if (!DBA::isResult($r)) {
1535                         return false;
1536                 }
1537
1538                 return $r;
1539         }
1540
1541         /**
1542          * @brief Fetch server list from remote servers and adds them when they are new.
1543          *
1544          * @param string $poco URL to the POCO endpoint
1545          */
1546         private static function fetchServerlist($poco)
1547         {
1548                 $curlResult = Network::curl($poco . "/@server");
1549
1550                 if (!$curlResult->isSuccess()) {
1551                         return;
1552                 }
1553
1554                 $serverlist = json_decode($curlResult->getBody(), true);
1555
1556                 if (!is_array($serverlist)) {
1557                         return;
1558                 }
1559
1560                 foreach ($serverlist as $server) {
1561                         $server_url = str_replace("/index.php", "", $server['url']);
1562
1563                         $r = q("SELECT `nurl` FROM `gserver` WHERE `nurl` = '%s'", DBA::escape(normalise_link($server_url)));
1564
1565                         if (!DBA::isResult($r)) {
1566                                 Logger::log("Call server check for server ".$server_url, Logger::DEBUG);
1567                                 Worker::add(PRIORITY_LOW, "DiscoverPoCo", "server", $server_url);
1568                         }
1569                 }
1570         }
1571
1572         private static function discoverFederation()
1573         {
1574                 $last = Config::get('poco', 'last_federation_discovery');
1575
1576                 if ($last) {
1577                         $next = $last + (24 * 60 * 60);
1578
1579                         if ($next > time()) {
1580                                 return;
1581                         }
1582                 }
1583
1584                 // Discover Friendica, Hubzilla and Diaspora servers
1585                 $curlResult = Network::fetchUrl("http://the-federation.info/pods.json");
1586
1587                 if (!empty($curlResult)) {
1588                         $servers = json_decode($curlResult, true);
1589
1590                         if (!empty($servers['pods'])) {
1591                                 foreach ($servers['pods'] as $server) {
1592                                         Worker::add(PRIORITY_LOW, "DiscoverPoCo", "server", "https://" . $server['host']);
1593                                 }
1594                         }
1595                 }
1596
1597                 // Disvover Mastodon servers
1598                 if (!Config::get('system', 'ostatus_disabled')) {
1599                         $accesstoken = Config::get('system', 'instances_social_key');
1600
1601                         if (!empty($accesstoken)) {
1602                                 $api = 'https://instances.social/api/1.0/instances/list?count=0';
1603                                 $header = ['Authorization: Bearer '.$accesstoken];
1604                                 $curlResult = Network::curl($api, false, $redirects, ['headers' => $header]);
1605
1606                                 if ($curlResult->isSuccess()) {
1607                                         $servers = json_decode($curlResult->getBody(), true);
1608
1609                                         foreach ($servers['instances'] as $server) {
1610                                                 $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
1611                                                 Worker::add(PRIORITY_LOW, "DiscoverPoCo", "server", $url);
1612                                         }
1613                                 }
1614                         }
1615                 }
1616
1617                 // Currently disabled, since the service isn't available anymore.
1618                 // It is not removed since I hope that there will be a successor.
1619                 // Discover GNU Social Servers.
1620                 //if (!Config::get('system','ostatus_disabled')) {
1621                 //      $serverdata = "http://gstools.org/api/get_open_instances/";
1622
1623                 //      $curlResult = Network::curl($serverdata);
1624                 //      if ($curlResult->isSuccess()) {
1625                 //              $servers = json_decode($result->getBody(), true);
1626
1627                 //              foreach($servers['data'] as $server)
1628                 //                      self::checkServer($server['instance_address']);
1629                 //      }
1630                 //}
1631
1632                 Config::set('poco', 'last_federation_discovery', time());
1633         }
1634
1635         public static function discoverSingleServer($id)
1636         {
1637                 $r = q("SELECT `poco`, `nurl`, `url`, `network` FROM `gserver` WHERE `id` = %d", intval($id));
1638
1639                 if (!DBA::isResult($r)) {
1640                         return false;
1641                 }
1642
1643                 $server = $r[0];
1644
1645                 // Discover new servers out there (Works from Friendica version 3.5.2)
1646                 self::fetchServerlist($server["poco"]);
1647
1648                 // Fetch all users from the other server
1649                 $url = $server["poco"] . "/?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation";
1650
1651                 Logger::log("Fetch all users from the server " . $server["url"], Logger::DEBUG);
1652
1653                 $curlResult = Network::curl($url);
1654
1655                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
1656                         $data = json_decode($curlResult->getBody(), true);
1657
1658                         if (!empty($data)) {
1659                                 self::discoverServer($data, 2);
1660                         }
1661
1662                         if (Config::get('system', 'poco_discovery') > 1) {
1663                                 $timeframe = Config::get('system', 'poco_discovery_since');
1664
1665                                 if ($timeframe == 0) {
1666                                         $timeframe = 30;
1667                                 }
1668
1669                                 $updatedSince = date(DateTimeFormat::MYSQL, time() - $timeframe * 86400);
1670
1671                                 // Fetch all global contacts from the other server (Not working with Redmatrix and Friendica versions before 3.3)
1672                                 $url = $server["poco"]."/@global?updatedSince=".$updatedSince."&fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation";
1673
1674                                 $success = false;
1675
1676                                 $curlResult = Network::curl($url);
1677
1678                                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
1679                                         Logger::log("Fetch all global contacts from the server " . $server["nurl"], Logger::DEBUG);
1680                                         $data = json_decode($curlResult->getBody(), true);
1681
1682                                         if (!empty($data)) {
1683                                                 $success = self::discoverServer($data);
1684                                         }
1685                                 }
1686
1687                                 if (!$success && (Config::get('system', 'poco_discovery') > 2)) {
1688                                         Logger::log("Fetch contacts from users of the server " . $server["nurl"], Logger::DEBUG);
1689                                         self::discoverServerUsers($data, $server);
1690                                 }
1691                         }
1692
1693                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1694                         DBA::update('gserver', $fields, ['nurl' => $server["nurl"]]);
1695
1696                         return true;
1697                 } else {
1698                         // If the server hadn't replied correctly, then force a sanity check
1699                         self::checkServer($server["url"], $server["network"], true);
1700
1701                         // If we couldn't reach the server, we will try it some time later
1702                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1703                         DBA::update('gserver', $fields, ['nurl' => $server["nurl"]]);
1704
1705                         return false;
1706                 }
1707         }
1708
1709         public static function discover($complete = false)
1710         {
1711                 // Update the server list
1712                 self::discoverFederation();
1713
1714                 $no_of_queries = 5;
1715
1716                 $requery_days = intval(Config::get('system', 'poco_requery_days'));
1717
1718                 if ($requery_days == 0) {
1719                         $requery_days = 7;
1720                 }
1721
1722                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
1723
1724                 $gservers = q("SELECT `id`, `url`, `nurl`, `network`
1725                         FROM `gserver`
1726                         WHERE `last_contact` >= `last_failure`
1727                         AND `poco` != ''
1728                         AND `last_poco_query` < '%s'
1729                         ORDER BY RAND()", DBA::escape($last_update)
1730                 );
1731
1732                 if (DBA::isResult($gservers)) {
1733                         foreach ($gservers as $gserver) {
1734                                 if (!self::checkServer($gserver['url'], $gserver['network'])) {
1735                                         // The server is not reachable? Okay, then we will try it later
1736                                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1737                                         DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1738                                         continue;
1739                                 }
1740
1741                                 Logger::log('Update directory from server ' . $gserver['url'] . ' with ID ' . $gserver['id'], Logger::DEBUG);
1742                                 Worker::add(PRIORITY_LOW, 'DiscoverPoCo', 'update_server_directory', (int) $gserver['id']);
1743
1744                                 if (!$complete && ( --$no_of_queries == 0)) {
1745                                         break;
1746                                 }
1747                         }
1748                 }
1749         }
1750
1751         private static function discoverServerUsers(array $data, array $server)
1752         {
1753                 if (!isset($data['entry'])) {
1754                         return;
1755                 }
1756
1757                 foreach ($data['entry'] as $entry) {
1758                         $username = '';
1759
1760                         if (isset($entry['urls'])) {
1761                                 foreach ($entry['urls'] as $url) {
1762                                         if ($url['type'] == 'profile') {
1763                                                 $profile_url = $url['value'];
1764                                                 $path_array = explode('/', parse_url($profile_url, PHP_URL_PATH));
1765                                                 $username = end($path_array);
1766                                         }
1767                                 }
1768                         }
1769
1770                         if ($username != '') {
1771                                 Logger::log('Fetch contacts for the user ' . $username . ' from the server ' . $server['nurl'], Logger::DEBUG);
1772
1773                                 // Fetch all contacts from a given user from the other server
1774                                 $url = $server['poco'] . '/' . $username . '/?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation';
1775
1776                                 $curlResult = Network::curl($url);
1777
1778                                 if ($curlResult->isSuccess()) {
1779                                         $data = json_decode($curlResult["body"], true);
1780
1781                                         if (!empty($data)) {
1782                                                 self::discoverServer($data, 3);
1783                                         }
1784                                 }
1785                         }
1786                 }
1787         }
1788
1789         private static function discoverServer(array $data, $default_generation = 0)
1790         {
1791                 if (empty($data['entry'])) {
1792                         return false;
1793                 }
1794
1795                 $success = false;
1796
1797                 foreach ($data['entry'] as $entry) {
1798                         $profile_url = '';
1799                         $profile_photo = '';
1800                         $connect_url = '';
1801                         $name = '';
1802                         $network = '';
1803                         $updated = DBA::NULL_DATETIME;
1804                         $location = '';
1805                         $about = '';
1806                         $keywords = '';
1807                         $gender = '';
1808                         $contact_type = -1;
1809                         $generation = $default_generation;
1810
1811                         if (!empty($entry['displayName'])) {
1812                                 $name = $entry['displayName'];
1813                         }
1814
1815                         if (isset($entry['urls'])) {
1816                                 foreach ($entry['urls'] as $url) {
1817                                         if ($url['type'] == 'profile') {
1818                                                 $profile_url = $url['value'];
1819                                                 continue;
1820                                         }
1821                                         if ($url['type'] == 'webfinger') {
1822                                                 $connect_url = str_replace('acct:' , '', $url['value']);
1823                                                 continue;
1824                                         }
1825                                 }
1826                         }
1827
1828                         if (isset($entry['photos'])) {
1829                                 foreach ($entry['photos'] as $photo) {
1830                                         if ($photo['type'] == 'profile') {
1831                                                 $profile_photo = $photo['value'];
1832                                                 continue;
1833                                         }
1834                                 }
1835                         }
1836
1837                         if (isset($entry['updated'])) {
1838                                 $updated = date(DateTimeFormat::MYSQL, strtotime($entry['updated']));
1839                         }
1840
1841                         if (isset($entry['network'])) {
1842                                 $network = $entry['network'];
1843                         }
1844
1845                         if (isset($entry['currentLocation'])) {
1846                                 $location = $entry['currentLocation'];
1847                         }
1848
1849                         if (isset($entry['aboutMe'])) {
1850                                 $about = HTML::toBBCode($entry['aboutMe']);
1851                         }
1852
1853                         if (isset($entry['gender'])) {
1854                                 $gender = $entry['gender'];
1855                         }
1856
1857                         if (isset($entry['generation']) && ($entry['generation'] > 0)) {
1858                                 $generation = ++$entry['generation'];
1859                         }
1860
1861                         if (isset($entry['contactType']) && ($entry['contactType'] >= 0)) {
1862                                 $contact_type = $entry['contactType'];
1863                         }
1864
1865                         if (isset($entry['tags'])) {
1866                                 foreach ($entry['tags'] as $tag) {
1867                                         $keywords = implode(", ", $tag);
1868                                 }
1869                         }
1870
1871                         if ($generation > 0) {
1872                                 $success = true;
1873
1874                                 Logger::log("Store profile ".$profile_url, Logger::DEBUG);
1875
1876                                 $gcontact = ["url" => $profile_url,
1877                                                 "name" => $name,
1878                                                 "network" => $network,
1879                                                 "photo" => $profile_photo,
1880                                                 "about" => $about,
1881                                                 "location" => $location,
1882                                                 "gender" => $gender,
1883                                                 "keywords" => $keywords,
1884                                                 "connect" => $connect_url,
1885                                                 "updated" => $updated,
1886                                                 "contact-type" => $contact_type,
1887                                                 "generation" => $generation];
1888
1889                                 try {
1890                                         $gcontact = GContact::sanitize($gcontact);
1891                                         GContact::update($gcontact);
1892                                 } catch (Exception $e) {
1893                                         Logger::log($e->getMessage(), Logger::DEBUG);
1894                                 }
1895
1896                                 Logger::log("Done for profile ".$profile_url, Logger::DEBUG);
1897                         }
1898                 }
1899                 return $success;
1900         }
1901
1902 }