src/Worker/DiscoverPoCo.php

   1 <?php
   2 /**
   3  * @file src/Worker/DiscoverPoCo.php
   4  */
   5 namespace Friendica\Worker;
   6
   7 use Friendica\Core\Cache;
   8 use Friendica\Core\Config;
   9 use Friendica\Core\Logger;
  10 use Friendica\Core\Protocol;
  11 use Friendica\Core\Worker;
  12 use Friendica\Database\DBA;
  13 use Friendica\Model\GContact;
  14 use Friendica\Model\Contact;
  15 use Friendica\Model\GServer;
  16 use Friendica\Network\Probe;
  17 use Friendica\Protocol\PortableContact;
  18 use Friendica\Util\DateTimeFormat;
  19 use Friendica\Util\Network;
  20 use Friendica\Util\Strings;
  21
  22 class DiscoverPoCo
  23 {
  24         /// @todo Clean up this mess of a parameter hell and split it in several classes
  25         public static function execute($command = '', $param1 = '', $param2 = '', $param3 = '', $param4 = '')
  26         {
  27                 /*
  28                 This function can be called in these ways:
  29                 - dirsearch <search pattern>: Searches for "search pattern" in the directory. "search pattern" is url encoded.
  30                 - checkcontact: Updates gcontact entries
  31                 - suggestions: Discover other servers for their contacts.
  32                 - server <poco url>: Searches for the poco server list. "poco url" is base64 encoded.
  33                 - update_server: Frequently check the first 250 servers for vitality.
  34                 - update_server_directory: Discover the given server id for their contacts
  35                 - PortableContact::load: Load POCO data from a given POCO address
  36                 - check_profile: Update remote profile data
  37                 */
  38
  39                 $search = "";
  40                 $mode = 0;
  41                 if ($command == "dirsearch") {
  42                         $search = urldecode($param1);
  43                         $mode = 1;
  44                 } elseif ($command == "checkcontact") {
  45                         $mode = 2;
  46                 } elseif ($command == "suggestions") {
  47                         $mode = 3;
  48                 } elseif ($command == "server") {
  49                         $mode = 4;
  50                 } elseif ($command == "update_server") {
  51                         $mode = 5;
  52                 } elseif ($command == "update_server_directory") {
  53                         $mode = 6;
  54                 } elseif ($command == "load") {
  55                         $mode = 7;
  56                 } elseif ($command == "check_profile") {
  57                         $mode = 8;
  58                 } elseif ($command !== "") {
  59                         Logger::log("Unknown or missing parameter ".$command."\n");
  60                         return;
  61                 }
  62
  63                 Logger::log('start '.$search);
  64
  65                 if ($mode == 8) {
  66                         if ($param1 != "") {
  67                                 GContact::updateFromProbe($param1, true);
  68                         }
  69                 } elseif ($mode == 7) {
  70                         if (!empty($param4)) {
  71                                 $url = $param4;
  72                         } else {
  73                                 $url = '';
  74                         }
  75                         PortableContact::load(intval($param1), intval($param2), intval($param3), $url);
  76                 } elseif ($mode == 6) {
  77                         PortableContact::discoverSingleServer(intval($param1));
  78                 } elseif ($mode == 5) {
  79                         self::updateServer();
  80                 } elseif ($mode == 4) {
  81                         $server_url = $param1;
  82                         if ($server_url == "") {
  83                                 return;
  84                         }
  85                         $server_url = filter_var($server_url, FILTER_SANITIZE_URL);
  86                         if (substr(Strings::normaliseLink($server_url), 0, 7) != "http://") {
  87                                 return;
  88                         }
  89                         $result = "Checking server ".$server_url." - ";
  90                         $ret = GServer::check($server_url);
  91                         if ($ret) {
  92                                 $result .= "success";
  93                         } else {
  94                                 $result .= "failed";
  95                         }
  96                         Logger::log($result, Logger::DEBUG);
  97                 } elseif ($mode == 3) {
  98                         GContact::updateSuggestions();
  99                 } elseif (($mode == 2) && Config::get('system', 'poco_completion')) {
 100                         self::discoverUsers();
 101                 } elseif (($mode == 1) && ($search != "") && Config::get('system', 'poco_local_search')) {
 102                         self::discoverDirectory($search);
 103                         self::gsSearchUser($search);
 104                 } elseif (($mode == 0) && ($search == "") && (Config::get('system', 'poco_discovery') != PortableContact::DISABLED)) {
 105                         // Query Friendica and Hubzilla servers for their users
 106                         PortableContact::discover();
 107
 108                         // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
 109                         if (!Config::get('system', 'ostatus_disabled')) {
 110                                 GContact::discoverGsUsers();
 111                         }
 112                 }
 113
 114                 Logger::log('end '.$search);
 115
 116                 return;
 117         }
 118
 119         /**
 120          * @brief Updates the first 250 servers
 121          *
 122          */
 123         private static function updateServer() {
 124                 $r = q("SELECT `url`, `created`, `last_failure`, `last_contact` FROM `gserver` ORDER BY rand()");
 125
 126                 if (!DBA::isResult($r)) {
 127                         return;
 128                 }
 129
 130                 $updated = 0;
 131
 132                 foreach ($r AS $server) {
 133                         if (!PortableContact::updateNeeded($server["created"], "", $server["last_failure"], $server["last_contact"])) {
 134                                 continue;
 135                         }
 136                         Logger::log('Update server status for server '.$server["url"], Logger::DEBUG);
 137
 138                         Worker::add(PRIORITY_LOW, "DiscoverPoCo", "server", $server["url"]);
 139
 140                         if (++$updated > 250) {
 141                                 return;
 142                         }
 143                 }
 144         }
 145
 146         private static function discoverUsers() {
 147                 Logger::log("Discover users", Logger::DEBUG);
 148
 149                 $starttime = time();
 150
 151                 $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url`, `network` FROM `gcontact`
 152                                 WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 153                                         `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 154                                         `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 155                                 DBA::escape(Protocol::DFRN), DBA::escape(Protocol::DIASPORA),
 156                                 DBA::escape(Protocol::OSTATUS), DBA::escape(Protocol::FEED));
 157
 158                 if (!$users) {
 159                         return;
 160                 }
 161                 $checked = 0;
 162
 163                 foreach ($users AS $user) {
 164
 165                         $urlparts = parse_url($user["url"]);
 166                         if (!isset($urlparts["scheme"])) {
 167                                 DBA::update('gcontact', ['network' => Protocol::PHANTOM],
 168                                         ['nurl' => Strings::normaliseLink($user["url"])]);
 169                                 continue;
 170                          }
 171
 172                         if (in_array($urlparts["host"], ["twitter.com", "identi.ca"])) {
 173                                 $networks = ["twitter.com" => Protocol::TWITTER, "identi.ca" => Protocol::PUMPIO];
 174
 175                                 DBA::update('gcontact', ['network' => $networks[$urlparts["host"]]],
 176                                         ['nurl' => Strings::normaliseLink($user["url"])]);
 177                                 continue;
 178                         }
 179
 180                         $server_url = Contact::getBasepath($user["url"]);
 181                         $force_update = false;
 182
 183                         if ($user["server_url"] != "") {
 184
 185                                 $force_update = (Strings::normaliseLink($user["server_url"]) != Strings::normaliseLink($server_url));
 186
 187                                 $server_url = $user["server_url"];
 188                         }
 189
 190                         if ((($server_url == "") && ($user["network"] == Protocol::FEED)) || $force_update || GServer::check($server_url, $user["network"])) {
 191                                 Logger::log('Check profile '.$user["url"]);
 192                                 Worker::add(PRIORITY_LOW, "DiscoverPoCo", "check_profile", $user["url"]);
 193
 194                                 if (++$checked > 100) {
 195                                         return;
 196                                 }
 197                         } else {
 198                                 DBA::update('gcontact', ['last_failure' => DateTimeFormat::utcNow()],
 199                                         ['nurl' => Strings::normaliseLink($user["url"])]);
 200                         }
 201
 202                         // Quit the loop after 3 minutes
 203                         if (time() > ($starttime + 180)) {
 204                                 return;
 205                         }
 206                 }
 207         }
 208
 209         private static function discoverDirectory($search) {
 210
 211                 $data = Cache::get("dirsearch:".$search);
 212                 if (!is_null($data)) {
 213                         // Only search for the same item every 24 hours
 214                         if (time() < $data + (60 * 60 * 24)) {
 215                                 Logger::log("Already searched for ".$search." in the last 24 hours", Logger::DEBUG);
 216                                 return;
 217                         }
 218                 }
 219
 220                 $x = Network::fetchUrl(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 221                 $j = json_decode($x);
 222
 223                 if (!empty($j->results)) {
 224                         foreach ($j->results as $jj) {
 225                                 // Check if the contact already exists
 226                                 $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", Strings::normaliseLink($jj->url));
 227                                 if (DBA::isResult($exists)) {
 228                                         Logger::log("Profile ".$jj->url." already exists (".$search.")", Logger::DEBUG);
 229
 230                                         if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) &&
 231                                                 ($exists[0]["updated"] < $exists[0]["last_failure"])) {
 232                                                 continue;
 233                                         }
 234                                         // Update the contact
 235                                         GContact::updateFromProbe($jj->url);
 236                                         continue;
 237                                 }
 238
 239                                 $server_url = Contact::getBasepath($jj->url);
 240                                 if ($server_url != '') {
 241                                         if (!GServer::check($server_url)) {
 242                                                 Logger::log("Friendica server ".$server_url." doesn't answer.", Logger::DEBUG);
 243                                                 continue;
 244                                         }
 245                                         Logger::log("Friendica server ".$server_url." seems to be okay.", Logger::DEBUG);
 246                                 }
 247
 248                                 $data = Probe::uri($jj->url);
 249                                 if ($data["network"] == Protocol::DFRN) {
 250                                         Logger::log("Profile ".$jj->url." is reachable (".$search.")", Logger::DEBUG);
 251                                         Logger::log("Add profile ".$jj->url." to local directory (".$search.")", Logger::DEBUG);
 252
 253                                         if ($jj->tags != "") {
 254                                                 $data["keywords"] = $jj->tags;
 255                                         }
 256
 257                                         $data["server_url"] = $data["baseurl"];
 258
 259                                         GContact::update($data);
 260                                 } else {
 261                                         Logger::log("Profile ".$jj->url." is not responding or no Friendica contact - but network ".$data["network"], Logger::DEBUG);
 262                                 }
 263                         }
 264                 }
 265                 Cache::set("dirsearch:".$search, time(), Cache::DAY);
 266         }
 267
 268         /**
 269          * @brief Search for GNU Social user with gstools.org
 270          *
 271          * @param string $search User name
 272          * @return bool
 273          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 274          * @throws \ImagickException
 275          */
 276         private static function gsSearchUser($search) {
 277
 278                 // Currently disabled, since the service isn't available anymore.
 279                 // It is not removed since I hope that there will be a successor.
 280                 return false;
 281
 282                 $url = "http://gstools.org/api/users_search/".urlencode($search);
 283
 284                 $curlResult = Network::curl($url);
 285                 if (!$curlResult->isSuccess()) {
 286                         return false;
 287                 }
 288
 289                 $contacts = json_decode($curlResult->getBody());
 290
 291                 if ($contacts->status == 'ERROR') {
 292                         return false;
 293                 }
 294
 295                 /// @TODO AS is considered as a notation for constants (as they usually being written all upper-case)
 296                 /// @TODO find all those and convert to all lower-case which is a keyword then
 297                 foreach ($contacts->data AS $user) {
 298                         $contact = Probe::uri($user->site_address."/".$user->name);
 299                         if ($contact["network"] != Protocol::PHANTOM) {
 300                                 $contact["about"] = $user->description;
 301                                 GContact::update($contact);
 302                         }
 303                 }
 304         }
 305 }