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