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