include/discover_poco.php

   1 <?php
   2
   3 use Friendica\Core\Config;
   4 use Friendica\Core\Worker;
   5 use Friendica\Database\DBM;
   6 use Friendica\Network\Probe;
   7
   8 require_once 'include/socgraph.php';
   9 require_once 'include/datetime.php';
  10
  11 function discover_poco_run(&$argv, &$argc) {
  12
  13         /*
  14         This function can be called in these ways:
  15         - dirsearch <search pattern>: Searches for "search pattern" in the directory. "search pattern" is url encoded.
  16         - checkcontact: Updates gcontact entries
  17         - suggestions: Discover other servers for their contacts.
  18         - server <poco url>: Searches for the poco server list. "poco url" is base64 encoded.
  19         - update_server: Frequently check the first 250 servers for vitality.
  20         - update_server_directory: Discover the given server id for their contacts
  21         - poco_load: Load POCO data from a given POCO address
  22         - check_profile: Update remote profile data
  23         */
  24
  25         if (($argc > 2) && ($argv[1] == "dirsearch")) {
  26                 $search = urldecode($argv[2]);
  27                 $mode = 1;
  28         } elseif (($argc == 2) && ($argv[1] == "checkcontact")) {
  29                 $mode = 2;
  30         } elseif (($argc == 2) && ($argv[1] == "suggestions")) {
  31                 $mode = 3;
  32         } elseif (($argc == 3) && ($argv[1] == "server")) {
  33                 $mode = 4;
  34         } elseif (($argc == 2) && ($argv[1] == "update_server")) {
  35                 $mode = 5;
  36         } elseif (($argc == 3) && ($argv[1] == "update_server_directory")) {
  37                 $mode = 6;
  38         } elseif (($argc > 5) && ($argv[1] == "poco_load")) {
  39                 $mode = 7;
  40         } elseif (($argc == 3) && ($argv[1] == "check_profile")) {
  41                 $mode = 8;
  42         } elseif ($argc == 1) {
  43                 $search = "";
  44                 $mode = 0;
  45         } else {
  46                 die("Unknown or missing parameter ".$argv[1]."\n");
  47         }
  48
  49         logger('start '.$search);
  50
  51         if ($mode == 8) {
  52                 if ($argv[2] != "") {
  53                         poco_last_updated($argv[2], true);
  54                 }
  55         } elseif ($mode == 7) {
  56                 if ($argc == 6) {
  57                         $url = $argv[5];
  58                 } else {
  59                         $url = '';
  60                 }
  61                 poco_load_worker(intval($argv[2]), intval($argv[3]), intval($argv[4]), $url);
  62         } elseif ($mode == 6) {
  63                 poco_discover_single_server(intval($argv[2]));
  64         } elseif ($mode == 5) {
  65                 update_server();
  66         } elseif ($mode == 4) {
  67                 $server_url = $argv[2];
  68                 if ($server_url == "") {
  69                         return;
  70                 }
  71                 $server_url = filter_var($server_url, FILTER_SANITIZE_URL);
  72                 if (substr(normalise_link($server_url), 0, 7) != "http://") {
  73                         return;
  74                 }
  75                 $result = "Checking server ".$server_url." - ";
  76                 $ret = poco_check_server($server_url);
  77                 if ($ret) {
  78                         $result .= "success";
  79                 } else {
  80                         $result .= "failed";
  81                 }
  82                 logger($result, LOGGER_DEBUG);
  83         } elseif ($mode == 3) {
  84                 update_suggestions();
  85         } elseif (($mode == 2) && Config::get('system','poco_completion')) {
  86                 discover_users();
  87         } elseif (($mode == 1) && ($search != "") && Config::get('system','poco_local_search')) {
  88                 discover_directory($search);
  89                 gs_search_user($search);
  90         } elseif (($mode == 0) && ($search == "") && (Config::get('system','poco_discovery') > 0)) {
  91                 // Query Friendica and Hubzilla servers for their users
  92                 poco_discover();
  93
  94                 // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
  95                 if (!Config::get('system','ostatus_disabled'))
  96                         gs_discover();
  97         }
  98
  99         logger('end '.$search);
 100
 101         return;
 102 }
 103
 104 /**
 105  * @brief Updates the first 250 servers
 106  *
 107  */
 108 function update_server() {
 109         $r = q("SELECT `url`, `created`, `last_failure`, `last_contact` FROM `gserver` ORDER BY rand()");
 110
 111         if (!DBM::is_result($r)) {
 112                 return;
 113         }
 114
 115         $updated = 0;
 116
 117         foreach ($r AS $server) {
 118                 if (!poco_do_update($server["created"], "", $server["last_failure"], $server["last_contact"])) {
 119                         continue;
 120                 }
 121                 logger('Update server status for server '.$server["url"], LOGGER_DEBUG);
 122
 123                 Worker::add(PRIORITY_LOW, "discover_poco", "server", $server["url"]);
 124
 125                 if (++$updated > 250) {
 126                         return;
 127                 }
 128         }
 129 }
 130
 131 function discover_users() {
 132         logger("Discover users", LOGGER_DEBUG);
 133
 134         $starttime = time();
 135
 136         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url`, `network` FROM `gcontact`
 137                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 138                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 139                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 140                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
 141                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
 142
 143         if (!$users) {
 144                 return;
 145         }
 146         $checked = 0;
 147
 148         foreach ($users AS $user) {
 149
 150                 $urlparts = parse_url($user["url"]);
 151                 if (!isset($urlparts["scheme"])) {
 152                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 153                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 154                         continue;
 155                  }
 156
 157                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 158                                                         "identi.ca", "alpha.app.net"))) {
 159                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 160                                         "facebook.com" => NETWORK_FACEBOOK,
 161                                         "twitter.com" => NETWORK_TWITTER,
 162                                         "identi.ca" => NETWORK_PUMPIO,
 163                                         "alpha.app.net" => NETWORK_APPNET);
 164
 165                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 166                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 167                         continue;
 168                 }
 169
 170                 $server_url = poco_detect_server($user["url"]);
 171                 $force_update = false;
 172
 173                 if ($user["server_url"] != "") {
 174
 175                         $force_update = (normalise_link($user["server_url"]) != normalise_link($server_url));
 176
 177                         $server_url = $user["server_url"];
 178                 }
 179
 180                 if ((($server_url == "") && ($user["network"] == NETWORK_FEED)) || $force_update || poco_check_server($server_url, $user["network"])) {
 181                         logger('Check profile '.$user["url"]);
 182                         Worker::add(PRIORITY_LOW, "discover_poco", "check_profile", $user["url"]);
 183
 184                         if (++$checked > 100) {
 185                                 return;
 186                         }
 187                 } else {
 188                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 189                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 190                 }
 191
 192                 // Quit the loop after 3 minutes
 193                 if (time() > ($starttime + 180)) {
 194                         return;
 195                 }
 196         }
 197 }
 198
 199 function discover_directory($search) {
 200
 201         $data = Cache::get("dirsearch:".$search);
 202         if (!is_null($data)) {
 203                 // Only search for the same item every 24 hours
 204                 if (time() < $data + (60 * 60 * 24)) {
 205                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 206                         return;
 207                 }
 208         }
 209
 210         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 211         $j = json_decode($x);
 212
 213         if (count($j->results)) {
 214                 foreach ($j->results as $jj) {
 215                         // Check if the contact already exists
 216                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 217                         if (DBM::is_result($exists)) {
 218                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 219
 220                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) &&
 221                                         ($exists[0]["updated"] < $exists[0]["last_failure"])) {
 222                                         continue;
 223                                 }
 224                                 // Update the contact
 225                                 poco_last_updated($jj->url);
 226                                 continue;
 227                         }
 228
 229                         $server_url = poco_detect_server($jj->url);
 230                         if ($server_url != '') {
 231                                 if (!poco_check_server($server_url)) {
 232                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 233                                         continue;
 234                                 }
 235                                 logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 236                         }
 237
 238                         $data = Probe::uri($jj->url);
 239                         if ($data["network"] == NETWORK_DFRN) {
 240                                 logger("Profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 241                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 242
 243                                 if ($jj->tags != "") {
 244                                         $data["keywords"] = $jj->tags;
 245                                 }
 246
 247                                 $data["server_url"] = $data["baseurl"];
 248
 249                                 update_gcontact($data);
 250                         } else {
 251                                 logger("Profile ".$jj->url." is not responding or no Friendica contact - but network ".$data["network"], LOGGER_DEBUG);
 252                         }
 253                 }
 254         }
 255         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 256 }
 257
 258 /**
 259  * @brief Search for GNU Social user with gstools.org
 260  *
 261  * @param str $search User name
 262  */
 263 function gs_search_user($search) {
 264
 265         // Currently disabled, since the service isn't available anymore.
 266         // It is not removed since I hope that there will be a successor.
 267         return false;
 268
 269         $a = get_app();
 270
 271         $url = "http://gstools.org/api/users_search/".urlencode($search);
 272
 273         $result = z_fetch_url($url);
 274         if (!$result["success"]) {
 275                 return false;
 276         }
 277
 278         $contacts = json_decode($result["body"]);
 279
 280         if ($contacts->status == 'ERROR') {
 281                 return false;
 282         }
 283
 284         /// @TODO AS is considered as a notation for constants (as they usually being written all upper-case)
 285         /// @TODO find all those and convert to all lower-case which is a keyword then
 286         foreach ($contacts->data AS $user) {
 287                 $contact = Probe::uri($user->site_address."/".$user->name);
 288                 if ($contact["network"] != NETWORK_PHANTOM) {
 289                         $contact["about"] = $user->description;
 290                         update_gcontact($contact);
 291                 }
 292         }
 293 }