include/discover_poco.php

   1 <?php
   2
   3 use \Friendica\Core\Config;
   4
   5 require_once('include/socgraph.php');
   6 require_once('include/datetime.php');
   7
   8 function discover_poco_run(&$argv, &$argc) {
   9
  10         /*
  11         This function can be called in these ways:
  12         - dirsearch <search pattern>: Searches for "search pattern" in the directory. "search pattern" is url encoded.
  13         - checkcontact: Updates gcontact entries
  14         - suggestions: Discover other servers for their contacts.
  15         - server <poco url>: Searches for the poco server list. "poco url" is base64 encoded.
  16         - update_server: Frequently check the first 250 servers for vitality.
  17         - update_server_directory: Discover the given server id for their contacts
  18         */
  19
  20         if (($argc > 2) && ($argv[1] == "dirsearch")) {
  21                 $search = urldecode($argv[2]);
  22                 $mode = 1;
  23         } elseif (($argc == 2) && ($argv[1] == "checkcontact")) {
  24                 $mode = 2;
  25         } elseif (($argc == 2) && ($argv[1] == "suggestions")) {
  26                 $mode = 3;
  27         } elseif (($argc == 3) && ($argv[1] == "server")) {
  28                 $mode = 4;
  29         } elseif (($argc == 2) && ($argv[1] == "update_server")) {
  30                 $mode = 5;
  31         } elseif (($argc == 3) && ($argv[1] == "update_server_directory")) {
  32                 $mode = 6;
  33         } elseif ($argc == 1) {
  34                 $search = "";
  35                 $mode = 0;
  36         } else {
  37                 die("Unknown or missing parameter ".$argv[1]."\n");
  38         }
  39
  40         logger('start '.$search);
  41
  42         if ($mode == 6) {
  43                 poco_discover_single_server(intval($argv[2]));
  44         } elseif ($mode == 5) {
  45                 update_server();
  46         } elseif ($mode == 4) {
  47                 $server_url = base64_decode($argv[2]);
  48                 if ($server_url == "") {
  49                         return;
  50                 }
  51                 $server_url = filter_var($server_url, FILTER_SANITIZE_URL);
  52                 if (substr(normalise_link($server_url), 0, 7) != "http://") {
  53                         return;
  54                 }
  55                 $result = "Checking server ".$server_url." - ";
  56                 $ret = poco_check_server($server_url);
  57                 if ($ret) {
  58                         $result .= "success";
  59                 } else {
  60                         $result .= "failed";
  61                 }
  62                 logger($result, LOGGER_DEBUG);
  63         } elseif ($mode == 3) {
  64                 update_suggestions();
  65         } elseif (($mode == 2) AND get_config('system','poco_completion')) {
  66                 discover_users();
  67         } elseif (($mode == 1) AND ($search != "") and get_config('system','poco_local_search')) {
  68                 discover_directory($search);
  69                 gs_search_user($search);
  70         } elseif (($mode == 0) AND ($search == "") and (get_config('system','poco_discovery') > 0)) {
  71                 // Query Friendica and Hubzilla servers for their users
  72                 poco_discover();
  73
  74                 // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
  75                 if (!get_config('system','ostatus_disabled'))
  76                         gs_discover();
  77         }
  78
  79         logger('end '.$search);
  80
  81         return;
  82 }
  83
  84 /**
  85  * @brief Updates the first 250 servers
  86  *
  87  */
  88 function update_server() {
  89         $r = q("SELECT `url`, `created`, `last_failure`, `last_contact` FROM `gserver` ORDER BY rand()");
  90
  91         if (!dbm::is_result($r)) {
  92                 return;
  93         }
  94
  95         $updated = 0;
  96
  97         foreach ($r AS $server) {
  98                 if (!poco_do_update($server["created"], "", $server["last_failure"], $server["last_contact"])) {
  99                         continue;
 100                 }
 101                 logger('Update server status for server '.$server["url"], LOGGER_DEBUG);
 102
 103                 proc_run(PRIORITY_LOW, "include/discover_poco.php", "server", base64_encode($server["url"]));
 104
 105                 if (++$updated > 250) {
 106                         return;
 107                 }
 108         }
 109 }
 110
 111 function discover_users() {
 112         logger("Discover users", LOGGER_DEBUG);
 113
 114         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url` FROM `gcontact`
 115                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 116                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 117                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 118                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
 119                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
 120
 121         if (!$users) {
 122                 return;
 123         }
 124         $checked = 0;
 125
 126         foreach ($users AS $user) {
 127
 128                 $urlparts = parse_url($user["url"]);
 129                 if (!isset($urlparts["scheme"])) {
 130                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 131                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 132                         continue;
 133                  }
 134
 135                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 136                                                         "identi.ca", "alpha.app.net"))) {
 137                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 138                                         "facebook.com" => NETWORK_FACEBOOK,
 139                                         "twitter.com" => NETWORK_TWITTER,
 140                                         "identi.ca" => NETWORK_PUMPIO,
 141                                         "alpha.app.net" => NETWORK_APPNET);
 142
 143                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 144                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 145                         continue;
 146                 }
 147
 148                 if ($user["server_url"] != "") {
 149                         $server_url = $user["server_url"];
 150                 } else {
 151                         $server_url = poco_detect_server($user["url"]);
 152                 }
 153                 if (($server_url == "") OR poco_check_server($server_url, $gcontacts[0]["network"])) {
 154                         logger('Check user '.$user["url"]);
 155                         poco_last_updated($user["url"], true);
 156
 157                         if (++$checked > 100) {
 158                                 return;
 159                         }
 160                 } else {
 161                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 162                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 163                 }
 164         }
 165 }
 166
 167 function discover_directory($search) {
 168
 169         $data = Cache::get("dirsearch:".$search);
 170         if (!is_null($data)) {
 171                 // Only search for the same item every 24 hours
 172                 if (time() < $data + (60 * 60 * 24)) {
 173                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 174                         return;
 175                 }
 176         }
 177
 178         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 179         $j = json_decode($x);
 180
 181         if (count($j->results)) {
 182                 foreach($j->results as $jj) {
 183                         // Check if the contact already exists
 184                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 185                         if ($exists) {
 186                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 187
 188                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 189                                         ($exists[0]["updated"] < $exists[0]["last_failure"])) {
 190                                         continue;
 191                                 }
 192                                 // Update the contact
 193                                 poco_last_updated($jj->url);
 194                                 continue;
 195                         }
 196
 197                         $server_url = poco_detect_server($jj->url);
 198                         if ($server_url != '') {
 199                                 if (!poco_check_server($server_url)) {
 200                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 201                                         continue;
 202                                 }
 203                                 logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 204                         }
 205
 206                         $data = probe_url($jj->url);
 207                         if ($data["network"] == NETWORK_DFRN) {
 208                                 logger("Profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 209                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 210                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 211                         } else {
 212                                 logger("Profile ".$jj->url." is not responding or no Friendica contact - but network ".$data["network"], LOGGER_DEBUG);
 213                         }
 214                 }
 215         }
 216         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 217 }
 218
 219 /**
 220  * @brief Search for GNU Social user with gstools.org
 221  *
 222  * @param str $search User name
 223  */
 224 function gs_search_user($search) {
 225
 226         // Currently disabled, since the service isn't available anymore.
 227         // It is not removed since I hope that there will be a successor.
 228         return false;
 229
 230         $a = get_app();
 231
 232         $url = "http://gstools.org/api/users_search/".urlencode($search);
 233
 234         $result = z_fetch_url($url);
 235         if (!$result["success"]) {
 236                 return false;
 237         }
 238         $contacts = json_decode($result["body"]);
 239
 240         if ($contacts->status == 'ERROR') {
 241                 return false;
 242         }
 243         foreach($contacts->data AS $user) {
 244                 $contact = probe_url($user->site_address."/".$user->name);
 245                 if ($contact["network"] != NETWORK_PHANTOM) {
 246                         $contact["about"] = $user->description;
 247                         update_gcontact($contact);
 248                 }
 249         }
 250 }