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         - poco_load: Load POCO data from a given POCO address
  19         */
  20
  21         if (($argc > 2) && ($argv[1] == "dirsearch")) {
  22                 $search = urldecode($argv[2]);
  23                 $mode = 1;
  24         } elseif (($argc == 2) && ($argv[1] == "checkcontact")) {
  25                 $mode = 2;
  26         } elseif (($argc == 2) && ($argv[1] == "suggestions")) {
  27                 $mode = 3;
  28         } elseif (($argc == 3) && ($argv[1] == "server")) {
  29                 $mode = 4;
  30         } elseif (($argc == 2) && ($argv[1] == "update_server")) {
  31                 $mode = 5;
  32         } elseif (($argc == 3) && ($argv[1] == "update_server_directory")) {
  33                 $mode = 6;
  34         } elseif (($argc > 5) && ($argv[1] == "poco_load")) {
  35                 $mode = 7;
  36         } elseif ($argc == 1) {
  37                 $search = "";
  38                 $mode = 0;
  39         } else {
  40                 die("Unknown or missing parameter ".$argv[1]."\n");
  41         }
  42
  43         logger('start '.$search);
  44
  45         if ($mode == 7) {
  46                 if ($argc == 6) {
  47                         $url = base64_decode($argv[5]);
  48                 } else {
  49                         $url = '';
  50                 }
  51                 poco_load_worker(intval($argv[2]), intval($argv[3]), intval($argv[4]), $url);
  52         } elseif ($mode == 6) {
  53                 poco_discover_single_server(intval($argv[2]));
  54         } elseif ($mode == 5) {
  55                 update_server();
  56         } elseif ($mode == 4) {
  57                 $server_url = base64_decode($argv[2]);
  58                 if ($server_url == "") {
  59                         return;
  60                 }
  61                 $server_url = filter_var($server_url, FILTER_SANITIZE_URL);
  62                 if (substr(normalise_link($server_url), 0, 7) != "http://") {
  63                         return;
  64                 }
  65                 $result = "Checking server ".$server_url." - ";
  66                 $ret = poco_check_server($server_url);
  67                 if ($ret) {
  68                         $result .= "success";
  69                 } else {
  70                         $result .= "failed";
  71                 }
  72                 logger($result, LOGGER_DEBUG);
  73         } elseif ($mode == 3) {
  74                 update_suggestions();
  75         } elseif (($mode == 2) AND get_config('system','poco_completion')) {
  76                 discover_users();
  77         } elseif (($mode == 1) AND ($search != "") and get_config('system','poco_local_search')) {
  78                 discover_directory($search);
  79                 gs_search_user($search);
  80         } elseif (($mode == 0) AND ($search == "") and (get_config('system','poco_discovery') > 0)) {
  81                 // Query Friendica and Hubzilla servers for their users
  82                 poco_discover();
  83
  84                 // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
  85                 if (!get_config('system','ostatus_disabled'))
  86                         gs_discover();
  87         }
  88
  89         logger('end '.$search);
  90
  91         return;
  92 }
  93
  94 /**
  95  * @brief Updates the first 250 servers
  96  *
  97  */
  98 function update_server() {
  99         $r = q("SELECT `url`, `created`, `last_failure`, `last_contact` FROM `gserver` ORDER BY rand()");
 100
 101         if (!dbm::is_result($r)) {
 102                 return;
 103         }
 104
 105         $updated = 0;
 106
 107         foreach ($r AS $server) {
 108                 if (!poco_do_update($server["created"], "", $server["last_failure"], $server["last_contact"])) {
 109                         continue;
 110                 }
 111                 logger('Update server status for server '.$server["url"], LOGGER_DEBUG);
 112
 113                 proc_run(PRIORITY_LOW, "include/discover_poco.php", "server", base64_encode($server["url"]));
 114
 115                 if (++$updated > 250) {
 116                         return;
 117                 }
 118         }
 119 }
 120
 121 function discover_users() {
 122         logger("Discover users", LOGGER_DEBUG);
 123
 124         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url` FROM `gcontact`
 125                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 126                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 127                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 128                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
 129                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
 130
 131         if (!$users) {
 132                 return;
 133         }
 134         $checked = 0;
 135
 136         foreach ($users AS $user) {
 137
 138                 $urlparts = parse_url($user["url"]);
 139                 if (!isset($urlparts["scheme"])) {
 140                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 141                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 142                         continue;
 143                  }
 144
 145                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 146                                                         "identi.ca", "alpha.app.net"))) {
 147                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 148                                         "facebook.com" => NETWORK_FACEBOOK,
 149                                         "twitter.com" => NETWORK_TWITTER,
 150                                         "identi.ca" => NETWORK_PUMPIO,
 151                                         "alpha.app.net" => NETWORK_APPNET);
 152
 153                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 154                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 155                         continue;
 156                 }
 157
 158                 if ($user["server_url"] != "") {
 159                         $server_url = $user["server_url"];
 160                 } else {
 161                         $server_url = poco_detect_server($user["url"]);
 162                 }
 163                 if (($server_url == "") OR poco_check_server($server_url, $gcontacts[0]["network"])) {
 164                         logger('Check user '.$user["url"]);
 165                         poco_last_updated($user["url"], true);
 166
 167                         if (++$checked > 100) {
 168                                 return;
 169                         }
 170                 } else {
 171                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 172                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 173                 }
 174         }
 175 }
 176
 177 function discover_directory($search) {
 178
 179         $data = Cache::get("dirsearch:".$search);
 180         if (!is_null($data)) {
 181                 // Only search for the same item every 24 hours
 182                 if (time() < $data + (60 * 60 * 24)) {
 183                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 184                         return;
 185                 }
 186         }
 187
 188         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 189         $j = json_decode($x);
 190
 191         if (count($j->results)) {
 192                 foreach($j->results as $jj) {
 193                         // Check if the contact already exists
 194                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 195                         if ($exists) {
 196                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 197
 198                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 199                                         ($exists[0]["updated"] < $exists[0]["last_failure"])) {
 200                                         continue;
 201                                 }
 202                                 // Update the contact
 203                                 poco_last_updated($jj->url);
 204                                 continue;
 205                         }
 206
 207                         $server_url = poco_detect_server($jj->url);
 208                         if ($server_url != '') {
 209                                 if (!poco_check_server($server_url)) {
 210                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 211                                         continue;
 212                                 }
 213                                 logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 214                         }
 215
 216                         $data = probe_url($jj->url);
 217                         if ($data["network"] == NETWORK_DFRN) {
 218                                 logger("Profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 219                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 220                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 221                         } else {
 222                                 logger("Profile ".$jj->url." is not responding or no Friendica contact - but network ".$data["network"], LOGGER_DEBUG);
 223                         }
 224                 }
 225         }
 226         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 227 }
 228
 229 /**
 230  * @brief Search for GNU Social user with gstools.org
 231  *
 232  * @param str $search User name
 233  */
 234 function gs_search_user($search) {
 235
 236         // Currently disabled, since the service isn't available anymore.
 237         // It is not removed since I hope that there will be a successor.
 238         return false;
 239
 240         $a = get_app();
 241
 242         $url = "http://gstools.org/api/users_search/".urlencode($search);
 243
 244         $result = z_fetch_url($url);
 245         if (!$result["success"]) {
 246                 return false;
 247         }
 248         $contacts = json_decode($result["body"]);
 249
 250         if ($contacts->status == 'ERROR') {
 251                 return false;
 252         }
 253         foreach($contacts->data AS $user) {
 254                 $contact = probe_url($user->site_address."/".$user->name);
 255                 if ($contact["network"] != NETWORK_PHANTOM) {
 256                         $contact["about"] = $user->description;
 257                         update_gcontact($contact);
 258                 }
 259         }
 260 }